摘要:

J. Chem. Soc., Faraday Trans. I , 1984,80, 2459-2466 Infrared Study of the Adsorption of Propan-2-01 on Rutile at the Solid/Vapour and Solid/Heptane Interfaces BY COLIN H. ROCHESTER* Department of Chemistry, The University, Dundee DDl 4HN AND JOHN GRAHAM AND ROBERT RUDHAM Department of Chemistry, The University, Nottingham NG7 2RD Received 16th November, 1983 Infrared spectra of propan-2-01 adsorbed on rutile have shown that non-dissociatively adsorbed propan-2-01 is coordinatively liganded to Lewis-acidic surface sites. Dissociative chemisorption occurs at exposed Ti4+ cations, particularly for dehydroxylated rutile, and generates isopropoxide anions. Surface hydroxyl groups on rutile were replaced by isopropoxide anions with the concomitant formation of water. Hydroxyl groups at sub-surface lattice sites were unaffected by propanol adsorption. Multilayer adsorption of propan-2-01 at high surface coverages involved the formation of molecular aggregates which were weakly bound by hydrogen-bonding interactions and which contained water molecules resulting from the chemisorptive reactions.Heat treatment of adsorbed propan-2-01 resulted in the formation of adsorbed carboxylate and possibly carbonate species. The implications of the results with respect to the dehydration of propan-2-01 catalysed by rutile are briefly discussed. Rutile catalyses the decomposition reaction of propan-2-01 to acetone, propene and di-isopropyl ether1 and is also effective in the photocatalytic oxidation of propan-2-01 to acetone.2 Interactions between the rutile surface and a~etone,~ propene4 and di-isopropyl ether4 have been characterized by infrared spectroscopy.The present paper reports infrared spectra of propan-2-01 adsorbed on an hydroxylated rutile surface. Munuera and Stone5 found no evidence for alcohol molecules coordinatively bound to exposed Ti4+ cation sites in spectra of propan-2-01 on partially hydroxylated rutile and concluded that propan-2-01 was adsorbed non-dissociatively through interactions involving an adjacent hydroxyl group and oxide ion sites or pairs of adjacent hydroxyl groups in the surface. A single spectrum of propan-2-01 adsorbed on titanium dioxide (of unspecified crystal structure) which had been preheated at 673 K suggested that a chemisorptive reaction had led to the formation of surface propan-2-oxide ions.6 Alcohols are believed to adsorb on anatase to give alkoxide anions,’ although one report suggested that the predominant interaction involved alcohol molecules coordinatively liganded to Lewis-acidic surface sites.8 The development of cells which enable infrared spectra of rutile immersed in liquid hydrocarbons to be recorded9? lo coupled with an interest in the photocatalytic oxidation of liquid propan-2-01 by rutile2 prompted an infrared study, reported here, of rutile immersed in liquid propan-2-01 and solutions of propan-2-01 in heptane.24592460 I.R. STUDY OF PROPANOL + RUTILE EXPERIMENTAL Rutile (code CL/D 338, Tioxide International Ltd) with a surface area of 30.3 m2 g-l was freed from surface impurities, pressed into self-supporting discs and pretreated as before,'O.l1 the final treatment involving heating in vacuum at 400 K for 1 h. A recently described celllo was used to record spectra of discs immersed in solutions of propan-2-01 and also served in the experiments involving the adsorption and desorption of propan-2-01 vapour. Discs were at ca. 306 K during spectroscopic examination. AnalaR propan-2-01 was doubly distilled and stored over molecular sieve. Heptane was shaken with calcium sulphate for 24 h, triply distilled (twice from calcium sulphate, once from anhydrous copper sulphate) and stored in the vacuum apparatus where it was freed from permanent gases by cold pumping and finally vacuum distilled. RESULTS The adsorption of increasing amounts of propan-2-01 on rutile led to the progressive disappearance of infrared bands at 3725, 3680(sh) and 3655cm-l due to surface hydroxyl groups12 and the growth of a broad absorption maximum centred at ca. 3350 cm-l (fig.1). The effects were time dependent [fig. 1 (b) and (c)]. The appearance of a band at 1620 cm-l showed that water was a product of the adsorption reaction, although water was apparently displaced from surface sites as the vapour pressure of propan-2-01 was increased [fig. l(d)]. The growth of bands at 3610 and 3520 cm-l for low coverages of the surface by propan-2-01 [fig. l(b)] and the subsequent weakening of these bands at high coverages support this conclusion. The maxima at 3610 and 3520cm-l have been assigned to OH-stretching vibrations of surface hydroxyl groups which are formed on rutile in the presence of a high concentration of non-dissociatively adsorbed water molecules.12 Maxima appearing in spectra at 2970, 2937, 2835, 1470, 1390(sh), 1383, 1368 and 1333 cm-l could be ascribed to vibrations of isopropyl groups predominantly in adsorbed species but partly in gaseous propan-2-01 molecules [fig. 1 (e) and 01.The shoulder at 1390 cm-l could not be discerned for rutile in the presence of high pressures of propan-2-01. Hydroxyl groups responsible for the absorption band at 3410 cm-l [fig. 1 (a)] probably exist at sub-surface lattice sites in rutile9-11v13 and were unaffected by the adsorption of propan-2-01. Decreases in the intensities of the band at 2970, 2937, 2835, 1470, 1383, 1368 and 1333 cm-l due to adsorbed species were caused by brief evacuation of the cell with the sample at ambient temperature [fig.2(a)]. A concomitant decrease in intensity of the broad maximum centred at ca. 3350 cm-l was not accompanied by the appearance of bands due to vibrations of unperturbed hydroxyl groups. The results suggest that hydrogen bonded aggregates of propan-2-01 molecules were weakly adsorbed in equilibrium with high vapour pressures of propan-2-01 and were easily desorbed by evacuation. A simultaneous decrease in absorption intensity at 1620 cm-l suggests that water molecules were also present in the aggregates. Prolonging the evacuation time led to futher desorption of propan-2-01 and the growth of a doublet of bands at 3640 and 3620 cm-l due to unperturbed or only very weakly perturbed hydroxyl groups.A maximum also appeared at 3530 cm-l [fig. 2(b)]. Residual bands due to CH stretching and deformation vibrations of isopropyl groups showed that evacuation at ca. 306 K was insufficient to desorb completely the products of adsorption from the rutile surface. Subsequent evacuation at 373 K had little further effect but increasing the temperature to 423 K caused the disappearance of the weak band at 1390cm-l, the maximum at 3530cm-l and the broad band envelope centred at 3350 cm-1 [fig. 2(c)]. A maximum at 3640 cm-l due to unperturbed surface hydroxylC. H. ROCHESTER, J. GRAHAM AND R. RUDHAM 246 1 3800 3400 3000 1 1 1 1600 1400 wavenumber/cm -' Fig. 1. Spectra of rutile (a) in vacuum, (b) and (c) in contact with propan-2-01 vapour at ca.306 K for (b) 5 min and (c) 16 h and (6) exposed to S.V.P. of propan-2-01 at ca. 306 K. Spectra of propan-2-01 vapour at (e) a higher vapour pressure than that in the cell when spectra (b) and (c) were recorded and v> S.V.P. 3000 2800 1600 1 LOO wavenum ber/cm-' Fig. 2. Spectra of rutile after exposure to propan-2-01 (s.v.P., 16 h, ca. 306 K) and evacuation at (a) ca. 306 K (5 min), (b) ca. 306 K (1 h), (c) 423 K (30 min), (6) 523 K (30 min) and (e) 573 K (30 min). groups was enhanced in intensity. Apart from the loss of a weak shoulder at 2895 cm-l [fig. 2(b)], the bands due to CH-stretching vibrations of isopropyl groups were unchanged in intensity suggesting that evacuation at 423 K had led to a small amount of redistribution of surface species containing isopropyl groups but that negligible desorption had occurred. In contrast, surface species containing isopropyl groups were largely desorbed after evacuation at 523 K [fig.2(d)]. An infrared band appearing at2462 I.R. STUDY OF PROPANOL+ RUTILE 1 I " l J 3700 3300 I ~ I 3000 281 30 1600 1400 wavenum berlcm-' Fig. 3. Spectra of rutile (a) immersed in heptane, (b)-(d) immersed in solutions of propan-2-01 at mole fractions in heptane of (b) (5 min), (e)-(h) after subsequent evacuation for 30 min at (e) ca. 306, v) 423, (g) 503 and (h) 563 K. (5 min), (c) (16 h) and (d) 2.3 x 1445 cm-l was enhanced in intensity after evacuation at 573 K [fig. 2(e)] and, together with a very weak maximum at ca. 15 10-1 560 cm-l, suggested that adsorbed carboxylate anions had been l1 Spectra of rutile immersed in solutions of propan-2-01 in heptane are shown in fig.3. The maximum at 3605 cm-l due to surface hydroxyl groups (band at 3655 cm-l for rutile in vacuum) weakly perturbed by interaction with heptane molecules9 decreased in intensity with increasing concentrations of propan-2-01. The losses of intensity were time dependent [fig. 3 (b) and (c)], suggesting that a slow chemisorption reaction involving the hydroxyl groups was taking place. The appearance of a weak band at 1600-1640 cm-l showed that water was a reaction product. Increases in absorption intensity in the broad range 3000-3550 cm-l were similar to corresponding increases for rutile in contact with propan-2-01 vapour. Unfortunately no bands due to vibrations of hydrocarbon groups in adsorbed species could be detected because of intense maxima in the same spectral regions due to vibrations of liquid heptane.Drainage of the liquid phase and subsequent evacuation at a series of increasing temperatures gave spectroscopic effects [fig. 3 (e)-(h)] which were similar to those recorded after the adsorption of propan-2-01 vapour on rutile (fig. 2). Identical effects to those shown in fig. 2 were also recorded after immersion of rutile in pure liquid propan-2-01, drainage and evacuation at increasing temperatures. The results showed that hydroxyl groups responsible for the band in spectra at 3410 cm-l were completely unreactive, even towards liquid propan-2-01. DISCUSSION Bands at 1392, 1383 and 1368 cm-l in spectra of propan-2-01 adsorbed on rutile may be compared with bands at.1390 and 1380 cm-l, ascribed to non-dissociatively adsorbed di-isopropyl ether on rutile, and at 1383 and 1368 cm-1 in spectra of surface,C.H. ROCHESTER, J . GRAHAM AND R. RUDHAM 2463 isopropoxide ions on rutile formed by the chemisorption of di-isopropyl ether.4 Isopropyl groups give two bands in this region of the spectrum because the maximum due to the symmetrical vibrations of methyl groups is split into a doublet. The bands were at 1393 and 1383 cm-l for propan-2-01 vapour. The present results after adsorption of low vapour pressures of propan-2-01 [fig. 1 (b)] or after adsorption and evacuation at ambient temperatures [fig. 2(a)] may be ascribed to two doublets, one at 1392 and 1383 cm-l due to non-dissociatively adsorbed propan-2-01 molecules coordinatively liganded to Lewis-acidic surface sites,149 l5 and the other at 1383 and 1368 cm-l due to surface isopropoxide ions.Dissociative adsorption to give isopropoxide ions occurred more readily for propan-2-01 than for di-isopropyl ether.4 The growth of absorption intensity at 1383 and 1368 cm-l for high vapour pressures of propan-2-01 [fig. 1(4] is, by comparison with corresponding maxima in spectra of liquid propan-2-01, attributed to multilayer adsorption16 involving the aggregation of propan-2-01 molecules through hydrogen-bonding interactions. The aggregated molecules were desorbed by evacuation at ambient temperature [fig. 2 (a)]. Hydrogen-bonding interactions between surface hydroxyl groups on rutile and adsorbed molecules constituted a less significant mode of adsorption for propan-2-01 than for di-isopropyl ether4 because propan-2-01 underwent a slow chemisorptive reaction which led to the replacement of hydroxyl groups by isopropoxide ions.Jackson and Parfitt17 similarly observed that ethanol, butan-1-01 and hexan-1 -01 reacted with surface hydroxyl groups to form adsorbed alkoxide ions on rutile. In contrast, surface hydroxyl groups formed hydrogen bonds with phenol molecules adsorbed from the vapour at pressures < 133 N m-2.1s Reaction to adsorbed phenoxide ions was only observed for a dehydroxylated rutile surface. Primet et aZ.18 reported a lack of reactivity between phenol and hydroxyl groups responsible for an infrared band at 3410 cm-l in spectra of rutile.The lack of reactivity of these hydroxyl groups towards adsorbate molecules appears to be general3? 4* 9-119 1 3 9 1 9 9 2o and has been attributed to the existence of the groups at sub-surface lattice sites.g* l1 The results for propan-2-01 conformed to the general pattern of behaviour, the only detectable effect being a reversible broadening of the band at 3410 cm-l similar to that previously recognized from spectra of ethyl acetatelo and triethylamine13 on rutile. Water which is adsorbed through non-dissociative coordinative interactions with Lewis-acidic sites in the exposed {loo} and { lOl} surface planes of rutile cannot be desorbed by evacuation at room temperature.12 Water molecules formed by the chemisorption of propan-2-01 on rutile were desorbed by evacuation [fig.2 (a)], suggesting that the water was not adsorbed on Lewis-acidic sites but was involved in hydrogen bonding interactions with surface hydroxyl groups or with aggregates of propan-2-01 molecules. The adsorption of water on Lewis-acidic sites was precluded by the adsorption of propan-2-01. Munuera and Stone5 found that water was not able to displace propan-2-01 from the surface of rutile. Bickley et aL21 reported that water was displaced by propan-2-01. Similarly Jackson and Parfitt17 observed, in accordance with the present results, that water was displaced from the rutile surface by the adsorption of alcohols. Alcohols also displace water from the surface of anatase.22 The adsorption of propan-2-01 on Lewis-acidic sites in the { 100) and { 101) planes of rutile is probably, as for water,12 predominantly non-dissociative and involves coordinative interactions with exposed Ti4+ cations.23 However, Jaycock and W a l d ~ a x ~ ~ suggested that water could adsorb dissociatively on the { 100) planes of rutile, and this led Griffiths and Rochester12 tentatively to attribute bands at 3680, 3610 and 3520 cm-l in infrared spectra of rutile to hydroxyl groups formed by this mode of adsorption.Evidence for a similar mode of adsorption for propan-2-01 was provided by the existence of bands at 3620 and 3530 cm-l in spectra despite negligible evidence2464 I.R. STUDY OF PROPANOL + RUTILE [fig. 2(b)] for non-dissociatively adsorbed water, which was a prerequisite for the appearance of the bands in the case of water adsorption.By analogy to the results for water, it is therefore proposed that a small proportion of the total number of adsorbed propan-2-01 molecules were dissociated on exposed Ti4+ cations to give adsorbed isopropoxide anions at the cationic sites. The proton from each dissociated molecule combined with an adjacent surface oxide anion to form a hydroxyl group. The suggestion that this mechanism occurs specifically on { 100) planes remains tentative. l2 However, a similar mechanism must constitute the predominant mode of adsorption of alcohols on { 110) planes of rutile which have been dehydroxylated by high temperature treatment before adsorption of an alcohol. The resulting alkoxide anions complete the coordination of row A Ti4+ cations in the (1 lo} planes,12?23 and the hydroxyl ions exist at sub-surface ~ i t e s , ~ ? ' ~ probably liganded to row B Ti4+ cations.This mechanism of chemisorption of alcohols on the ( 100) planes is analogous to the proposed mechanism for water a d s ~ r p t i o n , ~ ~ ~ ~ and is consistent with the obser~ationl~ that the adsorption of alcohols on dehydroxylated rutile leads to the appearance of an infrared band at 3410cm-l but no band at 3655cm-l due to hydroxyl groups on row A sites. In the present study of alcohol adsorption on an hydroxylated rutile surface the final outcome was the same because row A hydroxyl groups (band at 3655 crn-l) were replaced by isopropoxide ions and the sub surface hydroxyl groups were unaffected by the adsorption process. Jackson and Parfittl' reported infrared spectra of ethanol, butan- 1-01 and hexan-1 -01 on rutile and concluded that chemisorptive reactions led to the formation of surface alkoxide species which were precursors of the appropriate alkenes generated by thermal activation.Two general mechanisms of chemisorption, represented by reactions (1) and (2), are consistent with the previous and present spectroscopic data: -0Pr Ti4+02- + PrOH(g) -+ Ti4+0H- -OH -0Pr Ti4+ + PrOH(g) -+ Ti4+ + H,O. Reaction (1) occurs on dehydroxylated { 1 lo} planes, the product hydroxyl group giving an infrared band at 3410 cm-l. Reaction (2) involves row A hydroxyl groups on hydroxylated (1 lo} planes, and also hydroxyl groups giving bands at 3725 and 3700cm-l in spectra of rutile. Groups responsible for the latter two bands are particularly reactive towards a variety of adsorbates including a~etone,~ ethyl acetatell and hexafluoroacetone.20 The weak band at 3725 cm-l [fig.1 (a)] has been attributed to surface silanol groups resulting from a low silica impurity level in the rutile.12 The extent to which reaction (1) goes to completion at equilibrium on surfaces other than { 1 lo} planes can only be conjectured. However, a stable intermediate in the reaction might involve each propan-2-01 molecule coordinatively liganded uia its oxygen atom to a Ti4+ cationic site and with its hydrogen atom simultaneously linked by a hydrogen bond to a vicinal surface oxide ion. The geometry of the { 100) planes is particularly favourable for this type of intera~tion,~~ which would explain why a broad infrared band at ca.3400 cm-l remained in spectra after weakly adsorbed water and propan-2-01 had been desorbed by evacuaton [fig. 2(b)]. The disappearance of this band after evacuation at 423 K [fig. 2(c)] was accompanied by the simultaneous disappearance of the maximum at 1390 cm-l attributed to a vibration of propan-2-01 adsorbed on Lewis-acidic surface sites. The absence of change in the intensities of bands due to CH-stretching vibrations suggests that the thermal treatment had promoted the decomposition of propan-2-01 molecules on Lewis sites to give isopropoxide anions. No decomposition products other than water would have been expected to have beenC. H. ROCHESTER, J. GRAHAM AND R. RUDHAM 2465 desorbed at this temperat~re.~J~ At higher temperatures propene would be a significant product,l, 6* l7 possibly formed by the decomposition of isopropoxide Gentry et a1.l reported rates of decomposition of propan-2-01 to propene, acetone and di-isopropyl ether catalysed by rutile at 483-519 K.The infrared spectra show that under these conditions Lewis-acidic sites in the (100) and (101) surface planes are unlikely to be occupied by non-dissociatively adsorbed propan-2-01 molecules but that some sites might be occupied by isopropoxide anions, the population of which would become increasingly small with increasing temperature. Only a very small proportion of the available sites would be occupied at the upper end of the temperature range [fig. 2(4]. The spectroscopic evidence would be consistent with a mechanism for propene formation in which propan-2-01 is decomposed at Lewis-acidic sites to give isopropoxide ions which subsequently break down to propene and a hydroxyl group.The latter can condense with a further hydroxyl group prior to desorption as water. Propene is readily desorbed as a reaction product.* The low yields of di-isopropyl ether as a product ofpropan-2-01 dehydration may be linked to the fact that di-isopropyl ether itself undergoes decomposition to isopropoxide ions on rutile at elevated temperat~res.~ The possibility of a carbonium ion mechanism for the propan-2-01 dehydration reaction1 prompted a search for infrared bands due to adsorbed propyl cations. The propyl cation in (CH,),CH+SbF, gives strong infrared bands at 2730, 1490 and 1260 cm-1.26 No similar bands were discernible in the present spectra, although a hint of a very weak band existed at 2720 cm-1 for rutile in the presence of propan-2-01 vapour.However, this band corresponds to shoulders at 2750 cm-l reported17 (but not interpreted) in spectra of primary alcohols on rutile where carbonium-ion formation is less likely than for propan-2-01. No clear evidence for carbonium ions at ambient temperature was therefore detected, although this does not preclude their possible involvement in the mechanism of the catalytic dehydration reaction. Jackson and Parfitt" reported that the thermal decomposition of ethanol adsorbed on rutile led to adsorbed carbonate species.27 However, their figured spectra also resembled spectra of acetic acid on rutilell and therefore could at least partly be interpreted in terms of the formation of acetate species. The thermal decomposition of propan-2-01 on rutile [fig.2(e)] gave weaker but similar infrared bands to those resulting from ethanol decomposition and can be attributed to the formation of surface carboxylate, possibly with some further oxidation to carbonate anions. Comparison of the spectra of adsorbed ethano117 and propan-2-01 [fig. 2(4] after thermal activation at 523 K shows that surface isopropoxide ions on rutile are more readily decomposed than surface ethoxide ions. The lower extent of formation of adsorbed carboxylate or carbonate species after isopropoxide ion decomposition suggests that the hydrocarbon species were desorbed, predominantly as propene,lg 6 y l7 and supports the contention that at least one possible mechanism for propene formation may involve isopropoxide anions as intermediates in the reaction.s* 7 9 1 7 9 25 The formation of carboxylate species by oxidation of propan-2-01 on rutileZ1 may involve adsorbed acetone as a reaction 21v 28* 2B Spectra of titanium dioxide exhibited a band at 1690 cm-l, attributed to coordinatively adsorbed acetone, after the adsorption of propan-2-01 and subsequent heat treatment in oxygen at 423 K.6 A similar band at 1700 cm-l has been reported after the adsorption of propan-2-01 on anatase at 473-573 K.7 No bands attributable to adsorbed acetone3 were observed in the present work.anions.6,7,17,25 We thank Tioxide International Ltd for the award of a studentship to J.G.2466 I .R .STUDY OF PROPANOL + RUTILE S. J. Gentry, R. Rudham and K. P. Wagstaff, J. Chem. SOC., Faraday Trans. I , 1975, 71, 657. P. R. Harvey, R. Rudham and S. Ward, J. Chem. SOC., Faraday Trans. I , 1983, 79, 1381. D. M. Griffiths and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1978, 74,403. J. Graham, R. Rudham and C. H. Rochester, J. Chem. SOC., Faraday Trans. 1, 1984, 80, 895. G. Munuera and F. S. Stone, Discuss. Faraday SOC., 1971, 52, 205. T. Nakajima, H. Miyata and Y. Kubokawa, Buff. Chem. SOC. Jpn, 1982, 55, 609. Y. M. Shchekochikhin, V. N. Filimonov, N. P. Keier and A. M. Terenin, Kinet. Catal., 1964, 5, 94. G. Munuera and I. Garrizosa, Acta Cient. Venez., Supl., 1973, 24, 226. J. Graham, C. H. Rochester and R. Rudham, J. Chem. SOC., Faraday Trans. I , 1981, 77, 2735. 77, 2845. lo A. D. Buckland, J. Graham, R. Rudham and C. H. Rochester, J . Chem. SOC., Faraday Trans. I , 1981, l1 J. Graham, C. H. Rochester and R. Rudham, J. Chem. SOC., Faraday Trans. I , 1981, 77, 1973. l 2 D. M. Griffiths and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1977,73, 1510. l3 J. Graham, R. Rudham and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1983, 79, 2991. l4 G. D. Parfitt, J. Ramsbotham and C. H. Rochester, Trans. Faraday SOC., 1971, 67, 841. l5 G. D. Parfitt, J. Ramsbotham and C. H. Rochester, Trans. Faraday SOC., 1971, 67, 1500. l6 R. E. Day, G. D. Parfitt and J. Peacock, Discuss. Faraday SOC., 1971, 52, 215. l7 P. Jackson and G. D. Parfitt, J. Chem. SOC., Faraday Trans. I , 1972, 68, 1443. l9 D. M. Griffiths and C. H. Rochester, J. Chem. SOC., Faraday Trans. 1, 1977, 73, 1988. 2o D. M. Griffiths and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1977, 73, 1913. 21 R. I. Bickley, G. Munuera and F. S. Stone, J. Catal., 1973, 31, 398. 22 I. Carrizosa and G. Munuera, J. Catal., 1977, 49, 174. 23 P. Jones and J. A. Hockey, Trans. Faraday SOC., 1971, 67, 2679. 24 M. J. Jaycock and J. C. R. Waldsax, J. Chem. SOC., Faraday Trans. I , 1974, 70, 1501. 25 Y. Soma, T. Onishi and K. Tamaru, Trans. Faraday SOC., 1969, 65, 2215. 26 J. C. Evans, J. Am. Chem. SOC., 1964,86, 1371. 27 P. Jackson and G. D. Parfitt, J. Chem. SOC., Faraday Trans. I, 1972, 68, 896. 28 A. V. Deo, T. T. Chuang and I. G. Dalla Lana, J. Phys. Chem., 1971,75,234. 29 M. I. Zaki and N. Sheppard, J. Cataf., 1983, 80, 114. M. Primet, P. Pichat and M-V. Mathieu, J. Phys. Chem., 1971, 75, 1221. (PAPER 3/2055)

ISSN:0300-9599    

DOI:10.1039/F19848002459

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I , 1984,80,2467-2478 Polarirnetric Study of the Interaction of Electrolytes with Mannitol and Sorbitol BY J. GRAHAM DAWBER* AND GAIL E. HARDY Department of Chemistry and Biology, North Staffordshire Polytechnic, Stoke-on-Trent ST4 2DE Received 25th November, 1983 The effects of various electrolytes in forming complexes with mannitol and sorbitol have been studied using polarimetry at a wavelength of 436 nm. In general, the ease of interaction with electrolytes is greater for sorbitol than for mannitol and the effects are greater for both polyols with anions rather than cations. The Group I1 cations are more effective than the Group I cations in producing changes in optical rotation of the polyols. In the case of interaction with anions, a tetrahedral symmetry containing-OH groups appears to be effective in complexing with the polyols, as is also the case when the anion is of complex form such as B&O;-, B,Og- and Mo,O!;.The interactions of aqueous electrolytes with polysaccharides play an essential role in studies of plant cell-wall polysaccharides, and the fractionation and discriminatory properties of these materials play an important part in their function in living systems. Although some studies have been made recently using soluble polysaccharides to study ionic binding,l other studies have adopted an alternative approach using soluble low-molecular-weight polyols as model compounds for the more complex yet insoluble polysaccharides.2y Two such compounds which have received some recent attention are mannitol and s ~ r b i t o l .~ - ~ Despite their identical molecular formulae a number of studies have shown that mannitol and sorbitol have different hydration properties in solution.'-lo It has been suggested6 that the planar structure of mannitol is more compatible with the ordered water structure in solution than the non-planar, sickle-bent structure of sorbito1,llY l2 and as a result it was thought6 that the -OH groups of mannitol, compared with those of sorbitol, are less free for interactions with sodium chloride as an aqueous cosolute. The purpose of this work was to study the interactions of a variety of electrolytes with mannitol and sorbitol in order to compare the behaviour of both polyols towards a given cosolute and also to compare the different cosolutes with one another.The optical rotations of mannitol and sorbitol are fairly small and yet they are known to change markedly in some situations, e.g. in borax solutions and acidified ammonium molybdate solutions in which they may be characterised. It was therefore thought that polarimetry would be a suitable method for studying a wide range of electrolyte-polyol interactions. EXPERIMENTAL MATERIALS The materials used and their grades of purity were as follows: mannitol(GPR), sorbitol(GPR), NaCI(AR), KCl(AR), CsCl(GPR), MgCl,(AR), CaCl,(AR), SrCl,(GPR), BaCl,(AR), NaH,PO,(GPR), H,PO,(AR), NaBF,(GPR), disodium tetraborate(b0rax) (AR), sodium 24672468 POLARIMETRIC STUDY OF ELECTROLYTES AND POLYOLS metaborate(GPR), NaOH(AR), sodium molybdate(Na,MoO, .2H,O) (GPR), ammonium molybdate [(NH,),Mo,O,, -4H,O] (GPR), Na,WO, - 2H,O(GPR), NaClO,(GPR), H,BO,(AR), MIO,(GPR) and disodium phenyl phosphonate (PhPO,Na,)(GPR). Solutions of NaB(OH), were prepared by dissolving the required amount of H,BO, in NaOH solution. Solutions of ZnC1, were prepared by dissolving ZnO(AR) in the required amount of HCl solution. Solutions of NaAl(OH), were prepared by dissolving AlCl, in 2 mol dm-3 NaOH solution, the final stock solution containing 0.8 rnol dm-, of Al. The [a18 for 0.5moldm-3 solutions of mannitol and sorbitol were measured with a photoelectric polarimeter (see below) using a sodium lamp in conjunction with a solution of K,Cr,O, to remove unwanted lines of lower wavelength which would elevate the reading. The values obtained were mannitol [a]g = -0.36 and sorbitol [a]g = - 1.68.Various values are quoted in the literature, e.g. mannitol [a]E = -0.4913, l4 and -0.208,15 sorbitol [a18 = - 1.98137 l5 and -2.01 [a]g = - 1.73.14 Our studies were at 20 "C and were made with a photoelectric polarimeter. It was thought that the polyol samples were of satisfactory purity and they were used without further purification. Stock solutions of mannitol and sorbitol were made up containing 0.50 mol drn-,. Aliquots of these solutions were diluted with aliquots of the electrolyte solutions such that the final polyol concentration in a series of experiments was kept constant either at 0.25 moldm-, or for a few experiments at 0.05 mol dm-,. The concentrations of polyol used in a given series of experiments is indicated in the results.POLARIMETRIC MEASUREMENTS The optical rotations of the solutions were measured with a Bellingham and Stanley model A photoelectric polarimeter using a 100 mm polarimeter tube. Angular rotations could be estimated to 0.001". The measurements were made at a wavelength of 435.8 nm using a low-pressure Hg lamp with the unwanted wavelengths filtered out by a cobalt glass plus a solution of NaNO, in a 20 mm glass cell.lS This wavelength was used rather than the sodium D lines since it provided larger differences in optical rotation between successive solutions, thereby giving better discrimination. The temperature was 20 1 "C. RESULTS The optical-rotation results were converted to molar optical rotation, a,, by a, = a/cZ, where a is the measured optical rotation in a polarimeter tube of rn metres for a concentration of c mol m-3.The units of a, are thus O m2 mol-l. The experiments in a given run were conducted with c kept constant (usually 0.25 mol dm-3, i.e. 250 mol m-3) and the concentration of electrolyte (m) varied. In order to compare the effects of the different electrolytes upon the optical rotations of the polyols the change in molar rotation, Aa,, was calculated for each solution. The molar rotations of mannitol and sorbitol at 436 nm, based upon the measurements with 0.25 mol dm-3 aqueous solutions, were ( - 2.75 _+ 0.15) x m2 mol-l, respectively. and ( - 6.04 _+ 0.19) x DISCUSSION INTERACTIONS OF MANNITOL AND SORBITOL WITH DIFFERENT CATIONS The changes in molar rotation, Aa,, for mannitol and sorbitol produced by various mono- and di-valent metal chlorides are plotted as a function of electrolyte concentration in fig, 1-4.Although the general trend is for Aa, to increase positively there are small negative changes at low concentrations for NaCl and KCl with mannitol and for KCl with sorbitol; it is not clear what these changes are caused by. If one compares the effects of the electrolytes at a concentration of 1.0 mol dm-3 then the effect upon the molar rotation for mannitol is in the order NaCl c MgC1, < KCl < CsCl c BaCl, < CaCl, < ZnC1, < SrCl,, whereas for sorbitol theJ. G . DAWBER AND G . E. HARDY 2469 1 . o 0.8 0.6 " I - c1 E :: 0.4 2 a I -$ 0.2 0 -0.2 t 0.2 0.4 0.6 0.8 [salt 1 lmol dm-3 Fig. 1. Effects of metal chlorides upon optical rotation of mannitol: 0, NaCl; x , KC1; A, CsCl.I I 0.2 0.4 0.6 0.8 [salt] /mol dm-3 Fig. 2. Effects of metal chlorides upon optical rotation of mannitol: 0, MgCl,; 0, BaC1,; x , CaCl,; +, ZnC1,; A, SrC1,. order is slightly different, viz. NaCl < KCl < MgCl, c CsCl < ZnCl, < BaCl, < SrCl, d CaCl,. In general, a given electrolyte produced a substantially greater effect upon sorbitol than upon mannitol; this was particularly the case for the divalent cations. The exception to this was ZnCl,, which produced almost identical effects upon mannitol and sorbitol. The ratio of the Aam results for sorbitol/mannitol at 1.0 mol dm-3 for a given2470 POLARIMETRIC STUDY OF ELECTROLYTES AND POLYOLS 1.0 0.8 0.6 c1 I - PI E EJ 0.4 E 0.2 0 1 a Q t 0.2 0.4 0.6 0.8 [salt ] /mol dm-j Fig. 3.Effects of metal chlorides upon optical rotation of sorbitol: 0, NaCl; x , KC1; A, CsCl . 10 0 t I I I I I I I I 1 1 0.2 0.4 0.6 0.8 [salt] /mol dm-j Fig. 4. Effects of metal chlorides upon optical rotation of sorbitol: 0, MgC1,; +, ZnCl,; 0, BaCl,; A, SrCl,; x , CaC1,. electrolyte was NaCl 2.4, KCl 0.8, CsCl 1.1, MgCl, 1.3, CaCl, 6.1, SrCl, 3.5, BaCl, 2.5. It is interesting that while Ca2+ and Sr2+ interact most with each polyol, it is for these two cations that discrimination between the polyols is the greatest. Although the effects of the Group I metal chlorides follow cationic size, this is not the case for the Group I1 cations. Thus although Na+ and Ca2+ have approximately the same ionic size (95 and 99 pm) they have vastly different effects upon a,; the same can be said for K+ and Ba2+ (1 33 and 133 pm).Although ionic size may be important in its relation to fitting the -OH group conformation of the polyol, the hydration shell of the cation and the way in which this interacts with the hydration shell of the polyol may be equally important. The free energies of hydration" of the Group I1 cations are very much greater thanJ. G. DAWBER AND G. E. HARDY 247 1 5 4 3 N . E l - 2 - 3 90 80 70 60 % v/v water Fig. 5. Effects of organic solvents on optical rotation of mannitol and sorbitol: 0, mannitol+ MeOH; A, rnannitol+ acetone; 0, sorbitol+ MeOH; x , sorbitol+ acetone. those of the Group I cations, viz. (in kJ mol-I) Na+( -405), K+( - 331), Cs+( -278), Mg2+( - 1893), Ca2+( - 1581), Sr2+( - 1436) and Ba2+( - 1307), showing much more extensive hydration of the Group I1 cations.These free energies, however, do not correlate with the order of interaction of the cations with a given polyol as judged by the dam results. Hence the interaction of the cations with the polyols is likely to be a more subtle process than merely direct complexation with the -OH groups of the polyol and will involve a combination of factors including the size and shape of the ionic hydration shell and its interaction with the hydration sheath of the polyol, which will be related to the conformation of the polyol -OH groups, which themselves may be related to the water activity in the solution. The Actm results, of course, not only reflect the extent of interaction of thespolyol with a cosolute but also the value of the optical rotation of the complex formed.One possibility considered in interpreting the results was that the added electrolyte caused desolvation of the polyol because of competition for water molecules, and that this caused changes in the conformation of the polyol. However, addition of methanol, and also acetone, to aqueous solutions of mannitol and sorbitol in an effort to produce dehydration produced completely opposite effects (see fig. 9, i.e. more negative optical rotation for mannitol and more positive rotation for sorbitol, the effect being greater for sorbitol. Thus the organic component may become involved in the solvation sheath of the polyol in addition to displacing the water and hence altering the conformation.In the case of sorbitol it is thought that the -OH groups are more accessible to cosolutes.62472 POLARIMETRIC STUDY OF ELECTROLYTES AND POLYOLS 0.1 0.2 0.3 [borax I /mol dm-3 Fig. 6. Effect of borax on optical rotation of 0.05 mol dm-3 mannitol(0) and 0.05 mol dm-3 sorbitol (0). INTERACTION OF MANNITOL AND SORBITOL WITH DIFFERENT ANIONS In this series of experiments the majority of the electrolytes were sodium salts and thus differences between them reflected the different behaviour of the anions towards the polyol. The changes in molar optical rotation, Aa,, for all the electrolytes as a function of concentration were calculated. In the majority of cases the values of Aa, were very much greater than for the metal chlorides indicating a much greater involvement of anions with the polyols.However, no interaction was observed with either polyol for PhPO,Na, or NaClO, and very little effect with KIO,. For many of the electrolytes the plots of Aa, as a function of concentration of electrolyte were similar to those shown in fig. 6 and 7, where the values of Aam appear to be levelling out to a constant value, which would correspond to complete formation of a complex between the polyol and the electrolyte. The exception to this behaviour was for ammonium molybdate, where the changes in a, were very large indeed and involved maxima in the plots of Act, against concentration (see fig. 8). The curves illustrated in fig. 6 and 7 are similar to Langmuir adsorption isotherms10 8 L 6 g 4 s o 2 - 2 lE -4 *E 4 -6 -8 -1 0 J.G . DAWBER AND G . E. HARDY 2473 0.1 0.2 0.3 0.4 (NaA1(OH)4]/mol dm-3 Fig. 7. Effect of NaAl(OH), on optical rotation of mannitol (0) and sorbitol (A). 150 00 50 0.02 0.04 0.06 [ammonium molybdate] /mol dm-3 Fig. 8. Effect of ammonium molybdate upon optical rotation of 0.05 mol dm-3 mannitol (0) and 0.05 mol dm-3 sorbitol (A).2474 POLARIMETRIC STUDY OF ELECTROLYTES AND POLYOLS and it was thought that a similar approach might be useful where the -OH groups of a polyol are considered as possible adsorption sites for the anion k , P+A- =PA-. For a constant initial amount of polyol, P, and varying amounts (m) of the anion, A-, the rate of complexation will be proportional to the fraction of P which is not complexed, i.e.(1 - y), and also the amount of electrolyte. The rate of disproportion- ation of the complex will be proportional to the fraction complexed and thus at equilibrium we may write k-i k,(l -y)m = k-,y (1) = Km/(l + Km) k1m from which I) = k-,+k,m where K = k/k-,. The change in molar rotation will be proportional to y and thus we can write Aam = K'y, where K' will be the a, of the fully formed complex ( y = 1). Thus K'Km Aam = - 1+Km (3) which on rearrangement gives m/Aa, = m/K'+ l/KK'. (4) A plot of m/Aa, against m should be linear and have a gradient of 1/K' and an intercept of I/KK' from which K' and K could be evaluated. In general, such plots were linear although occasionally there was deviation from linearity at high m. Although the value of K will be a measure of the interaction between the electrolyte and the polyol it is not a proper equilibrium constant.Of more interest are the values of K' since this equals the difference between the molar optical rotations of the complex (PA) and the polyol (P), i.e. am(F,)-am(p), from which a,(,,, can be calculated. The measured molar rotation, a,, will be given by a m = (1 -Y)am(p)+Yarn(p,)* Thus arn-am(P) = darn = Y ( ~ ~ ( P A ) -arn(~)) from which y = Aa,/K'. ( 5 ) If we assume that a 1 : 1 complex is formed between the polyol and the electrolyte then the equilibrium constant for complexation will be which leads to Thus a plot of m-yc against y/(l -y) should be linear passing through the origin and having a gradient of l/Kc from which Kc may be calculated. Values of y were calculated for each solution using eqn ( 5 ) and the values of K, obtained from the plots based upon eqn (6) (see for example fig.9) are given in table 1. In the above derivation the assumption was made that the complex was of the 1 : 1J. G. DAWBER AND G. E. HARDY 2475 0.4 0.3 I E 0.2 0.1 0.2 0.4 0.6 0.8 rKl -7) Fig. 9. Determination of K, for mannitol+ H3BO3. Table 1. Comparison of the effects of different anions on mannitol and sorbitol mannitol sorbitol 'm(PA) 'rn(PA) c/mol / O m2 110-30 m2 substance dm-3 mol-l K, Kd n mol-l Kc Kcl n NaOH Na2Mo0, NaB( OH), borax sodium NaBF, NaAl(OH), NaH,PO, H3BO3 metaborate HPO4 0.25 0.25 0.05 0.05 0.25 0.05 0.25 0.25 0.25 0.25 0.25 - 32.5 - 86.0 27.9 136.6 388 131 82.0 - - 20.9 - - 5.3 0.08 1.98 37.0 2.08 82.4 4.4 - 11.4 - - 4.5 0.09 1.17 36.5 30.0 2.05 60.3 4.8 - 9.2 - - 0.85 1.18 0.80 1.6 0.9 0.99 0.72 1.04 - 0.80 - - - 66.3 - 9.3 47.3 92.4 42.8 23.1 1.93 3.12 6.1 - 4.68 -3.17 0.2 0.1 0.45 0.05 0.08 1.15 6.7 6.4 1.00 70.9 68.0 1.6 42.5 0.8 3.70 2.67 0.89 151.0 70.0 0.73 10.5 11.4 1.04 0.13 0.13 0.99 28.2 33.1 1.08 1.6 2.1 1.20 0.8 0.8 1.092476 POLARIMETRIC STUDY OF ELECTROLYTES AND POLYOLS type.If, however, the equilibrium involves n molecules of anion then the equilibrium constant for the formation of the complex will be (7) from which (8) In order to evaluate n we assume that ln(m - nyc) approaches In (m - yc) at low values of m and y and plots of ln[y/(l - y)] against ln(m-yc) have gradients of n and intercepts of In&. In most cases the plots were linear at low m, and for a few there was evidence of two linear portions.The values obtained for n and Kd are given in table 1. It can be seen from table 1 that in most cases the values of K, and Kd for a given substance are reasonably compatible. With the exception of NaOH, the interaction of the anions was greater for sorbitol than for mannitol, which complements the behaviour of the metal chlorides. The influence of NaOH upon the polyols may be because of the effects of desolvation rather than direct interaction. Most of the added substances caused changes in optical rotation in a positive sense, i.e. a,(,,, was more positive than a,(p). The exceptions were (a) NaOH and (b) sodium aluminate (in 2 mol dm-3 NaOH) with mannitol, for which a, changed more negatively.In most cases n was close enough to unity to allow us to conclude that most substances formed a 1 : 1 complex with the polyols. The main exceptions were for borax and sodium tetrahydroxyborate and these results are discussed below. The largest changes in optical rotation were produced by the addition of ammonium molybdate (see fig. S), but unfortunately the shapes of the graphs were such that values of K, and Kd could not be evaluated. Ammonium molybdate contains the isopolyanion Mo,O:, in which six MOO, octahedra are joined by sharing edges to form a hexagonal anulus which provides six oxygen atoms to coordinate to a central Mo atom.ls The shape of the anion is such that the corner oxygen atoms of the unshared edges of the octahedra are in positions which would facilitate hydrogen bonding with the -OH groups of several polyol molecules. At higher molybdate concentrations there will be greater competition for complex formation with the polyol and hence a change in the coordination number could account for the drastic change in the Aam results at higher m.In contrast, sodium molybdate produced an appreciable laevo rotation in mannitol and a small laevo rotation in sorbitol, but in both cases the extents of interaction ( K , and Kd) were small. The alkali-metal molybdates have the tetrahedral MOO:- anion19 and this shape may be such that it complements the polyol conformation. Sodium tungstate (Na,WO,) produced considerable changes in optical rotation in both polyols in a positive sense (see fig. lo), but the variation of a, with m was linear over the range studied and hence the values K' and therefore K, and Kd could not be evaluated, Thus in spite of the similarities between WOi- and MOO:- they produced opposite effects on the optical rotations of the polyols.Mannitol (and also sorbitol) is known to complex with boric acid and this reaction is commonly used in the titration of boric acid in volumetric analysis. The interaction could be either with the undissociated acid or with the anion, the tetrahydroxyborate ion B(OH);, although it is likely to be the latter. The values of K, and Kd for H3B0, and NaB(OH), with both polyols suggest that it is the tetrahydroxyborate ion which complexes and that the complex with sorbitol is more stable than that with mannitol. The B(0H); is tetrahedral and this may be in its favour for complexing with the polyols, but this in itself is not sufficient since, although tetrahedral W0:- complexes well, complexation with MOO:- is moderate to weak, and complexation with Cl0:- Y Kd = (1 -Y)(m--nyc)n In [ y / ( 1 - y)] = n In (m - nyc) +In Kd.J.G. DAWBER AND G. E. HARDY 2477 30 - I - 0 E “E 20 0 rn 2 \ E a Q 10 0.1 0.2 0.3 0.4 0.5 0.6 0.7 [ Na2W04] /mol dm-3 0.1 mol dm-3 sorbitol (A). Fig. 10. Effect of Na,WO, upon optical rotation of 0.1 mol dm-3 mannitol (0) and not detectable. There was some suggestions in the results when evaluating Kc and Kd for the NaB(OH),-polyol systems that the stoichiometry of the complex was changing with composition which showed itself as two distinct slopes for the plots based upon eqn (6) and (8).The problem of the stoichiometry of these particular complexes has been described previously4. and the effect was evident in the optical-rotation results. The complexes of borax with carbohydrates are often used in their characterisation. The optical-rotation results observed in this work indicate that the interaction between borax and the polyols is considerable and that the values for the complexes are very different from the am(P) values for the polyols themselves. The anion in borax is B,H,O;-, which has a bicyclic structure containing four -OH groups.2o The arrangement of -B(OH)----(OH)- around the structure is such that interaction with the polyols via hydrogen bonding is highly favourable. However, the results of the present work show that the composition of the complex may be variable in that the values of K,, Kd and n changed when the polyol/borax ratio was changed markedly (see table 1).Nevertheless the results do show that the interaction of borax with sorbitol is greater than with mannitol. The solute sodium metaborate contains the anion B,O:-, which is a cyclic structure and containing the -BO-0-BO- unit, which is very similar to the borax anion. For the same concentration of polyol (0.25 mol drn-,) the interaction of this anion with the polyols appeared to be greater than the borax anion as judged by the values of K, and Kd. Not all the boron anions complexed with the polyols. Thus the interaction of the BF, ion (which is isomorphous with C10,) with sorbitol, as judged by K,, is quite small.Therefore tetrahedral symmetry is not solely a prerequisite for complexation with the polyol. The tetrahydroxyaluminate ion, Al(OH);, is formed in solutions more basic than pH 13 and with aluminium concentration < 1.5 mol dm-3.21 This was ensured in the2478 POLARIMETRIC STUDY OF ELECTROLYTES AND POLYOLS present work by having 0.8 mol dm-3 A1 in 2 mol dm-3 NaOH, and hence the anion should be present as the tetrahedral Al(0H); ion. The complexation of the ion with both polyols can be seen to be considerable, although in the case of mannitol is more negative than the molar rotation of the polyol itself and will be complicated by the possible effects of the NaOH, as discussed above. Phosphoric acid has C3* symmetry about the P atom, giving a squashed tetrahedral structure containing three -OH groups, and the interaction with sorbitol is significant although is little different from the molar rotation for sorbitol.The interaction is roughly halved in the case of the H,PO, ion, where there are now only two -OH groups, and in the case of the PhPOi-, in which there are no -OH groups, no interaction was detectable. Thus the general conclusions with regard to the interaction of anions with mannitol and sorbitol in aqueous solution seem to be that a tetrahedral anion has a shape which is compatible with the polyol structure, but that the corners of the tetrahedron should possess -OH groups as opposed to 0 or F atoms. The very favourable interaction observed with the borax anion and the metaborate anion is likely to be as a result of the shape of the B-0-B segments of these cyclic anions matching the -OH group conformation of the polyols.I. T. Norton, D. M. Goodall, E. R. Morris and D. A. Rees, J. Chem. Soc., Faraday Trans. I , 1983, 79, 2475. F. Franks and M. D. Pedley, J. Chem. Soc., Faraday Trans. 1, 1983, 79, 2249. R. M. Williams and R. H. Atalla, in Solution Properties of Polysaccharides, ed. by D. A. Brant (American Chemical Society, Washington D.C., 1981). J. G. Dawber and D. H. Matusin, J. Chem. Sac., Faraday Trans. 1 , 1982,78, 2521. H. B. Davis and C. J. B. Mott, J. Chem. Soc., Faraday Trans. I , 1980, 76, 1991. R. V. Jasra and J. C. Ahluwalia, J. Chem. Soc., Faraday Trans. I , 1982, 78, 1677. D. P. Wilson and W. Y. Wen, J. Phys. Chem., 1976, 80, 431. G. Dipaola and B. Belleau, Can. J. Chem., 1977, 55, 3825. lo 0. D. Bonner and P. J. Cerutti, J. Chem. Thermodyn., 1976, 8, 105. l1 D. Horton and M. J. Miller, J . Org. Chem., 1965, 30, 2457. S. A. Angyl and R. Lefur, Carbohydr. Res., 1980, 84, 201. l3 Handbook of Chemistry and Physics, ed. R. C . Weast (C.R.C. Press, Columbus, Ohio, 60th edn, 1980). l4 Dictionary of Organic Compounds (Eyre and Spottiswoode, London, 1965). l5 Handbook of Chemistry and Physics, ed. R. C. Weast (C.R.C. Press, Columbus, Ohio, 45th edn, 1964). l6 J. G. Dawber, J . Chem. SOC., Faraday Trans. I , 1978, 74, 960. 17 D. A. Johnson, Some Thermodynamic Aspects of Inorganic Chemistry (Cambridge University Press, l8 R. B. Heslop and K. Jones, Inorganic Chemistry (Elsevier, Amsterdam, 1976). l9 F. X. N. M. Kools, A. S. Koster and G. D. Rieck, Acta Crystallogr., Sect. B, 1970, 26, 1974. *O K. F. Purcell and J. C. Kotz, Inorganic Chemistry (Saunders, Philadelphia, PA, 1977). 2 1 M. Yoshio, H. Waki and N. Ishibashi, J. Inorg. Nucl. Chem., 1970, 32, 1365; R. J. Moolenar, C. J. Evans and L. D. McKeever, J . Phys. Chem., 1970, 74, 3629. ' F. Franks, D. S. Reid and A. Suggett, J. Solution Chem., 1973, 2, 99. Cambridge, 1968). (PAPER 3/2894)

ISSN:0300-9599    

DOI:10.1039/F19848002467

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. I, 1984, 80, 2479-2488 Factors Influencing the Temperature-programmed Reduction Profiles of Vanadium Pentoxide BY HANS BOSCH,* BERT J. KIP, JAN G. VAN OMMEN AND PAUL J. GELLINGS Twente University of Technology, Department of Chemical Technology, P.O. Box 217, 7500 AE Enschede, The Netherlands Received 28th November, 1983 The temperature-programmed reduction (t.p.r.) of bulk V,O, has been examined as part of a study of the reducibility of V,O,-containing catalysts. T.p.r. profiles have been studied as a function of flow rate, heating rate and sample weight. From experiments at different flow rates it is concluded that the order of the reduction rate in hydrogen is low or even zero. A rule of thumb has been derived to provide an easy check on possible exhaustion of hydrogen in the feed.The influence of sample weight and heating rate is explained in terms of the formation of water in the sample during reduction. The reduction of bulk V,O, to V,O, proceeds in several steps; intermediate species include V,O,, and VO,. The apparent activation energy of ca. 200 kJ mol-l indicates that solid-state diffusion influences the reduction process of V,O,. Vanadium oxide based catalysts are used in many industrial oxidation processes. Among many other authors Roozeboom et al., studied the selective oxidation of methanol over such catalysts. Recently, van Hengstum et al.3 reported results on the selective oxidation of toluene over V,O,/TiO,. According to the oxidation-reduction rnechani~m,~ lattice oxygen plays an important role.Sachtler et al., showed that the rate at which the bond strength of the lattice oxygen increases with the degree of reduction influences the selectivity. In the present study temperature-programmed reduction6 (t.p.r.) has been used to investigate the different types of oxygen present in vanadium oxide which may be involved in the oxidation process over such catalysts. T.p.r. is a relatively new technique used to characterise reducable materials. The reduction is measured by monitoring H, consumption while increasing the temperature of the sample at a constant rate. In this way, one or more reduction peaks occur at different temperatures and the reduction profile or ‘spectrum’ can be obtained easily. In the case of heterogeneous catalysts, the compounds to be reduced are present on the surface or are potentially available to the surface. This method can therefore serve as a powerful tool to obtain ‘fingerprint’ characterisations of the reducibility of catalysts.’ Hurst et a1.8 have recently given an excellent review of both the theoretical background and the applications of the technique.However, it is difficult to compare t .p.r. spectra obtained by different investigators. The conditions reported in the literature differ widely in H, concentration and flow rate and in sample size and pretreatment. Some significant examples are given in table 1 . The parameters mentioned do not affect t.p.r. profiles independently but it is very likely that the optimum conditions differ from one system to another because of large differences in the reducibilities of the materials investigated.The information obtained from t.p.r. profiles is merely comparative. A quantitative approach is difficult because of the complexity of the bulk oxide reduction processes thought to be taking place in the reduction of different types of catalyst^.^ 24792480 TEMPERATURE-PROGRAMMED REDUCTION OF VANADIUM PENTOXIDE Table 1. Experimental conditions used in t.p.r. studies sample size volume flow (H, consumed rate/cm3 / P m a pretreatment min-l H, (%) 1 h a t 453 Kin 5% 0 , b 9 d 5b 4 4 , d , C j 9 1 , m 40-60a 2 h at 373 K in vacuoa 10a,V.i 88 70-100b 1 h at 773 K in dry aire, f 18-20e* h* 15f 1000-17OOc 4 h at 773 K in 0,Q 37k* 66e, h , Q 1 h at 823 K in dry air" a Ref. (6); ref. (21); ref.(22); ref. (23); ref. (10); f ref. (11); V ref. (24); ref. (13); ' ref. (25); 3 ref. (7); ref. (14); ref. (27); ref. (12); ref. (26). The aim of the present work is to study the reduction behaviour of bulk V,O, as part of the reduction of vanadium oxide containing catalysts. This paper presents a systematic study of the parameters which may govern the resolution of t.p.r. profiles, such as sample mass, heating rate, volume flow rate, grain size and the presence of water. EXPERIMENTAL APPARATUS The t.p.r. apparatus used in this study was constructed according to the descriptions given with the following changes being made. First, the reference flow through the thermal-conductivity detector (t.c.d.) was regulated separately and all gas lines were provided with flow and pressure controllers.Secondly, the H,+Ar stream from the reactor and dryer was diluted by an additional argon flow, typical volume flow rates being: H, + Ar 10 cm3 min-I, Ar (diluent) 10 cm3 min-' and Ar (reference) 20 cm3 min-l [all gas volumes are expressed at room temperature (298 K) and atmospheric pressure]. This set-up minimizes the change in flow rate caused by H, uptake. Ar instead of N, was used to avoid possible nitride formation, e.g. in the case of Fe-containing samples.1o Finally, the quartz reactor was made as small as possible (i.d. 4 mm, reactor dead space 3 cm3, total dead space 12 cm3) to avoid peak-widening by back-mixing. The temperature was measured with a thermocouple outside the reactor, its tip positioned near the sample, to prevent reduction of the thermocouple wall.Blanks were carried out to correct for the temperature difference of 9 K involved. MATERIALS All gases used were of the highest purity available commercially. Traces of oil were removed by activated charcoal filters and moisture was removed by molecular sieves. All experiments were carried out with 9% H, in argon. V,O, for use in bulk oxide experiments was supplied by Merck (pro analysis) or freshly prepared by decomposition of NH,V03 (Merck) or vanadyl acetylacetonate (Baker). Some t.p.r. experiments were performed starting with VO, (Jansen Chemicals). The main pretreatment of the samples consisted of heating in air at 773 K for 1 h, as recommended by ThomasLo and Yao," to assure good reproducibility. ADDITIONAL TECHNIQUES The results of some t.p.r.experiments were compared with the weight losses in similar thermogravimetric experiments (Du Pont 95 1 therrnogravimetric analyser). In some cases, X-ray diffraction was used to identify the phases present after partial or complete reduction (Philips PW 1025.25 diffractometer).H. BOSCH, B. J. KIP, J. G. VAN OMMEN AND P. J. GELLINGS 248 1 1 I I I I 800 900 1000 1100 1200 T/K Fig. 1. Influence of heating rate on t.p.r. spectra. Volume flow rate, 10 cm3 min-l, 9% H,. Run (12) sample weight 14.8 mg, heating rate 4.6 K min-l; run (8) sample weight 16.2 mg, heating rate 5.6 K min-l; run (15) sample weight 14.9 mg, heating rate 9.8 K min-'; run (81) sample weight 10.9 mg, heating rate 16.4 K min-' ; run (1 15) sample weight 4.3 mg, heating rate 16.1 K min-l.RESULTS The reduction of bulk V205 proceeds in at least four different steps. Typical results are given in fig. 1. At suitable conditions a slow rise in H, consumption is followed by two sharp, well resolved peaks. At 8 K min-l and 10 cm3 min-l the temperatures for the maximum reduction rates of the first two peaks are 930f2 and 964+3 K, respectively. The next two peaks do not show such good separation, but their positions can be reproduced within & 5 and f 8 K, respectively. The increase in reduction rate at the onset of the third peak is significantly higher than that at the first peak. This phenomenon is discussed in the Discussion section. A series of experiments was carried out to check the possible influence of heat-transport limitations.10 mg V205 was either used undiluted or mixed with 65-270 mg of quartz. All the peaks appeared at the same temperatures, within the limits given above. Thus heat-transport limitations are not present. Similarly, no significant change in peak positions was observed when t.p.r. results frommeasurements with grains of diameter 0.3-0.6mm were compared with those for V205 powder (particle diameter < 50 pm); hence there are no mass-transport limitations either. The residence-time distribution was measured by injection of hydrogen at the top 81 FAR 12482 TEMPERATURE-PROGRAMMED REDUCTION OF VANADIUM PENTOXIDE Table 2. T.p.r. results as a function of heating rate heating flow sample temperature of peak maximum/K rate/K /cm3 weight run min-1 min-l /mg 1 st 2nd 3rd 4th remarks 12 27 13 8 87 29 14 23 89 15 78 17 25 79 26 81 115 4.6 4.7 5.3 5.6 5.8 6.1 6.7 7.5 8.1 8.3 9.1 9.4 10.1 10.6 11.2 12.6 14.6 16.1 10 20 10 10 10 20 10 20 10 20 10 10 10 20 10 20 10 10 14.8 11.0 17.3 16.2 10.0 11.1 14.6 10.0 1 1 + 1 10.9 14.9 9.7 10.7 11.1 9.9 11.5 10.8 4.3 874 906 998 101 1 - 895 924 1005 900 932 1009 1024 - 920 952 1036 1052 - 914 949 1029 1041 - 924 96 1 1033 1045 - 923 956 - 93 1 968 1038 1054 - 930f2 964+3 1042f5 1056k8 average - - __ - 930 932 934 936 94 1 936 943 936 93 1 962 968 968 973 975 974 979 969 958 1032 1058 1054 1059 1056 1067 1056 1034 1028 1055 1076 1074 1075 1067 1084 1095 1116 1070 of 5 runs 5th peak at 1088 K 5th peak at 1095 K 5th peak at 1093 K 5th peak at 1130 K 5th peak at 1091 K of the catalyst bed. The peak widening, caused by diffusion, was < 2 K.Thus, plug flow can be assumed to occur. All temperatures mentioned in this paper have been corrected for the residence time (50 s). This correction was usually 10-20 K. Changing the sample size from 5 to 43 mg did not change the position of the first two peaks. With samples of > 20 mg, the third and fourth peaks shifted to higher temperatures. Similarly, at heating rates of 7 K min-l and higher, variation of volume flow rates from 10 to 20 cm3 min-l did not change the peak positions either; some examples of such experiments are given in table 2. The lower flow rates seem to shift the first peak to a slightly lower temperature, e.g. compare runs 17 and 25. Since typical H, conversions amount to 20-60%, changes in hydrogen concentration because of variations in sample weight and flow rate do not influence these results.This suggests that, although the inlet hydrogen concentration was not varied, the order in hydrogen is low or even zero. Some t.g. experiments were carried out to identify the compounds formed successively in the different reduction steps. In these experiments samples were cooled in nitrogen after the first peak appeared and then examined by X.r.d. In subsequent runs, similarH. BOSCH, B. J. KIP, J. G. VAN OMMEN AND P. J. GELLINGS 2483 Table 3. Compounds detected by X.r.d. at various stages in the reduction of V,O," ~ ~~~~ t.g.a., 10 K min-l, 20.24 mg V,O,; v,o,, V6013, vo,, v,o,, weight change/mg compounds ortho- mono- mono- hexag- detected rhombic clinic clinic onal observed theoretical starting S n.0.n.0. n.0. 1.15 1.187 reduction n.0. S W n.0. 0.625 0.593 compound terminated before 2nd peak terminated before 3rd peak at 1123 K total 3.55 total 3.56 reduction n.0. n.0. S n.0. 1.775 1.78 final product n.0. n.0. n.0. S a s, strong; w, weak; n.o., not observed. samples were cooled after the second and fourth peaks. The compounds identified by X.r.d. before and after these peaks are given in table 3. All the observed diffraction peaks could be labelled. Mainly V6013 is formed after the first reduction step. In the second step, V6013 is completely reduced to VO, before subsequent reduction takes place. In the final product, formed at 1123 K, only V,O, can be detected. From this reduction stiochiometry the theoretical weight changes after the first, second and fourth peak were calculated.These theoretical values are also given in table 3. They show good agreement with the values from the t.g. experiments. A number of t.p.r. runs are summarized in table 2. The positions of the peaks move ca. 70 K toward higher temperatures when heating rates are increased from 5 to 16 K min-l. Gentry et aZ.12 have given a method for calculating the apparent activation energy of the reduction from such changes. Assuming zero-order behaviour in hydrogen and that longitudinal mixing is absent (i.e. there is plug flow), they arrive at the equation: In (T&//3) = E / R T, + constant (1) where T, is the temperature of the peak maximum (K), is the heating rate (K s-l) and E is the apparent activation energy (kJ mol-I).In the derivation of eqn (l), the underlying assumptions are that a purely chemical process is described, its rate being governed by an Arrhenius equation, and that a number of parameters are constant during the reduction of a particular compound. Application of eqn (1) to the positions of the first peak results in a surprisingly high value of the apparent activation energy of ca. 200 kJ mol-l at heating rates of 5.6-10.1 K min-l. At lower as well as higher heating rates a marked deviation from the best straight line occurs. The heating rate influences not only the peak positions but also the form of the t.p.r. profiles. Some significant examples are given in fig. 1. A heating rate of < 5 K min-l results in very poor resolution of the first two peaks.A fifth peak can be observed with a rate > 10 Kmin-l although this is not well resolved. At 16 K min-l, the change in the t.p.r. profile is even more pronounced [fig. 1, run (Sl)]: the second, third and fourth peaks have become smaller and the fourth and fifth peaks are now completely resolved. All profiles mentioned so far in relation to fig. 1 were measured with a sample weight of ca. 11-15 mg. T.p.r. experiments at 16 K min-l with 4.3 mg [fig. 1, run (1 15)] resulted in profiles similar to those obtained at 10 K min-l. 81-22484 TEMPERATURE-PROGRAMMED REDUCTION OF VANADIUM PENTOXIDE I I I I I I 1 1 TIK Fig. 2. Influence of (a) crystal structure and (b) prereduction on the onset of the first reduction peak. (a,) Reduction of V,O, and (a,) reduction of VO,.(b,) First reduction peak of calcined V205 and (b,) reduction peak of prereduced and reoxidised V,05. 500 700 9 00 11 00 Thus, sample weight and heating rate are not independent experimental parameters. Combined they determine the possibility of exhaustion of the feed and/or a possible influence of water, resulting in spectra which do not fit the usual pattern. The first two t.p.r. peaks from the reduction of freshly prepared V,O, (from ammonium vanadate or vanadyl acetylacetonate) are similar to those mentioned above, the only difference being a shift to a higher temperature: 953 and 944 K, respectively, as compared with 930 K for the sample of V,O, supplied by Merck. However, the third peak is broad and does not show resolution into two distinct peaks.Obviously the pretreatment was insufficient to produce similar structures. Some t.p.r. experiments were performed starting from VO, instead of V,O, [fig. 2(a)]. The first reduction peak differs from the corresponding peak starting from V,O, in position and form and is observed at 1027 K as compared with 1037-1047 K for the third peak of the V,O, profile. The reduction of VO, starts relatively slowly as compared with that of VO, obtained during the reduction of V,O,. However, the low-temperature part of both profiles in fig. 2(a) is similar. The latter results, as well as the high value of the apparent activation energy, suggest that the structure of the solid phase plays an important role. To investigate this, the following t.g. experiments were conducted.A regular reduction experiment was carried out with V,O,, but the sample was cooled as soon as the first reduction peak was completed [fig. 2(6)]. The sample was reoxidized until constant weight, the final oxidation temperature not exceeding the temperature of the peak maximum. A starting weight corresponding to V,O, was always obtained. After this, the reoxidized sample was reduced again under the same conditions as before [fig. 2, run (6,)) Although the peak maxima of runs (b,) and (6,) appeared at the same temperature, the onset of the reduction in run (b,) started at a markedly higher temperature. This behaviour is similar to that for the reduction of VO, as mentioned above: reduction of VO, as an intermediate product proceeds in a similar way to the reduction of V,O, in run (b,).H.BOSCH, B. J . KIP, J. G. VAN OMMEN AND P. J. GELLINGS 2485 A n I I I I I I I I 500 7 00 9 00 11 00 1300 TIK Fig. 3. Influence of water vapour on t.p.r. profiles. (a) Heating rate 10 K min-l; (b) heating rate 20 K min-l; (c) heating rate 10 K min-', reduction with Ar+H, with 3% H,O. Finally, the influence of water on the reduction profile was investigated by t.g. experiments using H, + Ar saturated at room temperature with water. The presence of water has a considerable affect on the reduction profile, as can be seen in fig. 3 by comparing run (a) (no H20) and run (c) (H,O added). The number of peaks is increased and the temperature at which the reduction is completed is higher. A similar result can be obtained by just increasing the heating rate, as can be seen from fig.3, run (b). DISCUSSION Temperature-programmed reduction has been claimed to be a successful ' fingerprint method' for the characterisation of reducible species. This section will focus on the significance of the ' fingerprints' obtained. Reported t.p.r. spectra for V205 appear to differ notably from our results. Roozeboom et aL.l3 found one single reduction peak at ca. 800 K (5 K min-l, 2 mg sample weight as calculated from their data). Bond1* obtained spectra from V20, (Merck) similar to ours, the first two peaks shifted only slightly, the last of these not being resolved, as in our case, into at least two peaks (5 K min-l, 100 mg). Possible reasons for these different results are discussed below. It has already been shown in the Results section that the reduction probably proceeds as follows: The first two peaks (fig.1) correspond to the first two steps. The third and last t.p.r. peaks together represent the last step, which clearly comprises more than one step; V,Oll is one of the possible intermediate compounds. V,05 --+ $V,Ol, --+ 2V0, -+ V20,.2486 TEMPERATURE-PROGRAMMED REDUCTION OF VANADIUM PENTOXIDE Thus, distinct reduction steps show up as different peaks in the t.p.r. profile. When the place of a particular peak depends solely on kinetic parameters, the peak shift caused by changes in heating rate contains information on the apparent activation energy. Although such factors as surface composition, defects and geometry affect the reactivity of the reduction of even simple metallic oxide~,~ it is very difficult to account for these factors in a non-isothermal case.8 Gentry et al.12 have shown, for the case of the reduction of copper-exchanged zeolites, that the influence of H, concentration and flow rate can be described satisfactorily by means of a simple kinetic equation.The high value we obtained using this method for the reduction of bulk V,O, suggests that the behaviour of this system is more complicated. Peaks other than the first one show that there are compounds which are not present in the starting material but are formed during the reduction. The local rate of formation will be different because of local oxygen gradients in the particles, depending on experimental conditions such as heating rate and volume flow rate.Thus that part of the spectrum not only contains information relating to the properties of the starting material but also reflects how the reduction proceeds, as determined by kinetics as well as by experimental parameters. It is well documented how oxygen is removed from the V,O, matrix.l5?l6 Oxygen species to be removed leave behind anionic vacancies which are not randomly distributed but lie in particular planes. Therefore, beyond a certain degree of reduction, the structure may change by just small displacements along those vacancy- rich planes. The resulting structure is known as a shear s t r ~ c t u r e . ~ ~ Reasoning along these lines, it can be understood why a relatively low heating rate does not resolve the first two peaks, in contrast to the results at higher heating rates.At a certain temperature the degree of reduction of the sample in run (12) (see fig. 1) is higher compared with that in run (1 5 ) because of its lower heating rate. Therefore, in the first case a second phase can be formed at lower temperatures. Because of the lower temperature, the conversion proceeds relatively slowly and the resulting peak is quite broad : shear structures might therefore be present in a larger proportion of the sample instead of a progression of stoichiometric compounds, as described above. Local regions may be converted into a new shear structure, whereas other parts are still in the initial structure. This might also explain the poor reproducibility of the peak positions at low heating rates (table 2) and also the deviation from an Arrhenius plot on the lower-temperature side.Apart from the change in crystal structure because of the formation of shear structures, it has to be remembered that the temperature region in which the first peak appears (see table 2) is not far from the melting point of V,O, (963 K). This also gives rise to enhanced mobility of the ions in the matrix, which might contribute to the changes in crystal structure. However, this is probably not a significant process since prereduction and reoxidation in this temperature region do not influence the position of the subsequent rereduction peak (fig. 2). The influence of structural parameters may be reflected by a small shift of 10-20 K to higher temperatures starting from two kinds of freshly prepared V,O, and one of 10-20 K to lower temperatures starting from VO,.Crystal structure and orientation affect the reducibility,13 but this is far more pronounced in supported systems, which may show shifts of several hundred degrees, as will be demonstrated in a subsequent paper. With an increasing heating rate, the first peak shifts close (20 K) to the melting point of V,O, and the second to the region where V,O,, is not stable.18 These two peaks show an upper limit and cannot shift further upon increasing the heating rate. At those heating rates no meaningful values bf the apparent activation energy can be calculated, leaving only a relatively small range of 5-10 K min-l in which the method of GentryH. BOSCH, B. J. KIP, J. G. VAN OMMEN AND P.J. GELLINGS 2487 ut aZ. can be applied. However, the value of E obtained (200 kJ mol-l) is an indication of the occurrence of a solid-state diffusional process. All experiments presented in this paper are thought to be free from heat- and/or (pore-diffusional-) mass-transport limitations. This does not hold for run (8 l), which is presented in fig. 1 to show that it is the combination of heating rate and sample weight which determines possible exhaustion of the feed H,. Comparison of run (81) (10 mg) and run (1 15) (4.3 mg) show that the same form of t.p.r. profile can be obtained, even at relatively high heating rates, provided that a suitable sample weight is used. The hydrogen feed rate should be high enough to match the highest possible oxygen removal rate.Regular checks should thus be made at the highest reduction rate to ensure that the gradient of the hydrogen concentration (along the reactor and across the particles) is sufficiently small. An equation which relates the maximum allowable sample weight and the amount of oxygen removed during a particular reduction step (a particular peak) can be derived as follows. Starting from the boundary condition 4Hg > rredGM0, (where &2 is the flow rate of H,. rred is the specific reduction rate and GMOn is the total sample weight) and assuming that all the oxygen involved in the particular step considered is removed at a constant rate during an interval At = A7;,//?, this leads to the following constraint where A q is the peak width at half height, O/M is the molar ratio of oxygen removed and metal present and MMOn is the molar weight of MO,.If hydrogen consumption reaches loo%, the peak would assume a flat top. The use of eqn (2), however, indicates the region of near exhaustion, which is only the case in the deviant run (81). A similar criterium is derived recently by Monti and Baiker.lg The influence of sample weight, however, cannot be understood simply in terms of possible exhaustion of the feed. Gentry et aZ.12 observed a considerable change in their t.p.r. spectra with increasing sample weight. The two-step reduction of Cu2+ could not be resolved above a certain sample weight. Both at higher sample weights and at higher heating rates, a higher partial pressure of water vapour will be built up. Since the presence of water is known to enhance sintering processes,2o its presence might also be responsible for structural changes during reduction.Although Rooze- boom et aZ.13 have shown that water saturation of the feed did not cause a shift of their t.p.r. profiles, these profiles were found at a temperature > 200 K lower. The influence of water vapour at higher temperatures cannot be ruled out. In fact, the presence of water affects our t.p.r. profiles markedly, as shown in fig. 3, which also shows that too high a product of heating rate and sample weight (p G,,,) affects the t.p.r. spectrum in the same way as does the addition of water vapour. This leads to the conclusion that at a choosen heating rate the sample weight should be as small as possible to keep a low partial pressure of water vapour.A remarkable phenomenon is shown in fig. 2. The starting temperature of the first reduction peak is higher after prereduction and subsequent reoxidation of V,O, ; however, the temperature at which the maximum reduction rate occurs is the same in both cases. Analogously, reduction of bulk VO, starts at a relatively low temperature, while the reduction peak of VO, formed as an intermediate product from reduced V,O,, does not show this behaviour. The fact that the peak maximum of the prereduced sample [fig. 2, run (b,)] does not shift excludes the possibility that this phenomenon reflects a rate-determining step before that involving the reduction of the bulk phase, otherwise the peak would have shifted to a lower temperature after this step was removed by the prereduction.The nature of this process, which will have2488 TEMPERATURE-PROGRAMMED REDUCTION OF VANADIUM PENTOXIDE a very low activation energy (concluded from the slope of the lower-temperature part), is not yet clear. Experiments on supported V,05, not yet reported, show that this onset at a lower temperature is never observed in ‘monolayer’ catalysts. In the case of the reduction of NiO, Monti and Baikerlg describe this behaviour to an induction period during which stable nuclei are formed on the surface of the oxide. However, then the peak position would be affected by the pretreatment described above. The finding that the peak position is not influenced suggests that the virgin V205 (and also the VO,) contains structural irregularities which are removed by the preceeding reduction-oxidation cycle.CONCLUSIONS Calculations of apparent activation energies show that the reduction of bulk V205 is limited by solid-state diffusion. The highest resolution of the t.p.r. spectra is obtained at a relatively high heating rate and a small sample weight. It must be ascertained that no exhaustion of the feed takes place at the highest reduction rate and a rule of thumb has been supplied to provide for this. A pronounced influence of water, formed during the reduction, on the t.p.r. profiles of V,05 is observed only at relatively high values of the product of heating rate and sample weight. We thank Prof. G. C. Bond for many stimulating discussions, Prof. J. R. H. Ross for his helpful advice during the preparation of this manuscript and J.Boeysma for his skilful measurements of the X.r.d. patterns. D. B. Dadyburjor, S. S. Jewur and E. Ruckenstein, Catal. Rev. Sci. Eng., 1979, 19, 293. F. Roozeboom, P. D. Cordingly and P. J. Gellings, J. Catal., 1981, 68, 293. A. J. van Hengstum, J. G. van Ommen, H. Bosch and P. J. Gellings, Appl. Catal., 1983, 8, 369. P. Mars and D. W. van Krevelen, Chem. Eng. Sci., 1954, 3, 41. W. H. M. Sachtler, G. J. H. Dorgelo, J. Fahrenfort and R. J. H. Voorhoeve, Proc. 4th Inf. Congr. Catal. (Akademai Kiado, Budapest, 1971), vol. I, p. 454. S. D. Robertson, B. D. McNicol, J. H. de Baas and S. C. Cloet, J. Catal., 1975, 37, 424. J. W. Jenkins and B. D. McNicol, Chem. Technol., 1977, 7 , 316. * N. W. Hurst, S. J. Gentry, A. Jones and B. D. McNicol, Catal. Rev. Sci. Eng., 1982, 24, 233. M. Bulens, Ann. Chim., 1976, 13. lo R. Thomas, Thesis (University of Amsterdam, 198 1). l1 H. C. Yao, J. Catal., 1979, 59, 365. l 2 S. J. Gentry, N. W. Hurst and A. Jones, J. Chem. SOC., Faraday Trans. I , 1979, 75, 1688. l3 F. Roozeboom, M. C. Mittelmeyer-Hazeleger, J. A. Moulijn, J. Medema, V. H. J. de Beer and P. J. l4 G. C. Bond, personal communication. l5 G. L. Simard, J. F. Steger, R. J. Arnott and L. A. Siegel, Znd. Eng. Chem., 1955, 47, 1424. l6 E. Gillis, C. R. Acad. Sci., 1964, 258, 4765. l7 F. S. Stone, Oxide Crystal Chemistry and Catalysis, in Chemistry and Chemical Engineering of Catalytic Processes, ed. R. Prins and G. C. A. Schuyt (Sijthoff and Noordhoff, Alphenaan den Ryn, 1980), p. 477. G. Anderson, Ada Chem. Scand., 1954, 8, 1599. Gellings, J. Phys. Chem., 1980,84, 2783. Is D. A. M. Monti and A. Baiker, J. Catal., 1983, 83, 323. 2O G. C. Kuczynski, L. Aberethy and J. Allan, in Kinetics of High-temperature Processes, ed. W. D. Kingery (Technology Press of M.I.T., Wiley, New York and Chapman and Hall, London, 1959), chap. 12. 21 N. Wagstaf and R. Prins, J. Catal., 1979, 59, 434. 22 V. C. F. Holm and A. Clark, J. Catal., 1968, 11, 305. 23 P. J. G. Koopman, A. P. G. Kieboom and H. van Bekkum, J. Catal., 1981, 69, 172. 24 T. Paryjczak, J. Rynkowski and S. Karski, J. Chromatogr., 1980, 188, 254. 25 F. Mahoney, R. Rudham and J. V. Summers, J. Chem. SOC., Faraday Trans. I , 1979,75, 314. 26 A. J. Roosmalen, D. Koster and J. C. Mol J . Phys. Chem., 1979, 84, 3075. 27 A. Lycourghiotis, C. Defosse, F. Delanney, J. Lemaitre and B. Delmon, J. Chem. Soc., Faraday Trans. I, 1980, 76, 1677. (PAPER 3/21 12)

ISSN:0300-9599    

DOI:10.1039/F19848002479

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1984,80, 2489-2507 Non-isothermal Surface Barrier Model for Gas Sorption Kinetics on Porous Adsorbents BY SHIVAJI SIRCAR* AND RAVI KUMAR Air Products and Chemicals Inc., Box 538, Allentown, Pennsylvania 18105, U.S.A. Received 6th December, 1983 A review of the progressive development of the models for kinetics of adsorption is given and a new non-isothermal surface barrier model is proposed. The present model assumes that the adsorbate mass-transfer resistance is confined at the surface of the adsorbent particle and that heat transfer from the adsorbent mass is controlled by both its effective thermal conductivity and the external film heat-transfer resistance. Analytic equations are derived for describing adsorbate uptake and adsorbent temperature profiles.The model successfully describes the adsorption of iso-octane on 13X, n-pentane on 5 A and carbon dioxide on 4 A zeolites using a set of self-consistent parameters. The internal and the external heat-transfer coefficients evaluated from the uptake data agree well with those calculated using available theories. Design of adsorbers for gas separation requires both the adsorption equilibrium and the kinetics data for the adsorbates of interest. The equilibrium adsorption isotherms can be measured without defining either the structure of the adsorbent or the adsorption mechanism. Both of these properties, on the other hand, determine the mass-transfer characteristics of an adsorbate. Thus a mathematical model of the adsorption process is needed for estimation of the adsorbate mass-transfer coefficient or diffusivity from the experimental adsorbate uptake data.Numerous models have been published in an attempt to achieve this goal. A brief review of the key models is given below. REVIEW OF THE ADSORPTION KINETICS MODELS The earlier models assumed that the adsorption process is isothermal and that the mass transfer could be described by Fick's law of diffusi0n.l According to this model, the mathematical equations for adsorption into a spherical adsorbent particle of radius R, are: n = n(C) (2) fi(0) = no(Co>, N 4 = %&,) (4) where C and n are, respectively, the gas-phase adsorbate concentration within the 24892490 GAS SORPTION ON POROUS ADSORBENTS adsorbent pores and the corresponding equilibrium adsorbate loading at radius r of the adsorbent particle at time t during the uptake process, A is the average adsorbate loading in the particle at time t, no and n, are, respectively, the initial and the final equilibrium amounts absorbed at pressures Po and P,, E is the void fraction of the adsorbent and p p is its density, D is the adsorbate diffusivity and M is equal to EC or np, depending on whether the adsorption mechanism is the diffusion of the gaseous adsorbate within the pores of the adsorbent or the diffusion of the adsorbed molecules on the adsorbent surface.Eqn (2) represents the equilibrium adsorption isotherm in a functional notation. For a physically homogeneous adsorbent, defined by a constant D, eqn (1)-(3) can be solved analytically1 to obtain the uptake curve (E against t ) for (a) the surface diffusion mechanism irrespective of the nature of the equilibrium isotherm by assuming that ppn % EC and (b) the gas-phase diffusion mechanism with a linear adsorption isotherm (n cc C).Solutions of the model are available for both constant- pressure (gravimetric) and variable-pressure, constant-volume (volumetric) kinetic tests. Although these solutions were most frequently used prior to 1970 for their mathematical simplicity, many variations of the above isothermal Fickian diffusion model were proposed to account for (a) non-spherical particle geometry (cylindrical and cubic),l? (b) non-uniform particle-size distribution of the adsorbent mass used in the test,2 (c) concentration dependence of the adsorbate diffusivity3 and ( d ) polydisperse adsorbent pore structure having a micro- and macro-pore diffusivity.*-* Unfortunately, the uptake process is seldom isothermal because the heat of adsorption cannot be removed from the adsorbent as fast as it is generated in a conventional gravimetric or a volumetric apparatus.Consequently, the adsorbent temperature ( T ) rises during the initial portion of the uptake process, reaches a maximum and then declines to the starting temperature (T,) when the true thermal equilibrium is reached. This is shown in fig. 1, which plots dimensionless temperature (Y), uptake ( F ) and pressure (0) as functions of dimensionless time (z) and adsorbent particle radius (x). The definition of these dimensionless variables are given later in the text.Various investigators have measured such a general trend in the adsorbent temperature during the uptake p ~ o c e s s . ~ - ~ ~ Since both the equilibrium amount adsorbed and the adsorbate diffusivity can be strong functions of temperature depending on the magnitudes of the heat of adsorption and the diffusional activation energy of the adsorbate-adsorbent system, the assumption of adsorbent isothermality during uptake can lead to severe errors in the estimation of the mass-transfer coefficients . In order to circumvent this problem, several authors have proposed the use of a differential adsorption test (d.a.t.) where the step change (P, -Po) in the adsorbate pressure and the adsorbate loading (n, -no) during the uptake process are kept small so that the change in the adsorbent temperature (T- T,) is small at any time t and thus adsorbent isothermality can be a s s ~ m e d .l ~ - ~ ~ However, it has been shown that even a very small adsorbent temperature change can introduce large errors in the estimation of the adsorbate diffusivity by the isothermal models unless the step change in the equilibrium adsorbate loading (n, -no) during the d.a.t. is much greater than the differential change in n , [n,(T, P,) ---n,(T,,, P,)] caused by the change in the adsorbent temperature.207 21 Such a condition may be difficult to achieve for systems with a moderate heat of adsorption and a relatively fast rate of adsorption. Adsorbate diffusivities estimated using the d.a.t. and the isothermal models have often resulted in a U-shaped concentration dependence,22 which may actually be an aberration due to the non-isothermal effect^.^^-^^ The d.a.t., on the other hand, provides an important simplification in theS.SIRCAR AND R. KUMAR 249 1 1 .o 0 7 Fig. 1. Schematic of the uptake curve and the adsorbent temperature profiles. mathematics of the kinetic models because it allows linearization of the equilibrium isotherm with respect to P and T within the bounds of the test conditions: where n is the equilibrium amount adsorbed at P and T and a and b are, respectively, the pressure and the temperature coefficients of the equilibrium isotherm evaluated at the final conditions of the test. The importance of the adsorbent non-isothermality during uptake has been recognized in recent years and several models are proposed for the constant-pressure d.a.t.by assuming that the heat transfer from the adsorbent mass is controlled by an external film heat-transfer resistance.20* 239 2 4 9 28-28 The corresponding heat-balance equation is given by (8) where Tis the adsorbent temperature at time t, c p is the heat capacity of the adsorbent, q is the isosteric heat of adsorption at no3 and T, and h is the external film heat transfer coefficient. Analytic equations for the non-isothermal uptake (A against t ) have been obtained for the Fickian surface diffusion of the adsorbed molecules into a cylindrical2s and a spherical2' adsorbent particle using eqn (8) for heat balance. A solution is also available for the case where the mass transfer is controlled by the Fickian diffusion of the adsorbate through the interparticle void space of a cylindrical adsorbent mass.28 dT dfi dt dt c - = q--h(T- T,)2492 GAS SORPTION ON POROUS ADSORBENTS These solutions typically give ri as a summation of a series of time-dependent exponential terms with transcendental functions of the transfer coefficients as pre-exponents.Simpler analytic solutions for uptake also have been obtained for spherical adsorbents by assuming that the mass transfer can be described by a lumped-up gas film resistance at the outside surface of the adsorbent particle20 or a solid film resistance within the adsorbent particle.24 This assumption allows the use of a linear driving force model (1.d.f.) for the adsorbate mass transfer as shown below: dn - = k,[n,(P,, T)-A] dt solid film = k , [ L - PSI gas film (9 b) where n is the amount adsorbed at time t for both cases, Ps is the adsorbate pressure at the surface of the adsorbent at time t for the gas film model and k, and k, are lumped-up mass-transfer coefficients.According to eqn ( 5 ) A is given by (n--nm) = b(T- T,) solid film (104 ( ~ - n ~ ) = a(P,-P,)+b(T-T,) gas film. (lob) Both the Fickian diffusion and the linear driving force models for mass transfer have been successfully used to describe the experimental uptake data for various systems. These analyses show that the initial uptake is primarily mass-transfer controlled and the latter part of the uptake is heat-transfer controlled. A key assumption in the above models is that the temperature ( T ) is uniform within the adsorbent mass at any time t .Experiments, on the other hand, have shown that there can be a significant thermal gradient within the adsorbent mass, as shown in fig. 1 .Q9 lo, 1 4 9 l6 The temperatures at the centre and the outside surface of the adsorbent mass rise together to approximately the same maximum temperature level and then the outside surface of the adsorbent mass cools down faster than the centre. This indicates that the internal thermal resistance of the adsorbent mass may be comparable to the external film heat-transfer resistance, and the assumption of uniform Tmay lead to erroneous analysis of the kinetic data. Models to account for the internal thermal resistance have been p r ~ p o s e d ~ ~ - ~ ~ by assuming that the internal heat transfer can be described by the Fourier law of conduction, which yields the following heat-balance equation for a spherical adsorbent mass of radius R : h (g)r-o = 0, r;) r-R = --[T(R, k€! t ) - T,] where n and T are, respectively, the amount adsorbed and the temperature at radius r within the adsorbent mass at time t , p is the density of the adsorbent mass and k , is its thermal conductivity. Eqn (12) describes the boundary conditions for eqn (1 1) accounting for the temperature symmetry at the centre of the mass and the external film heat-transfer resistance at the outside surface.Numerical solutions of eqn (1)-(4) and (11) have been obtained for an integral adsorption test using the Langmuir equilibrium isotherm by assuming that k , < h and D is independent of T.31 These calculations are, however, tedious and time consuming.S.SIRCAR AND R. KUMAR 2493 SURFACE BARRIER MODEL The nuclear magnetic resonance (n.m.r.) field gradient technique has recently been used to directly measure the adsorbate diffusivity within the zeolite The results show that the diffusivities measured by this method are several orders of magnitude higher than those obtained by the conventional adsorption kinetic tests for many systems. It is also found that the diffusivities measured by the kinetic tests are often inversely proportional to the zeolite crystal 34 These results suggested that the resistance to the mass transfer may be confined within a skin at the crystal surface. Direct evidence of such a surface barrier to mass transfer has also been claimed from n.m.r.and small-angle X-ray diffraction ~ t u d i e s . ~ ~ * ~ ~ In view of these observations, we propose a non-isothermal surface barrier model for adsorption kinetics as follows : (a) the adsorbent mass in the kinetic test is a sphere of radius R (cm) consisting of an assembly of spherical zeolite crystals of radius Re (cm) and density pc (g ~ m - ~ ) . (b) The adsorbate mass transfer is controlled by a barrier at the surface of the crystals and a linear driving force model can be used to describe the transfer process. (c) The heat transfer from the adsorbent mass is governed by the Fourier conduction within the mass in conjunction with the external film heat transfer. There is no temperature gradient across a crystal and the crystal temperature during uptake depends on its position in the adsorbent mass and time.( d ) The kinetics test is a constant-pressure d.a. t. where a step change in the adsorbate gas-phase pressure is made from Po to P, (atm) at t = 0. The initial and the final equilibrium adsorbate loadings are, respectively, no and n , (g-mol g-l) and the corresponding temperature is T,(K). According to this model, the rate of adsorbate uptakej (g-mol g-l s-l) into a crystal located at radius r (0 d Y d R) of the adsorbent mass at time t may be written as where k (g-mol cm-2 s-l atm-l) is the mass-transfer coefficient, the quantity 3/R,pc represents the surface area per gram of crystal and P is the gas pressure within the crystal at time t. An adsorbate mass balance within the crystal at radius r gives: where E, is the crystal void fraction, n (g-mol g-l) is the amount adsorbed in the crystal at time t which is in equilibrium with the adsorbate at pressure P and temperature Tat that time, R is the gas constant and n is related to P and T by eqn ( 5 ) and (7).The heat balance for the adsorbent mass may be described by eqn (1 1) and (12), which can be combined with eqn (5)-(7) and (1 3) and (14) to obtain the following dimensionless partial differential equations (p.d.e.) describing the uptake process :2494 GAS SORPTION ON POROUS ADSORBENTS where the dimensionless groups are as follows: It was assumed that a % E,/R in the derivation. Eqn ( 1 5 ) and (16) are coupled p.d.e. describing the variations of the dimensionless pressure (0) and the temperature (Y') with respect to the dimensionless radius (x) of the adsorbent mass and time (T) during the uptake process.The boundary conditions for these variables are given by eqn (1 7) and the following: I 0(x, T = 0) = 1 Y(x, T = 0) = 0 0 ( X , T = 00) = 0 Y(x,z = co) = 0. The parameters of the model are K, /?, A and S. /? is a dimensionless equilibrium property of the adsorbate-adsorbent system and can be obtained from the adsorption thermodynamics as :24 A (modified Lewis number) and S (Biot number) are dimensionless transport properties which are proportional to the ratio of the effective thermal conductivity of the adsorbent mass to the adsorbate mass-transfer coefficient and the ratio of the external to the internal heat-transfer coefficients of the adsorbent mass, respectively.The overall fractional adsorbate uptake, F [ = (fi-nn,)/(na -no)], of the adsorbent mass at time t may be written analogous to eqn ( 3 ) by combined eqn (5) and (6) as: F(T) = 3 [ - @ + Q y r ] ~ ~ d ~ . s,' Solutions for the 0 and Y functions from eqn (1 5 ) and (16) may be substituted into eqn (20) to calculate F. The simultaneous solution of these equations is, however, complicated and may require a numerical technique. We simplified the problem by assuming that the adsorbent mass can be considered to be adiabatic in the initial part of the uptake. This assumption is valid when k, is low and kis large so that A is small and the second terms on the right-hand side of eqn (15) and (16) are negligible compared with the first terms at small t .In other words, the rate of heat generation within the adsorbent mass in the early part of the uptake is much faster than the heat loss from it to the surroundings. This behaviour is experimentally supported as mentioned earlier by the temperature profiles within the adsorbent mass. The temperature gradient within the adsorbent during the initial portion of the uptake process is very small until Y reaches its maximum value and then the outside surface of the adsorbent cools down faster than the centre, as shown in fig. 1 . Thus, for the initial portion of the uptake process 0 and Y are practically independent of x and eqn ( 1 5) and (1 6) may be approximated to : dY (%) z 0.S. SIRCAR AND R. KUMAR 2495 Eqn (21) and (22) can be integrated to obtain 0 and Y profiles for the initial part of uptake as: 0 = exp[-(1 -&I (23) 1 -exp[-(1 -p)z] 1 -P Y Z Eqn (23) shows that 0 decays exponentially with time and for relatively fast adsorption (large E) 0 approaches zero during the adiabatic uptake, as shown in fig.1. In other words, local mass equilibrium is approximately attained during this period and further adsorbate uptake is controlled by the shift of the equilibrium isotherm due to the adsorbent cooling. Eqn (23) can then be substituted into eqn (16) and integrated using the boundary conditions of eqn (1 7) and (1 8) by Laplace transform- ation and inversion by the method of residue to obtain the following overall temperature profile in the adsorbent mass during the uptake process: where p , are the roots of the equation pncotpn= 1-S n = 1,2, .. . . (26) Eqn (23) and (25) may then be substituted into eqn (20) to obtain x {exp ( - 4G. 4 - exp - (1 - P ) TI> (27) which represents the overall uptake as a function of t by the proposed model. ESTIMATION OF THE MODEL PARAMETERS According to the proposed model, the initial adsorbate uptake during the adiabatic portion of the process may be obtained by combining eqn (20), (23) and (24) as l-exp[-(l-P)z] 1 -D F(z) = which may be rewritten in t domain as In[l -(I -/?)F(t)] = (1 -fi)&t. (29) Eqn (29) shows that a plot of In [ 1 - (1 -p) F(t)] against t should yield a straight line with a slope equal to (1 --p)&. Since p can be independently evaluated from the equilibrium adsorption isotherms at the conditions of the d.a.t.using eqn (19), &can be estimated from the initial uptake data by the eqn (29). The parameters A and S of the model can then be estimated by curve fitting the entire uptake and the adsorbent temperature profiles using eqn (25)-(27). In the absence of the experimental Y profiles, which may be difficult to measure in a d.a.t., A and S can be evaluated from the uptake curve alone as follows. Eqn (27) can be approximated at large z as2496 GAS SORPTION ON POROUS ADSORBENTS where p1 is the first root of eqn (26). Eqn (30) shows that a plot of In [ 1 - F(t)] against t should yield a straight line with a slope equal to -Apf kand an intercept at t = 0 equal to Thus, A and S and subsequently h and k, can be estimated from the slope and the intercept of this line and known values of p and k.Such linear behviour has been experimentally observed for many systems. 27 EXPERIMENTAL VERIFICATION OF THE MODEL We used published uptake data for the sorption of (a) iso-octane on 13X,38 (b) n-pentane on 5 A27 and (c) carbon dioxide on 4 A to test the proposed model. These data were measured using an assembly of zeolite crystals under the conditions of the d.a.t. described in table 1. The equilibrium adsorption isotherms for these systems were available28$ 3 7 9 38 for the estimation of the parameter p. A c p of 0.22 cal g-l K-l was used for all the zeolites. A summary of the results is given below. ISO-OCTANE ON 13x The uptake data for this system were measured at 403 K using four different differential pressure levels. The same crystal size (30pm) and the adsorbent mass (12 mg) were used in these tests.Table 1 gives the test condition and the corresponding p values. The isosteric heat of adsorption for the system was calculated to be 18.5 x lo3 cal g-mol-l. Fig. 2 shows the plots of the initial uptake data for this system according to eqn (29). The linear plots verify the assumption of the adiabaticity of the adsorbent mass during the initial uptake. Values of the parameter k for this system were calculated at different pressure levels from the slopes of these linear plots and they are given in table 1. kmay be written as Eqn (31) shows that R should be proportional to Po/p and inversely proportional to R, when the other variables of that equation are held constant during the uptake experiment.Eqn (31) also shows that k should be independent of the size of the adsorbent mass (R). Fig. (3) shows a plot of k against Po/p for the iso-octane-13X system. A linear relationship is observed as demanded by the model. Fig. 4 shows the best fit of the entire uptake data for this sytem by eqn (27), which indicates that the model describes these data very well. The values of the transport properties are summarised in table 1. The thermal conductivity of the adsorbent mass (k,) was found to be practically independent of the adsorption pressure but the external film heat-transfer coefficient (h) decreased as P, was lowered. This behaviour is consistent because h is proportional to the gas-phase thermal conductivity ( k i ) which decreases with decreasing pressure at the very low pressure levels of these Fig.5 shows the temperature profiles at the centre of the adsorbent mass (x = 0) for this system as calculated by eqn (25). The initial rate of the temperature change of the adsorbent mass can be shown from eqn (24) to be (32) dn, -no) k CPTable 1. Test conditions and model parameters sample crystal weight diameter system /ms Pm iso-octane- 12 13X, 403 K 12 12 12 n-pentane- 13 43 5 A, 523 K 21 carbon dioxide- 4 A , 371 K 39.0 39.0 39.0 39.0 3.6 3.6 3.6 7.3 21.5 34.0 P, /Torr 0.008 0.020 0.033 0.064 21.0 21.0 21.0 75.0 75.0 75.0 pa3 /Torr 0.020 0.033 0.064 0.1 13 27.0 27.0 27.0 85.0 85.0 . 85.0 n0 12, /mmol g-l /mmol ggl p 0.390 0.600 -1.090 0.600 0.700 - 1.000 0.700 0.820 -0.760 0.820 0.900 -0.480 0.300 0.350 -0.290 0.300 0.350 -0.290 0.300 0.350 -0.290 0.145 0.155 -1.140 0.145 0.155 -1.140 0.145 0.155 -1.140 0.024 0.036 0.103 0.270 0.340 0.340 0.340 0.470 0.160 0.100 k,/ 10-5 10-4 cal cm-' s-1 K-1 cal cm-2 s-' K-' estimated theory estimated theory E 2.3 2.6 1.5 - $ zi 13.0 - 13.0 13.0 ?J 13.0 12.0 14.0 12.0 5 F 2.1 2.5 1.2 - 0 2.4 2.8 2.6 3.8 3.0 4.6 6.4 0 - - 15.0 12.0 12.0 6.5 - 62.0 - 56 4.6 7.0 59.0 5.7 4.0 - 16.0 -2498 1 .o 0.80 0.6 0 0.40 4 0.20 G - W I - 0.10 0.08 0.06 0.04 GAS SORPTION ON POROUS ADSORBENTS I 1 1 1 1 0 10 20 30 40 50 60 t l s Fig.2. Initial uptake of iso-octane on 13X zeolite. x , P = 0.0083-0.020 Torr, /? = - 1.09 and k = 0.024 s-l; a, P = 0.020-0.033 Torr, /? = - 1.00 and k= 0.036 s-l; 0, P = 0.033-0.064 Torr, /? = -0.76 and k= 0.103 s-l.1 .o 0.80 0.60 0.40 0.2 0 I; 0.10 0.08 0.06 0.04 0.0 2 0.01 I I I t t I I I l l 1 I I I 1 1 1 1 0 -01 0.10 1 .o Po lP Fig. 3. Plot of k against Po//? for the iso-octane-13X system.S. SIRCAR AND R. KUMAR 6 - 2499 120 160 200 240 280 0 40 80 t l s Fig. 4. Uptake curves for adsorption of iso-octane on 13X zeolite. Effect of adsorbate pressure. T = 403 K, sample weight = 12 mg and crystal diameter = 39 pm. P = x , 0.0083-0.020; 0, 0.020.033; 0, 0.0334.064; a, 0.064-0.1 125 Torr. v 6 c F - 4 --- I 2 1 0 50 100 150 200 2 50 t l s Fig. 5. Temperature profiles at the centre of the adsorbent mass during uptake of iso-octane on 13X zeolite. F = (a) 0.0141, (b) 0.0265, (c) 0.0485 and ( d ) 0.0882 Torr.2500 GAS SORPTION ON POROUS ADSORBENTS 0 40 80 120 160 200 240 280 tls Fig. 6.ERects of various heat-transfer resistances on uptake of iso-octane on 13X. IF = 0.103 s-l, h = 2.62 x cal cm2 s--l K-' and k, = 2.44 x cal cm-l s-l K-l; (a) present model, (b) k, = rxj and (c) isothermal model. The plots in fig. 5 follow the order set by eqn (32). The cooling of the adsorbent is controlled by the magnitudes of k , and h. For this system, k, being about the same for all the runs, the lower the value of h (or lower F) the longer the system takes to cool down. The relative effects of the various heat-transfer resistances for this system are shown in fig. 6. The graph compares the uptake curve of one of the present cases with the corresponding isothermal uptake and the uptake for the case where the external film is controlling the heat transfer ( k , -+ GO) assuming that Eand h are the same as the present case."The times taken to reach 95% of saturation uptake are, respectively, 20, I45 and 192 s for the three cases.S. SIRCAR AND R. KUMAR 250 1 n-PENTANE ON 5 A Adsorption kinetics for this system were measured at 523 K with the same step change in the pressure on the same adsorbent crystals (R, = 3.6 prn) but using three different adsorbent mass sizes (13, 21 and 43 mg). The isosteric heat of adsorption for the system was 14 x lo3 cal g-mol-l. Fig. 7 shows the plot of the initial uptake data according to eqn (29). The data for all three cases fall on the the same straight line, indicating that k is independent of R, as shown by eqn (3 1). The values for this system are given in table 1.and Fig. 7. Initial uptake of n-pentane OR 5 8, zeolite. 7 = 523 K, P = 21-27 Torr, E = 0.339 s-l. Sample weight: 0, 13: x , 21 and A, 43 mg. = -0.295 and Fig. 8 shows the fit of the entire uptake curves for this system by the model. Again, the fit is very satisfactory. The estimated k , values were found to be independent of the adsorbent mass size as expected. Fig. 9 shows the Y profiles at the centre of the adsorbent mass for this system calculated by eqn (25). The initial temperature rises for the three cases are identical because they depend only on 8 and k. The larger adsorbent mass, however, takes a longer time to cool because of the larger quantity of heat to be removed.2502 I .o 0.00 0.6 0 0.4 0 0.2 0 7 .3 0.1 0 0.00 0.06 0.04 0.02 0.0 1 GAS SORPTION ON POROUS ADSORBENTS 1 t 1 1 1 1 0 10 20 30 4C 50 60 70 t i s Fig.8. Uptake curves for adsorption of n-pentane on 5 A zeolite. Effect of adsorbent mass. T= 523 K, Sample weight: 0, 13; x , 21 and A, 43 mg. 1.00 i 0 10 20 30 40 t l s Fig. 9. Temperature profile at the centre of the adsorbent mass during uptake of n-pentane on 5 A zeolite (a) 43, (b) 21 and (c) 13 mg.S. SIRCAR AND R. KUMAR 2503 CARBON DIOXIDE ON 4 A The experimental data for this system were measured at 371 K with the same step change in the adsorbate pressure but using different crystal sizes (7.3, 21.5 and 34.0pm). Fig. 10 shows that the model describes the uptake data very well. The estimated k values for this case reported in table 1 were found to be inversely proportional to R, as demanded by eqn (31).The k, and h values were found to be practically independent of R, except for the case of the largest crystal size where a lower value of h was estimated. 1 .o 0.80 0.60 0.40 0.20 7 - 0.10 0.08 0.06 0.04 0.02 0.01 0 20 40 60 80 100 120 140 t / s Fig. 10. Uptake curves for adsorption of carbon dioxide on 4 %, zeolite. Effect of crystal size. T = 371 K, P = 75-85 Torr and /3 = - 1.14. x , crystal diameter = 7.3 pm and k= 0.47 s-l; 0, crystal diameter = 21.5 pm, k= 0.16 s-l; A, crystal diameter = 34 pm and E = 0.10 s-l. The above results demonstrate that the proposed surface barrier model for non-isothermal adsorption kinetics can successfully describe the uptake of various adsorbates on different zeolites measured by the d.a.t.using a consistent set of parameters. The model incorporates various basic characteristics of the adsorption process experimentally observed in recent years.2504 GAS SORPTION ON POROUS ADSORBENTS Note that the uptake data for the above systems can also be described fairly well by several of the previously discussed non-isothermal kinetic models which ignore the internal thermal resistance of the adsorbent mass and use different mass-transfer mechanisms, such as (a) the Fickian diffusion of adsorbed molecules, (b) the Fickian diffusion of adsorbate through interparticle void space of the adsorbent mass or (c) the solid-phase linear driving force model. For example, all three models have been used to describe the data for the iso-octane-13X 2 4 9 28 while models (a) and (c) were used for the n-pentane-5 A data.24y27 This suggests that there can be significant interplay between various mass and heat-transfer resistances and more direct experimental measurements such as the n.m.r., X-ray scattering and the temperature gradient within the adsorbent mass are needed to establish the true mechanism of the adsorption kinetics.THEORETICAL CALCULATION OF k, AND h No experimental data on k, and h could be found in the literature for the systems described above. However, several methods are available for the calculation of k, for an assemblage of microparticles. We used the method of Masamune and Smith,40 which gives B(Y + 1) (34) where a is the thermal accommodation coefficient, f is the void fraction of the assemblage, 6 is the fraction of solid-solid contact in the assemblage, 0 is the ratio of the effective path length between microparticles to the particle diameter (dp), k: and kg are, respectively, the gas thermal conductivity at the pressure of the system and the gas thermal conductivity within the voids of the assemblage, k, is the solid-phase thermal conductivity, 3, is the absorbate mean free path, Y is the ratio of the gas heat capacities at constant pressure and constant volume and B is a function of the thermal accommodation coefficient.Methods of calculation for a, 6 and 0 are given in ref. (40). The key point is that k, can be significantly lower than k: if 3, - dp (Knudsen regime). This can occur when dp is small, the gas pressure is low and the temperature is high.The Knudsen effect has also been observed at higher pressures where 3, < dP.,l ki is usually insensitive to P except at very low P ( < 1 Torr) where it decreases rapidly with decreasing pressure.39 The net result is that k, can be very low for an assemblage of small particles although the thermal conductivity of the solid may be relatively large. h, on the other hand, depends on the adsorbent sample configuration, type of contact between the adsorbent and its container used in the kinetic test, the shape and the size and the material of the container, exposed surface area of the adsorbent, emissivity of the adsorbent and the container, volume ratio of adsorbent to container, etc. We calculated an order of magnitude of h by adding the conductive and the radiative contributions of heat-transfer coefficients as2' +5.42 x 10-'2 eT3 (35) where R is the radius of the sphere of an equivalent volume of the adsorbent mass and e is the effective emissivity of the system.The mean free paths for the adsorbates under the test conditions were calculated to be 250-1600, 1.51 and 70.71 pm for iso-octane, n-pentane and carbon dioxide,S. SIRCAR AND R. KUMAR 2505 respectively. They were comparable to the zeolite crystal diameters used in the tests, indicating that the Knudsen regime existed for these systems. k, and h values were calculated by the above described methods using f= 0.5, k, = 2.5 x cal cm-l s-l K-1,40 e = 0.8. B = 1 and the literature values for ki and Y. I .o 0.90 0.80 0.70 0.60 F 0.50 0.40 0.30 0.20 0.10 0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 ti IS+ Fig.11. Effect of Eon the shape of the uptake curves for the surface barrier model. S = 2 and p = - 1.0. 0, k = 0.01 and A = 0.50; x , E = 0.10 and A = 0.05; 0, k= 1.0 and A = 0.005; 0, k = 10.0. Table 1 compares the theoretical values of k , and h with those estimated from the analysis of the uptake curve. The agreement is surprisingly good considering the uncertainty in the theories and the values of the parameters cf, B, e, k,) used. The theoretical h for the iso-octane-13X system was calculated by using ki at the atmospheric pressure. Thus it is larger than the h values estimated from the uptake data in the very-low-pressure region where ki is a strong function of pressure. The carbon dioxide4 A system exhibited the largest difference between the theoretical and the estimated h values.2506 GAS SORPTION ON POROUS ADSORBENTS SHAPE OF THE UPTAKE CURVE One interesting aspect of the proposed model is that the uptake curve is convex towards the time axis at the limit of t -+ 0.Mathematically, this can be shown from eqn (28) as (3 = - (1 -P) (6)2. (36) t + o Since P is a negative quantity, (i32F/W)t,o < 0. Fig. 11 shows the plots of the uptake against fl for various values of k using the same values for P and S. It may be seen that the initial curvature of the uptake curve is pronounced for relatively small values of k but it is not very apparent for large 6. The figure also shows that a considerable portion of the uptake curve can be approximately described as being proportional to fl Such initial curvature has been exhibited by various systems where data are available at very small 43 For some other system, the linear plots of Fagainst *appear to intersect the time axis at a non-zero value of t33 similar to the low k cases of fig. 11.For the case wherc the uptake is controlled by heat transfer within the adsorbent mass [k, 4 k, or h, Ap2, -g (1 +), S B 13, the roots of eqn (26) are given by p n = nn and eqn (27)transforks toz5 F(t) = 1 +(-) 6P C, 1 -P ( n 4 2 (37) Eqn (37) has a similar mathematical form as the solution of the isothermal Fickian diffusion model [eqn (l)], which has the mass diffusivity D / R i instead of the thermal diffusivity [k,/R2 (1 -P) pc,] in the exponential terms of eqn (37). Thus eqn (37) also exhibits a linear fi dependence of F at small t.However, in this case the linear plot will intersect the Faxis at 1 /( 1 -P) when t = 0. Such behaviour has also been exhibited by some systems.44 The above analysis show that the non-isothermal surface barrier mechanism can also describe the apparent linear fi dependence of the initial uptake curve, which is often accepted as evidence of the Fickian diffusion mechanism of adsorption. CONCLUSIONS The proposed non-isothermal surface barrier model could successfully describe the uptake of various adsorbates on different zeolites measured by the constant-pressure differential adsorption test. The data could be interpreted using a self-consistent set of parameters. The initial uptakes for these systems were found to be primarily mass-transfer controlled and the adsorbent mass remained approximately adiabatic during that period.The latter part of the uptake was essentially heat-transfer controlled. Both the internal thermal resistance of the adsorbent mass and the external film heat-transfer resistance were important in determining the adsorbent cooling. The former resistance could be significantly low if the conditions for the Knudsen regime existed during the kinetic test. The model could also describe the apparent linear fl dependence of the initial uptake curve. Various mechanistic models may be capable of describing the same uptake data. More detailed experimental investigation of the system, such as n.m.r., X-ray analysis and the temperature gradient within the adsorbent mass during uptake are needed to establish the true adsorption mechanism.We thank Air Products and Chemicals Inc. for their kind permission to publish this work.S. SIRCAR AND R. KUMAR 2507 J. Crank, The Mathematics of Diflusion (Oxford University Press, Oxford, 1970). D. M. Ruthven and K. F. Loughlin, Chem. Eng. Sci., 1971,26, 1145. D. R. Garg and D. M. Ruthven, Chem. Eng. Sci., 1972, 27,417. E. Ruckenstein, A. S. Vaidyanathan and G. R. Youngquist, Chem. Eng. Sci., 1971, 26, 1305. M. Koricik and A. Zikanova, Ind. Eng. Chem., Fundam., 1974, 13, 347. M. M. Dubinin, I. T. Erashko, 0. Kadlec, V. I. Ulin, A. M. Voloshchuk and P. P. Zolotarev, Carbon, 1975, 13, 193. L. K. Lee, H. Yucel and D. M. Ruthven, ACS Symp. Ser., 1977, 40. A. V. Lykov, KolloidZ., 1935, 71, 333; 1936, 74, 179.’ Y. H. Ma and T. Y. Lee, AIChE J., 1976,22, 147. lo E. Wicke, KolloidZ., 1939, 86, 167. l 1 D. M. Anderson and G. Sposito, Nature (London), 1963, 1085. l 2 E. F. Kondis and J. S. Dranoff, AIChE Symp. Ser. no. 117, 1971,67, 25. l 3 J. D. Egan, B. Kind1 and R. B. Anderson, Ado. Chem. Ser, 1971, 102, 164. l4 D. P. Timofeeve and I. T. Erashko, Russ J. Phys. Chem. (Engl. Transl.), 1971, 45, 359. l5 H. J. Doelle and L. Riekart, ACS Symp. Ser., 1977,40, 401. l6 J. Ilavsky, A. Brunovska and V. Hlavacek, Chem. Eng. Sci. 1980, 35, 2475. l7 K. F. Loughlin, R. I. Derrah and D. M. Ruthven, Can. J. Chem. Eng., 1971, 49, 66. Is D. M. Ruthven and R. I. Derrah, Can. J. Chem. Eng., 1972, 50, 743. l9 K. F. Loughlin and D. M. Ruthven, Chem. Eng. Sci, 1972, 27, 1401. 2o K. Chihara, M. Suzuki and K. Kawazoe, Chem. Eng. Sci., 1976,31, 505. 21 S. Sircar, Carbon, 1981, 19, 285. 22 D. 34. Ruthven, Sep. Purif Methoak, 1976, 5, 189. 23 L. K. Lee and D. M. Ruthven, J . Chem. SOC., Faraday Trans. I , 1979, 75, 2406. 24 S. Sircar, J. Chem. SOC., Faraday Trans. I , 1983, 79, 785. 25 S. Sircar and R. Kumar, ACS Symp. Ser., 1983, 223, 171. 26 A. A. Armstrong and V. Stannet, Makromol. Chem., 1966,90, 145. 27 D. M. Ruthven, L. K. Lee and H. Yucel, AIChE J., 1980, 26, 16. 28 D. M. Ruthven and L. K. Lee, AIChE J., 1981, 27, 654. 29 P. P. Zolotarev, Izv. Akad. Nauk SSSR, Ser. Khim., 1970, 1421; 1970, 2831. 30 P. P. Zolotarev and L. V. Radushkevich, Dokl. Akad. Nauk SSSR, 1970, 195, 1361. 31 A. Brunovska, V. Hlavacek, J. Ilavsky and J. Valtyni, Chem. Eng. Sci., 1978, 33, 1385. 32 J. Karger and J. Caro, J. Chem. SOC., Faraday Trans. I , 1977, 73, 1363. 33 M. Bulow, P. Struve, G. Finger, C. Redszus, K. Ehrhardt, W. Schirmer and J. Karger, J . Chem. SOC., 34 J. Karger, J. Car0 and M. Bulow, Izv. Akad. Nauk SSSR, Ser. Khim., 1977, 2666. 35 J. Karger, W. Heink, H. Pfeifer, M. Rauscher and J. Hoffmann, Zeolites, 1982, 2, 276. 36 M. Bulow, P. Struve and S. Pikus, Zeolites, 1982, 2, 267. 37 A. P. Vavlitis, D. M. Ruthven and K. F. Loughlin, J . Colloid Interface Sci., 1981, 84, 526. 38 H. Yucel and D. M. Ruthven, J . Colloid Interface Sci., 1980, 74, 186. 3g S. Dushman, Scient$c Foundation of Vacuum Technique (Wiley, New York, 1962). 40 S. Masamune and J. M. Smith, AIChE J., 1965, 11, 34. 41 R. G. Deissler and J. S. Boegll, Trans. ASME, 1958, 1417. 42 L. K. Lee, Ph.D. Thesis (University of New Brunswick, Canada, 1977). 43 A. P. Vavlitis, M S c . Thesis (University of New Brunswick, Canada, 1978). 44 H. Yucel, Ph.D. Thesis (University of New Brunswick, Canada, 1978). Faraday Trans. I , 1980, 76, 597. (PAPER 3/2157)

ISSN:0300-9599    

DOI:10.1039/F19848002489

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1984, 80, 2509-2524 The Triplet State of N-(n-Butyl)-5-nitro-2-furamide by Laser Flash Photolysis Spectrum, Lifetime, Energy and Electron-transfer Reactions BY LUIS J. A. MARTINS* Instituto Superior de Engenharia de Coimbra, 3000 Coimbra, Portugal AND TERENCE J. KEMP Department of Chemistry and Molecular Sciences, University of Warwick, Coventry CV4 7AL Received 19th December, 1983 Upon nanosecond laser flash photolysis (347.1 nm) of N-(n-butyl)-5-nitro-2-furamide (BNFA) in solution (at 298 K), a short-lived transient absorption has been observed with A,,, = 500 f. 5 nm, which shows a small bathochromic shift on going to hydroxylic solvents (irrespective of polarity) and which is attributed to its lowest excited triplet state (3BNFA). A long-lived species, formed in high yield in alkanols, is assigned to the radical anion (BNFA'-) and has A,,, d 360 nm with a weaker band peaking CQ.510 5 nm in propan-2-01. The lifetime of 3BNFA depends on both the solvent and the ground-state concentration of BNFA, having its smallest value in water (ca. 19 ns). Sensitization and quenching experiments enabled the triplet energy, ET, to be estimated as 238 & 4 kJ mo1-I; this value, together with the ground-state reduction potential, E:, yields E;(3BNFA/BNFA'-) = 2.23 V us NHE for the one-electron reduction potential of 3BNFA in neutral aqueous solution. 3BNFA is rather oxidizing, thus it abstracts a hydrogen atom from diphenylmethanol with k, = (3.01 & 0.12) x lo7 dm3 mol-l s-l in acetone solution, and the second-order rate constants for the one-electron oxidation of a number of substrates by 3BNFA have been measured in 1 : 4 (v/v) water + acetonitrile mixtures and range from ca.2 x lo8 for CE-13CO; to ca. 6 x lo9 dm3 mol-1 s-l for I-. Evidence for complete electron transfer arises, in several cases, from the observation of the semi-oxidized substrate. The correlation of the logarithm of k, with the standard free-energy change, obtained from the estimated thermodynamic potentials of the reactants, are discussed in terms of the Rehm-Weller and Marcus free-energy relationships. Absolute rate constants for the one-electron reduction of the lowest triplet states of duroquinone, 2-nitrothiophen and 5-nitro-2-fyroic acid are also included in this correlation. Pulse photolysis of a solution of 2-nitropyrrole showed no absorbing transient in the spectral region 390-700 nm.The implications of some of these findings for nucleophilic photosubstitution reactions of 5-membered nitro- heterocyclic compounds are discussed. Considerable interest has been devoted to nitrofurans in connection with both their redox properties, which make them useful electron-affinic radiosensitizers,'? and antibacterial activity, which is related to the displacement of the nitro group by a n~cleophile.~ The latter type of reaction can be induced by light, and it was shown3 that upon irradiation of N-(n-butyl)-5-nitro-2-furamide (BNFA) in methanol the nitro group is replaced by methoxy, while photolysis of BNFA in water leads to the destruction of the furan ring.No experimental data aimed at elucidation of the multiplicity of the reacting excited state have been reported. However, sensitization and quenching studies in the photoreaction of 2-nitrofuran with CN- in water led to 25092510 TRIPLET STATE OF N-(n-BUTYL)-5-NITRO-2-FURAMIDE the suggestion that the nucleophilic photosubstitution proceeds though the triplet ~ t a t e . ~ Meanwhile, the lowest excited triplet state of 5-nitro-2-furoic acid was shown5 to interact with nucleophiles via one-electron transfer to the triplet state. Recently it was proposed677 that the precursor of the substitution product is the radical cation formed in the oxidation of the triplet state by a ground-state molecule (via the triplet excimer). Evidence in favour of a radical mechanism in the photosubstitution reaction of 4-methylquinoline-2-carbonitrile in ethanol has also been presented whereby the solvent free radical( 1-methyl-1 -hydroxymethyl) is supposedly formed, in part, by hydrogen abstraction by the triplet state of the heterocyclic compound.* It therefore appears opportune to characterize the triplet states of such compounds, with special reference to their reactivity towards added solutes, with the aim of gaining some insight into the mechanism of the photoreactions in which they might engage.The object of the present study is to provide both spectroscopic and kinetic data concerning the lowest excited state of the title compound. EXPERIMENTAL MATERIALS BNFA was synthesized from 5-ni tro-2-furoic acid following a modificationg of an earlier procedure.1° The final product was recrystallized twice from mixtures of ethanol and water, yielding white crystals, as needles, with melting point 89-90 "C (uncorrected) (lit.89-90 "C,l0 92.5-93.5 "C9). The U.V. spectI1 m gave Amax = 308 nm (E = 13 212 dm3 mol-l cm-l) in spectro- scopic acetonitrile; the i.r. spectrum showed v(amide) = 3245, 1650 cm-l, and v(-NO,) = 1357, 1518 cm-l (measured in a KBr disc) as expected." The lH n.m.r. spectrum gave the following chemical shifts, S(ppm): 6.75(1 H, br s, -NH-), 0.95(3H,t,-CH,-CH,), 1.41(2 H, sextet, -CH,-CH,-CH,), 1.60(2 H, quintet, -CH,-CH,-CH,-), 3.46(2 H, quartet, -NH-CH,-CH,,, 7.36(1 H, d, 4-H on ring), 7.23(1 H, d, 3-H on ring), J34 = 3.75k0.1 Hz, as expected.', Elemental analysis gave: calculated (%) C, 50.94; H, 5.70; N, 13.20; found (%) C, 50.91; H, 5.70; N, 13.26.2-Nitropyrrole was prepared by nitration of pyrrole (freshly distilled) following a previously published method.', Both lH n.m.r. and U.V. spectra gave the same results as those reported before for 2-nitropyrr01e.l~ Doubly distilled water, acetonitrile, chloroform and acetone (all of A.R. grade) were subjected to further purification according to previously published methods.15 All the other chemicals were of the highest purity commercially available and used as supplied. METHODS IH n.m.r. spectra were run at 200 MHz in a Perkin-Elmer R34 spectrometer using CDC1, as the solvent and tetramethylsilane as reference. All solutions were prepared immediately before pulse irradiation and exposed to the analysing light for the minimum possible time.De-aeration was accomplished by bubbling the solution with high-purity argon (B.O.C., 99.999%) for 20 min. Pulses of ca. 100 mJ (347.1 nm, ca. 50 ns duration) were delivered from a frequency- doubled Q-switched ruby laser (Applied Photophysics, London). The monitoring system consisted of a pulsed xenon arc lamp (250 W), a high-radiance monochromator and a Hamamatsu R666S photomultiplier tube, as described before. l6 Typical concentrations of BNFA were in the range (0.5-1.0) x mol dm-3. All experiments were performed at 298 & 1 K, except where otherwise indicated. First- and second-order rate constants were obtained by a least-squares method.L. J. A. MARTINS AND T. J. KEMP 251 1 RESULTS AND DISCUSSION TRANSIENT SPECTRA, TRIPLET ENERGY AND LIFETIME When a de-aerated solution of BNFA is subjected to laser flash photolysis, a short-lived transient is observed with an absorption centred at A,,, = 500 & 5 nm, as shown in fig.1. Amax shows a small (ca. 15 nm) bathochromic shift on going from acetonitrile or CCl, to hydroxylic solvents [table 1 (a)]. On the other hand, A,,, does not appear to be affected by changing the solvent polarity. Following laser flash photolysis of BNFA in alkanols (methanol, ethanol, propan- 1-01 and propan-2-01) the short-lived transient decays to a residual absorption due to a longer-lived species, as presented in fig. 2. A dependence of the lifetime of the short-lived transient on both the solvent (tables 1 and 2) and ground-state concentration of BNFA (see below) is observed.The short-lived transient is assigned to the lowest excited triplet state of BNFA, hereafter denoted 3BNFA, in view of its efficient quenching by, e.g., O,, ferrocene and azulene, which are well established triplet-energy quenchers. In addition, both the observed broad absorptions and lifetimes (measured in several solvents) are similar to those of the triplet state of other nitroaromatic molecules studied in solution by pulsed-irradiation methods. 17-,0 The longer-lived species, formed in high yield in alkanols, has a strong absorption below 390 nm and a secondary weaker band centred around 5 10 5 nm, and it decays slowly over tens of milliseconds. It is attributed to the radical anion of BNFA formed as a result of the one-electron oxidation of the 1-hydroxyalkyl radical, this being in turn generated by H-atom abstraction from the alkanol by ,BNFA, viz. ,BNFA + R,R,CHOH + BNFAH' + RIR,cOH (1) (2) R,R,cOH + BNFA + RIR,CO + H+ + BNFA'-. This assignment is based on the fair agreement with the absorption spectrum of BNFA'- determined21 in buffered neutral aqueous solution by pulse radiolysis, which exhibits a strong absorption with A,,, = 360 nm (E,,, z 1.2 x lo4 dm3 mol-l cm-l) and a weaker band at ca.500 nm ( E , , ~ ;t: 530 dm3 mol-1 cm-l). It appears that the former band is shifted to shorter wavelength in propan-2-01, while the weaker band shows a bathochromic shift (ca. 10 nm) in this solvent as compared with that in water. This trend was observed in the spectrum of the radical anion of 5-nitro-2-furoic acid (which resembles that of BNFA'-) in the two solvents mentioned, as found in a pulse-radiolysis study.22 Now there is good general agreement between the spectra of BNFA'- in water and propan-2-01, but if we assume that the extinction coefficients at A,,, are approximately the same in both solvents then the absorbance at ca.500 nm in propan-2-01, which is twice that observed in water, might be indicative of the presence of another species absorbing in the same region. This other species could be related with subsequent reaction of BNFA'- and propan-2-01 (or a radical derived from it), and such a possibility is being investigated further. As no emission could be detected from BNFA either in solution or in EPA glass at 77 K, a set of quenching and sensitization experiments was performed in order to determine the triplet energy of this molecule.The results obtained from such a study are collected in table 3 together with literat~re,~-~, values for the triplet energy of the different acceptors and donors used. The energy, ET, of the lowest triplet state of BNFA lies between those of 1 -naphthaldehyde and of chrysene; because the quenching of the chrysene triplet state occurs at an almost diffusion-controlled rate ( k , z 5 x lo9 dm3 mol-1 s-l), while no quenching of the triplet state of 1 -naphthaldehyde by BNFA was observed, we may conclude33 that, in either case,2512 TRIPLET STATE OF N-(n-BUTYL)-5-NITRO-2-FURAMIDE Table 1. Triplet-state absorption maximum (Amax) and lifetime (t) of ,BNFA in different solvents ~ ~ [BNFA] solvent Amax/nm t/ns E~ at 293 K /lo-, mol dm-3 CD,CN CH,CN CH,CN CHCl,b CDCl,b CCl, CD,OD CH,OH (CH,),CHOH (CH3)2C0 D2O H2O 332 498 f 5 260 498 f 5 230 220 186 164 500 & 5 52 38 515f5 22 505 _+ 5 37 26 515+ 10 19 - - - - - - 37.5 37.5 37.5 20.7 4.8 1 4.81 2.24 32.70 32.70 19.92 80.20 80.30 0.60 0.6 1 0.91 1 .o 0.90 0.90 0.95 0.95 0.95 0.70 1.10 1.10 a E is the dielectric constant of the solvent from ref.(23). Aerated solution. Table 2. Lifetimea of ,BNFA in aqueous MeCN as a function of the water content H,O (mole fraction) t/ns 0.0 0.108 0.158 0.203 0.245 0.323 0.423 0.612 0.815 1 .o 26 1 303 339 28 1 247 181 124 60 31 19 a For [BNFA] = 3.5 x lop4 mol dm-3. the equilibrium between donor-triplet and acceptor-triplet does not involve comparable concentrations of these states.We are thus led to conclude that the triplet energy of BNFA lies approximately halfway between those of 1 -naphthaldehyde and chrysene, i.e. a value of ET = 238 2 kJ mol-I is obtained, although the estimated error should reflect those of the 'gating' molecules, i.e. be taken as ca. +4 kJ rno1-l. This value is, within the accuracy of the procedure, the same as that found for the lowest excited triplet state of 5-nitro-2-furoic acid.5* 34 The one-electron reduction potential of BNFA has been measured2' by pulse radiolysis in neutral aqueous solution (buffered) and is E: (BNFA/BNFA-) = -0.230 k0.005 V us the normal hydrogen electrode (NHE), which together with ET = 238 +4 kJ mol-1 yields (3BNFA/BNFA'-) x 2.23 V US NHE (see below).This high value for the one-electron reduction potential of 3BNFA accounts for its rather strongly oxidizing ability found both towards hydrogen-atomL. J. A. MARTINS AND T. J. KEMP 2513 0.30 0.20 P) s -e 4 % 0.10 400 450 500 550 600 650 700 h/nm Fig. 1. End-of-pulse spectrum of deaerated solutions of BNFA: ., 1.41 x acetonitrile; 0, 9.47 x lo-* mol dm-3 in carbon tetrachloride; V, 1.03 x mol dmP3 in mol dm-3 in water. ,-., ' \ 0.30 0.20 0 5 e 4 -2 0.10 350 400 450 500 550 600 650 700 X/nm Fig. 2. Spectrum of the transient obtained upon laser flash photolysis of a deaerated solution of BNFA (1.05 x mol dmb3) in propan-2-01, as measured immediately (.), 300 ns (V) and 4 ps (0) after the end of the laser pulse. The dashed line is the spectrum of the radical anion BNFA'- determined in neutral aqueous solution by pulse radiolysis [ref.(21)]. and one-electron donors. The second-order rate constants for reaction with methanol and propan-2-01 are ca. 6 x los dm3 mol-1 s-l, and for the former a deuterium kinetic-isotope effect k,/kD = 1.7 was measured, both facts suggesting an (n, n*) nature for 3BNFA. This is further supported by the fact that 3BNFA abstracts a 82 FAR 12514 TRIPLET STATE OF N-(n-BUTYL)-5-NITRO-2-FURAMIDE Table 3. Energy-transfer data for 3BNFA in solutiona at 298 K (a) Quenching of 3BNFA by acceptors k 2 / 1 O9 dm3 acceptor mol-l s-l E$/kJ mol-I oxygenb 0.83c 94.1 4d ferrocene 11.8k0.4 107.66e azulene 10.5k0.5 125.52f 1,4-benzoquinone 5.24 f 0.10 209.209 (6) Quenching by BNFA of triplet-state donors k,/ lo9 dm3 monitoring donor mol-l s-l E;/kJ mol-l wavelength/nm chryseneb 4.72 k 0.22 239.74h 600 p - terp hen y lb 3.46 k 0.10 244.39 450 1 -phenylnaphthalenem 5.60 k 0.0 1 246.02i 478 1 -bromonaphthalenem 5.3 1 0.03 247.69 418 2-br~monaphthalene~ 6.96 k 0.10 251.88i 414 nap h thalenem 8.49 & 0.78 254.801 412 ( c ) BNFA does not quench the triplet state of the following donors monitoring donor wavelength/nm E?/kJ mol-1 1 -nitronaphthalene 600 229.7Oi 1 -naphthaldehyde 500 235.56k ( d ) 3BNFA is not quenched by accept or E$/kJ mol-I nitro benzene > 251.2013 a Measured in MeCN, except where otherwise indicated. Measured in Me,CO. Corrected for the lifetime in this solvent and using [O,] = 2.4 x 1 0-3 mol dm-3 in aerated acetone at 298 K [from ref.(23)) Value from ref. (27). Value from ref. (28). Values from ref. (29). Values from ref. (30). Value from ref. (31). Value from ref. (32). Sensitized by benzophenone. Value from ref. (24). Value from ref. (25). f Value from ref. (26). hydrogen atom from diphenylmethanol (Ph,CHOH) with a rate constant k , = (3.01 kO.12) x lo7 dm3 mol-l s-l in acetone solution; when 3BNFA is fully converted its absorption is replaced by one that is identical with that for Ph,cOH (see fig. 3). The lifetime of 3BNFA in water shows a rather small isotope effect, k,/k, = 1.4, which may be taken as indicative of a hydrogen-atom abstraction with some charge-transfer character. This character may be imparted by the strong association that appears to exist between 3BNFA and water, as suggested by the initial enhancement of the triplet lifetime in water + acetonitrile mixtures as the mole fractionL. J.A. MARTINS AND T. J. KEMP 2515 0.15 r 450 500 550 6 00 h/nm I 650 Fig. 3. Transient spectrum formed on laser flash photolysis of a deaerated solution of BNFA (1.61 f mol dm-3) and diphenylmethanol (3.1 x 10-1 mol dm-3) in acetone, as measured ca. 1 ps after the end of the laser pulse. The full line is the spectrum of Ph,cOH taken from ref. (35). of water increases (table 2). The small red-shift of ;1,,,(3BNFA) observed on changing to hydroxylic solvents may be due to association, through hydrogen bonding, between the (n,n*) triplet state and the hydroxylic solvent molecule. Such an association is expectedl99 36 to reduce the mobility of the substituent groups, mainly the nitro group, and thus explain the initial lifetime enhancement as referred to above. On the other hand, exciplex formation through hydrogen bonding might facilitate reaction of 3BNFA with water, which clearly governs the triplet lifetime in this solvent.Some assisting mechanism is certainly operating in the photoreaction of BNFA with water; this is because the one-electron reduction potential of the relaxed triplet state of BNFA falls rather short of the half-cell potential37 (2.65 V us NHE) of HO' + H+ + e- e H,O. Steady-state photolysis of BNFA in water showed3 destruction of the furan ring, which was attributed to the formation of the stable tautomer of the hydroxy-substituted f ~ r a n . ~ In addition, quantum-yield measurements for the nucleophilic photosubstitu- tion reaction of 2-nitrofuran with CN- in water have shown4 that water competes with CN- as nucleophile; to account for this result it was proposed4y that the triplet state (thought to be the reacting excited state) undergoes ionization with formation of the radical cation and presumably the hydrated electron.4$ However, in the present investigation we found no experimental evidence of either the radical cation of BNFA or the hydrated electron.In fact, the only absorption observed, in the spectral region 365-720 nm, following laser flash photolysis of BNFA in water, is theT,-T, absorption of 3BNFA. Moreover, addition of CN- (ca. 2 x lo-, mol dmP3) to an aqueous solution of BNFA (ca. 8 x lop4 mol dm-3) suppresses the Tl-Tn absorption of 3BNFA and no new absorption [related to the formation of (CN);- or BNFA'-] could be detected from 365 to 720 nm.Conceivably the (CN);- produced in the photoreduction of BNFA could subsequently be involved in the oxidation of BNFA*- and/or of ground-state BNFA. The dependence of the lifetime, z, upon the ground-state concentration of BNFA was determined quantitatively in MeCN yielding the following relation : z-l = (3.71 k0.22) x 1O6+(5.53*0.19) x 108 [BNFA] S-l. Thus 3BNFA reacts with BNFA with a second-order rate constant k, = (5.53f0.19) x lo8 dm3 mol-1 s-l, and we found that for a highly concentrated solution the absorption of a longer-lived species develops in the spectral region 82-22516 TRIPLET STATE OF N-(n-BUTYL)-5-NITRO-2-FURAMIDE 390-600 nm, similar to that of BNFA'- radical anion (the yield of which increases with ground-state concentration).In view of this observation we propose that the self-quenching involves one-electron oxidation of 3BNFA by BNFA, possibly through excimer formation, uiz. 3BNFA + BNFA + (3BNFA * BNFA) + BNFA'+ + BNFA'-. (3) We could not attribute any absorption to the radial cation BNFA'+, which may be either because it absorbs in the same region of BNFA'- or because of a prohibitively small extinction coefficient. Unfortunately the oxidation potential in BNFA in MeCN is not available; however, a rough estimate of this can be made on the basis of the existing one for furan (in CH3C0,H+0.5 mol dm-3 CH3C02Na), E;(ox) = 1.70 V us the saturated calomel electrode (SCE),38 and following previous reasoning34 a value of E;(ox) = 1.91 V us SCE is estimated for 2-nitrofuran in MeCN.Now, considering that the one-electron potential of BNFA is higher than the corresponding one for 1,4-dinitrobenzene (E: = -0.257 V us NHE),39 one would expect that E;(BNFA/BNFA'-) 2 -0.70 V us SCE in MeCN, this being the half- wave reduction potential of 1,4-dinitrobenzene in MeCN.40 If we ignore both the effect of the amide group on the half-wave oxidation potential of BNFA and the entropy change between the So and T, states (of BNFA), we estimate E+(BNFA'+/ 3BNFA) z -0.56 V vs SCE in MeCN. This value, together with E;(BNFA/ BNFA-) 2 -0.70 V us SCE, predicts a rate constant k, 3 2 x los dm3 mol-1 s-l for the self-quenching process, when the Rehm-Weller empirical free-energy relation- ship is used41 [eqn (13) or ref.(41) with AGi3 = 10 kJ mol-,]. Despite the many approximations used, this calculation illustrates the thermodynamic feasibility of such a process. While self-quenching of 3BNFA was observed in other solvents (e.g. acetone) it could not be investigated in water on account of the limited solubility of the compound. It was also found that in water there is no appreciable effect of dissolved molecular oxygen on the lifetime of 3BNFA. This is expected on the basis of the concentration of O,(g) in water ([O,] = 2.7 x lo-* mol dm-3)23 and the rate constant for the reaction of 3BNFA with 02(g) in acetone [table 3(a)]. A decrease of ca. 0.1 ns in the lifetime would be expected under these conditions, which is well within experimental error.KINETIC AND SPECTROSCOPIC ASPECTS OF THE TRIPLET QUENCHING All the second-order rate constants k, for the bimolecular reaction of 3BNFA with added quenchers, Q, were obtained by considering the scheme below, which summarizes accessible channels for the decay of 3BNFA : 3BNFA 3 BNFA (and/or products) (4) ( 5 ) 3BNFA + Q 2 products reaction (4) representing all the first-order processes of deactivation of the triplet state (in a given solvent). According to this scheme, the differential rate expression for the decay of 3BNFA (completely formed at the end of the pulse) is = k, [3BNFA] + k, [Q] [3BNFA]. d[ BNFA] dt Under the experimental conditions prevailing in this work, [Q] % [3BNFA],,,L. J. A. MARTINS AND T.J. KEMP 2517 Table 4. Absolute rate constants for the reaction of 3BNFA with a number of electron donors (D) in solutiona D (numbering) k,/dm3 mol-1 s-l E"(D"/D)/V us NHEb CH,CO, CH,O- CI- CN- OCN- OH- Br- SCN- 1- Ag+ so;- s,o;- N3 (2.59+ 1.72) x los (2.68f0.10) x load (8.04f0.71) x lo8 (1.22 f 0.06) x lo9 (1.23f0.37) x lo9 (1.3250.15) x lo9 (1.55k0.20) x lo9 (1.95k0.10) x lo9 (2.17 f 0.08) x lo9 (2.77f0.16) x lo9 (5.50+0.29) x logd (6.25 f 0.14) x logd (2.00 f 0.27) x 109 2.32 f 0.02c 2.25e 2.20 * 0.02c 2.20e 2.15 k 0.02' 2.15 & 0.02c 2.18 k 0.02c 2.11 0.09e 1 .34e 1 .904e 1.37 0.22f 1.601e 1.33 k 0.038 a Measured in a 1 : 4 (v/v) water + acetonitrile solution, except where otherwise indicated. Measured in One-electron reduction potential in neutral aqueous solution.See text. acetonitrile solution. Values from ref. (46). f Value from ref. (69). Value from ref. (52). ([3BNFA]t,o being the initial concentration of 3BNFA), the quenching reaction ( 5 ) is pseudo-first-order in "NFA and the integrated form of eqn (6) is obtained as - In D, = (k, + k,[Q]) t + constant (7) where D, is the absorbance of 3BNFA at time t. Reactions obeyed pseudo-first-order kinetics over a range of quencher concentrations. The relatively high reduction potential of the triplet state, Ei(3BNFA/ BNFA'-) z 2.23 V us NHE, makes possible the oxidation of a number of substrates. In view of the current interest4* 7 9 42 in the nucleophilic substitution reactions of nitroaromatic molecules and because the intermediacy of the triplet state has been proposed4 in this type of reactions, we have undertaken the study of the interaction of 3BNFA with a number of widely used nucleophiles (table 4).In the cases of I- and SCN-, as quenchers, and when full conversion of 3BNFA is achieved, the absorption of the latter is replaced by those assigned to 1;- and (SCN);-, respectively; this assignment is made on the basis of the similarity with the previously p ~ b l i s h e d ~ ~ - ~ ~ spectra of these species and the known reactions of I' and SCN' following oxidation of the corresponding anion. Their spectra are shown in fig. 4. Only a small contribution from BNFA'- to these absorptions is expected at > 390 nm, since its extinction coefficient in this region is much smaller (8480 z 530 dm3 mol-1 cm-1)21 than those for both 1;- (8400 x 1.3 x lo4 dm3 mol-l cm-1)44 and (SCN),- (8500 x 7.1 x lo3 dm3 mol-1 ~ m - l ) .~ ~ Addition of KOCN (2.4 x lo-, mol dm-3) to a deaerated solution of BNFA (8.0 x mol dm-3) in methanol caused the T-T, absorption of 3BNFA to be replaced by that of BNFA'-, the yield of which is increased relatively to that found in the absence of quencher (OCN-), as presented in fig.5. Such spectroscopic data indicate that the interaction of 3BNFA with the nucleophiles examined involves complete electron transfer. In addition, the apparent correlation of log k, with the equilibrium oxidation potential of the electron donor is consistent2518 TRIPLET STATE OF N-(n-BUTYL)-5-NITRO-2-FURAMIDE 0.20 0.15 0.10 0.0 5 X/nm Fig. 4. Spectrum of the transient obtained upon laser flash photolysis of deaerated solutions of 0, BNFA (6.92 x mol dm-7 and KSCN ( I .7 x 10-l mol dm-3) in MeCN, as measured ca.200 ns after the end of the pulse; and 0, BNFA (6.78 x lop4 mol dm-3) and KI (1.22 x mol dm-3) in MeCN, as measured ca. 1 ps after the end of the laser pulse. L a, G z 0.10 - 'I 2 -2 0.00 1 350 400 450 500 550 600 650 700 X/nm Fig. 5. Spectrum of the transient observed following laser flash photolysis of deaerated solutions of: BNFA (7.99 x mol dm-3) in MeOH as measured immediately (0) and 20 p s (0) after the end of the pulse; and BNFA (7.99 x mol dm-3) in mol drnw3) and KOCN (2.42 x MeOH as measured 20 ps (V) after the end of pulse. with such a mechanism. We have shown5? 46 that the same mechanism operates in the ion-induced quenching of the lowest excited triplet states of both 5-nitro-2-furoic acid5 and 2-nitr0thiophen,~~ a conclusion based on the direct observation of transient species formed by one-electron oxidation of the quencher.Reductive quenching of the triplet state of some aromatic ketones by a number of the nucleophiles used in this work was also (on kinetic grounds) to proceed through one-electron transfer to the triplet state. In particular, the N; anion was shown to parti~ipate~~ in the quenching of the triplet states of some dyes, the anion undergoing one-electron oxidation as demonstrated by the observation of the resulting dye semiquinone radicals. Moreover, spin-trapping experimentsPg provided evidence for the involvement of azide radicals in the quenching of singlet oxygen (lo,) by N;, presumably via one-electron oxidation of N; by lo,.As appears from table 4, there is a general dependence of k, on the oxidation potential of the donor, E"(D'+/D); in order to correlate log k, with AGZ3 (the standard free-energy change for the actual electron-L. J. A. MARTINS AND T . J. KEMP 2519 transfer step) in the framework of current models for electron-transfer reactions, we may adopt the reaction scheme41 where k30 represents all possible models by which the radical ion pair BNFA- - * D'+ can disappear and z, is the triplet lifetime in the absence of quencher. Application of the steady-state approximation to the concentrations of both the reactants and products encounter complexes, and the introduction of the arguments and simplifying assumptions of the model of Rehm and Weller,41 leads to eqn (9) for the experimental second-order quenching constant, k, : k - (9) where KO is the frequency factor [assumed to be the same for all electron-transfer steps considered in scheme (8)] and AGi3 and AG23 are the free energy of activation and the standard free-energy change of the forward electron-transfer step.According to the same model AGH, is assumed to be a monotonic function of AG23 as expressed k12 - 1 + (k21/K0) [exP (AG&/RT) +exP (AGdRT)] AGi3(0) being the intrinsic energy barrier, i.e. the value of AG;, for AG23 = 0, arising from the application of the Franck-Condon principle to electron-transfer reaction^.^, The free-energy relationship, eqn (lo), together with eqn (1 2) predicts that for a series of homogeneous electron-transfer reactions [in which case k12, k21, KO and AGi,(O) are constant] a plot of log k2 against AG23 approaches a plateau value (the diffusion- controlled limit) for sufficiently exoergonic reactions, and decreases sharply when AG23 becomes positive, with a limiting slope of - 1 /2.303RT for sufficiently endoergonic reactions.Now AG23 can be estimated from the reduction potentials of the triplet state, J!Z;(~BNFA/BNFA'-), and of the donor, E"(D'+/D), by using AG23 = F[E"(D'+/D) - c(3BNFA/BNFA'-)] + wP - W , (1 1) where wp and w, are work terms* and F = 96485 C mol-1 is the Faraday constant. In this equation the reduction potential of 3BNFA is obtained from the relation q(3BNFA/BNFA'-) = C(BNFA/BNFA'-) +ET (12) where ET is the one-electron potential corresponding to the zero-zero spectroscopic energy of the excited triplet state, on the assumption that the difference between the entropies of the excited and ground states may be neglected.It is unfortunate that for the nucleophiles examined in the present work only estimates of the corresponding reduction potentials are available. However, some of these have been subjected to both theoretical and experimental scrutiny. Thus the reduction potential of the couple I*/I-, Eo(I'/I-) = 1.33 f 0.03 V us NHE, from the adjustment of a previous estimate, was subsequently and found to account satisfactorily for an experi- mentally observed linear free-energy relationship (on the basis of Marcus theory). The use of this value together with the relevant equilibrium constants made it possible to * The products' work term wp was calculated approximately using Debye-Hiickel theory51 for a mean separation distance of 7 A, while w,, due to ion-dipole interactions, was neglected.2520 TRIPLET STATE OF N-(n-BUTYL)-5-NITRO-2-FURAMIDE 7-01 6.0 - - I I I I - 200 -1 50 -1 00 - 50 0 + 50 AG23 /kJ mol-' Fig.6. Correlation of the logarithm of k, with AG23 for the one-electron reduction of the triplet states of @, BNFA (numbering as in table 3); 0, 2-nitrothiophen [numbering as in table 1 of ref. (46)]: 'I, 5-nitro-2-furoic acid [numbering as in table 2(b) of ref. (5) and reduction by TMPD (12) from ref. (56)]: V, duroquinone by I- (l), SCN- (2), [IrC1,,I3- (3), [Fe(CN),]*- (4), Br- (5), OH- (6), as in table 2 of ref.(46). The solid line is obtained from eqn (9) and (10) and the dashed line from eqn (9) and (13). (For details see text.) calculate the reduction potentials for the (SCN/SCN-), (Br*/Br-) and (S,O;/S,O:-) couples in aqueous Recently we the fair agreement existing between the experimental and the calculated rate constants for the reduction of the triplet state of 2-nitrothiophen in aqueous solution (pH 7, buffered); in view of this result it appeared reasonable to estimate the reduction potential of some of the nucleophiles from the corresponding experimental rate constants, under the prevailing conditions of measurement. Accordingly E"(D*+/D) values were estimated for the couples (CH,CO,/CH,CO,), (CH,O'/CH,O-), (OCN/OCN-), (CN'/CN-) and (SO;-/SO;-) as 2.32, 2.20, 2.18, 2.15 and 2.15 V us NHE, respectively, where a standard deviation of 0.025 V applies to all these values.For the last couple, where charged products are formed, wp was calculated for the ionic strength of the solution and taken into account in the calculation of E"(SO;-/SO;-). The figures obtained for the couples (CH,COJCH,CO;) and (CN'/CN-) compare with 2.41 and 1.9 & 0.3 V us NHE, re~pectively,~~, 54 as estimated from thermochemical data. In this connection, it is gratifying to find that a previous46 estimate of E"(CO;-/CO;-) = 2.15 V us NHE, made according to the same procedure as given above, is in excellent agreement with a recent calc~lation~~ based on thermochemical data (2.14 V us NHE). For the couple oH/OH- we have used a figure obtained before4,, (2.11 kO.09 V) which compares favourably with the reported5, value of 2.02 V us NHE.Using the available estimates for the reduction potential of the nucleophiles examined, it is possible to undertake the correlation log k, against G23, the free-energy change being calculated as prescribed in eqn (1 1). The result is shown in fig. 6, which also includes data for the one-electron reduction of the first excited triplet states of 2-nitr0thiophen,*~ duroq~inone~~ and 5-nitro-2-furoic acid (NFA)5 in neutral aqueous solution, aqueous methanol and aqueous (or neat) acetonitrile (for some quenchers two rate constants in two different solvents are presented). The recently measured55 one-electron reduction potential ofL. J. A. MARTINS AND T. J. KEMP 252 1 the radical cation TMPD'+ (TMPD = N,N,N',N',-tetramethyl-p-phenylenediamine), in aqueous solution at pH values of 7 and 13.5, as E$ (TMPD+/TMPD) = 0.266 V us NHE enabled the rate constant for the reduction of 3NFA by TMPD also to be included in this correlation.[The rate constant for this reaction was measured in MeCN56 and has the value k, = (7.90f0.10) x lo9 dm3 mo1-1 s-l; the observation of the resulting TMPD radical cation was also rep~rted.~] In fig. 6 the full line was calculated with the aid of eqn (9) and (10) with k,, = 1.0 x 1O1O dm3 mol-1 s-l, KO = 5 x 1O1O s-l, AGi3(0) = 10 kJ mol-1 and k,, 1.2 x 1O1O dm3 mol-l s-l, where the dissociation constant of the encounter complex is estimated through the Eigen equation57 k,, = k,, 3/4nr3N with r = 7 x dm, N being the Avogadro constant.While in general there is satisfactory agreement between experimental and calculated rate constants, thus consistent with an electron-transfer quenching mechanism, in a few cases the adherence to the calculated curve is not very good. This may be attributed to the approximations used and possibly to the use of mixed solvents in those instances. Indeed there is some indi~ation,~~ based on thermochemical data, that a number of anions used here are easier to oxidize in acetonitrile than in aqueous solution. The reorganization energy barrier, AG%,(O), obtained in this work has the same value as the one found by Rehm and Weller in their fluorescence quenching studies4, and is close to that adopted46 [AGI,(O) = 8 kJ mol-l] in correlating a series of electron-transfer reactions involving the triplet state of 2-nitrothiophen.Recently47 a value of 10.3 kJ mol-1 was obtained for this parameter in the anion-induced triplet-state quenching of some aromatic ketones. The broken line presented in fig. 6 was obtained by means of eqn (9) in conjunction with the parabolic free-energy relationship theoretically derived by Marcus5* (13) which predicts a decrease in the rate constant for highly exoergonic reactions, i.e. AG23 < -4AGi3(0); the region of the declining rate constant has been termed the 'Marcus inverted' region, and the search for its existence is a matter of much renewed interest, mainly because of the possibility of accessing that region in light-induced redox processes. However, in many 4 6 9 5 3 9 59 of highly exoergonic reactions a diffusion-limited rate constant is found for quite negative AG23, rather than the predicted 'inverted' behaviour.The same conclusion is reached in the present study, where the predicted decrease in the rate constant for AG23 < -40 kJ mol-1 is not observed for reactions as exoergonic as 185 kJ mol-l. Several explanations were recently60-62 discussed which may account for the experimental failure in finding the theoretically predicted decrease in the rate constant. Meanwhile it should be mentioned that a hyperbolic free-energy relationship first derived by Marcus, for atom- and proton-transfer has been used recently;25 however, while showing similar behaviour as eqn (10) for the entire range of AG23 values examined, when applied to electron-transfer reactions it has no theoretical advantages over that proposed by Rehm and Weller (used here).It is therefore concluded that both the spectroscopic evidence and the kinetic analysis presented above are consistent with a mechanism of complete electron transfer for the interaction of 3BNFA with the nucleophiles studied. AGi3 = AGi3(0) { 1 4- [AG23/4AGb3(0)11' CONCLUSIONS 3BNFA has been characterized in solution and shown to have a pattern of reactivity rather similar to that previously found5 for 3NFA, namely showing a propensity (towards appropriate donors) for engaging in hydrogen-atom abstraction reactions with rate constants of the magnitude expected for (n, z*) triplet states, while electron2522 TRIPLET STATE OF N-(n-BUTY L)-5-NITRO-2-FURAMIDE transfer occurs from a variety of anions with rates in accord with Rehm's and Weller's treatment.41 This similarity is particularly manifest in water, where both triplets exhibit their shortest lifetime, whereas the triplet state of 2-nitrothiophen, of (;n,z*) nature, has its longest lifetime in water and does not react with it.34946 Strong association of water with excited triplet molecules has been invoked to explain the initial enhancement of the triplet lifetime observed following addition of increasing amounts of water to acetonitrile (up to a certain limit).The effect of that association would be to decrease the radiationless decay of the triplet state.199 64 While steady-state photolysis work3* has clearly established that photoreaction with water (leading to substitution of the nitro group).occurs, we did not find support for the formation of the radical cation (and hydrated electron) via ionization of the triplet state as proposed b e f ~ r e .~ In fact we found that the absorption attributed to the triplet state is the only one observed upon flash-photolysing either BNFA or NFA in water; moreover, 3BNFA can be intercepted by addition of a nucleophile, e.g. CN-, although neither (CN);- nor BNFA'- could be detected. It is argued that (CN)',- reacts with ground-state BNFA (and possibly with BNFA'-) yielding the radical cation, which would subsequently react with either CN- or HzO, thus explaining the formation of both substitution p r o d ~ c t s . ~ In non-aqueous solvents (e.g. MeCN and Me,CO) we found that self-quenching of the triplet state occurs with generation of a radical cation (and BNFA-, the absorption of which is the only one detected) as a result of one-electron transfer [eqn (3)], a process that can be assisted by excimer formation rather as in the mechanisms recently proposed for aromatic photosubstitutions of both substituted6 and unsubstituted aromatic hydrocarbon^.^^ The self-quenching via eqn (3) might explain a previous observation3 that steady-state photolysis of BNFA in methanol (MeOH) leads to replacement of the nitro group by methoxy, viz.since MeOH has been shown66 to add to the radical cation of furans, preferentially at the 2- and 5-positions, yielding radical adducts and ultimately stable substituted products. Radical-cation intermediates are also 67 for the nucleophilic photosubstitution reactions of various methoxy-activated aromatic molecules which, in the cases examined, lead to a product composition similar to that obtained by anodic aromatic substit~tion.~~ The radical anion BNFA'- is formed in high yield in alkanols by a process represented by reactions (1) and (2) and, apart from the absorption of 3BNFA, another species appears to be produced, the spectrum of which closely resembles that of BNFA'-.A similar situation was found with NFA, for which it was shownz1 by pulse radiolysis that the second absorbing species is generated only in alkanol and not in water, where the radical anion is the only absorbing species. We tentatively assign this new absorption to a radical-adduct in view of a recent study6* in which it was shown that c H 3 radicals add to aci-nitroalkane anions yielding an adduct which exhibits a spectrum very similar to that of the corresponding nitroalkane radical anion.With the aim of understanding the mechanism of the nucleophilic photosubstitution reactions of 5-membered nitroheterocyclic c~mpounds,~ we have also subjected 2-nitropyrrole to laser flash photolysis (both in water and organic solvents), but found no absorbing transient in the spectral range 390-700 nm. This suggests that if the triplet state of the compound is populated then it is rather too short-lived to interact with suitable nucleophiles, which might explain the reported4 photostability of 2-nitropyrrole towards photosubstitution.L.J. A. MARTINS AND T. J. KEMP 2523 We thank Dr Peter Wardman of the Gray Laboratory, Mount Vernon Hospital, London, for performing a number of pulse-radiolysis experiments on our behalf. L. J. A. M. thanks IUBI (Portugal) for study leave. P. Wardman and E. D. Clark, Biochem. Biophys. Res. Commun., 1976, 69, 942. A. P. Reuvers, J. D. Chapman and J. Borsa, Nature (London), 1972, 237, 402. L. J. Powers, J. Pharm. Sci., 1971, 60, 1425. M. B. Groen and E. Havinga, Mol. Photochem., 1974, 6, 9. T. J. Kemp and L. J. A. Martins, J. Chem. SOC., Faraday Trans. I , 1981,77, 1425. J. P. Soumillion and B. De Wolf, J. Chem. SOC., Chem. Commun., 1981,436. C. Parkanyi, Pure Appl. Chem., 1983, 55, 331. N. Hata and M. Hokawa, Chem. Lett., 1981, 507. H. R. Snyder Jr, J. Med. Chem., 1967, 10, 737.lo H. Gilman and H. L. Yale, J. Am. Chem. SOC., 1950,72,3593. l1 J. R. Dyer, Applications of Absorption Spectroscopy of Organic Compounds (Prentice-Hall, Englewood l2 R. F. M. White, in Physical Methods in Heterocyclic Chemistry, ed. A. R. Katritzky (Academic Press, l3 P. Fournari and J. Tirouflet, Bull. SOC. Chim. Fr., 1963, 484. l4 K. J. Morgan and D. P. Morrey, Tetrahedron, 1966, 22, 57. l5 W. Armarego, D. R. Perrin and D. D. Perrin, Purijcation of Laboratory Chemicals (Pergamon Press, l6 T. J. Kemp and L. J. A. Martins, J. Chem. SOC., Perkin Trans. 2, 1980, 1708. l7 D. V. Bent and D. Schulte-Frohlinde, J. Phys. Chem., 1974, 78, 446. C. Capellos and G. Porter, J. Chem. SOC., Faraday Trans. 2, 1974, 70, 1 159. l9 C. A. G. 0. Varma, F. L. Plantenga, C.A. M. van den Ende, P. H. M. van Zeyl, J. J. Tamminga and J. Cornelisse, Chem. Phys., 1977, 22, 475. 2o C. Capellos and F. Lang, Znt. J. Chem. Kinet., 1977,9, 943. 21 P. Wardman, personal communication. 22 T. J. Kemp, L. J. A. Martins and P. Wardman, to be published. 23 S. L. Murov Handbook of Photochemistry (Marcel Dekker, New York, 1973). 24 G. Herzberg Molecular Spectra and Molecular Structure Z. Electronic Spectra and Electronic Structure of Diatomic Molecules (Van Nostrand, New York, 2nd edn, 1950), p. 560. 25 V. Balzani, F. Bolletta and F. Scandola, J. Am. Chem. SOC., 1980, 102, 21 52. 26 P. M. Rentzepis, Chem. Phys. Lett., 1969, 3, 717. 27 J. Saltiel and G. S. Hammond, J. Am. Chem. SOC., 1963, 85, 2515. 28 E. Clar and M. Zander, Chem. Ber., 1965, 89, 749.29 A. P. Marchetti and D. R. Kearns, J. Am. Chem. Soc., 1967,89, 769. 30 G. N. Lewis and M. Kasha, J. Am. Chem. SOC., 1944,66,2100. 31 W. G. Herkstroeter, A. A. Lamola and G. S. Hammond, J. Am. Chem. SOC., 1964,86,4537. 32 J. L. Charlton, C. C. Liao and P. de Mayo, J. Am. Chem. SOC., 1971,93,2463. 33 K. Sandros, Acta Chem. Scand., 1964, 18, 2355. 34 L. J. A. Martins and T. J. Kemp, J. Chem. Soc., Faraday Trans, 1, 1982, 78, 519. 35 H. D. Burrows, D. Greatorex and T. J. Kemp, J. Am. Chem. SOC., 1971,93, 2539. 36 C. A. G. 0. Varma, in Lasers in Chemistry, ed. M. A. West (Elsevier, Amsterdam, 1977), p. 354 and 37 H. A. Schwarz, J. Chem. Ed., 1981, 58, 101. 38 L. Eberson and K. Nyberg, J. Am. Chem. Soc., 1966, 88, 1686. 39 P. Neta, M. G. Simic and M. Z. Hoffman, J. Phys. Chem., 1976,80, 2018. 40 M. E. Peover, Trans. Faraday SOC., 1964, 60,479. dl D. Rehm and A. Weller, Zsr. J. Chem., 1970, 8, 259. 42 J. Cornelisse, G. Lodder and E. Havinga, Rev. Chem. Zntermed., 1979, 2, 231. 43 G. E. Adams, J. W. Boag and B. D. Michael, Trans. Faraday Soc., 1965,61, 1674. 44 J. K. Thomas, Trans. Faraday Soc., 1965, 61, 702. 45 J. H. Baxendale, P. L. T. Bevan and D. A. Stott, Trans. Faraday SOC., 1968, 64, 6389. 46 L. J. A. Martins, J. Chem. Soc., Faraday Trans. 1 , 1982,78, 533. 47 H. Shizuka and H. Obuchi, J. Phys. Chem., 1982,86, 1297. dB G. Winter, H. Shioyama and U. Steiner, Chem. Phys. Lett., 1981, 81, 547. 48 J. R. Harbour and S. L. Issler, J. Am. Chem. SOC., 1982, 104, 903. 50 R. A. Marcus, J. Chem. Phys., 1956,24, 966. 51 W. L. Reynolds and R. W. Lumry, Mechanisms of Electron Transfer (The Ronald Press, New York, Cliffs, N.J., 1965). New York, 1963), vol. 11, p. 124. Oxford, 1966). references therein. 1966), p. 123.2524 TRIPLET STATE OF N-(n-BUTYL)-5-NITRO-2-FURAMIDE 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 D. M. Stanbury, W. K. Wilmarth, S. Khalaf, H. N. Po and J. E. Byrd, Znorg. Chem., 1980,19, 2715. L. Eberson, Adv. Phys. Org. Chem., 1982, 18, 79 and references therein. V. M. Berdnikov and N. M. Bazhin, Russ. J. Phys. Chem., 1970, 44, 395. S . Steenken and P. Neta, J. Phys. Chem., 1982, 86, 3661. L. J. A. Martins, Ph.D. Thesis (University of Warwick, 1981). M. Eigen, Z. Phys. Chem. (N.F.), 1954, 1, 176. R. S. Marcus, Annu. Rev. Phys. Chem., 1964, 15, 155. V. Balzani, F. Bolletta, M. T. Gandolfi and M. Maestri, Top. Curr. Chem., 1978, 75, 1. P. Siders and R. A. Marcus, J. Am. Chem. SOC., 1981, 103, 741 ; 748. R. A. Marcus and P. Siders, J. Phys. Chem., 1982, 86, 622. R. A. Marcus, Znt. J. Chem. Kinet., 1981, 13, 865. R. A. Marcus, J. Phys. Chem., 1968, 72, 891. U. Dinur and B. Scharf, J. Chem. Phys., 1983,79,2600. H. J. Lemmetyinen, J. Chem. SOC., Perkin Trans. 2, 1983, 1269 and references therein. T. Majima, C. Pac, A. Nakasone and H. Sakurai, J. Am. Chem. SOC., 1981, 103, 4499. J. Den Heijer, 0. B. Shadid, J. Cornelisse and E. Havinga, Tetrahedron, 1977, 33, 779. D. Veltwisch and K-D. Asmus, J. Chem. SOC., Perkin Trans. 2, 1982, 1143. W. K. Wilmarth, D. M. Stanbury, J. E. Byrd, H. N. Po and C-P. Chua, Coord. Chem. Rev., 1983,51, 155. (PAPER 3/2233)

ISSN:0300-9599    

DOI:10.1039/F19848002509

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1984,80, 2525-2539 Studies of Reactions of Atoms in a Discharge-flow Stirred Reactor Part 4.-The 0 + NH, + NO/CO System BY DONALD L. BAULCH, IAN M. CAMPBELL* AND ROBERT HAINSWORTH Department of Physical Chemistry, The University, Leeds LS2 9JT Received 2 1st December, 1983 The reaction between O(,P) and NH, has been studied at 350K and a total pressure < 0.2 kPa in a discharge-flow stirred-reactor system using chemiluminescent emission (0 +NO, 0 + CO) intensities to measure concentrations. For small NH, additions ([NH,]/[O] w 10) a stoichiometry of -A[O]/A[NO] = 3.8k0.2 and a rate constant for the initial step of (3.2k0.3) x los dm3 mol-l s-l were established, both in reasonable agreement with previous measurements. When NO was added to the NH,+O system at concentrations at which, on the basis of rate constant data in the literature, the NO would be expected to compete with 0 atoms for any NH, or NH radicals in the system, no NO consumption was observed.On the other hand, when high ratios of [NH,]/[O] (< 100) are used so that NH, is generated by the secondary reaction net consumption of NO is observed on NO addition. These findings suggest that NH, and NH are not primary products of the attack of 0 atoms on NH,, contrary to previously postulated mechanisms. Addition of CO (< 60% replacement of the N, carrier gas) decreased the oxygen-atom consumption rates and stoichiometry, reflecting competition for OH produced between the reactions CO+OH-+CO,+H O+OH +O,+H. The observed effects corresponded to the generation of 1.3 kO.5 OH radicals per NH, reacted To explain these findings it is suggested that the first step in the reaction between 0 and NH, is associative leading to NH,O, which may rearrange to NH,OH.The latter is attacked by 0 atoms to produce H,O and HNO which appear to be the major immediate reaction products. In conjunction with the subsequent steps O+HNO+OH+NO O+OH+O,+H this mechanism leads to stoichiometries in acceptable agreement with those determined here and in previous studies. OH + NH, NH, + H,O at low [NH,I/[Ol. There have been two major studies of the mechanism of the O(,P) + NH, reaction in discharge-flow systems. Wong and Potter1 used a stirred-flow reactor coupled to a time-of-flight mass spectrometer to study the reaction in the temperature range 325-569 K.They detected production of H,, O,, H,O and NO, the last being the only nitrogen-containing product, and reported an average stoichiometric equation NH, +4.4 0 -+ NO+O.? H, + 1.2 0, + 1 .O H,O. 25252526 STUDY OF THE 0 + NH, + NO/CO SYSTEM Albers et aL2 studied the 0 +NH, reaction in a linear-flow system in the temperature range 330-470 K, using a combination of electron spin resonance and mass spectro- metric detection to determine an average stoichiometric equation NH, + 2.9 0 -+ 1.6 H +0.7 H,O + 1 .O NO +0.6 0,. It was established that H atoms (but not H,), H,O and NO were direct products appearing at short reaction times. Both of the above groups of worker& proposed that the initial step was hydrogen abstraction O+NH,-+NH,+OH.(1) O+NH,+NH+OH (2 a) O+NH+NO+H (3) O+OH -+O,+H (4) OH + NH, -+ H,O + NH,. ( 5 ) Albers et al., discounted reaction ( 5 ) as a significant source of H,O in their system since k, is not large enough compared with k, to divert substantial amounts of OH into reaction ( 5 ) when initial ratios of [NH,]/[O] were of the order of 10. Having detected HNO in the system (see Discussion section) and seeking a step forming H,O, they proposed that important steps subsequent to reaction (1) were Wong and Potter1 proposed that subsequent steps of significance were O+NH,-+HNO+H (2b) OH+HNO+H,O+NO (6) O+HNO-+OH+NO (7) O+OH +O,+H. (4) This leads to their observed stoichiometry of three oxygen atoms consumed per NH, molecule reacted, only if reaction ( 6 ) is substantially faster than reaction (7), under conditions where [OH]/[O] can be deduced to have been of the order of lo-, (reflecting the high value of k4), However, this cannot have been so, since Campbell and Handy3 established k,/k, < 4.4 for 7' 6 425 K from a study of the 0 + H, +NO/N, system in a similar reactor to that used in the present work.The mechanisms set out above have NH, as a primary intermediate and Wong and Potter' also proposed involvement of NH. A major aim in the present work was to apply new tests, based upon known kinetic data, for the generation of these intermediates in the 0 + NH, system. Both NH, and NH are known to react rapidly with NO forming nitrogen-nitrogen bonds: in both cases the rate constants for these reactions appear to be larger than those for the corresponding reaction with 0 atoms, even if the evidence is limited.For NH, reacting with NO, two pathways have been postulated to be of significance: NH,+NO -,N,+H,O + N,H +OH. The overall rate constant has been measured recently as k, = (5.8 1.4) x lo9 dm3 mol-' s-l,, (1.3 0.2) x 1O1O dm3 mol-1 s-l at 298 K and (1 .Of 0.3) xD. L. BAULCH, I. M. CAMPBELL AND R. HAINSWORTH 2527 1O1O dm3 mol-1 s-l at 295 K, the spread covering most previous measure- ments. The corresponding mean values at 350 K (our working temperature), calculated from the quoted temperature variations, are k, = 5.3 x loQ4 and 9.4 x lo9 dm3 mol-1 s - , . ~ Only one measurement of k,, and/or k,, (O+NH,) is available, 2.1 x lo9 dm3 mol-1 s-1 at 298 K, and k, = (5 f 1) x lo9 dm3 mol-l s-l was also measured in this Assuming that k,(k,, + kzb) has no strong temperature variation, these values predict k,/k, > 2.5 at 350 K.Thus, if NH, is a major intermediate in the O+NH3 system, additions of NO to produce [N0]/[0] > 0.4 should result in net consumption of NO rather than production. NH also reacts rapidly with NO in a reaction NH +NO -+ products (9) which must create a nitrogen-nitrogen bond to be exothermic. Measurements of k, at 298 K have yielded values of 2.3 x 10lo8 and (2.8k0.7) x 1O1O dm3 mol-1 s - ' . ~ Although there appear to be no measured values of k, (for O+NH) at ambient temperatures, extrapolation of the weakly temperature-dependent expression k, = 6.3 x 10, dm3 mol-l s-l derived from high temperature SystemslO~ l1 yields k, = 1.1 x 1Olo dm3 mol-1 s-l at 298 K. Thus, as for NH,, it appears likely that if NH is a major intermediate in the 0 +NH, system, even if NH, is not, the addition of NO should result in at least a dramatic change of the -A[O]/A[NO] stoichiometry, or even a switch from net NO production to consumption.In both of the proposed mechanisms the OH radical is a significant intermediate in the 0 + NH, system. Its presence can be established quantitatively by addition of a large excess of CO to the system, so that the reaction CO+OH -+CO,+H (10) can be made to compete for OH with reaction (4). The associated reduction in the consumption rate of oxygen atoms in association with the well established values of k, and k,, will then yield the OH-generation stoichiometry, which has not been measured before.EXPERIMENTAL The O+NH, reaction was studied using a stirred-flow reactor similar to that described in detail b e f ~ r e . ~ The central feature is a Pyrex sphere of internal volume 0.54dm3: this was internally coated with a thin layer of Teflon, using the procedure described by Berg and Kleppner,12 to inhibit heterogeneous removal of atoms. The sphere connected entry and exit tubings (1 1 mm i.d. Pyrex) along a common axis: the internal surfaces of these were also coated with Teflon. Inset into the entry tubing were two jets pointing upstream, J, located 340 mm and J, located 8 1 mm upstream of the entry to the sphere. A third jet, J,, also pointing upstream was inset into the exit tubing so that its tip was located just downstream from the exit from the sphere.These jets were used to add nitric oxide as required, at flowrates measured by calibrated capillary flowmeters. The viewing points, L, and L,, were located 46 and 51 mm downstream of J, and J,, respectively. Tubular aluminium light-guides, extending through the walls of the hot-box containing the sphere and tubings from above J, to below L,, were fitted to the tubes at L, and L,. An RCA IP28 photomultiplier in its housing could be attached to view L, or L, in turn through the optical filters described below. Thus the readings corresponding to the intensities of chemiluminescence, I, at L, and I, at L,, were measured on moving the photomultiplier from one light guide to the other with steady conditions maintained in the flow system.Ammonia was taken directly from a cylinder (BOC, Electra I1 grade, 99.9995% pure) and, at flowrates measured by a calibrated capillary flowmeter, it was added as a split-flow through two sidearms entering the sphere at points ca. 60" angularly displaced in the vertical plane from the entry tubing.2528 STUDY OF THE 0 + NH, + NO/CO SYSTEM Carbon monoxide (BOC, Technical grade) was purified as described before:', when used it was added at flowrates measured on a calibrated capillary flowmeter through a sidearm to the entry tubing located 167 mm downstream of J,. The general procedures for gas handling, purification and control were as described before.,' l3 The hot-box system, operated as described previ~usly,~ maintained a temperature of 350 f 2 K for all experiments.N atoms were produced by passage of N, through a microwave discharge at a flowrate typically of 80pmol s-l, reduced pro rata when CO was added. The resultant N(4S) atoms were titrated with NO at J,, completely converting them into O(3P) atoms via the very fast reaction N+NO -+ N,+O. This established the oxygen-atom concentration, [O],, at the entry to the sphere: no significant decay of [O] on the time-scale of ca. 30 ms between J, and the entry to the sphere would be expected when total pressures were kept at d 0.2 kPa suppressing three-body reactions. Upon addition of a small excess of NO the greenish air afterglow was produced, associated with the combination reaction O+NO+M +NO,+M with NO regenerated by the subsequent rapid reaction O+NO, +O,+NO.The signal IGl, produced when the photomultiplier viewed L, through a Wratten 61 filter (transmission range 478-602 nm, maximum at ca. 525 nm) is proportional to the product of the entry concentrations, [O],[NO],. The increase in IG1 when a known flowrate of NO was added through J, allowed calculation of [NO]. The signals measured similarly at L, were corrected by the methods described before, to obtain a value of IG2 which was directly comparable with IG1. When ammonia was added through the sidearms of the sphere IG2 required compensation for the dilution effect and also for the apparent quenching effect by NH, on the portion of the air afterglow transmitted by the filter. The compensation factor was established in separate experiments using rather faster total flows, in which NH, was added to the air afterglow through the sidearm in the entry tubing between J, and J,.The short time-scale involved before observation at L, precluded significant reaction of 0 with NH,. In the main experiments, in which the photomultiplier viewed L, through the Wratten 61 filter (no CO present), IG2 cc [O] [NO]. Previous experiments with this equipment operating under these pressure and flow conditions, have shown that the sphere behaves as an ideal stirred-flow reactor and hence the concentrations at the exit tubing are those throughout the sphere. The concentration of NO in the sphere, [NO], was deduced from the increase in when a known flowrate of NO was added through J,. Thereafter [O] can be evaluated using the calibrated proportionality factor for IG2 through IG1.Thus the changes in concentrations caused by reactions in the sphere are defined as A[NO] = [NO] -[NO], and - A[O] = [O] -[O],. The residence time, At, in the spherical reactor was calculated by dividing the volume of the sphere by the total volume flowrate of gases: this was typically ca. 0.6 s in our work. With stirred-flow operation the rate of removal of oxygen atoms, for instance, is then given by -A[O]/At. When CO partially replaced (< 60%) N, and the initial N atoms from the discharge were exactly titrated with NO at J, (i.e. [NO] = [NO], = 0), the bluish 0 + CO chemiluminescence was observed, associated with the combination reaction O+CO+M+CO,+M. This emission was viewed by the photomultiplier through an Oriel Optics G-774-3550 coloured glass filter (transmission range 290-410 nm, maximum at ca.360 nm). The signals produced with the photomultiplier viewing L, and L, are denoted by IB1 and Is,, respectively, the latter corrected to be directly comparable with the former and compensated for ammonia dilution and quenching effects as described above for IG2. These signals are proportional to the product of concentrations of oxygen atoms and carbon monoxide in the viewed region of tubing. Since [CO] is large and thus unchanged throughout, ZB1 and IB2 are proportional to [O], and [O], respectively. When CO was present in systems also containing significant concentrations of NO, theD. L. BAULCH, I. M. CAMPBELL AND R. HAINSWORTH Table 1. Selected values of the rate constant, k, measured at T = 350 K ~~ ~~ k/105 dm3 mol-l s-l [NH,]/[O] 3.2 3.1 3.4 3.1 3.3 3.2 8.7 10.3 12.4 14.1 16.7 18.9 2529 intensities transmitted through the optical filters contained components of both 0 +NO and 0 + CO chemiluminescence.In separate experiments in which 0 +NO or 0 + CO glows were established in isolation, the relative signals with each filter in the optical path in turn were measured. From these, linear equations were derived which enabled IBl(IB2) and IGI(IG2) to be extracted from the composite signals produced when the photomultiplier viewed 0 + NO/CO systems through the filters in turn. This procedure was aided by the very low transmittance of O+NO emission by the Oriel filter, so that IB1(IB2) accounted for all but a few per cent of the total signal for 0 + NO/CO systems at high NO concentrations.RESULTS Preliminary experiments were performed with 0 +NO + N, and 0 + H, +NO + N, systems at temperatures up to 425 K. As before,, the measured decay rates of oxygen atoms through the reactor indicated that the walls had effectively negligible activity for the heterogeneous removal of both 0 and H atoms in comparison with the gas-phase reaction rates concerned in this study. The principal sets of experiments were conducted with small additions of NH, (typically ca. 3% of the total gas flowrate) with the reactor maintained at a temperature of 350 K, under which conditions suitable rates of reaction were produced. The total pressures used, 5 0.2 kPa, were as small as compatible with stable operation of the discharge to produce sufficient partial dissociation of N, therein.Under this condition, the contributions of reaction cycles initiated by termolecular reactions were minimized so that the predominant components of the observed rates came from the bimolecular consecutive steps of the basic mechanism of the 0 + NH, reaction. According to the previous studies,lV there is a 1 : 1 stoichiometry between ammonia consumed and nitric oxide produced, i.e. -A[NH,] = A[NO]. Under these circum- stances and with relatively low values of [NH,]/[O] z 10 in the reactor, the applicable rate law will take the form A[NO]/At = k[O] [NH,] (9 where k may be presumed to be the rate constant for the first step of the mechanism and is likely to be rate-determining. A small correction to the measured value of A[NO] arises from the minor occurrence of reaction (5) between OH and NH,.This was calculated (see Discussion section) and hereafter A[NO] represents the corrected value. Table 1 shows a selection of values of k determined at 350 K on the basis of eqn (i) with the concentration ratio of [NH,]/[O] in the reactor. These show no trends in experiments in which no NO was added. Combination of all of our measured values gave k = (3.2 0.3) x lo5 dm3 mol-1 s-l for T = 350 K, where theerror limit represents2530 STUDY OF THE 0 + NH, + NO/CO SYSTEM 1.2 0.8 u 5! 0 = 0.4 X z -0.4 Fig. 1. Plot of A[NO]/(k[O][NH,]At) (see text) against [NO]/[O] for O/NH, systems with [NH,]/[O] < 12 with addition of NO and total pressures of ca. 0.2 kPa at 350 IS.The solid curve is the prediction from literature data. 10 8 6 4 n 2 0 -2 -4 -6 I /I I I I I I I I I I I 1 2 .o Fig. 2. Plot of the stoichiometry parameter n (see text) against [N0]/[0] for O+NH, systems with [NH,]/[O] < 12 with addition of NO and total pressures of ca. 0.2kPa at 350 K. The solid curves are the predicted variations from literature data, the dashed line indicating the predicted change from net NO production to consumption.4.0 3.0 n 2.0 D. L. BAULCH, I. M. CAMPBELL AND R. HAINSWORTH 253 I Fig. 3. Plot of the stoichiometry parameter n against [CO]/[O] for O+NH, systems with [NH,]/[O] < 15 without added NO but with CO partially replacing the N, carrier. Total pressures were ca. 0.2 kPa and T = 350 K. one standard deviation. As is pointed out in the Discussion section, this represents good agreement with most previous determinations.When excess amounts of NO were added at J,, 0 +NO + NH, + N, mixtures were created in the reactor with FJO]/[O] ratios in the range 0.2-2. On the basis of eqn (i), the effective NO stoichiometry parameter A[NO]/(k[O] [NH,] At) should have a value of 1 .O unless the addition of NO to the system introduces further reaction steps consuming NO and/or inhibiting nitrogen-containing intermediates from being converted eventually to NO. Fig. 1 shows experimentally determined values of this parameter as a function of FJO]/[O]. The points show no pronounced trend. The curve drawn in fig. 1 represents the variation which is predicted if O+NH,+NH,+OH (1) is the sole initiating reaction and the resultant NH, radicals are competed for by the reactions 0 + NH, + products (+ NO) (2) NO + NH, + N-N + other products (8) with k,/k, = 2.5, the minimum value (see Introduction).It is evident that this curve does not match with the experimental points which, in particular, show no tendency to switch to net NO consumption at the higher [NO]/[O]. Another result bearing on the same point is the variation of the combined stoichiometries of oxygen-atom consumption and NO production. The directly determined values of A[O] contain small contributions from reaction cycles initiated by three-body processes and arising from the minor fraction of OH radicals which attack NH, in reaction (5) rather than react with 0 atoms in reaction (4). The parameter, (A[O]),, is obtained after correction for these effects (see Discussion2532 1.8 0.2 ‘ I 1 I I I 20 40 60 80 100 [ NHB 1 /[ 0 1 Fig.4. Plot of A[NO]/(k[O] [NH,] At) (A[NO] is the experimental value) against pH,]/[O] in O f NH, systems with NO added to produce [NO]/[O] x 1.5. The solid line is the predicted variation if NH, does not react with NO but only with 0 atoms. Total pressures were ca. 0.2 kPa and T = 350 K. section): the difference between A[O] and (A[O]), amounted to at most 25% of A[O] at the highest [NO] under our conditions of low total pressure and [NH,]/[O] x 10. Fig. 2 shows the resultant values of the stoichiometry n = -(A[O]),/A[NO] over the range of [N0]/[0] used. As in fig. 1, the curves represent the expected variation if reaction (1) is the sole initiating step and k,/k, = 2.5.As [NO]/[O] increases, n shows only a very slight decrease and the curves do not in any way predict the variation. A short back-extrapolation to [N0]/[0] = 0 yields n = 3.8 rt 0.2. In a set of experiments with similar low additions of NH, but with CO addition to replace up to 60% of the N, carrier gas, the O+CO chemiluminescence (i.e. ZB1 and 1B2) was used to measure A[O]. Again it was necessary to make minor corrections to A[O] to obtain (A[O]),. The rate equation for the reaction under stirred-flow conditions can be written as n = - (A[OI)c/(H.OI “&I At) (ii) where n corresponds to the number of oxygen atoms removed per NH, molecule reacted (or per NO produced). Based on eqn (ii), fig. 3 shows the variation of n with [CO]/[O].A distinct trend is observed, with a pronounced decrease from the ordinate intercept of n = 3.8k0.2 to a value of n = 2.7k0.2 at high [CO]/[O]. Finally we also investigated 0 +NO + NH, + N, systems with higher NH, concen- trations (< 15% of the total gas) and [NH,]/[O] ratios of up to ca. 100. Under theseD. L. BAULCH, I. M. CAMPBELL AND R. HAINSWORTH 2533 circumstances, reaction of OH with NH, is expected to be a significant process producing NH, radicals : Fig. 4 shows the variation of the parameter AFJO]/(k[O] [NH,] At) (used in fig. 1) with [NH,]/[O] when mO]/[O] is maintained at 1.5. A distinct downward trend is apparent. The solid line shows the expected trend of the parameter if the NH, from reaction ( 5 ) did not react with NO but only underwent the catalytic cycles composed of reactions (2a) and (3) or (2b) and (7) (see Introduction). The divergence of this line from the experimental points indicates that the reaction of NH, with NO is of major importance under these conditions.It also bears out the fundamental point that the constancy of the value of A[NO]/(k[O] [NH,] At) at a value close to unity in fig. 1 for [NO]/[O] x 2 indicates that NH, is not a significant product from the reaction of 0 atoms with NH,. OH + NH, -+ NH, + H,O. ( 5 ) DISCUSSION The overall reaction stoichiometry at 330 K (close to our working temperature) reported by Albers et aL2 corresponded to the equation NH,+3.1 0 + 1.6H+0.7H20+ 1.00+~.70,. Albers et al. detected no H, as a direct product, only atomic H. In contrast, Wong and Potter1 found significant yields of H, which in their case may have arisen from surface recombination of H atoms, possibly in the mass-spectrometric sampling system.However, if it is assumed that the H atoms generated in the system of Wong and Potter appeared only as H,, the equation derived from their measured stoichiometries at 325 K is NH, + 4.0 0 --+ (0.8-0.9) H + (0.6-0.7) H,O + 1 .O NO + 1.3 0, with only 2.3 H atoms appearing among the products, pointing perhaps to difficulties in analysing for H, and/or H. Similar difficulties do not seem to have occurred in analysing for the 0-containing species; the 0 atoms' mass balance is adequate and their measured value of - A[O]/A[NO] at 304 K of ca. 4.0 (discounting two apparently erratic points shown on their diagram) is in good agreement with our determination of this ratio as 3.8 k0.2 at 350 K.of the reaction stoichiometry it appears that there is good agreement on the ratio of H,O and NO produced per NH, consumed but considerable discrepancy in the stoichiometric coefficients for H and 0, production. The only apparent source of 0, in the O+NH, system would appear to be the reaction Comparing then these two determinations'. O+OH-+O,+H (4) which must be the predominant mode of consumption of OH for the low [NH,]/[O] ratios used by Wong and Potter' (< 4) and Albers et aL2 (ca. 13.5) and in most of our work. On this basis, the stoichiometric coefficient of H atoms should not be less than that of 0, and the amount of OH generated should be approximately equal to that of 0,.The first requirement is not in accord with the stoichiometric equation deriving from the work of Wong and Potter, but again this could be accommodated if they underestimated the amount of €3, produced. Although the stoichiometric coefficients reported by Albers et aL2 meet the first requirement, because that for 0,2534 STUDY OF THE 0 + NH, + NO/CO SYSTEM (0.7) is considerably lower than that for H (1.6) this would imply that reaction (4) cannot be regarded as the sole source of H atoms in the system. As will be discussed below our work indicates that 1.3 k 0.5 OH radicals are generated per NH, molecule consumed. This value represents equality with the stoichiometry coefficient for 0, given by Wong and Potter1 but is in some disagreement with that given by Albers et aL2 It is thus clear that disparities remain to be resolved in the stoichiometry of the 0 + NH, reaction under discharge-flow conditions.Our measured rate constant of k = (3.2k0.3) x lo5 dm3 mol-1 s-l at 350 K repre- sents good agreement with values for the same temperature calculated from rate- constant expressions given in the literature, except for that given by Wong and Potter.' Table 2 summarizes these data, showing the Arrhenius parameters deduced in the other studies. In table 2, the result given by Kurylo et al.15 was produced on the assumption that 3.5 0 atoms were consumed per NH, molecule reacted, the simple average between the corresponding stoichiometries measured by Wong and Potter' and Albers et aL2 The contributions to concentration changes through the reactor of the reactions which are not part of the basic mechanism of the 0 + NH, reaction must be assessed in deriving (AIO])c from the measured A[O].Despite the low pressures used in this work, some cycles initiated by three-body combination reactions necessitate corrections to A[O]. In the reaction vessel there will be significant concentrations of 0, NO, O,, H and, when added, CO. Previous work in this laboratory3+ 1 3 9 l4 and literature sources quoted in these papers have provided the necessary values of rate constants for the quantitative assessment of such corrections. In calculating the rate of generation of M atoms in our reactor, we have accepted the A[H]/A[NO] stoichiometry ratio measured by Albers et a1.2 but, in any case, those cycles initiated by three-body reactions involving H atoms are of much less significance than those initiated by three-body reactions involving 0 atoms.The cycles assessed as having significant rates (producing corrections to A[O] above 3%) are as follows: O+NO+N, +NO,+N, (1 1) H+NO,+OH+NO O+OH +O,+H O+NO, -+ NO+O,. (13) At 350 K, k,, = 7.8 x lolo dm3 mol-1 s-l l6 and k13 = 4.5 x log dm3 mol-1 s-l:17 under typical conditions with [HI and [O] of the same order of magnitude in the reactor, reaction (12) will be the predominant consecutive step to reaction (1 1). An important point therefore is that this cycle generates a minor amount of the OH in the reaction system. Achieving similar effects in this respect but of rather lower rates than that of the cycle above are the two reaction sequences: H+NO+N,+HNO+N, (14) O+HNO+OH+NO (7) H+0,+N2 +HO,+N, O+HO,+OH+O, H + HO, + 20H.The last elementary step has been shown by Sridharan et a1.18 to be the predominant channel for reaction of H and HO,, and data from that study18 were incorporated intoD. L. BAULCH, I. M. CAMPBELL AND R. HAINSWORTH 2535 Table 2. Values of the rate constant, k, at T = 350 K and the corresponding Arrhenius parameters, A and E, k/105 dm3 mol-1 s-l A/lOs dm3 mol-1 s-l EJkJ mol-l ref. 8.7 1 .o 20.5 Wong and Potter' 2.7 1.5 25.1 Albers er aL2 3.0 f 0.7 4.0 f 0.9 27.6kO.l Kurylo er all5 3.2 f 0.3 - - this work the assessment of the correction to A[O]. The calculations show that when CO was present in the system, the cycles of reactions initiated by three-body combinations of H and 0 with CO make relatively insignificant contributions to A[O] under the conditions used on the basis of rate data given before.13 There is competition between 0 atoms and NH, for the OH generated in the system O+OH+O,+H (4) ( 5 ) OH +NH, + NH, + H,O.Mean values of the rate constants at 350 K derived from the literature are k, = 1.7 x 1O1O dm3 mol-1 s-l l9 and k, = 1.5 x lo8 dm3 mol-1 S - ' . ~ O For [NH,]/ [O] < 15 in our reactor (in the systems with low NH, additions), the fraction of the generated OH radicals reacting via reaction ( 5 ) is < 15%. However, reaction ( 5 ) generates NH, radicals (as discussed below) for which 0 and NO compete : reaction with NO can be important, particularly when FJO]/[O] is relatively large, which circumstance also produces the largest contributions to A[O] and A[NO] from the cycles initiated by the three-body reactions (1 1) and (14).A computer program was written to model the reaction system to derive the corrections to the measured values of A[O] and AFJO] from the cycles initiated by the three-body reactions and reaction (5). The largest corrections were assessed as ca. 25% of A[O] and ca. 13% of A[NO], the former being a downward correction to (A[O]),, the latter being an upward correction because of the predominant reaction of NH, with NO [reaction (S)] under high FJO]/[O] conditions. It is considered that any deviations in these corrections originating in uncertainties in the incorporated rate constant values are minor in comparison with random experimental variations in A[O] and A[NO].The relative insensitivity of the stoichiometry parameter in fig. 1 and 2 to the [N0]/[0] ratio demonstrates clearly that the predominant mechanism cannot involve any species with a substantial reactivity towards NO as compared with 0 atoms: this rules out NH,, NH or even N atoms as intermediates. Albers et aZ.2 detected HNO in the O+NH, system, although it must be noted that this was in the system at 1010 K with a high [NH,]/[O] ratio (ca. 100): formation of NH, via reaction ( 5 ) would be expected to be significant under these conditions and its subsequent conversion could have been the source of the HNO. However, HNO has a much higher reactivity towards 0 atoms than towards NO: the first has a rate constant of the order of 1 O 1 O dm3 mol-1 s-13 while the latter has a rate constant of 3 x 10, dm3 mol-l s-121 at ambient temperature.Thus we consider that HNO is the most plausible inter- mediate species satisfying the above requirements, produced by a reaction pathway not involving NH,. Since the completion of our experimental work, we have become aware of an unpublished experimental study on the 0 + NH, systems which supports our basic2536 STUDY OF THE 0 + NH, + NO/CO SYSTEM postulate that reaction (1) does not occur at moderate temperatures. Hancock and coworkers22 used a discharge-flow system at ambient temperature containing O+NH,+Ar mixtures with concentration ratios of 1 : 20-100: 300 at a total pressure of ca. 0.27 kPa. They used laser-induced fluorescence techniques in an unsuccessful attempt to detect NH,.On the basis of the assumption that reactions (1) and (2) were the sole initial steps and with the values of k, = 6.4 x lo4 dm3 mol-1 s-' from Albers et d2 and k, = 2.1 x lo9 dm3 mot1 s-' from Gehring et al.,' [NH,] z (0.2- 1.0) x mol dm-, should have existed in the system. The detection system was sensitive enough to have responded to this level of [NH,], yet no indication of its presence was obtained. However, this result is expected if reaction (1) is not a major reaction pathway. Turning now to the experiments in which CO was added to the system, fig. 3 shows that the stoichiometry parameter n falls as [CO]/[O] increases, i.e. fewer 0 atoms are consumed per NH, molecule reacted at high [CO]/[O]. The important reactions in this connection are CO+OH-,CO,+H (10) O+OH-+O,+H.(4) The mean rate constant value of k,, = 1 .O x lo8 dm3 mol-1 s-l is valid23 for low total pressures in the temperature range 300-500 K while k, = 1.7 x 1O1O dm3 mot1 s-l at 350 K as cited above. Thus when [CO]/[O] = 1000, reaction (10) will have a rate ca. 6 times larger than that of reaction (4), so accounting for the observed decrease in oxygen-atom consumption and quantitatively indicating the amount of OH generated within the basic mechanism of the 0 + NH, reaction. The progression to the right in fig. 3 is from n = 3.8f0.2 (deduced by extrapolation in conjunction with fig. 2) at [CO]/[O] = 0 to n = 2.7f0.2 when [CO]/[O] z 1000. This corresponds to the generation of 0.8-1.8 OH radicals per NH, molecule reacted and a mean value of 1.3.We have postulated that HNO is a major intermediate in the 0 + NH, system so that, if this has a unity stoichiometry with respect to NH, consumed, then the expected fate of HNO in the reaction 0 + HNO -+ OH +NO (7) could account for most of the OH produced in the system. Our measured mean stoichiometry of 1.3 OH per NH, reacted is also compatible with the stoichiometric coefficient of 0.7 for H,O, which comes from the work of both Wong and Potter' and Albert et ul.:, it indicates that together the H atoms (resulting from subsequent reaction of the OH) and H,O are the major repositories of hydrogen in the system. The data of fig. 4 show a significant decrease in the rate of NO production, corresponding to the parameter A[NO]/(k[O] [NH,] At), with increasing [NH,]/[O] ratios in the reactor when [NO]/[O] is maintained at the relatively high value of 1.5.The experimental points cannot be reconciled with the line which represents the situation expected if the NH, produced by the reaction of OH with NH, [reaction (5)] reacted only with 0 atoms. This confirms the viability of our kinetic test for the generation of NH,, which reacts predominantly with NO [reaction (8)] under these conditions and serves as supporting evidence that the near horizontal trend of the experimental points for [NO]/[O] G 2 in fig. 1 must discount NH, as a significant intermediate species in the basic 0 + NH, mechanism. No quantitative conclusions can be drawn from fig. 4 because of the lack of knowledge of relative third-body efficiencies of NH, and N, and the uncertainties in both absolute values of some of the rate constants and in the partitioning between pathways, the latter remarks applying particularly to reaction (8).D.L. BAULCH, I. M. CAMPBELL AND R. HAINSWORTH 2537 Table 3. Comparison of predicted and observed stoichiometries for the O+NH, system at T = 325-350 K stoichiometric relationships predicted observed ref. 4.0 1 .o 1 .o 1 .o 1 .o - A[ol/A”ol - A[Ol/AW,I OH generated per NH, H generated per NH, 0, generated per NH, H,O generated per NH, 2H,O +OH generated per NH3 3.0 3.8 k 0.2 3.1 4 1.3 k0.5 (A) 2 0.8 1.6 1.3 0.7 2 0.6 0.7 (B) this work Albers et al., Wong and Potter’ this work Wong and Potter’ (deduced) Albers et al., Wong and Potter1 Albers et al., Wong and Potter’ Albers et al., 2.7 mean values (A) and (B) It appears that the major initial steps of the 0 + NH, reaction under discharge-flow conditions must involve addition rather than abstraction steps to give rise to H20 and HNO as direct products.If the major net reaction accomplished by the first steps is represented by the stoichiometric equation 2 0 + NH, -+ HNO + H20 with reactions (7) and (4) as the subsequent steps we come close to matching the measured stoichiometries within the uncertainties imposed by divergences in literature values and by their likely error limits. The above proposals produce the ‘predicted’ values in table 3 in which they are compared with experimental values. The most apparent difference shown in table 3 is between the predicted coefficient of 1 .O for H,O and the two experimental values of 0.7 : this may imply the occurrence of another early step not leading to production of H20.It also needs to be remarked that the stoichiometric coefficients for H atoms and 0, molecules determined by Albers et aZ., are significantly above and below 1 .O, respectively: this implies that reaction (4) cannot be the only source of H atoms if both values are accepted. Our measured OH-generation stoichiometry can, within the error limit quoted, almost agree with either of these coefficients but cannot agree with both simultaneously. However, in conjunction with the almost reverse coefficients deriving from the work of Wong and Potter,l there is further confusion. It seems therefore that further measurements of these coefficients may be required before definitive statements can be made in this respect.Undoubtedly the most significant point in table 3 is that the stoichiometric coefficient for H20 formed, combined with our measured mean value for OH formed, comes close to accounting for the hydrogen content of the NH, consumed. Finally, we consider the proposed initial addition reaction in more detail. In a theoretical investigation of the 0 + NH, interactions Hart24 showed that O(,P) + NH, correlates with an antibonding NH,O(,E) potential surface. The bound NH,O(lA) species correlates with O(l0) + NH, and, according to Hart’s calculations, the minimum energy of this species as a function of N-0 bond length lies some 45 kJ mol-l above the energy level corresponding to separated O(,P) + NH,.Although the rate constant, k, shows a positive temperature coefficient, this corresponds to ca. 26 kJ mol-l only (table 2); any initial process involving O(,P)+NH, climbing the NH,O triplet surface to achieve crossing to NH,O(’A) would appear to be precluded under these circumstances. However, there is an obvious discrepancy in the results2538 STUDY OF THE 0 + NH, + NO/CO SYSTEM produced by Hart24 in that he reported that a similar calculation for hydroxylamine (NH,OH, isomeric with NH,O) gave its minimum energy as only 125.9 kJ mol-l below that of NH,O(lA). Heat of formation data are availablez5 which indicate an endothermicity corresponding to AH = +241 kJ mol-1 for the system NH,OH --+ NH, + O(,P). In combination with the above energy difference, this would place the minimum of the NH,O(lA) potential surface at 115 kJ mol-l below the energy of separated O(,P)+NH,.This would make the N-0 bond strength 305 kJ mol-1 [with respect to dissociation to O(lD)+NH,], comparable to the 273 kJ mol-1 for the bond in hydroxylamine, rather than the 145 kJ mol-1 indicated by Hart's calculation. It is perhaps reasonable to suggest that the difference in total energies of NH20H and NH,O(lA) indicated by similar theoretical calculations might be more secure than the individual absolute values. This deepening of the potential curve for NH,O('A) could then produce the possibility that the crossing between the NH,O(,E) and NH,O(lA) surfaces would occur at an energy more compatible with the activation energy asso- ciated with k .Hart24 points out that NH,O(lA) is likely to be highly unstable with respect to its' isomerization to NH,OH. If indeed a bound form of NH,O is formed in the initial stages of the O(,P) + NH, reaction and does undergo isomerization, even partially, prior to its attack by a second O(,P) atom, there appears to be a feasible route to the formation of H,O and HNO, as indicated below H H \ H-N +O-NHJOspecies+ " - 0 / / \ H H H 11 I . I . I . ' 0 , -HNO+H,O. H \ O + N - 0 -O-N---- H H' H / \ H An objection may exist to the involvement of NH,O(lA) in this scheme in that a spin conversion is required from NH,0(3E). A collision-induced process may be responsible but its effect is not manifested in the second-order, pressure-independent k for the overall reaction. However, it may be postulated, highly speculatively, that the reaction could go through a series of (unspecified) triplet states since HNO(a ,A"), the product for full spin conservation, is only 79 kJ mol-l above HNO(X1A')26 and the exother- micity available is 801 kJ mol-1 2 5 9 26 for the net reaction 2 0 + NH, + HNO + H,O.No study of the reaction of O(3P) with NH,OH has been reported so there is no evidence with which to compare this mechanistic proposal. Hydrogen-atom shifts in the course of bimolecular reactions are known. In the reaction of O(,P) with acetylene2' the major channel is represented by 0 + C,H, + CH, + CO. In the reaction of O(,P) with formaldehyde, Chang and Barker28 have proposed that a major part of the reaction is accomplished by a combination channel, with hydrogen-atom transfer effecting conversion between intermediate species prior to dissociation to products.This mechanism involved only triplet species however. The measurements that only 0.7 H,O is produced per NH, may suggest that < 30%D. L. BAULCH, I. M. CAMPBELL AND R. HAINSWORTH 2539 of the reaction may go via a different pathway to the above proposal. The slight downward trend towards higher [N0]/[0] in fig. 2 could be indicative of this, as could our mean measured OH generated per NH, and the measurement of the stoichiometric coefficient for H atoms by Albers et al.,, both being > 1.0. We cannot exclude NH, production from < 20% of the NH, consumed by reaction with O(,P) within our error limits.Initial attack of NH, by the H atoms present in the system can be excluded since shock-tube dataz9 show that this reaction has an activation energy of 73 kJ mol-l while Albers et aL2 found that, even at 800 K, the rate was too low for detection in a discharge-flow system. Beyond these observations, there appears little point in speculating on the minor aspects of the mechanism. In conclusion we may state that we have shown that existing mechanistic proposals for the O(3P) + NH, reaction under discharge-flow conditions are inadequate because they propose a major intermediate role for NH,. Our experiments with added NO definitely exclude NH,, NH and N as major intermediates and we propose that HNO has a central role in the main mechanism based on addition rather than abstraction in the initial step.One of us (R.H.) thanks the S.E.R.C. for a studentship. E. L. Wong and A. E. Potter, J. Chem. Phys., 1965,43, 3371. E. A. Albers, K. Hoyermann, H. G. Wagner and J. Wolfrum, Proc. 12th Znt. Symp. Combustion (The Combustion Institute, Pittsburgh, 1969), p. 313. I. M. Campbell and B. J. Handy, J. Chem. Soc., Faraday Trans. I , 1975,71, 2097. J. A. Silver and C. E. Kolb, J. Phys. Chem., 1982, 86, 3240. L. J. Stief, W. D. Brobst, D. F. Nava, R. P. Borkowski and J. V. Michael, J. Chem. SOC., Faraday Trans. I , 1982,78, 1391. P. Andresen, A. Jacobs, C. Kleinermanns and J. Wolfrum, Proc. 19th Znt. Symp. Combustion (The Combustion Institute, Pittsburgh, 1982), p. 11. M. Gehring, K. Hoyermann, H. Schacke and J. Wolfrum, Proc. 14th Int. Symp. Combustion (The Combustion Institute, Pittsburgh, 1973j, p. 99. S. Gordon, W. Mulac and P. Nangia, J. Phys. Chem., 1971, 75, 2087. I. Hansen, K. Hoinghaus, C. Zetsch and F. Stuhl, Chem. Phys. Lett., 1976, 42, 370. S. W. Benson, R. Shaw and R. W. Woolfolk, Proc. Stationary Source Combustion Symp. EPA-600/2-76/152a, 1976, vol. 1, p. 267. l1 J. M. Beer, M. T. Jacques, W. Faramayan and B. R. Taylor, Proc. 18th Znt. Symp. Combustion (The Combustion Institute, Pittsburgh, 1981 j, p. 101. l2 H. C. Berg and D. Kleppner, Retl. Sci. Znstrum., 1962, 33, 248. l3 I. M. Campbell and B. J. Handy, J. Chem. Soc., Faraday Trans. 1, 1978, 74, 316. l4 I. M. Campbell, J. S. Rogerson and B. J. Handy, J. Chem. Soc., Faraday Trans. I , 1978, 74, 2672. l5 M. J. Kurylo, G. A. Hollinden, H. F. LeFevre and R. B. Timmons, J. Chem. Phys., 1969, 51, 4497. J. V. Michael, D. F. Nava, W. A. Payne, J. H. Lee and L. J. Stief, J. Phys. Chem., 1979, 83, 281 8. l7 P. P. Bemand, M. A. A. Clyne and R. T. Watson, J. Chem. Soc., Furaday Trans. 2, 1974, 70, 564. U. C. Sridharan, L. X. Qiu and F. Kaufman, J. Phys. Chem., 1982, 86,4569. l9 R. S. Lewis and R. T. Watson, J. Chem. Phys., 1980,84, 3495. 2o J. A. Silver and C. E. Kolb, Chem. Phys. Lett., 1980, 75, 191. *l S. G. Cheskis, V. A. Nadtochenko and 0. M. Sarkisov, Znt. J. Chem. Kinet., 1981, 13, 1041. 22 G. Hancock, personal communication. 23 I. M. Campbell and B. J. Handy, Chem. Phys. Lett., 1977, 47, 475. 24 B. T. Hart, Aust. J. Chem., 1976, 29, 231, 25 S. W. Benson, J. Chem., Educ., 1965, 42, 502. z6 P. A. Freedman, Chem. Phys. Lett., 1976, 44, 605. 27 R. Lohp and P. Roth, Ber. Bunsenges. Phys. Chem., 1981,85, 153. *' J. S. Chang and J. R. Barker, J. Phys. Chem., 1979, 83, 3059. 2B J. E. Dove and W. S. Nip, Can. J. Chem., 1974, 52, 1171. (PAPER 3/2245)

ISSN:0300-9599    

DOI:10.1039/F19848002525

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1984, 80, 2541-2548 Influence of Pressure on Activity Coefficients of Electrolytes in Solution BY SEFTON D. HAMANN CSIRO Institute of Industrial Technology, G.P.O. Box 433 1, Melbourne, Victoria 3001, Australia Received 21st December, 1983 Direct measurements have been made by the solubility method of the activity coefficient of potassium picrate in water at 25 "C, at atmospheric pressure and at a pressure of 200 MPa, in the presence of medium to high concentrations of added lithium chloride or lithium sulphate. The results, and others derived indirectly from earlier volumetric data on aqueous solutions of sodium chloride and potassium sulphate, show that the mean ionic activity coefficients increase appreciably when the pressure is raised, to an extent that increases as the ionic strength goes up.Part of the trend may arise from an enhanced dissociation, under pressure, of ion pairs in the concentrated solutions. In recent years there have been many measurements made of the effects of high pressures on the rates and equilibria of chemical reactions in ~olution.l-~ They have almost all been carried out in conditions where the concentrations of the reactants were known but their activity coefficients were not, and in analysing the results the authors have generally assumed that the mixtures could be regarded as ideal, or at least that the pressure dependence of the activity coefficients could be ignored. The same assumptions have often been made in considerations of chemical equilibria in geological systems under high pressures.The supposition that activity coefficients are virtually independent of pressure is probably a valid one for most non-electrolyte mixtures: it is supported, for instance, by the calculations of Collings et u Z . , ~ which show that the activity coefficients for seven non-electrolyte systems change by only a few percent between atmospheric pressure and 100 MPa (1 MPa = 10 bar z 9.869 atm). However, many of the high-pressure chemical studies have been concerned with electrolyte solutions of quite high concentrations, where the departures from ideality can be much greater than those for non-electrolytes. The present paper will consider solutions of that kind and deal, in particular, with the high-pressure behaviour of the activity coefficients of some simple 1 : 1 and 1 : 2 electrolytes in aqueous solutions of high ionic strengths.EXPERIMENTAL There appear to have been no direct measurements made previously of ionic activity coefficients under pressure. The measurements to be described here were not intended to be highly accurate: they were meant to be exploratory rather than definitive and the results show some scatter, although the general trends are probably reliable. The experiments were aimed at determining the mean molal ionic activity coefficient, y * , of the 1 : 1 electrolyte potassium picrate, KPi, in water in the presence of added strong electrolytes. A solubility method5-' was used, in which the saturation concentration of the sparingly soluble KPi was measured at known molalities of added LiCl or Li,SO, up to m = 8.7 or m = 2.4, 254 12542 ACTIVITY COEFFICIENTS OF ELECTROLYTES respectively.[In this paper rn stands for the dimensionless molality defined by Guggenheim :* it is the (pure) number of moles of a solute present in 1 kg of water.] Those electrolytes were chosen because the salts formed by double decomposition, KCl, K,SO, and LiPi, are all sufficiently soluble to remain in solution at high concentrations of the lithium salts. The measurements were made at 25 "C, at atmospheric pressure (0.1 MPa) and at a pressure of 200 MPa. MATERIALS KPi was prepared by neutralizing a solution of analytical-grade picric acid, in hot ethanol, with aqueous KOH. The salt precipitated immediately and was filtered, washed with ethanol, recrystallized twice from hot water and dried in uacuo.The LiCl and Li,SO, were analytical reagents, and their solutions were made up by weight of the dried salts. APPARATUS AND PROCEDURE The high-pressure equilibrations were carried out in a bored stainless-steel cylinder, described previously. Ca. 25 cm3 of the solvent medium (water or an aqueous solution of LiCl or Li,SO,) were placed in the cylinder, together with sufficient solid KPi and a Teflon-coated stirring pellet. The mixture was then compressed by driving a plunger into the cylinder in a hydraulic press; the plunger was locked in place and the assembly was removed from the press.9 The mixture was then stirred in a temperature-controlled room at 25 f 0.5 "C by rotating the cylinder, head over heels, at 2 rev.min-l for several hours. The motion of the pellet produced very effective stirring and that period was more than sufficient to establish equilibrium. At the end of the time, the undissolved KPi was allowed to settle, the pressure was rapidly released, and a sample of the supernatant solution taken as quickly as possible through a sintered glass filter. The sample was diluted 100-fold with pure water and the resulting molality of KPi measured photo- metrically, using the strong absorption band of picrate ions at 357 nm; these analyses were accurate to ca. 1%. The measurements at atmospheric pressure were made in the same way, except that the cylinder was simply stoppered or replaced by a glass phial. RESULTS The method of sampling the solutions is open to the criticism that, if the solubility of KPi increased under pressure, there was a danger that some of the salt might have crystallized out and been lost during the decompression and sampling steps. However, in most cases the solubility decreased under pressure, and in these instances there was no real risk of additional solid dissolving after the release of pressure.In the cases where the solubility increased (at low ionic strengths), the degree of supersaturation was small and no evidence of recrystallization was ever found. The solubility of KPi in the absence of added salt was found to be m = 0.0242 at atmospheric pressure and m = 0.0282 at 200 MPa. To estimate the activity coefficients of KPi, the solubilities were extrapolated to zero ionic strength by the method of Lewis and Randall.l0 This gave values for the hypothetical solubility at zero ionic strength, m', equal to 0.0206 at atmospheric pressure and 0.0233 at 200 MPa.The mean molal ionic activity coefficients, y + , were then calculated from the relationship y+ = m'/m. The results are shown in fig. 1 in the form of plots of In y + - against the square root of the dimensionless ionic strength I, defined8 as I = i c z i m i i where zi is the charge number and mi is the dimensionless molality of each ionic species i.S. D. HAMANN 2543 Fig. and DHL - 1 . 5 I I I I I I 0 1 .o 2 .o 3.0 I4 1. Plots of In y k against I$ for KPi in water at 25 "C in the presence of (a) added LiCl (b) added Li,SO,. The numbers on the curves indicate the pressures in MPa and the lines labelled DHL represent the limiting Debye-Hiickel relationship [eqn (4)].EARLIER EXPERIMENTAL RESULTS Although there have been no previous direct measurements of ionic activity coefficients under pressure, it is possible to derive some values indirectly from volumetric data that Adams obtained for aqueous solutions of NaClll and of K2S04.12 Adams measured the densities and compressions of the solutions at 25 "C and at pressures up to 1000 MPa, at molalities m of 0.9-5.7 for NaCl and 0.15-0.64 for K,S04. From the results, he calculated 'fictive' specific volumes, which are pro- portional to the partial molar volumes of the salts. The partial molar volumes are related thermodynamically6-8 to the pressure dependence of the mean ionic activity coefficients, y + , - by the expression where R, T and p denote the gas constant, absolute temperature and pressure, respectively, v is the stoichiometric number of ions produced by dissociation of the electrolyte, V is its partial molar volume at the ionic strength of the solution and Vm is its extrapolated partial molar volume at zero ionic strength.The relationship (2) can be integrated to give In y$ -In y$ = ( I /vRT) J p ( V - P) dp (at constant T and composition) (3) Po2544 ACTIVITY COEFFICIENTS OF ELECTROLYTES Table 1. Partial molar volumes and activity coefficients of NaCl in water at 25 "C and at high pressures m = O m = 0.900 TI = 5.703 VW V In y$ - V In yP+ - p/MPa /cm3 mo1-I /cm3 molP In y l r$ /cm3 mol-1 In YS O.la 17.0 19.3 0 0.659b 23.3 0 0.95b 1 00 20.2 21.9 0.040 0.686 24.7 0.1 10 1.06 200 21.9 23.4 0.071 0.707 25.6 0.191 1.15 400 23.9 25.0 0.121 0.744 26.5 0.316 1.30 600 24.8 25.8 0.164 0.776 26.9 0.410 1.43 800 25.3 26.2 0.200 0.805 26.9 0.483 1.54 1000 25.5 26.3 0.233 0.832 26.9 0.544 1.64 a Atmospheric pressure.These two values of y l at atmospheric pressure were obtained by interpolation in the tables of Robinson and Stokes.' 0.5 0 -0.5 L / 1000 I / / L O O I I I I I - 0 . 5 -1 .o O.? 2L" -1.5. 4?O'DtiL I I I I 0 1 .o 2.0 3.0 14 Fig. 2. Plots of In y* against fi for (a) pure NaCl in water at 25 OC and (b) pure K,SO, in water at 25 "C. The numbers on the curves indicate the pressures in MPa, the lines labelled DHL represent the limiting Debye-Huckel relationship (4) and the curves marked DH are from the complete Debye-Huckel formula (6), with a = 0.48 nm.'S.D. HAMANN 2545 Table 2. Debye-Huckel parameters a and p for water and methanol at 25 "C and at high pressures water methanol p/MPa p / g ~ m - ~ E, a p/nrn-l p / g ~ m - ~ E, a plnrn-l 0.1" 0.9970 78.36 1.175 3.29 0.7867 32.80 3.854 4.51 100 1.0380 81.73 1.125 3.28 0.8514 35.68 3.533 4.50 200 1.0722 84.88 1.081 3.27 0.8922 37.73 3.327 4.48 400 1.1270 90.57 1.005 3.25 1000 1.2398 104.53 0.850 3.17 a Atmospheric pressure. where 7% and yP+ denote, respectively, the mean ionic activity coefficients at atmosph6ric pressure, p", and under a high pressure, p . Adams's values of V for NaCl, at the lowest and the highest concentrations that he studied, are shown in columns 3 and 6 of table 1, together with his extrapolated l2 P, in column 2.119 l2 If these quantities are inserted in eqn (3) and the integration is performed by Simpson's rule, they yield the values of In yP+ -In 7; that are listed in columns 4 and 7 of the table.Those differences, combined-with experimental values of y> taken from the tables of Robinson and Stokes,' give the high-pressure activity co&fficients 7% that are listed in columns 5 and 8 of table 1. The results for all the concentr&ons studied by Adams, but at just three high pressures, are shown in fig. 2, in the form of plots of In y+ against the square root of the ionic strength I ( I = m for NaCl and I = 3m for K2S0,). DISCUSSION It is seen in fig. 1 and 2 that the values of y + all increase with increasing pressure, particularly at high ionic strengths. At I = 4, for example, y+ for KPi in KPi + LiCl increases by 40% between atmospheric pressure (0.1 MP;) and 200 MPa.It is worthwhile to consider possible reasons for these trends. For very dilute solutions, the effect of pressure on In y+ can be predicted accurately from the Debye-Hiickel limiting law (DHL), providedthat the density, p, and the relative permittivity (dielectric constant), E,, of the solvent have been measured at high pressures. In Guggenheim's notation* the limiting Debye-Huckel formula is In?+ - = -alz+z-lfi (4) where z+ and z- denote the charge numbers of the ions of the electrolyte, I is the dimensionless ionic strength, defined in eqn (l), and a is a dimensionless parameter defined by ( 5 ) in which N is Avogadro's constant, e is the electronic charge, k is Boltzmann's constant, e, is the permittivity of a vacuum and Tis the absolute temperature.At any particular temperature the factor in square brackets is constant and the pressure dependence of a (and hence of In y + , since I is independent of pressure at constant composition) arises from the variation of the last term involving p and E,. Table 2 a2 = [(Ne6/32z2&; k3T3) (mol kg-l)] @ / E ; ) 83 FAR 12546 ACTIVITY COEFFICIENTS OF ELECTROLYTES shows some experimental values of p and E, for water and methanol at 25 "C and at several high pressures, p , along with derived values of a. The data for p and E , for water are taken from an earlier review1 and those for methanol are from the work of Ledwig and Wurflinger.13 Eqn (4) is represented in fig.1 and 2 by the straight lines labelled DHL, marked with the relevant pressures in MPa. In less dilute solutions, to I = 0.2, it might be expected that In y+ - would be described quite well by the complete Debye-Hiickel (DH) relationship In y+ - = - (al.z+z-lIi)/(1+ apIi) (6) where a is the distance of closest approach of the two ions and p stands for Guggenheim's8 factor 2a/s : p2 = ( 2 a / ~ ) ~ = [(2Ne2/&, k7) (mol kg-l)] @/E,). (7) Like a, varies with pressure in a calculable way, and table 2 lists some of its values at high pressures. Unfortunately, however, there is no a priori way of determining the quantity a in eqn ( 6 ) ; it is normally treated as an adjustable parameter chosen to fit the experimental values of In y+ at atmospheric pressure, but its behaviour at higher pressures cannot be predicted.'The curves labelled DH in fig. 2(a) are based on the simplest assumption that a is independent of pressure. A number of empirical relationships based partly on the Debye-Huckel formula have been proposed to describemoreconcentrated solutions(ofI > 0.2), but they all involve further parameters whose pressure dependences are not predictable. Fig.1 and 2 show that, although the limiting DHL lines do not of course have the right curvature, they do show the same kind of pressure dependence as the experimental results. The calculated values of In y + for particular values of I increase steadily with increasing pressure, and the reason for this lies in the increase of E, that occurs under pressure (see table 2).For water and most other solvents, E, is approximately proportional to p, and the factor PIE: in eqn ( 5 ) is nearly proportional to 1 / E ; , so that an increase of E, reduces a and raises y + . A physical interpretation of the experimental increase of y+ under pressure, at least at low ionic strengths, is therefore that an applied pressure raises the dielectric constant of the solvent and weakens the long-range electrostatic attractions of the ions. At the highest ionic strengths there may be a further contribution to the increase of y+ from the increased crowding of the ions under compression, with consequent enhancement of the repulsive part of the interaction free energy. Surprisingly, the full Debye-Hiickel theory, DH, is not as quantitatively satisfactory as the limiting approximation, DHL, in describing the influence of pressure in con- centrated solutions. This is demonstrated by the curves in fig.2(a) and, more clearly, in fig. 3, where values of In yP+ -In y: for NaCl at I = 5.703 are plotted against the pressure, p , assuming a in (6) to have a constant value of 0.48 nm.7 It is not obvious why the limiting law should predict the pressure effect more or less correctly, although it is quite wrong in its prediction of the absolute values of the activity coefficients at high ionic strengths. It must do so by accident. In fact, the apparent success of the limiting equation here is similar to its accidental success in describing the concentration dependence of the apparent molar volumes and partial molar volumes of concentrated 1 : 1 electrolytes in water at 25 OC.14 In that case, although the experimental volumes follow the limiting line quite closely at 25 "C, they deviate from it in opposite directionsl59l6 at lower and at higher temperatures, a fact which Millero16 explained in terms of a strongly temperature-dependent formation of ion pairs in the more concentrated solutions, without which the experimental volumes would always show large negative deviations from the limiting relationship.The sameS . D. HAMANN 2547 0.7 0.6 0.5 o + I 0.4 4 I 0 . 3 0.2 0.1 M 0 Fig. 3. Plots showing the influence of pressure on In y+ for NaCl in water at 25 "C, at an ionic strength of I = rn = 5.703. The dots represent experimental values derived from the volumetric measurements of Adams (see column 7 of table 1) and the curves are semi-theoretical ones, discussed in the text.type of explanation probably applies to the high-pressure results: an increase of pressure is known to enhance the dissociation of ion pairs1 and so it will tend to increase y + by more than the Debye-Hiickel theory predicts. Among the present salts, K,SO, is -particularly likely to show this effect because of the relatively strong association of K+ and SO!- ions. Fisher and have shown that an increase of pressure from atmospheric to 200 MPa roughly doubles the dissociation constant of K$3Oi- ion pairs in water at 25 "C, and that kind of change must certainly contribute to the trends of the curves in fig. 2(b). Included in fig. 3 are three curves based on empirical or semi-empirical relationships between y+ and p , taken from the literature.The curve labelled HO is calculated from a formula, (12412), that Harned and Owen6 obtained by inserting experimental values of partial molar volumes and partial molar compressibilities in equations equivalent to eqn (3) of this paper. The curve HKF is from eqn (173) of Helgeson, Kirkham and Flowers,l8 which is an extended form of the Debye-Huckel theory incorporating a term linear in I with a coefficient that varies with pressure in a manner determined empirically. Neither the HO nor the HKF relationship was intended to apply at very high pressures, and their failure here is understandable. The curve labelled HL is based on a formula proposed by Hamann and Linton19 [their eqn (5)], which has no theoretical basis except in very dilute solutions at low pressures, but which has the advantage of containing no adjustable parameters.Fig. 3 shows that it works quite well for NaCl at high concentrations and high pressures. That may be some justification for its use in other cases.l9 For non-aqueous solutions the pressure effects are likely to be considerably greater than for aqueous ones. This is suggested, for instance, by the values of a in table 2, which change under pressure by a much bigger factor for methanol than for water because of the proportionally larger increase of E , (which is characteristic of organic solvents). Some efforts were made in this work to measure y+ for KPi in methanol, in the presence of added LiCl. Although the results gave indications of large increases 83-22548 ACTIVITY COEFFICIENTS OF ELECTROLYTES of y+ under pressure, the experimental scatter was rather great, probably because the viscosities of the solutions became very high at the higher concentrations of LiCl, and equilibrium may not have been established.The solubility method is unsuitable for these solutions and it is hoped to make some measurements instead by an e.m.f. method. S. D. Hamann, in Modern Aspects of Electrochemistry, ed. B. E. Conway and J. O’M. Bockris (Plenum Press, New York, 1974), vol. 9, p. 47 ff. T. Asano and W. J. Le Noble, Chem. Rev., 1978,78,407. N. S. Isaacs, Liquid Phase High-pressure Chemistry (Wiley and Sons, New York, 1981). A. F. Collings, M. A. McCool and L. A. Woolf, Phys. Chem. Liquids, 1974,4, 65. S . Glasstone, Physical Chemistry (Van Nostrand, New York, 1946), p. 961. H. S. Harned and B. B. Owen, Physical Chemistry of Electrolyte Solutions (Reinhold, New York, 1963). E. A. Guggenheim, Thermodynamics (North Holland, Amsterdam, 5th edn, 1967). S. D. Hamann and I. W. McCay, AIChE J. 1966, 12,495. lo G. N. Lewis and M. Randall, J. Am. Chem. Soc., 1921,43, 1 1 12. L. H. Adams, J. Am. Chem. Soc., 1931,53, 3769. l2 L. H. Adams, J. Am. Chem. Soc., 1932, 54, 2229. l3 R. Ledwig and A. Wurflinger, 2. Phys. Chem. (N.F.), 1982, 132, 21. l4 0. Redlich and D. M. Meyer, Chem. Rev., 1964, 64, 221. l5 L. A. Dunn, Trans. Faraday Soc., 1968, 64, 2951. l6 F. J. Millero, J. Phys. Chem., 1970, 74, 356. l7 F. H. Fisher and A. P. Fox, J. Solution Chem., 1977, 6, 641. l8 H. C. Helgeson, D. H. Kirkham and G. C. Flowers, Am. J. Sci., 1981, 281, 1249. l9 S. D. Hamann and M. Linton, J. Chem. Soc., Faraday Trans. I , 1974, 70, 2239. ’ R. A. Robinson and R. H. Stokes, Electrolyte Solutions (Butterworths, London, 1959). (PAPER 3/2246)

ISSN:0300-9599    

DOI:10.1039/F19848002541

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. I , 1984, 80, 2549-2561 X-Ray Photoelectron Spectroscopic Study of the Surface of Borided Zirconium BY P. HUGH MIDDLETON,? ALAN J. PAUL AND PETER M. A. SEERWOOD* School of Chemistry, University of Newcastle upon Tyne, Newcastle upon Tyne NEl 7RU Received 29th December, 1983 X-ray photoelectron spectroscopy (X.P.S.) using depth profiling and angle-resolved studies has been used to investigate the surface of borided zirconium. The study shows that there is considerable surface oxidation which may seriously limit some practical applications of the borided sample. There has been considerable interest in the metal borides and related carbides and nitrides over the past thirty years.l They are all characterised by high melting points, hardness and chemical durability, while the metal borides have the added property of being electrical conductors.The uses of borides fall roughly into five areas: as refractories,l as catalysts in the hydrogenation of unsaturated hydrocarbons,2 as themionic emitter^,^ in case hardening of metals (particularly iron),* as cutting tools and abrasives' and as cathodes in the electrolytic production of al~minium.~ The electrical conductivity is particularly high in the transition-metal borides such as ZrB, and LaB,. ZrB, has a resistivity of 7-10 $2 cm' compared with the value of 1.7 pR cm for copper.6 The electrical conductivity has only been exploited in industrial electrodes and thermoionic emitters, and this study arose from an interest in ZrB, as a novel cathode material in electrochemistry. This boride was chosen because it forms a stable dodecaboride and diboride,l and the high boron content was thought to make for interesting electrochemical behaviour.The great hardness of the boride makes shaping of the material very difficult, requiring diamond-tipped tools or spark erosion; thus it was decided to prepare the material to the required shape by using the boriding (boronizing) process. The use of ZrB, prepared in this way as an electrode material requires an uncontaminated boride surface at the electrode/electrolyte interface. X-ray photoelectron spectroscopy (X.P.S.) is a well established technique for the study of electrode surface^.^^ There are a few X.P.S. and related studies of zirconium boride reported in the 1iteraturcg-l2 Studies of borides in general have shown the importance of surface oxidation of air-exposed sample^.^^ 13-18 In this paper we have investigated the surface of borided zirconium using X.P.S.combined with depth profiling and angle-resolved studies. EXPERIMENTAL All the spectra were obtained usinganAEI(Krat0s) ES200B X-ray photoelectron spectrometer, operated in the FRR mode, using unmonochromatised Mg Ka X-radiation. The data were collected using an Apple I1 microcomputer system linked to an IBM 370/168 computer, the latter being used for most of the data analysis and molecular-orbital calculations.1s The base t Present address: Corrosion and Protection Centre, UMIST, Manchester M60 1QD. 25492550 X.P.S. STUDY OF BORIDED ZIRCONIUM pressure in the sample chamber was ca.lo-* Tom.* Samples were etched using an Ion Tech B24 mechanically scanned saddle ion source operated at 5 kV and 2 mA. Calibration was based upon the C 1s electron peak (284.6 eV) due to residual carbon on the sample surface, with the internal calibrant provided by the Zr 3d5,, peak in ZrO, taken as 182.0 eV. Spectra were fitted using a non-linear least-squares method with a Gaussian/Lorentzian peak shape,20' 21 including the effect of radiation satellites. CNDO calculations were carried out using the program for transition-metal compounds discussed previously22 with 0.5 (I+A) values of 4.29, 1.89 and 4.30 eV for the 5s, 5p and 4d electrons of zirconium and niobium, values of 9 and 1 for zirconium and 9.28 and 1.75 for niobium for the 4d and 5s/5p electrons and Burn's exponents.23 Boron powder (95 % ,325 mesh) supplied by Alpha Chemical Ltd and zirconium sheet (99.9%, 1 mm thick) supplied by Goodfellow Metals Ltd were used.Zirconium boride was prepared using the boriding process.4' 24 This method uses the diffusion of boron atoms into the metal lattice at high temperatures, but significantly below the melting points of either component or boride. The metal plate used was cut out as 2 cm x 0.5 cm and placed in a layer of boron powder in a silica crucible. The composite was then compacted and placed in a Metals Research Vacseal VS2 furnace, evacuated to low4 Torr, flushed with high-purity argon at atmospheric pressure and heated to 1250 "C for 24 h while maintaining a steady flow of argon to remove any volatile matter.The furnace was then cooled to ambient temperature over a period of 12 h. The extracted borided sample was analysed after the removal of residual boron particles by light abrading with a Selvyt cloth. A sample of conventionally prepared boride was also studied. The boride samples were directly attached to a metal block in such a way that no part of the metal block was exposed to X-rays in the X.P.S. studies. Zirconium foil was mounted on a u.h.v. heatable, rotatable probe, and temperatures up to 600" C were achieved by radiant filament heating. Angle-resolved studies were achieved by varying the effective depth of the sample studied ( d ) with the angle of emergence of the photoelectrons (8)25 such that d represents a depth d sin 8 below the sample surface.Hence 8 = 90" corresponds to a bulk-sensitive angle and 0 = 5" is a surface-sensitive one. Boron samples were studied as powder mounted on double-sided Sellotape. RESULTS AND DISCUSSION It was important to record spectra from both boron and zirconium samples in addition to the study of the borides. These samples provided useful information about the various oxides that can be found on the surface of these materials. The binding energies of all the species encountered in this study are summarised in table 1. ZIRCONIUM-FOIL STUDIES Initial studies on zirconium foil revealed the presence of a tenacious oxide layer (ZrO,), as expected.2s Argon-ion etching failed to remove this layer and the resultant Zr 3d spectrum [fig. 1 (a)] shows mainly oxide, although small metal peaks are clearly discerna ble.Using residual carbon as calibrant gave an oxide Zr 3d5j2. binding energy of 182.00 eV with a 3d spin-orbit separation of 2.41 eV. It was decided to use this level as an internal calibrant for the boride studies. The 0 1s region for the ZrO, showed three peaks at 532.2, 53 1 .O and 529.4 eV corresponding to adsorbed oxygen and/or water, hydroxide and oxide, respectively. Scraping the foil surface produced large metal peaks in the Zr 3dregion, almost equal in intensity to those of the oxide. The metal 3d levels were shifted 4.25 eV relative to those of ZrO,, giving a binding energy of 177.75 eV for the 3d5,2 peak. The oxide overlayer was completely removed by heating in vacuo to 600 OC2' and * 1 Torr = 101 325/760 Pa.P.H. MIDDLETON, A. J. PAUL AND P. M. A. SHERWOOD 255 1 Table 1. Binding energies of compounds studied" treatment core level binding energy/eV f.w.h.m./eV assigned species B 1s B 1s B 1s 0 1s 0 1s 0 1s etched Zr 3d5p 0 1s 0 1s 0 1s scraped Zr 3dSl2 600 "C Zr 3d5/2 Zr 3s5/, 0 1s 0 1s Zr Zr 3 d 6 ~ 2 B 1s B 1s 0 1s B 1s B 1s B 1s 0 1s 0 1s 0 1s (a) boron powder 192.16 (0.04) 1.82 (0.10) 188.50 (0.03) 1.71 (0.03) 186.84 (0.03) 1.49 (0.03) 53 1.98 (0.06) 2.03 (0.04) 286.35 (0.04) 1.62 (0.02) (b) zirconium metal/oxide 182.00 (0.01) 1.46 (0.01) 532.16 (0.09) 1.57 (0.03) 53 1 .OO (0.09) 1.57 (0.03) 529.39 (0.03) 1.57 (0.03) 182.00 (0.01) 1.46 (fixed) 177.75 (0.02) 1.15 (fixed) 177.75 (0.01) 1.15 (0.01) 531.51 (0.19) 1.83 (0.16) 529.38 (0.07) 1.83 (0.16) 530.95 (0.20) 2.20 (0.11) (c) conventional zirconium boride 182.00 (0.01) 1.52 (0.03) 177.80 (0.05) 1.42 (0.06) 191.74 (0.11) 2.08 (0.10) 186.37 (0.08) 1.60 (0.13) 531.26 (0.21) 2.37 (0.35) 529.45 (0.46) 2.37 (0.35) ( d ) borided zirconium 182.00 (0.02) 1.65 (0.12) 177.86 (0.12)* 1.19 (0.1 1) 194.79 (0.56)* 2.92 (0.52) 191.82 (0.16)* 2.43 (0.36) 186.48 (0.10)* 1.48 (0.22) 533.65 (0.10)* 2.55 (0.09) 532.14 (0.33)* 2.55 (0.09) 529.46 (0.31)* 2.55 (0.09) (0.04) (0.08) (0.09) (0.08) (0.06) (0.09) (0.10) ZrO, -OH zfi2 Zr02 Zr Zr o/02/H20ads Oads Osub Zr02 ZrB, ZrB, Zr02 BOX O2/O/BOX ZrO, ZrB, BO BOX ZrB, Osub Zr02 a ads = Adsorbed species, sub = subsurface species.A greater degree of accuracy is attained when the calibrant and measured peak are within the same spectrum. The errors ( _+ 2 x standard deviation) are shown in parentheses.Starred values show ( k 2 x standard deviation) of binding energies averaged over a number of different spectra, which should be distinguished from the other values that refer to the accuracies of fitting single spectra.2552 X.P.S. STUDY OF BORIDED ZIRCONIUM n 165 180 binding energy/eV' Fig. 1. Zr 3d region for metal foil. (a) Argon-ion etched surface at room temperature; (b) at 600 "C after argon-ion etching; ( c ) as (a) but with a non-linear background removed. recording the spectrum at this temperature [fig. l(b)]. The oxide layer is removed at high temperature by the diffusion of oxygen from the oxide film into the bulk of the metal. The 0 1s spectrum showed two peaks at 531.5 and 529.4eV. The lower- binding-energy species does not correspond to oxide since the associated Zr 3 d levels would have been easily observable.The oxygen peaks are thought to correspond to both subsurface and chemisorbed o x ~ g e n . ~ ~ - ~ ~ The distinct asymmetry of the metal 3 d peaks is quite apparent and an exponential tail was included in the Gaussian/Lorentian fitting function to account for this. The asymmetric lineshape is not considered to arise from chemically shifted Zr 3 d intensity (due to the presence of subsurface and chemisorbed oxygen) since we have found, in many studies, such asymmetry to be associated with conducting species. Thus this feature always corresponds to a fixed percentage of the metal-peak intensity andP. H. MIDDLETON, A. J. PAUL AND P. M. A.SHERWOOD 2553 correspondingly varies in intensity as the metal peak varies when other chemically shifted peaks (from oxides etc.) are also present. Peak asymmetry has been widely noted in metallic systems and arises from the coupling of the final state hole to the conduction e l e c t r o n ~ . ~ ~ - ~ ~ We~theim~~ has shown the importance of including asymmetry effects in quantitative analysis. However, there is some doubt concerning the ability to distinguish between asymmetric broadening and inelastic background intensity.36 Inelastic background removal has been employed by a number of w o r k e r ~ ~ l - ~ ~ and fig. 1 (c) shows the effect of background on the metal Zr 3d spectrum. It is a prerequisite of this technique that the extreme ends of the spectral region are a true representation of the background alone.The magnitude of the calculated asymmetric tail at the high-binding-energy end of fig. 1 (b) is extremely small and will have little adverse effect on the background-subtraction method. It was not possible to use a spectral start point at even lower binding energy as the spectral intensity gradually rises because of possible plasmon excitation. It is clear from fig. 1 (c) that the overall line shape is still well fitted by the use of an exponential tail in the fitting function. There is, however, less asymmetry and therefore less intensity associated with the metal 3d levels. Background removal would therefore appear to be important in quantitative analysis, although the danger of removing true signal intensity is always present. In the case of the borides (vide infra) it was not possible to use inelastic background removal owing to the considerable complexity of the spectra. BORON STUDIES Boron powder gave a B 1s spectrum [fig.2(a)] which showed three peaks at 192.2, 188.5 and 186.8 eV corresponding to B,03, elemental boron and boron carbide (B4C). The oxide and carbide peaks were substantially reduced after a 10 min etch, indicating that they represented a considerable surface impurity on the boron particles. The 0 1s spectrum showed two peaks at 532.0 and 53 1 .O eV corresponding to adsorbed oxygen and/or water and B203. There is a range of B,03 B 1 s binding energies given in the literature (1 9 1.5-1 93.5 eV) using various calibrant~.~'-~~ Our B203 value is close to that of Wheeler.48 Our carbide value shows a separation of 1.7 eV from elemental boron, which we tentatively assign to B4C.This is supported by the presence of a corresponding carbide peak at 286.3 eV in the C 1s region.5o BORIDE STUDIES The B ls/Zr 3d spectrum of the conventionally prepared boride (ZrB,) sample is shown in fig. 2(6). The spectrum shows peaks due to boride (Zr 3d5,2 = 177.8 eV, B 1s = 186.4 eV), ZrO, (Zr 3d5,, = 182.0 eV) and Box (oxidised boron, B 1s = 191.7 eV). The oxidised boron (Box) has been observed at similar B 1s binding energies of many other metal b~rides.~? 48 The 0 1s region showed peaks at 53 1.3 and 529.4 eV which are assigned to adsorbed oxygen and ZrO,. There will be a small amount of oxygen intensity associated with Box which would be expected to lie under the 531.3 eV peak.The boride peaks were fitted using an exponential tail due to conduction-band interaction, as expected for conducting species. l3? 33-40 In addition to the main peaks described above the spectrum showed oxide satellite peaks at 14-15 eV higher binding energy from the ZrO, 3d peaks. Rao51 suggests a plasmon excitation process as a possible origin for these peaks. STUDIES OF THE BORIDED SAMPLE The sample of borided zirconium was analysed by depth profiling and recording spectra at different sample angles. Spectra were recorded at a bulk-sensitive angle,2554 X.P.S. STUDY OF BORIDED ZIRCONIUM A n 1 I I 1 I I 8 I 190 180 binding energy/eV Fig. 2. Spectra of reference compounds. (a) B 1s region for boron powder.(b) Zr 3d/B 1s region for conventionally prepared sample of ZrB,. 8 = 90, for all etch times and 8 = 5 and 45" after etch times of 45 and 165 min, respectively. It was apparent from the metal/oxide study that the metal 3d spectral peaks have a larger associated inelastic tail than those of the oxide. Because of differences in the associated inelastic tail one cannot use inelastic background subtraction to analyse effectively metal/oxide and indeed metal/boride mixtures. The background method cannot discriminate between varying inelastic contributions for different species in the spectra. The background routine depends on the variation of the total peak(s) area within a spectrum, and therefore the application of this method to analyse the boride spectra would be invalid.Fig. 3 shows the changes in the Zr 3d/B 1s spectra at 8 = 90" for all etch times. The Zr 3d5,, peaks had binding energies of 182.00 and 177.86 eV due to ZrO, and ZrB,. Careful fitting showed that the exponential tail on the borides was a real eKect, as expected. Joyner and Hercules have demonstrated considerable asymmetry effects in the 2p spectra of Fe, Fe,B and FeB.13P. H. MIDDLETON, A. J. PAUL AND P. M. A. SHERWOOD 2555 30 190 180 I I I 190 180 binding energy/eV Fig. 3. Variation of Zr 3d/B 1s region for borided zirconium. Spectra were recorded at 0 = 90" after etch times of: (a) 10, (b) 45, (c) 105 and (d) 165 min. A linear background was adopted ignoring any satellite structure from both oxide and boride peaks. A small amount of intensity would be expected for the satellites above the rising linear background, since the 3d5/2 satellite peak occurs at a binding energy of 196-197.0 eV.It was not possible to determine confidently the amount of satellite intensity above the background although its effect on the calculated intensity ratios would undoubtedly be small. Were it possible to evaluate correctly the spectral inelastic profile then the satellite intensities could be accounted for. The spectra showed B 1s peaks at 194.8, 191.8 and 186.5 eV. The two latter peaks are assigned to 'oxidised boron' and ZrB,. Alyeshin has reported the existence of an anomalous combination of boron and oxygen B,O, (binding energy = 194.8 eV) on the surface of borided iron powders.lB Zint15, has shown that high-temperature reactions of boron with alumina (1 300 "C) and ZrO, (1800 "C) produced boron monoxide (BO).This is the most commonly occurring boron s ~ b o x i d e ~ ~ although a wide variety of such species have been reported, e.g. B,O, BI3O2, B30, B,05 etc. The production of BO (empirical formula) by high-temperature reaction of metal oxides with elemental boron has been well established . 53-55 The zirconium sample used in this work would have possessed a film of ZrO, on the surface, and the reaction with boron at 1200 "C would be expected to produce BO (empirical formula); the peak at 194.8 eV is therefore assigned to this species. The 0 1s regions for all etch times (0 = 90°) are very similar in that they show three peaks, and the 10 and 105 min etch spectra are shown as examples [fig.4(a) and (b)].2556 X.P.S. STUDY OF BORIDED ZIRCONIUM 535 530 binding energy/eV Fig. 4. Selection of 0 1s spectra for borided zirconium. Etch times and angles are as follows: (a) 10 min and 8 = 90°, (b) 105 rnin and 8 = 90°, (c) 165 rnin and 8 = 90°, (d) 165 rnin and 8 = 45" and (e) 165 rnin and 8 = 5". The binding energies at 533.6, 532.1 and 529.5 eV correspond to subsurface oxygen, a mixture of surface boron oxides and adsorbed oxygen, and ZrO,. Clearly the highest-binding-energy species dominates the spectrum. Angle-resolved results for the 165 rnin etched sample [fig. 4(c)-(e)] show considerable changes in relative peak intensities. The highest-binding-energy species is mainly located in the bulk and is therefore assigned to dissolved oxygen.The presence of this peak is accounted for by the collapse of the oxide film at high temperature in addition to oxygen impurities originating from the argon atmosphere during the boride preparation. The considerable dissolution of oxygen in zirconium at elevated temperatures has been well 56-58 The other oxygen peaks appear to arise from species located close to the surface, as expected for the surface oxides of boron and zirconium.P. H. MIDDLETON, A. J. PAUL AND P. M. A. SHERWOOD 2557 190 190 180 binding energy/eV Fig. 5. Angular variation in Zr 3d/B 1s region for borided zirconium. Etch times and angles are as follows: (a) 45 min and 8 = 90", (b) 45 rnin and 8 = 45", (c) 45 min and 8 = 5", ( d ) 165 min and 8 = 90" and (e) 165 min and 8 = 5" with smoothed spectrum included.Angular variation studies for the Zr 3d region are shown in fig. 5 and clearly show BO and Box to lie on the surface. It would appear that Box is caused by surface oxidation of BO, possibly through the etching process, although it would not be unreasonable to expect other oxides of boron to be formed in the sample preparation. However, the large increase in the relative Box/BO intensity ratio [fig. 5(e)] favours oxidation by the etching process. The smoothed overlayer of this shows the changes in relative peak intensities more clearly. The presence of a small peak at the highest-binding-energy side of the most surface-sensitive spectra [fig. 5 (c) and (e)] may correspond to some other boron suboxide, produced by the argon-ion etching.Fig. 6 is a graph of the change in relative intensities with etch times for spectra recorded at 90°, and fig. 7 shows the variation of individual peak areas with respect to the total area in angle-variation studies of each species in the spectrum after 45 and 165 min.2558 X.P.S. STUDY OF BORIDED ZIRCONIUM f0 45 105 165 time/min Fig. 6. Graph showing variation of individual peak areas with respect to total area for each species in the Zr 3d/B 1s region as a function of etch time. 0, BO; 0, Box; ., ZrO,; A, Zr 3d of ZrB, and A, B 1s of ZrB,. The variations in relative intensities with both etch time and angle lead to an approximate representation of the surface as shown in fig. 8. The results clearly show the boron oxides located on the immediate surface, although some BO concentration is found as one penetrates the layers.There would appear to be considerable mixing of the zirconium oxide and boride layers, as was found for the conventionally prepared boride where in fact the oxide dominated. However, the boride/oxide ratio increases with both etch time and 8 and shows a greater proportion of boride with increasing distance into the sample. Approximate boride (ZrB,) stoichiometries can be calculated using appropriate intensity expression^^^^ 6o and cross-section information provided by Scofield.61 They range from x = 3.5 to 4.0 for the borided sample in most cases as compared with 2.4 for the conventionally prepared boride, but these figures may be distorted by differential sputtering effects. BORIDE BONDING There have been several papers on the nature of the bonding in ZrB, and borides in genera1.13* 62-64 There is some debate about the extent of charge transfer, which has r e ~ e n t l y l ~ 9 ~ ~ been concluded to be very small between the metal and boron centres. This suggestion is confirmed by a CNDO relaxation-potential model calculation22 which shows very small charge transfer (average B charge = -0.0303 and H charge = - 0.01 30) for a ZrB,,H,, neutral unit and gives a slightly negative chemical shift (-0.06 eV) with respect to the metal, which is close to the observed slightly positive (0.05 eV) chemical shift.70- 60- 50- 40- 30.- 20- 10..70- 60.- 50.- 40- 30.. 20- 10.- P. H. MIDDLETON, A. J. PAUL AND P. M. A. SHERWOOD 5 45 90 8 I" I I 5 45 90 0 I" 2559 Fig.7. Angular variation of individual peak area with respect to total area for each species in the Zr 3d/B 1s region after etch times of (a) 45 and (b) 165 min. Symbols as in fig. 6. Fig. 8. zirconium.2560 X.P.S. STUDY OF BORIDED ZIRCONIUM CONCLUSIONS The presence of surface oxidation on fabricated borides is clearly demonstrated in this work. This oxide formation must restrict the application of such borides when the surface must reflect the full boride properties, as in electrochemical applications. We thank the S.E.R.C. and the Royal Society for support. R. Thompson, in Progress in Boron Chemistry, ed. R. J. Brotherton and H. Steinberg (Pergamon Press, Oxford, 1970), vol. 2, chap. 5. Y. Okamoto, Y. Nitta, T. Imanaka and S. Teranishi, J. Chem. SOC., Faraday Trans.1, 1979,75,2027. J. M. Lafferty, J. Appl. Phys., 1951, 22, 299. W. Hodge, R. M. Evans and A. F. Haskins, J. Metals, 1955, 7, 824. C. E. Ransley, J. Metals, 1962, 14, 129. Handbook of Chemistry and Physics (The Chemical Rubber Co., Cleveland, Ohio, 1982-83), p. E-72. P. M. A. Sherwood, in Contemporary Topics in Analytical and Clinical Chemistry, ed. D. M. Hercules, G. M. Hieftje, L. R. Snyder and M. A. Evenson (Plenum Press, New York, 1982), vol. 4, pp. 205-293. P. M. A. Sherwood, Am. Lab., 1983,42, 1521. G. Mavel, J. Escard, P. Costa and J. Castaing, Surf. Sci., 1973, 35, 109. lo H. Ihara, M. Hirabayashi and H. Nakagawa, Phys. Rev. B, 1977, 16, 726. l1 T. Tanaka, Y. Ishizawa, E. Bannai and S. Kawai, Solid State Commun., 1978, 26, 879. l2 D. L. Johnson, B.N. Harman and S. H. Liu, J. Chem. Phys., 1980,73, 1898. l 3 D. J. Joyner, 0. Johnson and D. M. Hercules, J. Am. Chem. SOC., 1980, 102, 1910. l4 A. Berrada, J. P. Mercurio, J. Etourneau and P. Hagenmuller, Rev. Int. Hautes Temp. Refract., 1978, l5 J. N. Chazalviel, M. Campagna, G. K. Wertheim and P. H. Schmidt, Phys. Rev. B, 1976, 14, 4586. l6 A. Lebugle and G. Montel, Rev. Int. Hautes Temp. Refract., 1974, 11, 231. l7 R. J. Colton and J. W. Rabalais, Inorg. Chem., 1976, 15, 236. l8 V. G. Alyoshin, A. I. Kharkmov and V. M. Prokopenko, J. Solid State Chem., 1981, 38, 105. l9 P. M. A. Sherwood, J. Microsc. Spectrosc. Electron., 1980, 5, 475. 2o R. 0. Ansell, T. Dickinson, A. F. Povey and P. M. A. Sherwood, J. Electroanal. Chem., 1979, 98, 79. 21 P. M. A. Sherwood, in Practical Surface Analysis by Auger and Photoelectron Spectroscopy, ed.D. Briggs and M. P. Seah (J. Wiley, London, 1983), pp. 445475. 22 P. M. A. Sherwood, J. Chem. SOC., Faraday Trans. 2, 1976, 72, 1791. 23 G. Burns, J. Chem., Phys., 1964, 42, 1521. 24 R. Thompson, in Borides: their Chemistry and Applications (R.I.C. Lecture Series, London, 1965), 25 R. 0. Ansell, T. Dickinson, A. F. Povey and P. M. A. Sherwood, J. Electron Spectrosc. Relat. 26 B. W. Veal, D. J. Lam and D. G. Westlake, Phys. Rev. B, 1979, 19, 2856. 27 N. N. Berezina, D. P. Valyukhov and E. S. Vorontsov, Russ. J. Phys. Chem., 1981, 55, 1650. 28 H. Y. Hall and P. M. A. Sherwood, J. Chem. SOC., Faraday Trans. I , 1984, 80, 135. 29 A. Proctor and P. M. A. Sherwood, Sur. Interface Anal., 1980, 2, 191.30 S. Evans and J. M. Thomas, J. Chem. SOC., Faraday Trans. 2, 1975, 71, 313. 31 M. W. Roberts, Pure Appl. Chem., 1981, 53, 2269. 32 P. R. Norton, J. Catal., 1975, 36, 211. 33 S. Doniach and M. Sunjic, J. Phys. C, 1970, 3, 285. 34 P. Nozieres and C. T. De Dominicis, Phys. Rev., 1969, 178, 1097. 35 S. Hufner, G. K. Wertheim, D. N. E. Buchanan and K. W. West, Phys. Lett., 1974,46A, 420. 36 G. K. Wertheim and S. Hufner, J. Inorg. Nucl. Chem., 1976, 38, 1701. 37 G. K. Wertheim and S. Hufner, Phys. Rev. Lett., 1975, 35, 53. 38 S. Hufner and G. K. Wertheim, Phys. Rev. B, 1975, 11, 678. 39 S. Hufner and G. K. Wertheim, Phys. Rev. B, 1975, 11, 5197. 40 E. Antonangeli, A. Balzarotti, A. Bianconi, P. Perfett, P. Ascarelli and N. Nistico, Solid State 41 D. A. Shirley, Phys.Rev., B, 1972, 5, 4709. 42 M. 0. Krause, T. A. Carlson and R. D. Dismukes, Phys. Rev., 1968, 170, 37. 43 D. W. Fischer, Adv. X-ray Anal., 1969, 13, 159. 15, 115. no. 5. Phenom., 1977, 11, 301. Commun., 1977, 21, 201.P. H. MIDDLETON, A. J. PAUL AND P. M. A. SHERWOOD 256 1 44 A. Barrie and F. J. Street, J. Electron Spectrosc. Relat. Phenon., 1975, 7, 1. 45 N. S. McIntyre and D. G. Zetaruk, Anal. Chem., 1975, 49, 1521. 46 A. Proctor and P. M. A. Sherwood, Anal. Chem., 1982,54, 13. 47 D. J. Joyner and D. M. Hercules, J. Chem. Phys., 1980,72, 1095. 48 D. R. Wheeler, J. Vac. Sci. Technol., 1978, 15, 24. 49 D. N. Hendrickson, Znorg. Chem., 1970,9, 612. 50 D. T. Clark and A. Dilks, J . Polym. Sci., Polym. Chem. Ed., 1978, 16, 791. 51 D. D. Sarma and C. N. R. Rao, J. Electron Spectrosc. Relat. Phenom., 1980, 20, 25. 52 E. Zintl, W. Morawietz and E. Z. Gastenger, Z. Anorg. Chem., 1940, 245, 8. 53 D. Nicolls, in Mellor’s Comprehensive Treatise on Inorganic and Theoretical Chemistry (Macmillan, 54 A. W. Searcy and C. E. Myers, J. Phys. Chem., 1957, 61, 957. 55 A. W. Searcy, University of California, Radiation Laboratory Report, 1951, UCRL-1404. 56 L. M. Kovba, E. M. Kenina, I. I. Kornilov and V. V. Glazova, Dokl. Akad. Nauk SSSR, 1968, 180, 57 R. F. Domagala and D. J. McPherson, J. Metals, 1954, 6, 238. 58 G. N. Krishnan, B. J. Wood and D. Cubicciotti, J. Electrochem. SOC., 1981, 128, 191. 59 A. Proctor and P. M. A. Sherwood, Anal. Chem., 1980,52, 2315. 6o T. A. Carlson and G. E. McGuire, J. Electron Spectrosc. Relat. Phenon., 1972/3, 1, 259. 61 J. H. Scofeld, J. Electron Spectrosc. Relat. Phenom., 1976, 8, 129. 62 D. J. Joyner, Phys. Rev. B, 1981, 24, 3122. 63 V. G. Aleshin, T. Ya. Kosolapova and V. Nemoshkalenko, J. Solid State Chem., 1981, 38, 105. 64 V. G. Aleshin, T. I. Serebryakova and A. I. Kharlamov, Phys. Status Solidi B, 1977, 83, 537. London, 1962), vol. V, part A, sect. A5. 360. (PAPER 3/2271)

ISSN:0300-9599    

DOI:10.1039/F19848002549

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1984,80, 2563-2566 Tracer-diffusion Coefficients of Water in Aqueous Tetra-alkylammonium Chloride Solutions BY KAZUKO TANAKA The Institute of Physical and Chemical Research, Wako-shi, Saitama 35 1, Japan Received 3rd January, 1984 The tracer-diffusion coefficients of water in aqueous solutions of tetramethylammonium chloride, tetraethylammonium chloride and tetra-n-propylammonium chloride have been determined at 298.2 K by the diaphragm-cell method using deuterium as a tracer. Observed retardation of water diffusion is discussed in terms of the obstruction effect by applying Wang's theory. The experimental values of tracer-diffusion coefficients of water in aqueous solutions provide important information concerning the effects of ions on water structure, such as ion-solvent interactions.Retardation and acceleration of water diffusion, caused respectively by electrostatic hydration and by distortion of the water structure around ions, were discussed in a previous paper.' However, in addition to these, the effect of obstruction on the path taken by water molecules when diffusing may not be negligible when molecules of a relatively large size compared with water are introduced into the solutions; this may result in a retardation of water diffusion. In order to obtain information concerning this obstruction effect, the tracer-diffusion coefficients of water in the presence of tetramethylammonium chloride (Me,NCI), tetraethylammonium chloride (Et,NCl) and tetra-n-propylammonium chloride (PrtNCl) have been determined at 298.2 K by the diaphragm-cell method using deuterium as a tracer. Systematic data on tracer-diffusion coefficients of tetra- alkylammonium ions, bromide ions and water in aqueous tetra-alkylammonium bromides were reported recently.2 Although we hesitate to report yet another set of data, our experimental values for Me,NCI are slightly different from the values reported for the corresponding bromide.The results are discussed from the standpoint of the size effect of tetra-alkylammonium ions by applying Wang's the01-y.~ EXPERIMENTAL Tetra-alkylammonium chlorides were obtained from Eastman Kodak, recrystallized once from an appropriate solvent according to literature methods4 and dried in vacuo at 50 "C for a few days before use. D,O (99.7 atom % deuterium) used as a tracer was obtained from Junsei Pure Chemicals and diluted to 2 atom % with normal water, which was distilled three times.The diffusion measurements were made with a diaphragm cell which incorporated a conventional sintered glass disc. Details of the experimental procedure were as described in previous papers.'I5 The cell was calibrated by diffusion measurements on 0.5 mol dm-3 potassium chloride solutions diffusing into pure water at 298.2 K together with Stokes' data for the system.s The concentrations of potassium chloride were determined by weighing a potassium chloride residue obtained from evaporation of experimental solutions to dryness. 25432564 TRACER-DIFFUSION COEFFICIENTS OF WATER Table 1. Tracer-diffusion coefficients of water in aqueous solutions of Me,NCI, Et,NCl and PryNCl at 25 "C D,/ 1 OP9 m2 s-l C/mol dmP3 Me,NCl Et,NCl PrtNC1 0 2.24 2.24 2.24 2.18 0.05 - 0.1 2.22 2.19 2.1 1 0.2 2.21 2.13 1.96 0.5 2.16 1.98 - - The water isotope contents were determined densitometrically by means of an Anton Paar digital precision density meter.The water in the sample solutions was separated from the added electrolytes by distilling the solution twice, first under reduced pressure and secondly under normal pressure, before making the density measurements. Details of the density measurements were as described previ~usly.~ RESULTS AND DISCUSSION The tracer-diffusion coefficients of water in aqueous solutions of Me,NCl, Et,NCl and PrYNCl at several concentrations at 298.2 K are shown in table 1, Each value listed is the average of 3-5 measurements and the mean deviation from the average is < 1 %, From a comparison of these values with values given in the literature,2 good agree- ment was observed for water diffusion in the presence of 0.2 and 0.5 mol dm-3 Et4NC1 and 0.05 and 0.1 mol dm-3 PryNCl.However, a difference was observed in case of Me,NCl and Me4NBr;2 a decrease in the water diffusion with increasing concentration of Me,NCl was observed, while a maximum appeared in the presence of Me,NBr around 0.25 mol dm-3; the difference is presumably due to the effect of the anion on water diffusion. In fig. 1 the relative diffusion coefficient of water (D,/D",, where D, is the diffusion coefficient of water in the presence of an electrolyte and DG is that in pure water, is plotted against the concentration of electrolyte together with those in the presence of several other monovalent metal chlorides cited from our previous paper.l Over the concentration range investigated, water diffusion in the presence of tetra-alkylammonium chlorides decreases almost linearly with increasing electrolyte concentration.Although the retardation of water diffusion by Me,NCl was similar to that by alkali-metal chlorides such as LiCl and NaC1, more significant effects on water diffusion were observed in the presence of Et,NCl and PryNCl. Since chloride ions are assumed to have little effect on water diffusion,l*' the observed effects on water diffusion caused by these electrolytes are attributed mainly to the cations. The retardation effects of electrolytes on water diffusion may be caused mainly by the following reasons : (a) electrostatic hydration, (b) hydrophobic hydration and ( c ) obstruction effects.In the last, the diffusion path of the water molecule is blocked by large ions, and water molecules near the large ions have to diffuse along longer paths; this decreases the diffusion coefficient of water, even if the water molecule moves at the same speed as in pure water. In our previous paper' it was shown that the decrease in the diffusion coefficient of water with increasing concentration of sodium ion was explained in terms of electro- static hydration. The diffusion of water in the presence of Et,N+ and Pr;N+, how- ever, may be too slow to be explained by the electrostatic hydration effect alone.K.TANAKA l U i l 1 1 ,oo 0 % !? 0,95 d On90 2565 0 0,2 0,4 0,6 0,8 cfmol Fig. 1. Tracer-diffusion coefficients of water in aqueous solutions of several electrolytes at 25 "C. (a) NH4C1 and KC1, (b) NaCl, (c) Me4NC1, ( d ) LiCl, (e) Et4NCl and cf) PrtNC1. Wang3 proposed a theoretical equation for water diffusion in the presence of large molecules (such as a protein) whose Brownian motion is negligible: D, = Dk(1 -a+) (1) where 4 is the volume fraction of the obstructing particle and a is 1.5 when the particle is spherical. Assuming that eqn (1) may serve as a useful guide for small obstructing molecules or ions, we applied eqn ( 1 ) to the case of tetra-alkylammonium ions. In fig. 2 relative diffusion coefficients (D,/Dk) are plotted against the volume fraction of tetra-alkylammonium ions together with Wang's theoretical line of a = 1.5.The volume fraction of the tetra-alkylammonium ions was evaluated from the partial molar volumes of these ions and from their Pauling crystal radii using an equation proposed by Conway et aZ.;4 the former points are given by open circles and the latter by solid circles, respectively, in fig. 2. In the case of Et,N+, the theoretical line was sandwiched between the experimental points, as shown in fig. 2. A positive deviation for Me4N+ and a negative deviation for Pr:N+ from the theoretical line were observed. Since there exist several sources of experimental evidence (e.g. the viscosity B-coefficient* and the heat capacity of transfer from H,O to D209) which support the finding that Et,N+ has little overall effect on water structure, the main cause of the retardation of water diffusion in the presence of Et,N+ is the obstruction effect.It can thus be concluded that Wang's theory is applicable to smaller particles such as tetra-alkylammonium ions. The positive deviation from the theoretical line observed for Me,N+ may indicate either that this ion is not sufficiently large for an estimate2566 TRACER-DIFFUSION COEFFICIENTS OF WATER 0 2 4 6 8 10 lo2 l$ Fig. 2. Relationship between water diffusion and volume fraction of tetra-alkylammonium ions at 25 "C. (---) Wang's theoretical line; 0, volume fraction calculated using the Pauling crystallographic radius; , volume fraction calculated using partial molar volume.(a) Me,N+, (b) Et,N+ and (c) PryN+. of the obstruction effect to be made from Wang's theory, or that it acts as a so-called structure-breaking ion, so that water molecules around the ion diffuse faster than those of bulk water, thus compensating for the obstruction effect. The negative deviation observed for PrzN+ may suggest that water diffusion in the presence of this ion being slower than that expected from theory is due to hydrophobic hydration, in which water molecules in the vicinity of the ion move slower than in bulk water. I thank Drs W. Y. Wen and R. Tamamushi for valuable suggestions and dis- cussions during this work. K. Tanaka, J. Chem. Soc., Faraday Trans. I , 1975, 71, 1127. L. A. Woolf and H. Weingartner, Faraday Symp. Chem. Soc., 1982, 17, 41. J. H. Wang, J. Am. Chem. Soc., 1954,76,4755. B. E. Conway, R. E. Verrall and J. E. Desnoyers, Trans. Faraday Soc.. 1966. 62, 2738. K. Tanaka, J . Chem. Soc., Faraday Trans. I , 1978, 74, 1879. R. H. Stokes, J. Am. Chem. Soc., 1951, 73, 3527. L. Endon, H. G. Hertz, B. Thul and M. D. Zeidler, Ber. Bunsenges. Phys. Chem., 1967, 71, 1008. J. E. Desnoyers and G. Perron, J. Solution Chem., 1972, 1, 199. J. L. Fortier, P. R. Philip and J. E. Desnoyers, J . Solution Chem., 1974, 3, 523. (PAPER 4/0 14)

ISSN:0300-9599    

DOI:10.1039/F19848002563

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1984, 80, 2567-2579 The Structure of Ethylene Polymerisation Catalysts? Part 1 .-Fast-atom Bombardment-Secondary-ion Mass Spectrometry Studies of Chromium Clusters BY ALAN ELLISON School of Science, Humberside College of Higher Education, Cottingham Road, Kingston-upon-Hull HU6 7RT Received 1 1 th January, 1984 Chromium(6 + ) oxide was supported on y-alumina and silica by aqueous impregnation producing two series of samples (SiO,, 1-18% Cr; A1,0,, 1-12% Cr) dried by vacuum desiccation at 25 "C and calcined in air at temperatures from 100 to 600 "C. Fast-atom bombardment-secondary-ion mass spectroscopy studies have revealed contrasting development of chromium oxide clusters, so that clusters on silica are thin and widely spread relative to clusters of uneven thickness on alumina.These surface models are explained in terms of the long-range chromium oxide-support interaction evident in alumina-supported material. Chromium compounds deposited on alumina, silica-alumina and silica have been extensively investigated by a wide range of spectroscopic and magnetic techniques. In addition to their intrinsic interest, these materials are important as catalysts, particularly in the polymerisation of ethene. There is still great disagreement in the literature regarding the nature of the active site responsible for ethene polymerisation, and every chromium oxidation state from (2 +) to (6 +) has been cited. Early studies1 suggested that reduced catalysts were more active than unreduced samples. Later, kinetic and spectroscopic studies identified activity with high chromium oxidation states, (4+)2 or ( 5 +),3f or with mixed-valence phases of Cr3+ and Cr6+.59 These ' oxidised-phase' catalysts required an induction period prior to becoming active towards polymeri~ation.~7 In complete contrast, by reducing the chromium with carbon monoxide, Krauss and c ~ w o r k e r s ~ - ~ ~ established the role of Cr2+ as the active chromium oxidation state.Kazansky et aZ.13 claimed, however, that Cr3+ compounds were more active than Cr2+ compounds and that catalysts prepared from Cr6+ were more active after mild reduction in carbon monoxide than after extensive reduction. Beck and Lunsford14 proposed that specific Cr3+ sites were active as isolated ions, coordinated only to the silica lattice used as support and showing considerable coordinative unsaturation (cus).Current l6 apparently suggests that three cus--Cr2+ sites exist on carbon monoxide-reduced chromium on silica differing in their degree of cus-quality and in their polymerisation activity. Each relationship between activity and a specific oxidation state has been proposed from a good data base. The apparently confused situation may be rationalised by the mechanistic demand for facile redox behaviour of the supported chromium, as described in the reaction scheme of Zakharov17 and Ermakov.l* Meanwhile, the usually unstable chromium species must acquire stability from their environment. t Invited paper presented at the 184th A.C.S. National Meeting, Division of Colloid and Surface Chemistry, September 1982, Kansas City, U.S.A.25672568 CHROMIUM FAB-SIMS It is usual to argue that both Cr5+ and Cr2+ species occur as isolated ions, gaining stability from crystal-field effects. It is not obvious, however, that the parallel propositions of crystal-field stabilisation and of coordinative unsaturation can be reconciled, while there remains considerable doubt that regular crystal-field sites exist at the surfaces of these supports. To circumvent this difficulty, it has been claimedlg that, even for silica-supported chromium, a solid solution of chromium ions occurs in the support, despite experimental evidence that chromia and silica are cpmpletely immiscible, even in the liquid state, and that chromium ions do not find stability within the silica lattice.20-26 The assumption of isolation for species on these surfaces is perhaps based upon circumstantial evidence.The current literature shows that at concentrations as low as 0.5% chromium on silica, considerable clumping or clustering of the supported chromium occurs, so that two- or three-dimensional aggregates of chromium oxide pred~rninate.~' Krauss15 and Ghiotti and Garrone16 now suspect that ion pairs of Cr2+, Cr3+ or some combination of Cr2+ and Cr3+ contribute towards polymerisation activity. Rebenstorff and Larsson2* consider that Cr2+ species in reduced catalysts are active entirely as ion pairs and they propose a similar model for active Cr3+ catalysts. to be associated with a clumped, mixed-valence phase of chromium. Indeed, 0' Reilly and MacIver30 originally concluded that y-resonance arose from trapped electrons on Cr6+ centres which were partially associated with /?-phase resonance, i.e.with clusters of Cr3+. It would appear therefore that ethene polymerisation activity can be associated with the following: the chromium oxide surface; Cr3+, Cr6+ and y-phase chromium in oxidised systems; Cr2+ and Cr3+ in reduced systems; or the chromium occurring as small clusters. This paper describes the use of secondary-ion mass spectrometry (SIMS) to study the extent to which the supported chromium exists as clusters or aggregates on the surface of alumina and silica. The use of electron spin resonance spectroscopy (e.s.r.), used to study systematically the magnetic chromium phases in these oxidised systems, is described in Part 2.31 This combination of techniques identifies the character of the magnetic species and reveals details of their environment.For oxidised chromium on alumina, y-phase chromium has been EXPERIMENTAL The alumina support was prepared from high-purity electrolytic boehmite (Laporte Industries Ltd), surface area 138 m2 g-l, by calcination at 525 "C in air. This y-alumina had a surface area of 157 m2 g-l and pore radius 6.5 nm (mesopores). The silica support was a GASIL 35 micronised xerogel (Unilever Ltd), calcined at 525 "C in air, with a surface area of 350 m2 g-l and pore volume 1.0 cm3 g-l. Radiotracer studies showed that the y-alumina irreversibly adsorbed ca. 3 wt % Cr. The silica showed no adsorptive properties and all of the deposited chromium was removed by washing with cold water, < 0.001% Cr remaining.The supports were impregnated with aqueous chromium(6+) oxide. In the case of alumina, loadings < 3% were obtained by filtering the slurry and drying the solids without water washing. Samples were dried by vacuum desiccation at room temperature to try to reduce the decomposition of the supported Crs+. Small amounts of all samples were calcined in air at temperatures between 100 and 600 "C. Samples dried at 25 "C were analysed using neutron activation analysis (Physics and Radioisotope Services, I.C.I. Ltd, Billingham, Cleveland). Samples are designated according .to their chromium content (wt % ), the calcination temperature ("C) and the nature of the support; thus sample 2-300-Si contains 2 wt % Cr on silica and was calcined at 300 "C in air.A.ELLISON 2569 Thermogravimetry (t.g.a.) and differential thermal analysis (d.t.a.) were performed in air using a Stanton Redcroft thermobalance. Secondary-ion mass spectra (SIMS) were obtained by courtesy of SIMS Consultancy Ltd, UMIST, using a Vacuum Generators Analytical MM 12-12s system. Argon, fast-atom bombardment (FAB-SIMS) was used as the primary beam, current density lo-' A m-2, beam area 4 mm2. Spectra were optimised using a low-energy (2 kV) atom gun, final analyses being obtained as positive-bias spectra using a high-energy (4 kV) atom gun. Pre-calcined samples were presented as compressed discs, using a low pressure (< 5 ton in-2)* to try to prevent fracture of the silica discs. Six samples of chromium oxide on alumina and six samples of chromium oxide on silica were simultaneously outgassed in the SIMS analysis chamber to facilitate comparison between the samples under the same high-vacuum conditions. 185 -AT 520 + A T 1 EXOTHERM Fig.1. Different thermal analysis curves for the samples (a) 12-25-A1 and (b) 18-25% RESULTS Fig. 1 shows representative d.t.a. curves for samples 12-25-A1 and 18-25-Si. For chromium loadings between 6 and 12% on alumina, the two exotherms at ca. 400 and 520 "C are both accompanied by loss of chromium. The alumina loses water continuously over the whole temperature range and it was not possible to determine accurately the loss of mass due to chromium. For silica-supported chromium, between 7 and 18% Cr, an exotherm occurs at 400 "C, a similar temperature to that observed for chromium-alumina. At ca.495 "C, however, an endotherm occurs, in contrast with the 520 "C exotherm for chromium on alumina. Both thermal reactions are again accompanied by the loss of chromium. For silica, water loss is sharp and apparently complete before 250 "C so that the chromium mass loss could be measured; ca. 16% of the chromium is lost during the * 1 ton E 2240 x 0.45359237 kg; 1 in _= 2.54 x 10-2 m.Cr,OH02 I I I Cr3 A1 04 I I Cr Si OHO, I Cr OH (4 Cr30H02 i Cr20H I Cr202 / ( 4 Cr5Si05 I h Cr,OH 0 I I Cr, (OH),- CrzSi(OH)30 200 300 I 400 500 I mle 200 250 I 250 300 mle N ul 4 0 Fig. 2. FAB-SIMS positive-ion spectra: (a) low resolution, 1-100-A1; (b) low resolution, 18-100-Si; (c) high resolution, 1-100-A1 and ( d ) high resolution, 18-100-Si.A.ELLISON 257 1 Table 1. FAB-SIMS ion-current ratios for the fragments 52Cr+, 27Al+ and 28Sif sample 27A1+/52Cr+ sample z8Si+/52CrS 1 - 1 00-A1 16 1-25-Si 3.5 1 -400-A1 13 1 -400-Si 1.1 6- 1 00-A1 5.1 11-100-Si 0.7 6-300-A1 2.0 1 1-300-Si 0.7 6-400-A1 2.7 18- 1 00-Si 1.3 12- 1 00-A1 2.1 18-200-Si 0.6 12-400-A1 1.8 18-300-Si 0.7 Table 2. FAB-SIMS ion-current ratios for various fragments, Cr-Al,O, samples (a) Cr,: Cr, fragments Vr(OH); 136Cr20i 171Cr2(OH),0+ sample 87Cr(OH),H+ l2OCr20+ 172Cr2(OH)a 1 - 1 00-A1 2.6 0.20 3.0 1 -400-A1 2.4 0.24 1.5 6- 100-A1 1.5 0.43 5.5 6-300-A1 1.5 0.43 5.0 6-400-A1 1.5 0.38 5.5 12-400-A1 3.6 0.7 1 7.0 (b) Cr,: Cr, fragments 155Cr2(OH),f 205Cr,(OH)O~ sample lo3Cr(OH),f lZ0Cr20+ 1 - 100-A1 0.03 1 -400-A1 0.03 6- 1 00-A1 0.03 6-300-A1 - 6-400-A1 0.03 12-400-A1 0.01 0.8 0.9 0.9 0.8 0.9 0.9 (c) CrA1: Cr fragments l12CrA1(OH)O+ 129CrA1(OH),0+ sample 85Cr(OH)Of lo3Cr(OH)3f 1 - 1 00-A1 0.40 1 -400-A1 0.26 6- 1 00-A1 0.3 1 6-300-A1 0.31 6-400-A1 0.3 1 12-400-A1 0.54 0.12 0.10 0.56 0.56 0.40 0.782572 CHROMIUM FAB-SIMS Table 3.FAB-SIMS ion-current ratios for various fragments, Cr-SiO, samples (a) Cr, : Cr, fragments 139Cr,(OH),H+ lslCr,Si(OH)Ol sample 122Cr,(OH)Hf 165Cr,Si(OH)O+ 2-2543 2-400-Si 1 1 - 1 00-Si 1 1 -300-Si 11-400-Si I 8- 1 00-Si 18-200-53 18-300-Si 18-400-Si 1.2 1.1 1.2 0.9 1.1 1.5 1.6 1 .o 1.5 (b) Cr, : Cr, fragments 1 .o 1.3 1.2 1.2 1.1 1.1 1.6 1 .o - 205Cr3(OH)O~ 207Cr3(OH)i sample 153Cr2(OH)OQ 155Cr2(OH),+ 2-2543 2-400-Si 1 1 - 1 00-Si 1 1 -300-Si 1 1 -400-Si 18- 1 00-Si 18-200-Si 18-300-Si 18-400-Si 0.3 0.3 0.2 0.4 0.3 0.4 0.5 0.5 - (c) CrSi : Cr fragments 0.2 0.5 0.3 0.2 0.3 0.5 0.4 0.3 - 1soCr2Si(OH),0~ 249Cr,Si(OH)O~ sample 152cr,o,+ lg7Cr,Si(OH)O,+ 2-25-Si 2-400-54 1 1 - 1 00-Si 1 1 -300-Si 1 1 -400-Si 18- 100-Si 18-200-Si 18-300-Si 18-400-Si 0.06 0.04 0.05 0.04 0.06 0.04 0.03 0.03 0.04 0.5 0.2 0.2 0.1 0.15 0.3 0.15 0.15 - exothermic reaction at 400 "C, regardless of the original chromium loading.During the endothermic process at 495 O C , ca. 34% of the remaining chromium is sublimed. Fig. 2(a) and (b) show FAB-SIMS spectra for the 1-100-A1 and 18-100-Si samples, respectively, at fast scan or low resolution. Proper quantification of peak intensities proved to be impossible owing to variations in the distance between the sample surface and the detector/primary beam, caused by fractured sample discs.Although variationsA. ELLISON 2573 are observed for different chromium loadings and calcination temperatures, the above spectra are representative, and substantial differences were not detected within the separate groups of alumina and silica samples. Although single and clustered Cr ions, as well as cluster-ions containing both chromium and support atoms, are observed in both spectra, there are obvious differences between the spectra. All chromium-containing clusters are enhanced for chromium on silica, including those clusters containing Si from the support surface. This difference is clearly shown in the high-resolution spectra for the same samples, fig.2 ( c ) and (d). Identification of fragments was carried out in the following manner. After rationalisation of peak intensities with respect to amplification factors, a computer program was used to fit the observed mass of fragments, m, or mass-to-charge ratio, m/e, to the general formula Cr,Al,O,H, or Cr,Si,O,H,. Uniqueness of fit was complicated at low mass values by the presence of secondary-ion fragments from the support and at high mass values by formula coincidences. However, it became obvious during the analysis that the number of hydroxy groups in the fragment having the highest intensity in a related group of peaks was always between 1 and 3. Finally, to correlate ion fragments it was assumed that if the ratio of the intensities, for example of MO+ and M,O+, was constant then MO+ was related to the fragment M,0+.32 All of the fragments, including the support-background ions and those generated from coincident formulae, were subjected to this correlative procedure, so that only successfully correlated ion fragments were considered to be confirmed and their data reported here.The total yields of the species 52Cr+ and 27Al+ or 28Si+, as intensity ratios, are shown in table 1 . The ratio 27A1+/52Cr+ is always greater than the ratio 28Si+/52Cr+ at comparable Cr loadings, and is substantially larger at 1 % Cr. Table 2 (a) shows fragment-intensity ratios for some typical fragments for chromium on alumina. Constant values of fragment ratio were only achieved within each con- centration group.The fragment ratios for cluster types containing different numbers of chromium atoms, table 2(a), show constant values independent of chromium loading and calcination temperature, suggesting that these fragments arise from the same part of the chromium surface. In contrast, constant values of fragment ratio were realised only within the concentration groups for the fragment types CrO,H, and CrAlO,H,, table 2(c). The sparse data did not enable a meaningful relationship to be established between multiple-chromium clusters Cr,O,H, and Cr,AlO,H,. Table 3 (a) shows similar fragment-ratio data for the various cluster types observed for chromium on silica. Within each fragment type, constant values of fragment ratio were achieved over the whole set of silica samples, independent of chromium loading and calcination temperature.The fragment ratios for the species CrO,H, and Cr,O,H, are not constant, suggesting that single chromium fragments may not arise from the same sites as Cr, clusters. However, the ratios Cr,O,H, : Cr,O,H, and Cr,O,H, : Cr,O,H, [table 3 (b)] do show some constancy of value, showing that multiple-chromium-atom clusters may arise from the same cluster surface. The fragmentation ratios between the pairs CrO,H$-CrSiO,Hi, CrO,H,+- Cr,SiO,Hi, Cr,O,Hi-Cr,SiO,Hi and CrSi0,Hi-Cr,SiO,H, [table 3 (c)] are sufficiently constant to suggest that Cr,O,H$ fragments may originate from Cr,SiO,H,+ species. Finally, table 4 shows all of the species confirmed by the fragment-ratio method arranged as families of cluster ions.Table 4.FAB-SIMS positive cluster ions for chromium supported on alumina or on silica, arranged as cluster types A1203 SiO, Cr(0H)O c - - Cr,H Cr ,(OH)O Cr,(OH) Cr,(OH)O, - Cr,(OH)O Cr3(0H)02 - - - c Cr4(OHP3 Cr4(OH)04* Cr,(OH)O,* - - Cr5(OH)07* - cfi2 CrO, - Cr2 Cr202 Cr,O Cr203 - Cr302 Cr303 Cr304 Cr209* - - Cr404 Cr405 cr4012* Cr,04* Cr,O,* - cr5010* Cr(OH),H Cr(OH),O CWH), Cr(0H)302 n- - Cr5011* Cr,(OH)O,,* CrAl CrAlO CrA10, - CrAlH CrAl(0H) CrAl(0H)O CrAl(OH)O, CrAl(0H)O CrAl(OH), CrAl(OH),O CrAl(OH),O, Cr,Al(OH), - - - Cr,Al(OH),O, - - CrSi(0H)O CrSi(OH)O, CrSi(OH)O, Cr,Si(OH)O Cr,Si(OH)O, Cr,Si(OH)O, Cr,Si(OH)O, Cr,Si(OH)O,* Cr,Si(OH) Cr,Si(OH)O Cr,Si(OH)O, Cr,Si(OH)O,, Cr,Si(OH)O, Cr,Si(OH)O, Cr,Si(OH)O, - - - CrSi(OH), CrSi(OH),O CrSi(OH),O, Cr,Si(OH),O Cr,Si(OH),O Cr,Si(OH),O, Cr,Si(OH),O, Cr,Si(OH),O,* Cr,Si(OH),O, Cr,Si(OH),O, Cr,Si(OH),O,, Cr,Si(OH),O, Cr,Si(OH),O, Cr,Si(OH),O, Cr,Si(OH),O, Cr,Si(OH),O,, Cr,Si(OH),O, Cr,Si(OH),O,, Cr,Si(OH),O,, - - - CrSi(OH),H CrSi(OH), CrSi(OH),O Cr,Si(OH),H Cr,Si(OH), Cr,Si(OH),O - CrAl(OH),H CrAl(OH), CrAl(OH),O CrSiO, CrSiO, Cr,AlO Cr,AlO, - Cr,SiO, Cr,SiO, Cr,SiO, Cr,SiO, Cr,SiO,* Cr,SiO Cr,SiO, Cr,SiO, Cr,SiO,, Cr,SiO, Cr,SiO, - - - Cr,A10, - ? m r E z Is Cr,Si(OH),O, Cr,Si(OH),O Cr,Si(OH),O, - Cr,Al(OH),O, - Cr, Al(OH)O, - Cr,Si(OH),O, Cr,Si(OH),O,, Cr,Si(OH),O, Cr,Si(OH),O,, Cr,Si(OH),O, , Cr,Al(OH)O, Cr,Al(OH),O, - Cr, Al(OH),O, - Cr,SiO,, Cr,Si(OH)O, Cr,Si(OH),O,, - * Detection at only slow scan (low resolution).2576 CHROMIUM FAB-SIMS DISCUSSION The SIMS spectra for chromium on silica and alumina (fig. 2) appear to show that although the sputtering yield of Al+ is greater than that of Si+ (table l), silicon has a greater tendency to appear in chromium cluster fragments (table 4).to increase monotonically as the surface coverage of a supported species decreases from the monolayer value, suggesting that the yields of Al+ and Si+ found here are indicative of the relative amounts of exposed surface on the alumina and silica within each series of samples. It is, however, only possible to speculate from the data that, at all values of chromium loading, more alumina surface is exposed even though the silica has a larger surface area. The sputtering coefficients for Al+ and Si+, SA1+ and Ssi+, respectively, are in the ratio 20: 1 for the pure corresponding to the greater ease of ionisation of Al+.The presence of foreign ions at these support surfaces modifies the ionisation processes and the sputtering characteristics of the support metal ions.34 Consequently the fragmentation patterns of the pure supports and the chromium-modified supports will be different and therefore SAI+ and Ssi+ remain unknown. During the fast-atom bombardment in the SIMS experiment, the energy of the primary beam is dissipated by a cascade process. Momentum is lost through a variety of collisional processes, but components of momentum are always directed towards to the surface of the 36 In consequence sections of the surface are lifted and peeled away to form the secondary beam of ion fragments. The components of this beam may well dissociate further before detection but recombination of ion fragments does not occur.Thus secondary ions may be single ions or clusters, but if a cluster is emitted its constituent atoms must have come from adjacent surface s i t e ~ . ~ ~ ? ~ ’ The SIMS spectra, fig. 2 and table 4, show that the sizes of Cr and CrM clusters (where M is the support atom) released from the catalyst surfaces are consistently greater for chromium-silica samples. These data may indicate that the chromium oxide cluster surface is more extensive on silica or that the long-range interaction of the support with the chromium surface is greater in the case of alumina. for a variety of chromium oxides. The clusters may be artefacts created by the high-energy rupture of the surface which produces excited-state ions with short half-lives.Consequently the composition of cluster ions will not necessarily depend upon the valence and structural constraints ordinarily associated with ground-s ta te species. Consideration of the fragment-ratio data suggests that, in general for chromium- silica, the ratios are constant regardless of the chromium loading as though the surface texture of each chromium-silica sample is similar. For alumina, in general, the ratios are constant only within the concentration groups as though for each different chromium loading the surface composition is different. The ratios for fragments Cr,O,H, and Cr,O,H, on chromium-alumina are constant independent of chromium loading, showing that small amounts of multiple- chromium-atom clusters arise from the same region of chromium surface and occur at all chromium concentrations.In contrast, constant ratios for CrO,H, and CrAlO,H, fragments are found only within the concentration groups, as though the amount and the dispersion of the chromium clusters close to the alumina surface depend upon chromium loading. In summary, chromium clusters occur for all the chromium loadings studied, inchding the relatively low loading of 1 % , but the surface textures of the samples are different. The yield of the support metal ion has been Similar cluster ions have beenA. ELLISON 2577 Fig. 3. Schematic pictures of chromium clusters on the surface of (a) alumina and (b) silica. For chromium-silica the relationship between single-chromium-atom and multiple-chromium-atom fragments is unclear because of the coincidence of m/e values for Cr+ and silica fragments.However, all other fragment ratios, including Cr,O,H, :Cr,O,H,, are constant independent of chromium loading. The extensive chromium clusters are inextricably associated with the silica surface, and each chromium-silica sample has a similar surface texture. The e.s.r. spectra of these samples3' support the conclusion reached from SIMS data that clustering of chromium species occurs even at 1 % Cr loading, the e.s.r. spectrum for 1-600-A1 showing a broad, low-field resonance which may well arise from Cr3+ nearest-neighbours in small clusters, while for samples 1-1 50-Si, 1 -200-Si and 1 -300-Si distinctive /3-resonance is observed.The thermal gravimetrydifferential thermal analysis results reveal that on calcin- ation in air more chromium oxide sublimes from silica than from alumina. Observations suggest that local ' over-heating' may occur in particular chromium oxide regions during the exothermic reactions, causing the supported chromium to be redistributed onto 'cooler' sites. The chromium oxide is further 'spread-out' on the surface so that, regardless of the notational amount of chromium remaining on the silica surface, the texture, environment and therefore the properties of the chromium oxides are similar. Sublimation and redistribution of chromium oxide on alumina is limited in extent. Again the e.s.r. spectra of these samples3' provide evidence supporting these ideas.The development of y-resonance for chromium-alumina depends upon chromium concentration. Similarly, after calcination at temperatures > 400 "C, the y-resonance peak-to-peak width varies with chromium loading as different degrees of dipole-dipole broadening occur for y-phase chromium in different Cr3+ environments. In contrast, the y-resonance behaviour for chromium-silica is very similar for all chromium loadings, revealing a close similarity of cluster types. Pictures of the surfaces of these catalysts can be developed from these results (fig. 3). For chromium-alumina, an exposed alumina surface is observed together with chromium clusters whose depth depends upon chromium concentration. For chromium-silica, a less exposed silica surface is apparent, together with more extensive but 'thinner' chromium clusters, so that when the chromium surface is 'peeled away' through primary atom bombardment silicon often forms part of the ion fragment.which attract chromium species during impregnation to form an adherent surface chromium layer. In regions on the alumina surface there are specific adsorption 84 FAR 12578 CHROMIUM FAB-SIMS The adsorption forces at these specific sites are evidently long range: less chromium sublimes from the alumina surface on calcination; the tendency of the chromium to decompose is much less for that on alumina, even at relatively high concentrations (12% Cr) when a large amount of the chromium is distant from the alumina surface; Q2 e . ~ . r . ~ l appears at a lower temperature for chromium-silica and again decomposes at a lower temperature to give a normal Q3 resonance31 characteristic of Cr3+ clusters; for chromium-alumina the Cr3+ Q3 resonances are not observed at the highest calcination temperatures used, revealing the enhanced stability of even heavily clustered chromium on alumina.These long-range forces cause multilayer adsorption and hence thick clusters, while exposed alumina surface remains where the adsorption sites do not exist. For silica, in contrast, no adsorption takes placego and the chromium simply matts or clogs the surface during sample preparation to form spread-out and relatively thin clusters, a process aided by redistribution of chromium occurring for the higher chromium loadings on calcination. Although these surface models may be speculative and novel, recent work does support these ideas.At only 0.1 % Cr loading on the same Laporte alumina used in this report, FAB-SIMS studies reveal4' surface clusters of chromium oxides, demon- strating that the primary adsorption sites on the alumina surface are not evenly dispersed. Finally, nitrogen-gas adsorption, X-ray and FAB-SIMS studiesg2-,, of alkene metathesis catalysts show that at 5% loading, and above, of Re20, or NH,ReO, on a low-surface-area, mesoporous alumina, bulk rhenium oxide or ' crystalline ' NH,ReO, structures, respectively, predominate on the support surface. I thank the Royal Society, B.P. Chemicals Ltd, Croda Chemicals Ltd and the Humberside College of Higher Education for financial help and Drs J. A. van den Berg and J.C. Vickerman (UMIST) for help with FAB-SIMS studies. My thanks also go to A. K. Coverdale and M. Cawley for help and encouragement and to Dr R. B. Moyes (Hull University) for valuable discussions. A. Clark, 1. N. Finch and B. H. Ashe, Proc. 3rd Int. Congr. Catal. (North Holland, Amsterdam, 1965), (a) C. Eden, H. Feilchenfeld and Y. Haas, J. Catal., 1967, 9, 367; (6) C. Eden, H. Feilchenfeld and Y. Haas, J. Catal. 1968, 11, 263. V. B. Kazansky and J. Turkervitch, J. Catal., 1967,8, 231. D. D. Eley, C. H. Rochester and M. S. Scurrell, Proc. R. SOC. London, Ser. A , 1972, 329, 361. G. Vuillaume, R. Spitz, A. Revillon and A. Guyot, J . Macromol Sci., Chem., 1974, AS, 11 17. R. Spitz, A. Revillon and A. Guyot, J. Macromol Sci., Chem., 1974, AS, 1129.A. Clark, Catal. Rev., 1969, 3, 145. J. P. Hogan, J. Polym. Sci., 1970, 8, 2637. H. L. Krauss and H. Stack, Inorg. Nucl. Chem. Lett., 1968, 4, 396. vol. 2, p. 1010. lo H. L. Krauss, Proc. 5th Znt. Congr. Catal. (North Holland, Amsterdam, 1973), vol. 1, p. 207. l1 H. L. Krauss, B. Rebenstorff and U. Westphal, 2. Anorg. Chem., 1975, 39, 363. i2 H. L. Krauss and L. Mums, Z. Naturforsch., Teil B, 1979, 34, 1628. l3 L. K. Przhevalskaya, V. A. Shvets and V. B. Kazansky, J . Catal., 1975, 39, 363. l4 D. D. Beck and J. H. Lunsford, J. Catal., 1981, 68, 121. l5 H. L. Krauss, 184th ACS National Meeting, Div. Colloid and Surface Chem., Sept. 1982, Kansas City, U.S.A. G. Ghiotti and E. Garrone, 184th ACS National Meeting, Div. Colloid and Surface Chem., Sept. 1982, Kansas City, U.S.A.l7 V. A. Zakharov, V. N. Druzhkov and Y. I. Ermakov, Kinet. Katal., 1973, 14,998. i8 Y. I. Ermakov and V. A. Zakharov, Adv. Catal., 1975, 24, 173. l9 V. B. Kazansky, A. N. Pershin and B. N. Shelimov, Proc. 6th Int. Congr. Catal. (Elsevier, Amsterdam, 2o D. Comet and R. L. Burwell Jr, J. Am. Chem. Soc., 1968,90, 2489. 1980).A. ELLISON 2579 21 (a) E. N. Bunting, J. Res. Natl Bur. Stand., 1930, 5, 325; (b) E. N. Bunting, J. Res. Natl Bur. Stand., 22 V. M. Bums and R. G. Bums, Geochim. Cosmochim. Acta, 1975,39,903. 23 F. P. Glasser and E. F. Osborn, J. Am. Ceram. Soc., 1958,41, 358. 24 A. Muan, Geochim. Cosmochim. Acta, 1975,39, 78 1. 25 A. Navrotsky, Geochim. Cosmochim. Acta, 1975, 39, 819. 26 I. Zaplatynsky, NASA Tech. Memo, NASA TMX-2969 (1 974). 27 C. Groenveld, P. P. M. M. Wittgen, A. M. van Kersbergen, P. L. M. Mestrom, C. E. Nuijten and 28 B. RebenstorEand R. Larsson, 184th ACS National Meeting, Div. Colloid and Surface Chem., Sept. 29 A. Ellison and K. S. W. Sing, J. Chem. Soc., Faraday Trans.1, 1978, 74, 2807. 30 D. E. O’Reilly and D. S. MacIver, J. Phys. Chem., 1962, 66, 276. 31 A. Ellison, J. Chem. Soc., Faraday Trans. 1, 1984, 80,2581. 32 M. Barker, J. C. Vickerman and J. Wolstenholme, J. Chem. Soc., Faraday Trans. I , 1976, 72, 46. 33 J. A. van den Berg and J. C. Vickerman, personal communication. 34 M. Barber, P. K. Sharpe and J. C. Vickerman, J. Catal., 1976, 41, 240. 35 A. Benninghoven, Dechema Monogr., 1975, 78, 197. 36 A. Benninghoven and H. Bispcrinck, Modern Physics in Chemistry, 1979, 2, 391. 37 M. Barber and J. C. Vickerman, Surf. Defects Prop. Solid, 1976, 5, 162. 38 H. W. Werner, H. A. M. de Grefts and J. van den Berg, Advances in Mass Spectrometry, ed. 3s H. C. Chen, G. C. Ghillies and R. B. Anderson, J. Catal., 1980, 62, 367. 40 P. F. Dearing, personal communication. 41 A. Ellison, unpublished work. 42 A. K. Coverdale, P. F. Dearing and A. Ellison, J. Chem. Soc., Chem. Commun., 1983, 567. 43 A. Ellison, A. K. Coverdale and P. F. Dearing, Appl. Catal., 1983, 8, 109. 44 A. Ellison, A. K. Coverdale and P. F. Dearing, paper presented at 5th Int. Symp. Olefin Metathesis, 1931, 6, 947. G. C. A. Schuit, J. Catal., 1979, 59, 153. 1982, Kansas City, U.S.A. A. D. West (Applied Science Publishers, Barking, Essex, 1974), vol. 6, p. 693. August 1983, Graz, Austria; to be published in J. Mol. Catal. (PAPER 4/053)

ISSN:0300-9599    

DOI:10.1039/F19848002567

出版商:RSC    

年代:1984

数据来源: RSC