摘要:

Dehydrogenation and Dehydration of Ethyl Alcohol over a Polynaphthoquinone containing various amounts of FeC13 Selectivity of Formation of Acetaldehyde, Ethylene and Diethyl Ether BY YASUHIRO IWASAWA AND SADAO OGASAWARA Department of Applied Chemistry, Faculty of Engineering, Yokohama National University, Ooka-cho, Minami-ku, Yokohama AND TAKAHARU ONISHI AND KENZI TAMARU* Department of Chemistry, University of Tokyo, Hongo, Bun kyo-ku, Tokyo, Japan Received 1 1 th April, 1973 The decomposition of ethyl alcohol over polynaphthoquinones with various amounts of FeC13 in the temperature range from 150 to 230°C is reported. Adsorption of ethyl alcohol on poly- naphthoquinones containing FeCIJ is of a Langmuir type, ethyl alcohol being adsorbed selectively on the FeC1,. The mechanism and the selectivity of the formation of acetaldehyde, ethylene and diethyl ether are discussed in detail.The relative ratio of the formation of diethyl ether to ethylene in the dehydration is markedly affected by the FeCl, content of the catalysts. Polymers containing quinone groups are effective catalysts in dehydrogenation reactions. l The catalytic activity of polynaphthoquinones containing various metal halides, such as FeCl,, VC13, MoC1, and TiC14, has been investigated in a previous paper,2 where the polynaphthoquinones catalyzed various reactions, such as the dehydrogenation of formic acid, isopropyl alcohol and cyclohexene, and the isomer- ization of butene. Quinone compounds play an important role in living organisms, acting, in many cases, as hydrogen or electron acceptors.The organic polymers containing quinone groups are accordingly considered not only to be of biological importance, but also to be heterogeneous catalytic systems which have relatively well-defined active sites with uniform activity. It is of interest to study the catalytic activity as well as the selectivity of such catalysts with various compositions and also with the quinone groups coordinated with metal ions. As a model reaction the dehydration and the dehydrogenation of ethyl alcohol over the polynaphthoquinones containing various amounts of FeCl, have been studied. EXPERIMENTAL As previously reported,2 the polynaphthoquinone was obtained by the oxidation of naphthalene-1,7-diol by nitric acid in air, followed by heat treatment at 350°C.The poly- naphthoquinone containing FeCI, was prepared by mixing the polynaphthoquinone (0.1 g) and known amounts of FeCIJ from 0.0058 to 0.08 g in ethyl alcohol in the absence of air, followed by slow evaporation of the solvent at room temperature. The polynaphthoquinone 1-7 193194 REACTIONS ON POLYNAPHTHOQUINONES thus obtained was treated at 250°C for 6 h under vacuum prior to each run. The FeC13 in the catalysts was examined by Mossbauer spectroscopy showing its oxidation state to be high spin Fe3f. The polynaphthoquinone employed was not decomposed except at temp- eratures greater than 450°C ; a small amount of decomposition to form ethylene and carbon monoxide was observed at 500°C. The B.E.T. surface areas of the polynaphthoquinones were not affected by changing the temperature of the pre-treatment ; 73 m2/g (treated at 250°C), 70 m2/g (treated at 300"C), 73 m2/g (treated at 350"C), 70 m2/g (treated at 4OO"C), 69 m'/g (treated at 450°C).Ethyl alcohol was purified by repeated distillation after the removal of water by molecular sieves in the absence of air. The products were analyzed by gas chromatography, using two columns of dioctyl sebacate and diethyl phthalate in series at 40°C. The amounts of ethyl alcohol adsorbed on the catalysts were measured by a volumetric technique, taking the adsorption of the vapour by the apparatus into consideration. The reactions were carried out in a temperature range from 150 to 230°C in a closed circulating system (170 cm3) in a U- shaped glass reactor. RESULTS AND DISCUSSION 1.ADSORPTION MEASUREMENTS OF ETHYL ALCOHOL ON THE POLY- NAPHTHOQUINONES The adsorption of ethyl alcohol over both the polynaphthoquinone without FeCI, and that containing FeCI, was studied in the temperature range from 150 to 230°C by pressure measurements. Adsorption is rapid and complete in 8 min, not being affected by the reaction products. Typical results for ethyl alcohol adsorption over the polynaphthoquinones with or without FeCl, are given in fig. 1 ; the reciprocal of the amount adsorbed varies linearly with the reciprocal of the ethyl alcohol vapour 5 3 2 t 1 / 'LO .5 I 1 I 2.5 5.0 7.5 104P-' /m2 N-' FIG. 1 .-Langmuir plots for the adsorption of ethyl alcohol on the polynaphthoquinones containing various amounts of FeCl, at 200°C : 1, metal-free ; 2, 0.0058 g ; 3, 0.015 g ; 4, 0.018 g, 5,0.026 g ; 6, 0.036 g ; 7, 0.046 g ; 8, 0.056 g ; 9, 0.08 g : Vad, the adsorbed amount of ethyl alcohol P, the vapour pressure of ethylene alcohol.Y .IWASAWA, S . OGASAWARA, T . ONISHI AND K . TAMARU 195 pressure. The saturation amount adsorbed obtained from fig. 1 remains unchanged at different temperatures. Hence, adsorption of ethyl alcohol on the polynaphtho- quinones is of a Langmuir type ; vad = bKP/(l +KP), where Va, is the volume ad- sorbed and b and K are the saturation mass adsorbed and the equilibrium constant for the adsorption, respectively. The plot of log K against the reciprocal of temper- ature gives the heat of adsorption to be 3.6 x lo4 J mo1-, over the polynaphthoquinone without FeC1,.This value corresponds to the energy of one or two hydrogen bonds. The adsorption of ethyl alcohol is influenced by addition of FeCI, to the polynaphtho- quinone. The amount of adsorption increases with the amount of FeCI, added. The saturation mass adsorbed, b is proportional to the FeC1, content of the catalysts as shown in fig. 2, indicating that the FeCI, molecules in the catalyst behave as adsorption sites. t 3.0 a cf rr i.' 0 0.04 0.0 8 FeCl content / g FIG. 2.-The variation of saturation mass of the ethyl alcohol adsorption with FeCI3 content in the catalyst (polynaphthoquinone ; 0.1 g). 2. THE DEHYDROGENATION OF ETHYL ALCOHOL Ethyl alcohol is rapidly dehydrogenated to form acetaldehyde over the poly- naphthoquinone with or without FeCI,. No hydrogen is evolved during the dehydro- genation process and the reaction proceeds until the polynaphthoquinone surface is saturated with hydrogen, a characteristic property of quinone polymers.The surface hydrogen (as hydroquinone) is removed easily by oxygen, and when ethyl alcohol and oxygen are introduced simultaneously onto the catalyst, both acetaldehyde and water are formed as reaction products. The reaction is expressed by a similar scheme to that described in the previous paper, where hydrogen is transferred from ethyl alcohol to the quinone groups in the catalyst, forming hydroquinone groups and subsequently the hydroquinone groups are oxidized to quinone groups by oxygen. The initial rates of the acetaldehyde formation are plotted against the initial pressures of ethyl alcohol and its adsorbed amounts in fig.3, showing that the amount of acetaldehyde formed is proportional to the adsorbed amount of ethyl alcohol. Accordingly, the rate equation is expressed as follows ; d[acetaldehyde]/dt = k, [ethyl al~ohol],~, where [ethyl alcohol],, represents the amount of ethyl alcohol adsorbed on the surface of the polynaphthoquinone. From the adsorption measurements of ethyl alcohol, this also equals klbKP/(l + KP). Consequently, a surface unimolecular196 REACTIONS ON POLYNAPHTHOQUINONES x pressure of C&T,OH/N m-2 0.01 6 0.012 0.m a004 0 4.00 8.00 0.2 0.4 0 6 C2H50H(ad)/cm3 (s.t.p.) FIG. 3.-The dependence of initial rates FIG. 4.-The variation of initial rates of the ethylene of acetaldehyde formation on the initial formation with the initial pressures of ethyl alcohol (0) pressures of ethyl alcohol (0) and upon and with the adsorbed amounts of ethyl alcohol (A) at the adsorbed amounts of ethyl alcohol (0) 230°C on a polynaphthoquinone containing 0.036 g at 200°C over a polynaphthoquinone con- taining 0.036 g FeC13. FeCl,.reaction of ethyl alcohol adsorbed on the catalyst is rate-determining in the dehydro- genation of ethyl alcohol. The process of the dehydrogenation of ethyl alcohol over the polynaphthoquinone may be shown as follows ; c2 ,CH, CHO (1) and in the presence of oxygen, When oxygen is admitted onto the reduced polynaphthoquinone at room temperature hydrogen peroxide is formed (confirmed by analysis with a 1 : 1 mixture of aqueousY . XWASAWA, S. OGASAWARA, T. ONISHI A N D K .TAMARU 197 solutions of FeCl, (0.4 weight %) and K,Fe(CN), (0.8 weight %))., The hydrogen peroxide is very rapidly decomposed to form water. Although the mechanism of the decomposition of hydrogen peroxide is generally rat her c~mplex,~ two possible paths for the decomposition are given as follows ; 2H20 Oxygen also appears from the decomposition of hydrogen peroxide over the reduced catalyst. However, it should be noted that hydrogen peroxide is produced by the reaction of air with the reduced polynaphthoquinone at room temperature. The initial rates of the reaction and the rate constant, k l , and the activation energies over the polynaphthoquinones containing various amounts of FeCI, are given in table 1. It is shown in table 1 that the addition of FeCl, to the catalyst markedly TABLE THE FORMATION OF ACETALDEHYDE (A), ETHYLENE (E) AND DIETHYL ETHER (DE) AT 200°C rate constant1 (molecule min-1) ( 1 0 4 J mol-1) initial rate?/ (ratelsite)/ activation energy/ - ( 1020 motecule min-1) A E DE A E DE A E A E DE FeCl3 content*/ FeCl -free 0.0058 0.015 0.01 8 0.026 0.036 0.046 0.056 0.08 3.68 35.74 66.1 1 82.77 110.19 165.28 206.67 248.06 357.44 0.66 0.20 7.4 1.33 2.13 (7.6) (1.4) 9.62 10.9 12.59 6.07 4.14 66.7 6.11 10.6 118.9 10.9 5.44 7.86 9.62 9.65 6.88 83.3 7.29 11.3 114.2 10.0 5.19 7.32 7.70 13.79 11.58 105 8.77 12.3 123.7 10.3 5.19 7.32 7.70 17.63 17.09 107 9.14 12.6 121.1 10.3 5.19 7.32 7.45 21.20 23.41 112 9.63 13.0 119.4 10.2 5.02 7.32 7.45 30.37 42.73 122 10.4 13.5 120.4 10.2 5.19 7.03 7.11 2.77 - 54.2 4.17 - 166.2 12.9 6.82 8.58 - 7.18 - 78.9 6.84 - 123.9 10.7 5.19 7.32 - * per 0.1 g of polynaphthoquinone ; t u = k,[ethyl alCOhOl],d for the formation of A, u = k,[ethyl akOhOl],d for the formation of E, and v = k,[ethyl a l ~ o h o l ] ~ ~ for the formation of DE ; $ kl(103 dm3 min-I) for A, k2(103 dm3 min-I) for E, and k3(103 dm3 cm-, min-I) for DE.accelerates the dehydrogenation process. The initial rates are approximately proportional to the amount of FeCI,, indicating that all the FeCI, in the catalyst has a similar activity for the reaction. 3. DEHYDRATION OF ETHYL ALCOHOL O N THE POLYNAPHTHOQUINONES CONTAINING FeC1, Ethylene and diethyl ether are both produced by the dehydration of ethyl alcohol over the polynaphthoquinones. The decomposition of diethyl ether to form ethylene is negligible under these conditions.The initial rate of ethylene formation at 230°C is plotted against the initial pres- sures of ethyl alcohol and the adsorbed amounts of ethyl alcohol in fig. 4 ; a linear relation between the rate of the ethylene formation and the amount of ethyl alcohol198 REACTIONS ON POLYNAPHTHOQUINONES adsorbed is observed. The initial rate is expressed by the following equation: d[ethylene]/dt = k,[ethyl alcohol],, and is equal to k,bKP/(l + KP). In fact, the reciprocal of the initial rates gives a good linear correlation with the reciprocal of ethyl alcohol pressures. These results demonstrate that ethylene is formed by a unimolecular reaction of ethyl alcohol adsorbed on the catalyst, where a surface reaction involving the adsorbed ethyl alcohol is the rate-determining step.0.02 U I c 3 0.01 3 < k - - ' ,'d , / x, 00" o I ' 0 , , -0, . -6 0 0 0.04 0.08 FeC13 content/g pressure of ethyl alcohol = 3.07 x lo3 N m-'. FIG. 5,-The dependence of the initial rates of ethylene formation on the FeCI, content; initial The rate constants, kZ, and the activation energies on the polynaphthoquinones containing various amounts of FeCl, are given in table 1. Fig. 5 shows the relation between the initial rates of formation of ethylene and the FeCl, content in the catalysts ; the rates of formation of ethylene are markedly accelerated by adding FeCl, to the catalyst, the initial rates increasing linearly with the FeCl, content. The rates per active site, which were estimated from the FeCl, content, are independent of the amount of FeCl,, as shown in table 1.This indicates that all of the FeC1, behaves uniformly as active sites for the over-all reaction as well as for adsorption. The dependence of the rate of diethyl ether formation on the adsorbed amounts of ethyl alcohol is given in fig. 6. The initial rates of formation of diethyl ether are proportional to the square of the amounts of ethyl alcohol adsorbed on the catalysts. FIG. 0 0.4 0.8 1.2 CZH50H(ad)2/(cm3 (s.t.p.))' 6.-The dependence of the initial rates of the ether formation upon the square of the adsorbed amounts of ethyl alcohol over a polynaphthoquinone containing 0.046 g of FeC13 at 200°C.Y . IWASAWA, s. OGASAWARA, T. ONISHI AND K . TAMARU 199 The initial rate accordingly obeys the following equation : d[diethyl etherlldt = k,b2K2P2/(1 + KP)2.The rate constants, k3, and the activation energies on the polynaphthoquinones containing various amounts of FeCl, are given in table 1. The activation energies of the reaction do not depend on the amount of FeCl, contained in the catalysts. The initial rates of the formation of diethyl ether over polynaphthoquinone containing FeCl, are plotted against the FeCl, content in fig. 7, the initial rates showing a good correlation with the square of the FeCl, content. 0.02 1 rl 1 a .L.l E 0.01 0 FeCI, content /g FIG. 7.-The dependence of the initial rates of ether formation on the FeCl, content ; initial pressure of ethyl alcohol = 3.07 x lo3 N m-2. As described in section 1, ethyl alcohol adsorbs on the FeC1, in the catalyst. Consequently, it is concluded from these results that the mechanism of formation of diethyl ether from ethyl alcohol over the polynaphthoquinone containing FeCl, is of a Langmuir-Hinshelwood type, and a surface bimolecular reaction between ethyl alcohol molecules adsorbed on the catalyst is rate-determining.The schemes for the dehydration reactions of ethyl alcohol may be shown as follows : ,CH3 H2/ HZC, H 2 f--/ C' 'Om OH8 - H - C*H,+ H20 PI C 2 t i 50H (g 1 =,5,2> (111) (Iv) (V) (VI) The adsorption of ethyl alcohol on polynaphthoquinone containing FeCl, is represented by (111) or (V) in the scheme, where two ethyl alcohol molecules interact with each other to form diethyl ether. As is shown in (IV), the hydroxyl anion is200 REACTIONS ON POLYNAPHTHOQUINONES removed from ethyl alcohol by FeC1, acting as a Lewis acid, while a proton is removed by the quinone group, ethylene being formed.Diethyl ether is formed from two ethyl alcohol molecules adsorbed on the FeCl, through (V) and (VI). As is shown in (VI), the ethyl carbonium ion produced by removal of the hydroxyl anion by the FeCI, electrophilically attacks the oxygen atom of an adjacent ethyl alcohol molecule, while a proton is removed by the quinone group. Thus one diethyl ether molecule is produced from 2 molecules of ethyl alcohol. Pines et aL5 demonstrated that ethers formed from alcohols on supported Ni catalysts are produced from two molecules of alcohol, one adsorbing on an acid site of the catalyst and another on a basic site.On the polynaphthoquinone containing FeCl,, the FeCl, acts as a Lewis acid and the quinone group may correspond to the basic site. It is to be emphasized that the organic polymers containing quinone groups behave as typical heterogeneous catalysts on which active sites are uniformly distributed and relatively well-defined. 4. SELECTIVITY IN THE FORMATION OF ACETALDEHYDE, ETHYLENE A N D DIETHYL ETHER Dehydrogenation of ethyl alcohol over the polynaphthoquinone without FeCl, accounts for 91 % of the reaction at 150°C and this is almost independent of added FeCI, content, whereas in the case of isopropyl alcohol dehydrogenation decreases to 20 % as the FeCl, content increases. The selectivity of the reaction over the polynaphthoquinones containing FeCl, is markedly affected by the alcohols used.All of the quinone groups are not reduced, as described in the previous paper,2 but only one eighth of the quinone groups in the catalyst act as effective hydrogen acceptors. The dehydrogenation ceases when the active quinone groups in the catalyst have been reduced. Catalytic dehydrogenation proceeds in the presence of air or oxygen, redox reaction being repeated at each active site. However, dehydra- tion is much less influenced by the reduction of the quinone groups. Accordingly, dehydration rather than dehydrogenation is predominantly observed over the reduced polynaphthoquinones, thus altering the selectivity of the reaction by modifying the catalysts. As was shown in the section 2, acetaldehyde is formed by a unimolecular process via a transition state involving two hydrogen bonds. It is another characteristic property of this catalyst, different from other catalysts, that dehydrogenation may be explained as the reaction of the substrate with the quinone group.The selectivity of the formation of acetaldehyde and ethylene may be decided by the relative rates of the two reactions ; one reaction proceeds via (11) in the scheme and another by way of (IV). The dehydrogenation through (11) takes place more easily than dehydration through (IV) over the catalyst, while over the reduced catalyst ethylene rather than acetaldehyde was produced. It is necessary for the formation of diethyl ether that two molecules of ethyl alcohol are adsorbed close together, as is shown in (V) and (VI).The formation of diethyl ether, as is expected from the kinetic expression, takes place with difficulty with small amounts of adsorption or low pressure of ethyl alcohol, where the main product of dehydration is ethylene. The greater the amount of ethyl alcohol ad- sorbed, the higher the ratio of diethyl ether formed to that of ethylene. The addition of FeCI, to the catalyst increases the amount of ethyl alcohol adsorbed (see fig. 8). At FeCl, contents less than 0.046 g per 0.1 g of polynaphthoquinone, ethylene is predominantly produced, whereas formation of diethyl ether predominates at FeCl,Y . IWASAWA, S . OGASAWARA, T . ONISHI AND K . TAMARU 201 contents higher than 0.046 g. The cross-over point of the two lines for formation of ethylene and ethyl ether is shifted towards lower FeCI, contents by increasing the initial pressure of ethyl alcohol. FeCl, content /g FIG. 8.-The selectivity of formation of ethylene and diethyl ether during dehydration of ethyl alcohol over the polynaphthoquinone containing various amounts of FeCI3 at 200°C. 0, ethylene; 0, diethyl ether. In the dehydration of ethyl alcohol the ratio of the formation of diethyl ether to that of ethylene is markedly affected by the FeCI, content of the cataIysts. The selectivity of the formation of acetaldehyde, ethylene and diethyl ether may be in- fluenced by using other Lewis acids than FeCI,. J. Manassen and Sh. Khalif, J. Cufalysis, 1969, 13, 290. Y. Iwasawa, M. Soma, T. Onishi and K. Tamaru, J.C.S. Furuday I, 1972, 68, 1617. for example, W. C. Schumb, C. N. Satterfield and R. N. Wentworth, Hydrogen Peroxide (Reinhold, New York, 1955). for example, J. L. Bolland and P. Ten Have, Truns. Furaduy Soc., 1947,43,201 ; E. C. Horswill and K. U. Ingold, C a d . J. Chem., 1966,44,263, 269. J. Hensel and H. Pines, J. Catalysis, 1972, 24, 197.

ISSN:0300-9599    

DOI:10.1039/F19747000193

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Various Reactions Catalyzed by the Electron Donor-Acceptor Complex of Polynaphthoquinone with Potassium BY YASUHIRO IWASAWA, HIROSHI FUJITSU, TAKAHARU ONISHI AND KENZI TAMARU" Department of Chemistry, University of Tokyo, Hongo, Bunk y 0- ku , To k y 0, Japan Received 1 1 th April, 1973 The H2-D2 exchange reaction, the synthesis of ammonia, hydrogen exchange reaction of propene with deuterium and the isomerization of butene over the electron donor-acceptor (EDA) complexes of potassium with metal-free polynaphthoquinone and with polynaphthoquinones containing FeC13 are discussed. H2-D2 exchange over the EDA complexes is catalyzed even at -78°C. The inter- mediate in propene-deuterium exchange is the half-hydrogenated isopropyl species. The difference in the poisoning effect of oxygen for the formation of ammonia and to the H2-D2 exchange is also noted.The polynaphthoquinone obtained by oxidizing naphtha1ene-l77-diol behaves as an effective hydrogen acceptor in the dehydrogenation of various substrates, such as formic acid, cyclohexene, isopropyl alcohol and ethyl alcohol, as demonstrated previous1y.l. Though the pure polymer exhibited high catalytic activity for dehydro- genation and isomerization, molecular nitrogen or hydrogen were not dissociated to form nitrogen or hydrogen atoms on the quinone polymer, hence neither the hydrogen exchange reaction nor the formation of ammonia from nitrogen take place over the polynapht hoquinone. In the present study, high catalytic activities of the electron donor-acceptor (EDA) complexes of potassium with the polynaphthoquinones with or without FeCI, are reported for various reactions, such as H2-D2 exchange, synthesis of ammonia from nitrogen and hydrogen, hydrogen exchange of propene with deuterium and the isomerization of butene.EXPERIMENTAL The polynaphthoquinone employed was prepared as in the previous work.2 The chelate complexes of polynaphthoquinone with various metal halides, such as FeC13, FeC12, VC13 and MoC15, were prepared from mixtures of the metal-free polynaphthoquinone and known amounts of metal halides in ethyl alcohol followed by removal of the solvent by evaporation at room temperature. The polynaphthoquinones thus obtained were treated at 350°C for 12 h under vacuum before the EDA complexes with potassium were prepared. The EDA catalysts were prepared by adding potassium vapour to the metal-free polynaphthoquinone or polynaphthoquinones containing metal halides, followed by heating at 350°C under vacuum.The amount of potassium contained in the EDA complexes was estimated by measuring the pH of the aqueous solution prepared by dissolving the EDA complexes, in the presence of air. The ratio of potassium atoms to naphthoquinone monomer molecules was about 1 : 3. The EDA complex of the metal-free polynaphthoquinone with potassium was proposed to have the following structure and composition ; 202Y . IWASAWA, H. FUJITSU, T . ONISHI AND K . TAMARU 203 0 The e.s.r. spectrum of the polynaphthoquinone radical anion shows a strong sharp signal with a g value of 2.00 and peak width of 2.9 G, indicating that the radical is delocalized over the n-conjugated system of the polynaphthoquinone. In the EDA complex of potassium with the polynaphthoquinone containing Fe CI, Fe could be detected by Mossbauer spectroscopy using "FeC13 prepared from 57Fez03 dilute hydrochloric acid.The B.E.T. surface areas of the EDA catalysts of potassium with the metal-free and metal-containing polynaphthoquinones were 122 m2 g-l and 115 m2 g-l, respectively. Reactions were carried out over the EDA catalysts prepared from 0.1 or 0.4 g of poly- naphthoquinone in a closed circulating system (175 or 292cm3) with a glass U-shaped reaction vessel. The reaction products were analyzed by gas chromatography, and deuter- ated propene was analyzed by microwave and mass spectral methods.The reaction was also followed by measuring the pressure and i.r. spectra of the reaction products. RESULTS AND DISCUSSION 1. ISOMERIZATION OF BUTENE The isomerization of cis-but-2-ene to trans-but-2-ene and to but-1-ene proceeds over the EDA complexes with or without FeCl, in the temperature range from -8 to 50°C. The reaction proceeds rapidly and reaches equilibrium within 1 h at room temperature. The initial rates of formation of trans-but-2-ene are plotted against initial pressures of cis-but-2-ene in fig. 1. The logarithms of the rates vary linearly with the logarithms of the pressures of cis-but-2-ene and the slope of the straight line was unity, showing the initial rate to be proportional to the initial pressure of cis-but- 2-ene. Accordingly, the rate of formation of trans-but-2-ene is given by the equation ; d[trans-but-2-ene]/dt = k,[cis-but-2-ene).The formation of butadiene was not observed throughout the isomerization. The activation energies obtained for the reactions over the EDA catalysts are also shown in fig. 1. The isomerization takes place more readily on the EDA complex of the metal-free polynaphthoquinone with potassium than on the polynaphthoquinone complex containing FeCl,. The isomerization of but-1-ene to cis-but-2-ene and trans-but-2-ene over the EDA catalysts was also studied under similar conditions. The formation of cis-but-2-ene from but-1-ene reached a maximum value with reaction time, while that of trans-but-2- ene increases with time to reach its equilibrium value, indicating that isomerization of but- 1 -ene to trans-but-2-ene predominantly proceeds through cis-but-2-ene.In accord with this in the initial stage of the reaction the main product was cis-but-2-ene.204 REACTIONS ON POLYNAPHTHOQUINONES log(P/N m-2) 3.5 4.0 0.0030 0. b035 o.bo4o KIT FIG. 1.-Dependence of the rate of isomerization of cis-but-2ene to trans-but-2ene on the initial pressure of cis-but-2ene at 0°C and the activation energies over the EDA catalysts. 0, Fe-free, &, = 7.95 kJ mol-' ; 0, Fe-containing, EB = 16.3 kJ mol-' ; 0, t, = kt[cis-but-2ene]. 2. HYDROGEN EXCHANGE OF PROPENE WITH DEUTERIUM Hydrogen exchange between propene and deuterium was studied in the tempera- ture range from room temperature to 200°C by a microwave technique to examine the position of the hydrogen which participated in the reaction. As shown in table 1, the monodeuteriopropene produced by the exchange reaction consists of CHD=CHCH3(cis-[l-2H,] and tr~ns-[l-~H,])and CH2=CHCH,D([3-2H,]propene), very little [2-2H,]propene(CH2=CDCH3) being observed throughout the reaction.TABLE 1 .-PERCENTAGE DEUTERIUM DISTRIBUTION IN THE DEUTERIOPROPENE PRODUCED BY NAPHTHOQUINONE AT 60°C THE HYDROGEN EXCHANGE OVER THE EDA CATALYST OF POTASSIUM WITH METAL-FREE POLY- mean deuterium content* cis41 -2Hll tram-[ 1-2H11 [3-2Hi1 [2-2H11 1 .o 18 17 64.5 0.5 8.0 19 17 63.7 0.3 19.0 19 17 63.7 0.3 mean deuterium content* [l,l-*Hzl [3,3-2H21 [1 ,3-2H~1 21.0 8 29 63 6 6 f = O i = O * mean deuterium content = { X i[2Hi]/6x ['Hi]} x 100 %. In the initial stage of the reaction, three kinds of monodeuteriopropene, ~is-[l-~HJ, ?ran,~[l-~H,] and [3-2H,], are formed in the ratio 1 : 1 : 3 which remains unchanged throughout the exchange.If the reaction proceeds with a n-ally1 or o-ally1 species as an intermediate or through a concerted mechanism, [3-2H,]propene (100 %) will beY . IWASAWA, H . FUJITSU, T . ONISHI AND K . TAMARU 205 the only monodeuteriopropene produced in the initial stage of the exchange reaction, whereas if a n-propyl species is the reaction intermediate, [2-2Hl]-propene (100 %) will be produced. However, if an isopropyl species is the intermediate, ~is-[l-~H~]-, tran~-[l-~H~]- and [3-2H,]-propene will be produced in the ratio of 1 : 1 : 3, and the species, [l,l-2H2], [3,3-2H2], [1,3-2H2] in the ratio 1 : 3 : 6 .The results given in table 1 are in reasonable agreement with those expected from the mechanism involving the isopropyl species ; the half-hydrogenated state is the reaction intermediate of the hydrogen exchange reaction to the exclusion of the n-allyl, 0-ally1 and n-propyl species as intermediates and also the concerted mechanism. 3. HYDROGEN-DEUTERIUM EXCHANGE The H2-D2 exchange reaction was studied in a closed circulating system (175 cm3), using the EDA catalysts containing 0.1 g of polynaphthoquinone. The H2-D2 exchange over catalysts proceeds rapidly at room temperature and was observed even at -78"C, as shown in fig. 2. HD formation from a mixture of H2 and D2 over the EDA complex of /3-naphthoquinone with potassium is slow below room temperature, whereas the exchange reaction occurs readily on the EDA complex of potassium with violanthrone, a large condensed aromatic compound containing a ketone group.5 However, the EDA complex of polynaphthoquinone with potassium is more active than that of violanthrone.time/h FIG. 2.-HD formation from a mixture of HZ and D2 over the EDA catalysts (e, 0, Fe-free ; A, 0, Fe-containing) and their activation energies : 0 Ea = 24.3 kJ mol-' ; 0, Ea = 25.1 kJ inol-l. The initial rate of HD formation in a mixture of H2 and D2 is approximately half- order with respect to each of H2 and D2. When deuterium alone is introduced onto the EDA catalysts, the amount of HD formed is negligible, indicating that HD is produced from hydrogen and deuterium atoms derived from the gas phase on the catalyst surface though the adsorption of H2 or D2 is not appreciable.The formation of HD is accelerated by the addition of FeCl, to the EDA catalyst, whereas the206 REACTIONS ON POLYNAPHTHOQUINONES activation energies of the reaction are not influenced (fig. 2). The activation energies were found to be 2.51 x lo4 J m01-l on the EDA complex of the metal-free poly- naphthoquinone and 2.43 x lo4 J mol-1 over that of the polynaphthoquinone con- taining FeCI,. 4. CATALYTIC ACTIVITY OF THE EDA COMPLEXES OF POLYNAPHTHO- QUINONE WITH POTASSIUM IN AMMONIA SYNTHESIS Catalytic ammonia synthesis from a mixture of nitrogen and hydrogen was investigated over the EDA complexes (0.4 g polynaphthoquinone) in a closed circulat- ing system (292 cm3) with a liquid nitrogen trap. Ammonia was identified by i.r.spectroscopy and its amount estimated by pressure changes. The amount of am- monia formed at 300°C in 21 h is given in table 2. Formation of ammonia occurs over the EDA complex of potassium with the metal-free polynaphthoquinone. When the EDA complex containing FeCI, is used, however, formation of ammonia is accelerated by a factor of 13, while in the case of the EDA catalysts containing MoC1, and VCI,, it is not appreciably affected. TABLE 2.-hMONIA SYNTHESIS AND H2-D2 EXCHANGE OVER VARIOUS EDA CATALYSTS ammonia synthesis at 300°C Hz-Dz exchange reaction at 23°C ~- ~~ ~ partial pressure/l03 N m-2 partial pressure/ 103 N m-2 catalysts* Q-FeCI3 Q-K Q-FeCI3-K Q-FeC13-MoCI,-K Q-FeCI,-K Q-VCl3-K N2 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Hz 0 2 CO c 60.0 - - 60.0 - - 60.0 5.33 - 60.0 - - 60.0 5.33 - 60.0 - 6.67 60.0 - - 60.0 5.33 - 60.0 - 6.67 60.0 - - 60.0 5.33 - 60.0 - 6.67 60.0 - - 60.0 5.33 - 60.0 - 6.67 :onv.% t cm3/g 0 0 1.4 1.8 0.06 0.08 18.4 24.4 14.3 18.6 13.6 18.1 13.6 18.1 11.9 15.8 11.9 15.8 5.1 6.8 4.1 5.4 3.7 5.0 2.1 2.7 0.85 1.1 0.51 0.7 H 2 D2 conv. %$ 26.7 20.0 0 26.7 20.0 52 26.7 20.0 1.5 26.7 20.0 53 26.7 20.0 2.2 26.7 20.0 50 26.7 20.0 2.0 26.7 20.0 47 26.7 20.0 1.8 26.7 20.0 46 26.7 20.0 2.0 - - - - - - - - - - - - * metal halides (2.9 x mol), polynaphthoquinone (Q) (0.4 g, 2.5 x mol monomer), mol) ; t conversion of Nz to NH3 in 21 h ; $ conversion of D2 to HD in 10 min. potassium (0.9 x The Mossbauer spectra of the EDA catalysts containing FeCI, and FeCI, were recorded at 77 K in the absence of air to determine the oxidation state of iron con- tained in the catalysts.The sample was prepared in a dry box filled with helium gas after the formation of the EDA complex of potassium with the polynaphthoquinones containing FeCI, or Feel2 in a glass U-shaped tube under vacuum. The Mossbauer spectrum of the EDA complex obtained from potassium and the polynaphthoquinone containing Feel, gave six characteristic absorption peaks indicating the presence of Fe metal. Small amounts of FeCI2 and FeC1, in addition to Fe metal were also observed. Also, in the case of the EDA complex of the polynaphthoquinone con- taining FeCI,, a large amount of Fe metal was shown to be present. These resultsY . IWASAWA, H .FUJITSU, T . ONISHI AND K . TAMARU 207 suggest that the Fe metal may be an active site for the formation of ammonia, par- ticularly in the dissociation of nitrogen because the H2-D2 exchange reaction proceeds rapidly irrespective of the presence of Fe metal. The formation of ammonia from a mixture of nitrogen and hydrogen occurs even when oxygen (5.33 x lo3 N m-2) or carbon monoxide (6.67 x lo3 N m-’) are present over the EDA catalysts, as shown in table 2. In the case of the EDA catalyst of the polynaphthoquinone containing FeCI, the ammonia synthesis is much less affected by introducing oxygen or carbon monoxide, while in the case of the EDA complex of the metal-free polynaphthoquinone its catalytic activity decreases markedly. When oxygen is repeatedly admitted onto the EDA complexes, all of the EDA complexes employed are finally inactivated.The H2-D2 exchange reactions over these EDA complexes are drastically decreased by introducing oxygen onto the complexes, as shown in table 2. These results indicate that the dissociation of hydro- gen rather than nitrogen is most influenced by introducing oxygen into the system. The poisoning effect of oxygen on the formation of ammonia is markedly different from that of the H2-D2 exchange reaction over the EDA catalysts examined. The authors thank Dr. S . Naito for advice on microwave spectroscopy and Dr. M. Takeda for a discussion on Mossbauer spectroscopy. Y . Iwasawa, M. Soma, T. Onishi and K. Tamaru, J.C.S. Furuduy I, 1972, 68, 1617. Y . Iwasawa, S . Ogasawara, T. Onishi and K.Tamaru, J.C.S. Furuduy I, 1974, 70, 193. Y. Iwasawa, T. Onishi and K. Tamaru, Chem. Comm., 1972, 1051. K. Tamaru, Ann. N. Y. Acud. Sci., in press. M. Ichikawa, M. Soma, T. Onishi and K. Tamaru, Bull. Chem. SOC. Japan, 1967, 40, 1294 ; Y . Iwasawa, unpublished results. Various Reactions Catalyzed by the Electron Donor-Acceptor Complex of Polynaphthoquinone with Potassium BY YASUHIRO IWASAWA, HIROSHI FUJITSU, TAKAHARU ONISHI AND KENZI TAMARU" Department of Chemistry, University of Tokyo, Hongo, Bunk y 0- ku , To k y 0, Japan Received 1 1 th April, 1973 The H2-D2 exchange reaction, the synthesis of ammonia, hydrogen exchange reaction of propene with deuterium and the isomerization of butene over the electron donor-acceptor (EDA) complexes of potassium with metal-free polynaphthoquinone and with polynaphthoquinones containing FeC13 are discussed.H2-D2 exchange over the EDA complexes is catalyzed even at -78°C. The inter- mediate in propene-deuterium exchange is the half-hydrogenated isopropyl species. The difference in the poisoning effect of oxygen for the formation of ammonia and to the H2-D2 exchange is also noted. The polynaphthoquinone obtained by oxidizing naphtha1ene-l77-diol behaves as an effective hydrogen acceptor in the dehydrogenation of various substrates, such as formic acid, cyclohexene, isopropyl alcohol and ethyl alcohol, as demonstrated previous1y.l. Though the pure polymer exhibited high catalytic activity for dehydro- genation and isomerization, molecular nitrogen or hydrogen were not dissociated to form nitrogen or hydrogen atoms on the quinone polymer, hence neither the hydrogen exchange reaction nor the formation of ammonia from nitrogen take place over the polynapht hoquinone.In the present study, high catalytic activities of the electron donor-acceptor (EDA) complexes of potassium with the polynaphthoquinones with or without FeCI, are reported for various reactions, such as H2-D2 exchange, synthesis of ammonia from nitrogen and hydrogen, hydrogen exchange of propene with deuterium and the isomerization of butene. EXPERIMENTAL The polynaphthoquinone employed was prepared as in the previous work.2 The chelate complexes of polynaphthoquinone with various metal halides, such as FeC13, FeC12, VC13 and MoC15, were prepared from mixtures of the metal-free polynaphthoquinone and known amounts of metal halides in ethyl alcohol followed by removal of the solvent by evaporation at room temperature.The polynaphthoquinones thus obtained were treated at 350°C for 12 h under vacuum before the EDA complexes with potassium were prepared. The EDA catalysts were prepared by adding potassium vapour to the metal-free polynaphthoquinone or polynaphthoquinones containing metal halides, followed by heating at 350°C under vacuum. The amount of potassium contained in the EDA complexes was estimated by measuring the pH of the aqueous solution prepared by dissolving the EDA complexes, in the presence of air. The ratio of potassium atoms to naphthoquinone monomer molecules was about 1 : 3. The EDA complex of the metal-free polynaphthoquinone with potassium was proposed to have the following structure and composition ; 202Y .IWASAWA, H. FUJITSU, T . ONISHI AND K . TAMARU 203 0 The e.s.r. spectrum of the polynaphthoquinone radical anion shows a strong sharp signal with a g value of 2.00 and peak width of 2.9 G, indicating that the radical is delocalized over the n-conjugated system of the polynaphthoquinone. In the EDA complex of potassium with the polynaphthoquinone containing Fe CI, Fe could be detected by Mossbauer spectroscopy using "FeC13 prepared from 57Fez03 dilute hydrochloric acid. The B.E.T. surface areas of the EDA catalysts of potassium with the metal-free and metal-containing polynaphthoquinones were 122 m2 g-l and 115 m2 g-l, respectively. Reactions were carried out over the EDA catalysts prepared from 0.1 or 0.4 g of poly- naphthoquinone in a closed circulating system (175 or 292cm3) with a glass U-shaped reaction vessel.The reaction products were analyzed by gas chromatography, and deuter- ated propene was analyzed by microwave and mass spectral methods. The reaction was also followed by measuring the pressure and i.r. spectra of the reaction products. RESULTS AND DISCUSSION 1. ISOMERIZATION OF BUTENE The isomerization of cis-but-2-ene to trans-but-2-ene and to but-1-ene proceeds over the EDA complexes with or without FeCl, in the temperature range from -8 to 50°C. The reaction proceeds rapidly and reaches equilibrium within 1 h at room temperature. The initial rates of formation of trans-but-2-ene are plotted against initial pressures of cis-but-2-ene in fig.1. The logarithms of the rates vary linearly with the logarithms of the pressures of cis-but-2-ene and the slope of the straight line was unity, showing the initial rate to be proportional to the initial pressure of cis-but- 2-ene. Accordingly, the rate of formation of trans-but-2-ene is given by the equation ; d[trans-but-2-ene]/dt = k,[cis-but-2-ene). The formation of butadiene was not observed throughout the isomerization. The activation energies obtained for the reactions over the EDA catalysts are also shown in fig. 1. The isomerization takes place more readily on the EDA complex of the metal-free polynaphthoquinone with potassium than on the polynaphthoquinone complex containing FeCl,. The isomerization of but-1-ene to cis-but-2-ene and trans-but-2-ene over the EDA catalysts was also studied under similar conditions.The formation of cis-but-2-ene from but-1-ene reached a maximum value with reaction time, while that of trans-but-2- ene increases with time to reach its equilibrium value, indicating that isomerization of but- 1 -ene to trans-but-2-ene predominantly proceeds through cis-but-2-ene. In accord with this in the initial stage of the reaction the main product was cis-but-2-ene.204 REACTIONS ON POLYNAPHTHOQUINONES log(P/N m-2) 3.5 4.0 0.0030 0. b035 o.bo4o KIT FIG. 1.-Dependence of the rate of isomerization of cis-but-2ene to trans-but-2ene on the initial pressure of cis-but-2ene at 0°C and the activation energies over the EDA catalysts. 0, Fe-free, &, = 7.95 kJ mol-' ; 0, Fe-containing, EB = 16.3 kJ mol-' ; 0, t, = kt[cis-but-2ene].2. HYDROGEN EXCHANGE OF PROPENE WITH DEUTERIUM Hydrogen exchange between propene and deuterium was studied in the tempera- ture range from room temperature to 200°C by a microwave technique to examine the position of the hydrogen which participated in the reaction. As shown in table 1, the monodeuteriopropene produced by the exchange reaction consists of CHD=CHCH3(cis-[l-2H,] and tr~ns-[l-~H,])and CH2=CHCH,D([3-2H,]propene), very little [2-2H,]propene(CH2=CDCH3) being observed throughout the reaction. TABLE 1 .-PERCENTAGE DEUTERIUM DISTRIBUTION IN THE DEUTERIOPROPENE PRODUCED BY NAPHTHOQUINONE AT 60°C THE HYDROGEN EXCHANGE OVER THE EDA CATALYST OF POTASSIUM WITH METAL-FREE POLY- mean deuterium content* cis41 -2Hll tram-[ 1-2H11 [3-2Hi1 [2-2H11 1 .o 18 17 64.5 0.5 8.0 19 17 63.7 0.3 19.0 19 17 63.7 0.3 mean deuterium content* [l,l-*Hzl [3,3-2H21 [1 ,3-2H~1 21.0 8 29 63 6 6 f = O i = O * mean deuterium content = { X i[2Hi]/6x ['Hi]} x 100 %.In the initial stage of the reaction, three kinds of monodeuteriopropene, ~is-[l-~HJ, ?ran,~[l-~H,] and [3-2H,], are formed in the ratio 1 : 1 : 3 which remains unchanged throughout the exchange. If the reaction proceeds with a n-ally1 or o-ally1 species as an intermediate or through a concerted mechanism, [3-2H,]propene (100 %) will beY . IWASAWA, H . FUJITSU, T . ONISHI AND K . TAMARU 205 the only monodeuteriopropene produced in the initial stage of the exchange reaction, whereas if a n-propyl species is the reaction intermediate, [2-2Hl]-propene (100 %) will be produced.However, if an isopropyl species is the intermediate, ~is-[l-~H~]-, tran~-[l-~H~]- and [3-2H,]-propene will be produced in the ratio of 1 : 1 : 3, and the species, [l,l-2H2], [3,3-2H2], [1,3-2H2] in the ratio 1 : 3 : 6 . The results given in table 1 are in reasonable agreement with those expected from the mechanism involving the isopropyl species ; the half-hydrogenated state is the reaction intermediate of the hydrogen exchange reaction to the exclusion of the n-allyl, 0-ally1 and n-propyl species as intermediates and also the concerted mechanism. 3. HYDROGEN-DEUTERIUM EXCHANGE The H2-D2 exchange reaction was studied in a closed circulating system (175 cm3), using the EDA catalysts containing 0.1 g of polynaphthoquinone. The H2-D2 exchange over catalysts proceeds rapidly at room temperature and was observed even at -78"C, as shown in fig.2. HD formation from a mixture of H2 and D2 over the EDA complex of /3-naphthoquinone with potassium is slow below room temperature, whereas the exchange reaction occurs readily on the EDA complex of potassium with violanthrone, a large condensed aromatic compound containing a ketone group.5 However, the EDA complex of polynaphthoquinone with potassium is more active than that of violanthrone. time/h FIG. 2.-HD formation from a mixture of HZ and D2 over the EDA catalysts (e, 0, Fe-free ; A, 0, Fe-containing) and their activation energies : 0 Ea = 24.3 kJ mol-' ; 0, Ea = 25.1 kJ inol-l. The initial rate of HD formation in a mixture of H2 and D2 is approximately half- order with respect to each of H2 and D2.When deuterium alone is introduced onto the EDA catalysts, the amount of HD formed is negligible, indicating that HD is produced from hydrogen and deuterium atoms derived from the gas phase on the catalyst surface though the adsorption of H2 or D2 is not appreciable. The formation of HD is accelerated by the addition of FeCl, to the EDA catalyst, whereas the206 REACTIONS ON POLYNAPHTHOQUINONES activation energies of the reaction are not influenced (fig. 2). The activation energies were found to be 2.51 x lo4 J m01-l on the EDA complex of the metal-free poly- naphthoquinone and 2.43 x lo4 J mol-1 over that of the polynaphthoquinone con- taining FeCI,. 4. CATALYTIC ACTIVITY OF THE EDA COMPLEXES OF POLYNAPHTHO- QUINONE WITH POTASSIUM IN AMMONIA SYNTHESIS Catalytic ammonia synthesis from a mixture of nitrogen and hydrogen was investigated over the EDA complexes (0.4 g polynaphthoquinone) in a closed circulat- ing system (292 cm3) with a liquid nitrogen trap.Ammonia was identified by i.r. spectroscopy and its amount estimated by pressure changes. The amount of am- monia formed at 300°C in 21 h is given in table 2. Formation of ammonia occurs over the EDA complex of potassium with the metal-free polynaphthoquinone. When the EDA complex containing FeCI, is used, however, formation of ammonia is accelerated by a factor of 13, while in the case of the EDA catalysts containing MoC1, and VCI,, it is not appreciably affected. TABLE 2.-hMONIA SYNTHESIS AND H2-D2 EXCHANGE OVER VARIOUS EDA CATALYSTS ammonia synthesis at 300°C Hz-Dz exchange reaction at 23°C ~- ~~ ~ partial pressure/l03 N m-2 partial pressure/ 103 N m-2 catalysts* Q-FeCI3 Q-K Q-FeCI3-K Q-FeC13-MoCI,-K Q-FeCI,-K Q-VCl3-K N2 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 Hz 0 2 CO c 60.0 - - 60.0 - - 60.0 5.33 - 60.0 - - 60.0 5.33 - 60.0 - 6.67 60.0 - - 60.0 5.33 - 60.0 - 6.67 60.0 - - 60.0 5.33 - 60.0 - 6.67 60.0 - - 60.0 5.33 - 60.0 - 6.67 :onv.% t cm3/g 0 0 1.4 1.8 0.06 0.08 18.4 24.4 14.3 18.6 13.6 18.1 13.6 18.1 11.9 15.8 11.9 15.8 5.1 6.8 4.1 5.4 3.7 5.0 2.1 2.7 0.85 1.1 0.51 0.7 H 2 D2 conv. %$ 26.7 20.0 0 26.7 20.0 52 26.7 20.0 1.5 26.7 20.0 53 26.7 20.0 2.2 26.7 20.0 50 26.7 20.0 2.0 26.7 20.0 47 26.7 20.0 1.8 26.7 20.0 46 26.7 20.0 2.0 - - - - - - - - - - - - * metal halides (2.9 x mol), polynaphthoquinone (Q) (0.4 g, 2.5 x mol monomer), mol) ; t conversion of Nz to NH3 in 21 h ; $ conversion of D2 to HD in 10 min.potassium (0.9 x The Mossbauer spectra of the EDA catalysts containing FeCI, and FeCI, were recorded at 77 K in the absence of air to determine the oxidation state of iron con- tained in the catalysts. The sample was prepared in a dry box filled with helium gas after the formation of the EDA complex of potassium with the polynaphthoquinones containing FeCI, or Feel2 in a glass U-shaped tube under vacuum. The Mossbauer spectrum of the EDA complex obtained from potassium and the polynaphthoquinone containing Feel, gave six characteristic absorption peaks indicating the presence of Fe metal.Small amounts of FeCI2 and FeC1, in addition to Fe metal were also observed. Also, in the case of the EDA complex of the polynaphthoquinone con- taining FeCI,, a large amount of Fe metal was shown to be present. These resultsY . IWASAWA, H . FUJITSU, T . ONISHI AND K . TAMARU 207 suggest that the Fe metal may be an active site for the formation of ammonia, par- ticularly in the dissociation of nitrogen because the H2-D2 exchange reaction proceeds rapidly irrespective of the presence of Fe metal. The formation of ammonia from a mixture of nitrogen and hydrogen occurs even when oxygen (5.33 x lo3 N m-2) or carbon monoxide (6.67 x lo3 N m-’) are present over the EDA catalysts, as shown in table 2. In the case of the EDA catalyst of the polynaphthoquinone containing FeCI, the ammonia synthesis is much less affected by introducing oxygen or carbon monoxide, while in the case of the EDA complex of the metal-free polynaphthoquinone its catalytic activity decreases markedly. When oxygen is repeatedly admitted onto the EDA complexes, all of the EDA complexes employed are finally inactivated. The H2-D2 exchange reactions over these EDA complexes are drastically decreased by introducing oxygen onto the complexes, as shown in table 2. These results indicate that the dissociation of hydro- gen rather than nitrogen is most influenced by introducing oxygen into the system. The poisoning effect of oxygen on the formation of ammonia is markedly different from that of the H2-D2 exchange reaction over the EDA catalysts examined. The authors thank Dr. S . Naito for advice on microwave spectroscopy and Dr. M. Takeda for a discussion on Mossbauer spectroscopy. Y . Iwasawa, M. Soma, T. Onishi and K. Tamaru, J.C.S. Furuduy I, 1972, 68, 1617. Y . Iwasawa, S . Ogasawara, T. Onishi and K. Tamaru, J.C.S. Furuduy I, 1974, 70, 193. Y. Iwasawa, T. Onishi and K. Tamaru, Chem. Comm., 1972, 1051. K. Tamaru, Ann. N. Y. Acud. Sci., in press. M. Ichikawa, M. Soma, T. Onishi and K. Tamaru, Bull. Chem. SOC. Japan, 1967, 40, 1294 ; Y . Iwasawa, unpublished results.

ISSN:0300-9599    

DOI:10.1039/F19747000202

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

pH Effects on the Pulse Radiolysis of Deoxygenated Aqueous Solutions of Sulphur Dioxide BY TRYGVE E. ERIKSEN Department of Nuclear Chemistry, The Royal Institute of Technology, S-100 44 Stockholm 70, Sweden Receiued 8th June, 1973 Dilute deoxygenated aqueous solutions of sulphur dioxide and its related anions HSO, and SOg- in the pH-range 2-11 were pulse irradiated and the time dependence of the transient optical absorptions and electrical conductance recorded. Hydrogen atoms and OH radicals abstract hydrogen from HSO;. Also OH radicals react with SO2 and SO:- to form the radical ionqSO, with an absorption maximum at 255 nm. The SO, radical ions recombine to form sulphate (SO:-) and dithionate (S20,2-), the reaction rate and relative proportions of products being pH-dependent. In acid solutions the SO,.radical ion with an absorption maximum at 360 nm is also formed, presum- ably by reaction of hydrogen atoms with sulphur dioxide. The SO;- radical ions recombine to form dithionite (S20:-). Photochemical and pulse radiolytic studies of aqueous sulphite and dithionate solutions have been carried out by Dogliotti and Hay0n.l A transient absorption with Amax = 275 nm was observed and found to decay by second order kinetics. They also reported a strong absorption with Amax = 720 nm decaying by fist order kinetics. These absorptions were assigned to the radical ions SO,. and HS0,2-* respectively. A transient at about 720 nm was also obtained by Adams et aL2* in a 0.5 M oxygen- ated sulphite solution. However, the SO;. spectrum reported earlier by Dogliotti and Hayon suffers clearly from stray light effects and recently Hayon, Treinin and Wilf,4 in a paper published after this work was well along, have revised their earlier conclusions.The transient absorption has a maximum at il = 255 nm and decays by second order kinetics with 2k = (1.1 -t 0.2) x 109M-l s-l. The decay is hardly affected by pH in the range 4-10 and the spectrum shape also remains unchanged. The decay constant varies with sulphite concentration, as might be expected for a singly charged negative ion, and the recombination reaction gives dithionate as product. The absorption at 720nm is assigned to e; and no significant reaction between e&, and sulphite or hydrogen sulphite takes place. An oxidation reaction mechanism with SO;.and SO,. as chain carriers is also suggested. Kontges,’ however, claims that a mixture of sulphate and dithionate is formed by recombination of the radical ions SO 3 *. The autoxidation of alkaline sulphite solutions has also been studied by Zagorski, Sehested and Nielsen and a reaction mechanism with the radical ions SO 3 e, SO 4 and SO; as chain carriers is postulated. The recombination of the radical ions is assumed to give sulphate as product. In the present work, a pulse-radiolytic investigation of oxygen free solutions of sulphur dioxide and its related anions HSO, and SO$- in the pH-range 2-11 is described. The reactions were followed by spectrophotometric and conductometric methods. 208T. E. ERIKSEN 209 EXPERIMENTAL All chemicals of Merck p.a.quality were used without further purification. Triply distilled pre-irradiated water was made oxygen free by simple purging with Ar (Aga SR- quality) for several hours. N20 saturated solutions were obtained by bubbling N20 (Aga) through pre-purged water for about 15 min. The solutions were prepared by injecting small volumes (0.5-5 ml) of fairly concentrated degassed hydrogen sulphite or sulphite solutions into large volumes of water saturated with Ar or N20. All pH adjustments were made by addition of NaOH, H2S04 or HCIOs. No buffers were added. PULSE RADIOLYSIS The optical and pulsed d.c. conductometric equipments are described elsewhere.'. The microtron accelerator gives 7 MeV electrons and the beam current and pulse lengths were in the range 50-100 mA and 0.5-4 ps respectively.The cells used were of 2 cm optical path length and the irradiations were carried out at room temperature i.e., (22f 1)"C. The dose was monitored by a secondary emission chamber previously calibrated with the Fricke dosimeter. RESULTS AND DISCUSSION SPECTRA In the pH-range 4-10 a transient absorption with a maximum at 255 nm was obtained. The spectrum which is assigned to the radical ion SO;. is similar to spectra obtained earlier by other groups 4-6 in flash photolysis and pulse radiolysis work. FIG. 1.-pH dependence of initial absorption at 255nm produced from pL,e radiolysis of Ar- saturated hydrogen sulphite/sulphite solutions. C(S02- + HSO3-7 x M. The absorption intensity in Ar saturated solutions was found to increase with de- creasing pH in the pH-range 4-10 but otherwise the spectrum remained unchanged (fig.1). The absorption intensities in Ar and N20 saturated solutions at pH = 5 were nearly the same. In more acidic solutions a small shoulder on the low energy210 PULSE RADIOLYSIS OF AQUEOUS so2 side of the SO;. spectrum with a maximum at 360 nm appeared immediately after the pulse. The decay of this absorption was paralleled by the build up of a long- lived intense absorption with a maximum at 320 nm. Transient absorption spectra obtained at pH = 2.9 immediately and 400 ps after the electron pulse are shown in fig. 2. 0.10 0.OL 0.02 0.0 220 Z O 260 280 300 320 340 360 380 LOO A/nm FIG. 2.-Transient spectra produced from pulse radiolysis of aqueous hydrogen sulphite/sulphur dioxide solutions at pH = 2.9.0 initial absorption ; A absorption after 400 ps. Inserted : Oscilloscope traces 255 nm so;. 50 &division 300nm overlap SO;. arid SzOi- 50 &division 360 nm so,* 100 ps/division 220 240 260 280 300 320 340 360 380 400 A/nm FIG. 3.-Absorption spectra of dithionite ions and sulphur dioxide. 0, SzOz- ( ~ 3 2 0 = 8500); A, SOz aq ( ~ 7 6 = 600).T. E. ERIKSEN 21 1 The long-lived absorption at 320nm closely resembles the known spectrum of S 2 0 i - (fig. 3) and the precurser spectrum at 360 nm is therefore assigned to the SO,. radical ion. Similar spectra have been obtained by Hayon, Treinin and Wilf by photolysis of S2052- in fairly concentrated hydrogen sulphite solutions at pH = 4. In aqueous solutions of sulphur dioxide the following equilibria exist.g SO,+H,O+H++HSO, ; Kl = 1 .7 2 ~ M HSO;+SO$-+H+; K2 = lO-'M In very dilute solutions, S20; - can be neglected and only the first two equilibria need be considered. 2 HSO; +S2052- +H20 ; K3 = 7 x M-l. In alkaline solutions the SO;. radical ion is formed by reaction (1). OH- + SO $ - +OH-+ SO, (1) The pH dependence of the initial transient absorption at 255 nm (fig. 1) clearly indicates that hydrogen is abstracted from hydrogen sulphite by both OH. and He. OH- + HSO, +HzO + SO;* H*+HSO, *H2 +SO,* The reaction between e; and HSO, is known to be slow (k = (2f 1) x lo7 M-l s-l ) 4 9 5 and the most probable reaction is assumed to be ea; + HSO; +He + SO - . As no S 2 0 i - could be detected in neutral or slightly alkaline solutions, our results suggest that other possible reaction products can be excluded. Norman and Storey lo have reported the formation of SO;.by the one-electron reduction of hydrogen sulphite ions by Tirrr in acid solution Tir1' + HO-SO 2 -+TP + OH- + SO 2- and Hayon et aL4 infer from detection of S 2 0 i - after flashing 5 mM HSO; solutions at pH = 3.7 that the reaction H+HO-SO, +H2O+ SO,* may occur. This is, however, in contradiction to our results which indicate that hydrogen abstraction takes place. The structure of HSO, is not known but the presence of a peak at 2532 cm-1 in the infra-red spectrum of aqueous solutions confirms the presence of a S-H bond l1 and supports the structure H-SO;.. We therefore believe that the most likely path to the SO,. radical ion is the reaction Based on this assumption and taking into consideration the equilibrium H*+S02+H++S0,*.(4) HS20,+H++S20i- (K2 = 3.5 x " in calculating the yield of SO,. which has dimerized, we find from the competition between reactions (3) and (4) In order to accommodate the possibility of the reaction *H + HO-SO, + H2O + SO 2.212 PULSE RADIOLYSIS OF AQUEOUS so2 the present results would require a pH-dependent equilibrium and increasing stability of HO-SO; with decreasing pH. The molar absorption coefficients were found to be HSO, +HO-SO, E ~ ~ ~ ( ~ ~ , ~ ) = 1200k50; &36()@0;') 21 600 M-I cm-l. KINETICS The decay of the transient absorption at 255 nm was found to be second order and the rate constant to be decreasing with decreasing pH in the pH-range 7-1 1 (fig. 4). 3 4 5 6 7 8 9 10 I! PH FIG.4.-pH dependence of 2klc for reactiQn 2 SOs-+products, in aqueous hydrogen sulp_ite/sulphite solutions. The rate constants were found to be and 2k = (1.4k0.2) x lo9 M-l S-' 2k = (0.85k0.2) x lo9 M-l S-' at pH = 10 and pH = 5 respectively. The most probable recombination reactions are 2 s o ~ * + S 0 3 + S O ~ - (5) 2 S 0 , . - + S 2 0 ~ -. (6) 2 s0;.-+s20:-. (7) The decay of SO;. (360 nm) and build up of S20:- (320 nm) is second order An analysis of the decay of SO,. and parallel build up of SzO;- absorption gives 2k7 = (1.3f0.4) x lo9. This results is in marked disagreement with the value obtained by Hayon, Treinin and viz. 2k7 = (1 kO.1) x 1O1O M-l s-l.T. E. ERIKSEN 213 There is, however, a chance of cross recombination reactions between the radical ions SO;- and SO,.to give S20; - ; the discrepancy may be due partly to these reactions. Assuming reaction (7) to be diffusion controlled and taking Dso;. N cm2 s-l, the collision distance l2 corresponding to 2k = 1O1O M-ls-' is calculated to be 6.5A which seems rather long in comparison to the S-S length 2.4 A. A collision distance of 2.4A gives 2k- 1.2 x lo9 M-l s-l. CONDUCTIVITY MEASUREMENTS Measurements of the transient conductance were carried out in N,O-saturated alkaline solutions (pH- 10) and in Ar-saturated solutions in the pH-range 3.5-5. Two typical traces are inserted in fig. 5. IS -I I n 10 d, 05 0 0 50 100 150 200 250 300 tlPS FIG. 5.-Second order decay of signal voltage on the oscilloscope due to decrease in transient con- ductance in N20 saturated alkaline sulphite solutions (pH 9.5 ; C-2 x 2k = 1.4 X lo9 M-1 s-l .Inserted : Oscilloscope trace of signal voltage due to change inlconductance, in aqueous hydrogen sulphite/sulphite solutions. Upper trace : pH-9.5 ; N20 saturated solution 2 mV and 50 ps per large division C M ~ . ~ O - ~ M. Lower trace : pH = 4.5 ; Ar saturated solution 2 mV and M). 20 ps per large division C-2 x M. In the following discussion the molar conductances at 25°C are taken as l 3 The conductance in N20 saturated alkaline solutions was found to increase initially and then to decay by second order kinetics as shown in fig. 5. The saturation by N20 H+, 350; OH-, 198; 3 S,O;-, 93; 3 SO:-, 81 ; HSO, -SO;, 50; 3 SO:-, 72.214 PULSE RADIOLYSIS OF AQUEOUS SO2 ensures conversion of elq into OH- and OH- in < 0.1 p s and asg(H+) = g(e-) +g(OH-) the initial change in conductance due to the reactions e, + N20+N2 + OH- + OH.H++OH-+H20 OH* + SO 3 - -+OH-+ SO,* will be proportional to (g(0H) +g(e-)> (&OH-) + ;1(SO,-) - A ( S 0 3 -)> . Using the molar conductances tabulated above A(OH-)+;1(SO;*)-L(SO~-) = 104 R-l cm2 M-l. The recombination reaction (5) above will be followed by the very fast protolytic reactions l4 S03+H20-+2 H++SO:- Hf +OH--+H20 and changes due to reactions (5) and (6) will be and 4A(S2062-)-R(S03.) = 43 R-' cm2 M-l. Thus the overall change in conductance arising from the formation of SOi- and S20: - will be 9 and 147 R-I cm2 M-l respectively. As seen from fig. 5, the overall change in conductance clearly indicates that SO:- is formed at pH = 10.It is claimed by Kontges that the decrease in conductivity follows mixed order kinetics and the product is a mixture of SO;- and S20;-. The change in the product of concentration and equivalent conductivity can be written where AVs is the transient signal voltage, V, the voltage applied to the cell, R, a work- ing resistor and K, the cell constant. Fig. 5 shows the second order decay of the signal voltage due to the decrease in the transient conductance. Using the value -95 R-l cm2 M-l, the rate constant at pH = 10 was found to be 2k5 = 1.4 x lo9 M-I s-l, which is in very good agreement with the results from optical measurements. At pH = 4 the conductance was found to increase initially (fig. 5) and then to increase by second order kinetics. The most probable reactions are @(SOi-)+iA(SO$-)-R(OH-)-A(SO;*) = -95 R-I cm2 M-I ACL = AV~K, x 103/v,~~l4 2 so,*+so,+so~- SO,+H20-+2 H++SOi- SO$- + H+-+HSO; $A(SOi-) +3;1(H+) +QA(HSO;) -A(SO;*)-231 R-l cm2 M-' and 2 so,*-+s20,2 - 3 R ( S 2 0 , Z - ) - R ( S 0 ; .) = 43 R-1 cm2 M-l. The change in conductance was however found to be - 100 C2-I cm2 M-l. One possible explanation is that as the pH goes from pH = 10 to pH = 4 a mixture of dithionate and sulphate is formed, the relative amounts depending on the pH of the solution. This has also been confirmed in y-radiolysis experimenklT. E. ERIKSEN 215 The author wishes to thank Prof. Torbjorn Westermark for his interest in this work and the accelerator staff and Mr. S. 0. Engman for technical assistance.The financial support of the Swedish Atomic Research Council is gratefully acknowledged. L. Dogliotti and E. Hayon, J. Phys. Chem., 1968,72,1800; Nature, 1968, 218,949. G. E. Adams and J. W. Boag, Proc. Chem. Soc., 1964,112. G. E. Adams, J. W. Boag and B. D. Michael, Trans. Faraday SOC., 1965,61, 1674. E. Hayon, A. Treinin and J. Wilf, J. Amer. Chem. SOC., 1972,94 : 1,47. H. Kontges, Dissertation (Berlin, 1971). Z . P. Zagorski, K. Sehested and S. 0. Nielsen, J. Phys. Chem., 1971,75, 3510. ' T. Reitberger, to be published. T. E. Eriksen, to be published. H. V. Tartar and H. H. Garretson, J. Amer. Chem. SOC., 1941,63,808. l o R. 0. C. Norman and P. M. Storey, J. Chem. SOC. B, 1971,5,1009. l 1 Inorganic Sulphur Chemistry, ed. G. Nicless (Elsevier, Amsterdam, 1968), Chap.14. l2 Ionic Interactions, VoZ. 11 Kinetics and Structure, ed. S . Petrucci (Academic Press, N.Y., 1972), l 3 Landolt-Bornstein, Vol. 7, p. 258. l4 M. Eigen and L. De Maeyer in Investigation of Rates and Mechanisms of Reactions: Techniques l 5 K. D. Asmus, Int. J. Radiation Phys. chem., 1972, 4,417. l6 T. E. Eriksen, unpublished results. Chap. 7. of Organic Chemistry, Vol. VIII, part 11, ed. A. Weissberger. pH Effects on the Pulse Radiolysis of Deoxygenated Aqueous Solutions of Sulphur Dioxide BY TRYGVE E. ERIKSEN Department of Nuclear Chemistry, The Royal Institute of Technology, S-100 44 Stockholm 70, Sweden Receiued 8th June, 1973 Dilute deoxygenated aqueous solutions of sulphur dioxide and its related anions HSO, and SOg- in the pH-range 2-11 were pulse irradiated and the time dependence of the transient optical absorptions and electrical conductance recorded.Hydrogen atoms and OH radicals abstract hydrogen from HSO;. Also OH radicals react with SO2 and SO:- to form the radical ionqSO, with an absorption maximum at 255 nm. The SO, radical ions recombine to form sulphate (SO:-) and dithionate (S20,2-), the reaction rate and relative proportions of products being pH-dependent. In acid solutions the SO,. radical ion with an absorption maximum at 360 nm is also formed, presum- ably by reaction of hydrogen atoms with sulphur dioxide. The SO;- radical ions recombine to form dithionite (S20:-). Photochemical and pulse radiolytic studies of aqueous sulphite and dithionate solutions have been carried out by Dogliotti and Hay0n.l A transient absorption with Amax = 275 nm was observed and found to decay by second order kinetics.They also reported a strong absorption with Amax = 720 nm decaying by fist order kinetics. These absorptions were assigned to the radical ions SO,. and HS0,2-* respectively. A transient at about 720 nm was also obtained by Adams et aL2* in a 0.5 M oxygen- ated sulphite solution. However, the SO;. spectrum reported earlier by Dogliotti and Hayon suffers clearly from stray light effects and recently Hayon, Treinin and Wilf,4 in a paper published after this work was well along, have revised their earlier conclusions. The transient absorption has a maximum at il = 255 nm and decays by second order kinetics with 2k = (1.1 -t 0.2) x 109M-l s-l.The decay is hardly affected by pH in the range 4-10 and the spectrum shape also remains unchanged. The decay constant varies with sulphite concentration, as might be expected for a singly charged negative ion, and the recombination reaction gives dithionate as product. The absorption at 720nm is assigned to e; and no significant reaction between e&, and sulphite or hydrogen sulphite takes place. An oxidation reaction mechanism with SO;. and SO,. as chain carriers is also suggested. Kontges,’ however, claims that a mixture of sulphate and dithionate is formed by recombination of the radical ions SO 3 *. The autoxidation of alkaline sulphite solutions has also been studied by Zagorski, Sehested and Nielsen and a reaction mechanism with the radical ions SO 3 e, SO 4 and SO; as chain carriers is postulated.The recombination of the radical ions is assumed to give sulphate as product. In the present work, a pulse-radiolytic investigation of oxygen free solutions of sulphur dioxide and its related anions HSO, and SO$- in the pH-range 2-11 is described. The reactions were followed by spectrophotometric and conductometric methods. 208T. E. ERIKSEN 209 EXPERIMENTAL All chemicals of Merck p.a. quality were used without further purification. Triply distilled pre-irradiated water was made oxygen free by simple purging with Ar (Aga SR- quality) for several hours. N20 saturated solutions were obtained by bubbling N20 (Aga) through pre-purged water for about 15 min. The solutions were prepared by injecting small volumes (0.5-5 ml) of fairly concentrated degassed hydrogen sulphite or sulphite solutions into large volumes of water saturated with Ar or N20.All pH adjustments were made by addition of NaOH, H2S04 or HCIOs. No buffers were added. PULSE RADIOLYSIS The optical and pulsed d.c. conductometric equipments are described elsewhere.'. The microtron accelerator gives 7 MeV electrons and the beam current and pulse lengths were in the range 50-100 mA and 0.5-4 ps respectively. The cells used were of 2 cm optical path length and the irradiations were carried out at room temperature i.e., (22f 1)"C. The dose was monitored by a secondary emission chamber previously calibrated with the Fricke dosimeter. RESULTS AND DISCUSSION SPECTRA In the pH-range 4-10 a transient absorption with a maximum at 255 nm was obtained.The spectrum which is assigned to the radical ion SO;. is similar to spectra obtained earlier by other groups 4-6 in flash photolysis and pulse radiolysis work. FIG. 1.-pH dependence of initial absorption at 255nm produced from pL,e radiolysis of Ar- saturated hydrogen sulphite/sulphite solutions. C(S02- + HSO3-7 x M. The absorption intensity in Ar saturated solutions was found to increase with de- creasing pH in the pH-range 4-10 but otherwise the spectrum remained unchanged (fig. 1). The absorption intensities in Ar and N20 saturated solutions at pH = 5 were nearly the same. In more acidic solutions a small shoulder on the low energy210 PULSE RADIOLYSIS OF AQUEOUS so2 side of the SO;. spectrum with a maximum at 360 nm appeared immediately after the pulse.The decay of this absorption was paralleled by the build up of a long- lived intense absorption with a maximum at 320 nm. Transient absorption spectra obtained at pH = 2.9 immediately and 400 ps after the electron pulse are shown in fig. 2. 0.10 0.OL 0.02 0.0 220 Z O 260 280 300 320 340 360 380 LOO A/nm FIG. 2.-Transient spectra produced from pulse radiolysis of aqueous hydrogen sulphite/sulphur dioxide solutions at pH = 2.9. 0 initial absorption ; A absorption after 400 ps. Inserted : Oscilloscope traces 255 nm so;. 50 &division 300nm overlap SO;. arid SzOi- 50 &division 360 nm so,* 100 ps/division 220 240 260 280 300 320 340 360 380 400 A/nm FIG. 3.-Absorption spectra of dithionite ions and sulphur dioxide. 0, SzOz- ( ~ 3 2 0 = 8500); A, SOz aq ( ~ 7 6 = 600).T.E. ERIKSEN 21 1 The long-lived absorption at 320nm closely resembles the known spectrum of S 2 0 i - (fig. 3) and the precurser spectrum at 360 nm is therefore assigned to the SO,. radical ion. Similar spectra have been obtained by Hayon, Treinin and Wilf by photolysis of S2052- in fairly concentrated hydrogen sulphite solutions at pH = 4. In aqueous solutions of sulphur dioxide the following equilibria exist.g SO,+H,O+H++HSO, ; Kl = 1 . 7 2 ~ M HSO;+SO$-+H+; K2 = lO-'M In very dilute solutions, S20; - can be neglected and only the first two equilibria need be considered. 2 HSO; +S2052- +H20 ; K3 = 7 x M-l. In alkaline solutions the SO;. radical ion is formed by reaction (1). OH- + SO $ - +OH-+ SO, (1) The pH dependence of the initial transient absorption at 255 nm (fig.1) clearly indicates that hydrogen is abstracted from hydrogen sulphite by both OH. and He. OH- + HSO, +HzO + SO;* H*+HSO, *H2 +SO,* The reaction between e; and HSO, is known to be slow (k = (2f 1) x lo7 M-l s-l ) 4 9 5 and the most probable reaction is assumed to be ea; + HSO; +He + SO - . As no S 2 0 i - could be detected in neutral or slightly alkaline solutions, our results suggest that other possible reaction products can be excluded. Norman and Storey lo have reported the formation of SO;. by the one-electron reduction of hydrogen sulphite ions by Tirrr in acid solution Tir1' + HO-SO 2 -+TP + OH- + SO 2- and Hayon et aL4 infer from detection of S 2 0 i - after flashing 5 mM HSO; solutions at pH = 3.7 that the reaction H+HO-SO, +H2O+ SO,* may occur.This is, however, in contradiction to our results which indicate that hydrogen abstraction takes place. The structure of HSO, is not known but the presence of a peak at 2532 cm-1 in the infra-red spectrum of aqueous solutions confirms the presence of a S-H bond l1 and supports the structure H-SO;.. We therefore believe that the most likely path to the SO,. radical ion is the reaction Based on this assumption and taking into consideration the equilibrium H*+S02+H++S0,*. (4) HS20,+H++S20i- (K2 = 3.5 x " in calculating the yield of SO,. which has dimerized, we find from the competition between reactions (3) and (4) In order to accommodate the possibility of the reaction *H + HO-SO, + H2O + SO 2.212 PULSE RADIOLYSIS OF AQUEOUS so2 the present results would require a pH-dependent equilibrium and increasing stability of HO-SO; with decreasing pH.The molar absorption coefficients were found to be HSO, +HO-SO, E ~ ~ ~ ( ~ ~ , ~ ) = 1200k50; &36()@0;') 21 600 M-I cm-l. KINETICS The decay of the transient absorption at 255 nm was found to be second order and the rate constant to be decreasing with decreasing pH in the pH-range 7-1 1 (fig. 4). 3 4 5 6 7 8 9 10 I! PH FIG. 4.-pH dependence of 2klc for reactiQn 2 SOs-+products, in aqueous hydrogen sulp_ite/sulphite solutions. The rate constants were found to be and 2k = (1.4k0.2) x lo9 M-l S-' 2k = (0.85k0.2) x lo9 M-l S-' at pH = 10 and pH = 5 respectively. The most probable recombination reactions are 2 s o ~ * + S 0 3 + S O ~ - (5) 2 S 0 , .- + S 2 0 ~ -. (6) 2 s0;.-+s20:-. (7) The decay of SO;. (360 nm) and build up of S20:- (320 nm) is second order An analysis of the decay of SO,. and parallel build up of SzO;- absorption gives 2k7 = (1.3f0.4) x lo9. This results is in marked disagreement with the value obtained by Hayon, Treinin and viz. 2k7 = (1 kO.1) x 1O1O M-l s-l.T. E. ERIKSEN 213 There is, however, a chance of cross recombination reactions between the radical ions SO;- and SO,. to give S20; - ; the discrepancy may be due partly to these reactions. Assuming reaction (7) to be diffusion controlled and taking Dso;. N cm2 s-l, the collision distance l2 corresponding to 2k = 1O1O M-ls-' is calculated to be 6.5A which seems rather long in comparison to the S-S length 2.4 A.A collision distance of 2.4A gives 2k- 1.2 x lo9 M-l s-l. CONDUCTIVITY MEASUREMENTS Measurements of the transient conductance were carried out in N,O-saturated alkaline solutions (pH- 10) and in Ar-saturated solutions in the pH-range 3.5-5. Two typical traces are inserted in fig. 5. IS -I I n 10 d, 05 0 0 50 100 150 200 250 300 tlPS FIG. 5.-Second order decay of signal voltage on the oscilloscope due to decrease in transient con- ductance in N20 saturated alkaline sulphite solutions (pH 9.5 ; C-2 x 2k = 1.4 X lo9 M-1 s-l . Inserted : Oscilloscope trace of signal voltage due to change inlconductance, in aqueous hydrogen sulphite/sulphite solutions. Upper trace : pH-9.5 ; N20 saturated solution 2 mV and 50 ps per large division C M ~ . ~ O - ~ M. Lower trace : pH = 4.5 ; Ar saturated solution 2 mV and M). 20 ps per large division C-2 x M.In the following discussion the molar conductances at 25°C are taken as l 3 The conductance in N20 saturated alkaline solutions was found to increase initially and then to decay by second order kinetics as shown in fig. 5. The saturation by N20 H+, 350; OH-, 198; 3 S,O;-, 93; 3 SO:-, 81 ; HSO, -SO;, 50; 3 SO:-, 72.214 PULSE RADIOLYSIS OF AQUEOUS SO2 ensures conversion of elq into OH- and OH- in < 0.1 p s and asg(H+) = g(e-) +g(OH-) the initial change in conductance due to the reactions e, + N20+N2 + OH- + OH. H++OH-+H20 OH* + SO 3 - -+OH-+ SO,* will be proportional to (g(0H) +g(e-)> (&OH-) + ;1(SO,-) - A ( S 0 3 -)> . Using the molar conductances tabulated above A(OH-)+;1(SO;*)-L(SO~-) = 104 R-l cm2 M-l.The recombination reaction (5) above will be followed by the very fast protolytic reactions l4 S03+H20-+2 H++SO:- Hf +OH--+H20 and changes due to reactions (5) and (6) will be and 4A(S2062-)-R(S03.) = 43 R-' cm2 M-l. Thus the overall change in conductance arising from the formation of SOi- and S20: - will be 9 and 147 R-I cm2 M-l respectively. As seen from fig. 5, the overall change in conductance clearly indicates that SO:- is formed at pH = 10. It is claimed by Kontges that the decrease in conductivity follows mixed order kinetics and the product is a mixture of SO;- and S20;-. The change in the product of concentration and equivalent conductivity can be written where AVs is the transient signal voltage, V, the voltage applied to the cell, R, a work- ing resistor and K, the cell constant.Fig. 5 shows the second order decay of the signal voltage due to the decrease in the transient conductance. Using the value -95 R-l cm2 M-l, the rate constant at pH = 10 was found to be 2k5 = 1.4 x lo9 M-I s-l, which is in very good agreement with the results from optical measurements. At pH = 4 the conductance was found to increase initially (fig. 5) and then to increase by second order kinetics. The most probable reactions are @(SOi-)+iA(SO$-)-R(OH-)-A(SO;*) = -95 R-I cm2 M-I ACL = AV~K, x 103/v,~~l4 2 so,*+so,+so~- SO,+H20-+2 H++SOi- SO$- + H+-+HSO; $A(SOi-) +3;1(H+) +QA(HSO;) -A(SO;*)-231 R-l cm2 M-' and 2 so,*-+s20,2 - 3 R ( S 2 0 , Z - ) - R ( S 0 ; . ) = 43 R-1 cm2 M-l. The change in conductance was however found to be - 100 C2-I cm2 M-l. One possible explanation is that as the pH goes from pH = 10 to pH = 4 a mixture of dithionate and sulphate is formed, the relative amounts depending on the pH of the solution. This has also been confirmed in y-radiolysis experimenklT. E. ERIKSEN 215 The author wishes to thank Prof. Torbjorn Westermark for his interest in this work and the accelerator staff and Mr. S. 0. Engman for technical assistance. The financial support of the Swedish Atomic Research Council is gratefully acknowledged. L. Dogliotti and E. Hayon, J. Phys. Chem., 1968,72,1800; Nature, 1968, 218,949. G. E. Adams and J. W. Boag, Proc. Chem. Soc., 1964,112. G. E. Adams, J. W. Boag and B. D. Michael, Trans. Faraday SOC., 1965,61, 1674. E. Hayon, A. Treinin and J. Wilf, J. Amer. Chem. SOC., 1972,94 : 1,47. H. Kontges, Dissertation (Berlin, 1971). Z . P. Zagorski, K. Sehested and S. 0. Nielsen, J. Phys. Chem., 1971,75, 3510. ' T. Reitberger, to be published. T. E. Eriksen, to be published. H. V. Tartar and H. H. Garretson, J. Amer. Chem. SOC., 1941,63,808. l o R. 0. C. Norman and P. M. Storey, J. Chem. SOC. B, 1971,5,1009. l 1 Inorganic Sulphur Chemistry, ed. G. Nicless (Elsevier, Amsterdam, 1968), Chap. 14. l2 Ionic Interactions, VoZ. 11 Kinetics and Structure, ed. S . Petrucci (Academic Press, N.Y., 1972), l 3 Landolt-Bornstein, Vol. 7, p. 258. l4 M. Eigen and L. De Maeyer in Investigation of Rates and Mechanisms of Reactions: Techniques l 5 K. D. Asmus, Int. J. Radiation Phys. chem., 1972, 4,417. l6 T. E. Eriksen, unpublished results. Chap. 7. of Organic Chemistry, Vol. VIII, part 11, ed. A. Weissberger.

ISSN:0300-9599    

DOI:10.1039/F19747000208

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Electron Spin Resonance Detection of Radical Intermediates During Photo-oxidation by Metal Ions in Solution BY DAVID GREATOREX, RICHARD J. HILL, TERENCE J. KEMP* AND THOMAS J. STONE Department of Molecular Sciences, University of Warwick, Covehtry CV4 7AL, Warwickshire, England Received 9th July, 1973 Irradiation of fluid solutions of ceric and uranyl ions in various organic media in the temperature range 140-290 I( gives rise in many cases to well-resolved e.s.r. spectra of solvent-derived radicals, enabling characterisation of the photolytic process. Whilst UO3+ photo-oxidises predominantly, but not exclusively, by means of abstracting a hydrogen atom from carbon adjacent to an activating site such as -OH or -C02H, Ce(1V) attacks by a process of C-C fission ; for example RCOzH yields R- and R’CH20H yields, not R’CHOH as is customarily found at 77 K, but R’..The production of free radicals in the liquid phase in sufficient concentration (- pM) to enable electron spin resonance detection has depended upon two principal methods, namely (i) continuous-flow mixing of two reactants, one of which is either an exceptionally powerful oxidising agent, e.g. the Ti3+/H202 couple which behaves like OH- radical or a strong reducing agent, e.g. eFmm and (ii) continuous irradiation of a mixture of a photolabile molecule, such as H202 or di-t-butyl p e r ~ x i d e , ~ - ~ with a second molecule under study using an intense beam of u.-v. light or, alterna- tively, continuous or pulsed irradiation of a single molecule, or suitable solution of this molecule, with high energy election^.^ All of these methods depend upon mixing or irradiation processes taking place within the spectrometer cavity.Other developments include the use of triplet state sensitisers such as acetone * or benzo- phenone which absorb light to yield triplet states capable of abstracting hydrogen atoms from neighbouting solvent molecules. The demonstration, particularly by Kochi and his colleagues,1o* l 1 of the photo- lability of complexes of molecules such as alcohols and carboxylic acids with oxidising metal ions such as Pb(IV), Tl(III), Ce(Iv), Cu(II), etc., led us to attempt to characterise the free radical intermediates produced in the primary photochemical process, i.e. a one-electron transfer reaction. This proved successful, particularly with Ce(1V)‘ 2-14 and UO $+ ions,’ when dilute solutions of these in a solution of the organic molecule of interest were frozen to 77 K and then irradiated directly in the cavity : substantial concentrations of radicals derived from the organic solute (or solvent) were obtained, enabling ready characterisation. This technique depends on the immobility of the radicals once formed by photolysis, but suffers from two shortcomings, namely (i) the photolysis occurring at 77 K may not exactly parallel behaviour at ambient temperature, and (ii) the inherently broad nature of the solid-state e.s.r.absorption reduces the amount and precision of spectroscopic information, rendering character- isation equivocal or even impossible in some instances. In this paper we describe spectra obtained on irradiating liquid solutions of Ce(1v) and UO $ + ions ; the rapid rate of removal of organic radicals by processes of dimer- sation or secondary oxidation by residual oxidant being overcome in some, but by no 21 6D. GREATOREX, R .J . HILL, T . J . KEMP AND T . J . STONE 217 means all, cases by utilising both a high intensity u.-v. lamp (to boost the rate of production of radicals) and solvents (or co-solvents), retaining liquidity at ternpera- tures between 150 and 270 K. In general the photochemical pathway at temperatures of ca. 200 K does not differ from that at 77 K for oxidation by Ce(1V) or UOi+ ; however, whilst Ce(1V)- RCH20H complexes give radicals of type RCHOH at 77 K, the exclusive radical at the higher temperature is R e .A preliminary account of an application of this technique has been published.16 EXPERIMENTAL SOURCE The irradiation source was a Phillips CSX 900 W Xe-Hg point-source lamp mounted in a h i s s Ikon Xenoblock 111 unit of the type described in detail by Davidson and Wilson," which was itself maintained in a metal housing equipped with a cooling fan and coupled to a powerful extraction system to remove ozone. For some irradiations, the light output was filtered solely through continuously circulated water, but more usually a Chance- Pilkington OX7 filter was immersed in the water. SPECTROMETER The e.s.r. spectrometer has been described previously.'* An additional feature was the incorporation of a Biomac lo00 computer-of-average-transients (C.A.T.) supplied by Data Laboratories Ltd. of London to enable accumulation of weak e.s.r.signals. SAMPLES Variation of Ce(1V) concentration from to 5x M produced no change in the nature of the radical produced, but in the higher concentration range, the spectrum was reduced in intensity : in the lower concentration range the Ce(1V) was sometimes irradiated to exhaustion, and the optimum concentration was found to be 5 x M. Solutions were nitrogen-bubbled over a sinter for 15 min before being withdrawn into two 30 ml all-glass syringes. In order to run a spectrum over a suitably lengthy period, or to accumulate spectra in the C.A.T., it was found desirable to irradiate for up to 30min, when to prevent the consequent exhaustion of oxidant and the occurrence of secondary reactions, the sample was continuously replenished by flowing the reactant through the e.s.r.cell at a rate of ca. 2 ml min-'. At the low temperatures employed, the sample became highly viscous and the flow was effected by driving the horizontally-mounted.syringes with a 2 r.p.m. motor (Parvalux, 35 lb in.) coupled to a four-speed planet-type gearbox constructed by Mr. C. Worland of the departmental workshop. By a suitable choice of both output gear and size of syringe, flow rates of 0.5 to 15 ml min-' could be realised. The two syringes were connected via Teflon tubing and Teflon ball-and-socket joints (to allow flexibility) to a small mixer, based on the design of Dixon and Norman,19 situated at the bottom of the flow cell. Samples consisted of a solution of the metal ion in neat solvent : for solid samples, or for liquids that froze above 250 K methanol was used as a solvent or co-solvent. Ce(1V) was normally in the form of ceric ammonium nitrate.FLOW CELL This consisted of a mixer (M), an indented cooling: column (C), a flat section (F) and an exit chamber (fig. 1) : the last three sections were jacketed in glass tubing. Although the mixer was always used (to take the contents of two syringes), its use became significant only when thermally unstable mixtures e.g., of ceric salts and carboxylic acids were to be photo- lysed. The cooling was effected by passing pre-cooled nitrogen through the outer jacketing. The cell fitted into a modified variable temperature insert (Decca Radar Ltd.), and the temper- ature of the flowing solution was measured with a thermocouple (") (Comark Ltd.).Using this apparatus it was possible to study liquid phase reactions at temperatures of as low as 140 K.21 8 E . S . R . OF RADICAL INTERMEDIATES FIG. 1 .-Low-temperature e.s.r. cell for photo-irradiations of organic liquids. M, mixer ; indented cooling column ; F, flat cell ; T, electric thermometer. C, RESULTS ALCOHOLS With ceric ammonium nitrate as oxidant, irradiated methanol solution gave no signal. This was a significant result in the sense that other simple primary alcohols gave intense e.s.r. absorptions with this oxidant and also that irradiated solid solutions of ceric salts in methanol yielded the spectrum of .CH2OH. This photostability was useful, however, in providing an excellent solvent for other molecules of interest.Uranyl perchlorate (5 x lo-, M) in neat methanol gave intense, well-resolved spectra of CH20H in the temperature range 195-21 5 K ; although the relative intensity of the centre doublet decreased as the temperature was lowered, the temperature at which the couplings from the two protons appear inequivalent 2o was never reached because of the freezing of the solution. Details of coupling constants, etc. are given in table 1. With ethanol, whilst uranyl perchlorate yielded the spectrum of CH,cHOH at 185 K (table 1), ceric ammonium nitrate gave an intense 1 : 3 : 3 : 1 quartet at nearly the same temperature with a, 2.29 mT, i.e. the methyl radical is produced. To eliminate the possibility of some secondary oxidation involving the NO,.radical, a known product of ceric ammonium nitrate photolysis,2 the experiment was repeated using ceric perchlorate as oxidant ; the production of CH3- was, however, unaffected by this change. With ceric ammonium nitrate as oxidant, propan-1-01 and butan-1-01 yielded ethyl and n-propyl radicals respectively, although the y-proton coupling in the latter radical of 0.038 mT Uranyl perchlorate gave a less intense, but well- resolved, spectrum with propan-1-01 of the CH3CH2cHOH radical (table 1) ; Essentially the same pattern emerged with higher members of the series. was unresolved.TABLE SUMMARY OF E.S.R. DATA FOR PHOTO-OXIDATION OF ALCOHOLS BY METAL IONS substrate oxidant temp./K species (couplings in mT) CH3OH CH3CH20H CH3CH2CH2CH20H Ce(1V) U(V1) CH3CH2CH2CH2CH20H Ce(1V) (CH3)2CHCH20H U(VI) PhCH2CH20H U(V1) (CH3)zCHOH various 21 5 182 185 190 180 194 273 21 0 239 264 Ce(1V) 200 CH~*(UH 2.29) U(VI) 186 (CH3)ZCOH ( ~ ~ 1 .9 3 ) Broadening of central doublet observed at lower temperatures ; b spectrum accumulated in C.A.T.220 E.S.R. OF R A D I C A L INTERMEDIATES however, with butan-1-01, U(V1) photo-oxidation produced a complex spectrum which we have not fully analysed, but which exhibits a contribution from the butan-1-01 radical. Ce(1V) photo-oxidation of pentan-1-01 gave only a very poor spectrum of nine lines analysed in terms of coupling from two protons of 2.85 mT and from two further protons of 2.18 mT and assigned to n-butyl radical. Branching of the carbon chain, or the introduction of an aryl group, introduced a diversion from the pathway of RcHOH production for (UO 9 +)* : 2-methylpropan-1 -01 (isobutanol) yielded the spectrum of (CH3),&H20H whilst 2-phenylethanol gave benzyl radical (for details see table 1).Runs with secondary alcohols gave generally much inferior spectra compared with those described above. With propan-2-01, Ce(IV) gave methyl radical at 200 K whilst at 186 K U(V1) produced (CH,),COH. Butan-2-01 and pentan-2- and -3-01s gave weak signals indicating mixtures of radicals with U(V1). ETHERS Photolysis of tetrahydrofuran containing 5 x M uranyl perchlorate at 245 K yielded six groups of septets with the groups in intensity ratios of 1 : 1 : 2 : 2 : 1 : 1. The spectrum was analysed in terms of the radical with at = 1.34 (one proton), ar = 2.82 (two protons), a: = 0.10, a: = 0.19 mT.A room temperature spectrum of this radical affords at = 1.24, a; = 2.82, a! = 0.09, a: = 0.18 mT.22 A similar photo-oxidation of 1,Zdimethoxyethane at 248 K gave a spectrum in- dicating a mixture of radicals assigned to CH,0CH2CH20CH3 with aFH, = 1.71 and a&, = 0.22 mT and also CH30cHCH20CH3 with gH = 1.73, aEHz = 0.81 and A 2-methoxyethanol solution of U(V1) was examined from 233-273 K. The major species present was -CH20CH2CH20H appearing as a triplet of triplets with the centre group relatively broadened at 233 K, but in a normal 1 : 2 : 1 distribution at 273 K (details are summarised in table 2). The coupling constants agree with those given by Shiga et aZ.24 of aCHz = 1.67, aCH, = 0.20 mT. = 0.20 mT (in conformity with the data of Norman and colleague^).,^ CARBOXYLIC ACIDS The procedure here was to photolyse 2 M solutions of the acid in methanol containing 5 x lo-, M ceric ammonium nitrate.Intense and well-resolved spectra of the radical R. were obtained from RC0,H in the cases of acetic, propionic, isobutyric, pivalic and cyclohexane carboxylic acids, and with pivalic acid, where Re = t-butyl radical, the second-order structure of each line was well-resolved, the fifth line, for example, being split into four components with relative intensities of 8 : 28 : 49 : 36 and couplings of 23, 63 and 117 pT, in fair agreement with the theoretical values 25 of the intensities 1 : 8 : 27 : 48 : 42 and the couplings of 24, 71 and 142 pT. Tri- fluoroacetic acid gave no resonance on photo-oxidation by Ce(1V).Other experi- mental data are compiled in table 3 and the solution spectrum of cyclohexyl radical is shown in fig. 2 as representative of those we have dealt with. It was possible to obtain reasonably strong resonance from irradiated solutions of uranyl perchlorate in neat acetic and propionic acids at room temperature and 255 K respectively : the radicals were, respectively, CH2C02H and CH3tHCO2H (for details see table 3).substrate tetrahydro furan TABLE 2.-sUMMARY OF E.S.R. DATA FOR PHOTO-OXIDATION OF ETHERS BY METAL IONS ? E c oxidant temp. / K species (couplings in mT) U(W 245 OCH2CH2CH2CH (uH(=) 1.34, aH(B) 2 . 8 2 , U H ( , ) 0 . 1 9 ) 1 J r a Central group of lines'relatively broadened at 233 K. mTABLE 3.-sUMMARY OF E.S.R.DATA FOR PHOTO-OXIDATION OF CARBOXYLIC ACIDS AND HYDROXYACIDS AND THEIR ESTERS BY METAL IONS a substrate oxidant temp./K species (couplings in mT) CH3C02H Ce(1V) 190 CH3* ( ~ ~ 2 . 2 9 ) UWI) 293 CH2C02H (QH 2.1 1) P h2C(OH)C02H CH3CH(0H)C02Bun 234 c~s-CH~C(OH)CO~CH~CH~CHCH~ (@H3 1.668, Q ~ H 0.265, 0.165) + ~~W~S-CH~C(~H)CO~CH~CH~CH~CH~ 1.656, agH 0.236, @HZ 0.1 16) d 2 Ceric ammonium nitrate photo-oxidations were performed in methanolic solution : uranyl perchlorate photo-oxidations were carried out in neat acid solution ; 6 second order couplings given io. text ; C we thank Mr. M. J. Welbourn for carrying out this expximent ; d Fessenden et aL2' give cis- CH~~(OH)CO~CH~CHJ (&I, 1.674, a& 0.223, c&, 0.151) ; trans-isomer (a&3 1.641, a& 0.192, 0.117).D. GREATOREX, R .J . HILL, T . J . KEMP AND T. J . STONE 223 , 4.0 mT , FIG. 2.-E.s.r. spectrum of cyclohexyl radical obtained by irradiating methanolic ceric ammonium nitrate containing cyclohexanecarboxylic acid at 203 K. HYDROXYACIDS Neat lactic acid at room temperature underwent photo-oxidation by U(V1) perchlorate to giveaquartet of doublets assigned to CH,t(OH)CO,Hwith a&, 1.66mT, u& 225 pT, in agreement with Norman et aZ.26 A methanolic solution M) of benzilic acid was photolysed in the presence of Fe(II1) perchlorate M) and also uranyl perchlorate to give a well-resolved spectrum of the benzophenone ketyl neutral radical, Ph2COH, with couplings listed in table 3 in good agreement with those of Wilson who obtained this radical by irradiation of benzophenone in hydrogen- donor solvents.CARBOXYLIC ESTERS Both of the esters examined presented difficulties of interpretation. Photo- oxidation of n-butyl lactate and diethyl succinate as neat liquids by U(V1) perchlorate at 234 and 258 K respectively gave spectra of reasonable quality. The spectrum from the di-ester could not be analysed, but that from n-butyl lactate (fig. 3) was interpreted as arising from a mixture of cis- and trans-conformations of the radical CH3e(OJ3)CO2CH2CH2CH2CH3 ; following an analogous interpretation by Fessenden et aZ.,27 we have been able to extract coupling constants for both confor- meric forms of the radical, and our results are in good agreement with theirs (table 3). The assignment of each member of the mixture spectrum to a particular conformation is also based on the arguments of Fessenden et aZ.27 A number of other esters were examined with U(V1) as photo-oxidant, including ethyl acetate, dimethyl succinate, ethyl lactate and methyl propionate, but no signals were observed in any of these cases.224 E .S . R . OF RADICAL INTERMEDIATES 0.4mT 'I - FIG. 3.-(u) Full e.s.r. spectrum of cis- and trans- forms of CH3c(OH)CO2Bun obtained during U(V1) photo-oxidation at 234 K ; (6) computer simulation of (a) based on coupling constants given in table 3 ; (c) third group of lines on expanded field scale. DISCUSSION PHOTOCHEMICAL MECHANISMS It is evident from the results that detectable concentrations of radicals can be realised during the photo-irradiation of solutions of Ce(IV) and U(V1) in liquid organic media using the experimental arrangement described, i.e., incorporating an intense source, low temperatures and sample renewal.However, one drawback, particularly apparent with U(VI), is that a number of substrates give no resonance for no obvious reason, thus the lower primary alcohols RCHzOH give RcHOH with U(V1) as photo-oxidant, whilst RIR,CHOH yields rather weak spectra in all cases other than propan-2-01 : again, very few esters yield any absorption with U(VI), but n-butyl lactate gives an intense, complex and well-resolved spectrum. Another generalisation which can be made is that U(V1) functions by abstracting hydrogen atoms from C-H bonds adjacent to some activating group, for example, alcohols (with the exception of 2-phenylethanol) yield hydroxyalkyl radicals, carboxy- lic acids RCH2C02H produce RCHCO,H, ethers afford -CH,OcH- species, and RCH(OH)CO,H similarly gives Rc(OH)CO,H, although Ph,C(OH)CO,H undergoes oxidative decarboxylation to give the benzophenone ketyl radical, which we also characterised optically by means of a flash photolysis experiment.16 In some instances more than one C-H site is attacked and two radicals are formed, e.g.from ethylene glycol dimethyl ether. U(V1) resembles, therefore, in its action such species as OH- and ButO- and its behaviour at temperatures between 140 and 300 K closely mirrors that we have found on photo-irradiation in rigid matrices at 77 K. One feature of note is the behaviour of (UO; +)* towards methanol. The sole radical *CH20CH2CH20CH3 + CHSOCHCH20CH3D.GREATOREX, R. J . HILL, T. J . KEMP AND T. J . STONE 225 we observe at both 77 K and 215 K is CH20H; however, Ledwith et aL2' have succeeded in spin-trapping CH30- during the photo-oxidation of neat methanol by uranyl nitrate at room temperature, a result which implies that our own observation relates to a secondary radical formed by the fast attack of CH30- upon CH30H even at 77K, (it should be pointed out that in aqueous methanol, the atbck of (UOt,+)* is upon C-H bonds, as evinced by a primary kinetic isotope effect in fluorescence quenching of (UO2,+)* by dilute, aqueous CH30H and CD30H,29 and, moreover, only CH20H radicals were spin-trapped in 2 : 1 (v/v) watermethanol by Ledwith et aL2'). With ethanol and higher alcohols, our observation of RcHOH accords well with the appreciable kinetic isotope effect observed in both the photochemical oxidation by uranyl ion of CH3CH20H and CD3CD20H 30 and the Stern-Volmer quenching constants of uranyl ion fluorescence by the two substrate^.^ Again, spin-trapping experiments indicate that U(V1) nitrate photo-oxidations of ethanol and butan-1-01 yield exclusively the hydroxyalkyl radical.28 C-C cleavage occurs with the alcohols we examined solely in the case of 2- phenylethanol (to give benzyl radical), reproducing the behaviour established in the solid state.15 This fragmentation might be concerted, viz. hv PhCH2-CH2-OH -+ PhCH2* + CH2=O + Hf + U(V) 1 W I ) or it could involve fission of a primarily formed 2-phenylethanoxy radical, PhCH2- CH20*.Ledwith et aL2' make a brief reference to spin-trapping of both PhCH2- and PhCH2CH20- in the oxidation of 2-phenylethanol by Ag+/S20,2- couple, but the latter oxidant is particularly prone to attacking 0-H bonds in primary and secondary alcohols to give alkoxyl radicals rather than hydroxyalkyl radicals.28 Ceric ion behaves in a rather unprecedented fashion in its photo-oxidation of primary alcohols. Thermal oxidation of these substrates by Ce(1V) is generally considered to proceed through a breakdown of a 1 : 1 Ce(IV)-alcohol complex involving C-H fission to give an intermediate hydroxyalkyl and our studies l2 of the photo-induced breakdown of such complexes at 77 K confirmed the production of RcHOH from RCH20H, although very minor production of a second radical displaying the relatively narrow lines of small alkyl radicals in the solid state was perceptible (fig.l(d) and (e) in ref. (12)). At temperatures of ca. 200 K, however, the exclusive radical found is that corresponding to loss of *CH20H from RCH20H, implying a pathway of C-C cleavage established for the thermal and photo-oxidation of tertiary alcohols by Ce(IV),33 viz. hv R-CH2-OH -+ Re + CH,=O + H+ + Ce(II1). 1 Ce(1V) Evidently the two photochemical pathways, ix., to give Re and RCHOH, have different energies of activation and the former, presumably of higher E,, predominates when T>200 K. In its behaviour with carboxylic acids, Ce(1V) reproduces the behaviour found at 77 K in oxidatively decarboxylating the acid RC02H hv R-C-OH -+ Re + C02 + H+ + Ce(II1) II 1 0 Ce(IV)226 E .S . R . OF RADICAL INTERMEDIATES COUPLING CONSTANTS The vast majority of the radicals we have described are commonplace and require no comment. The spectrum of cyclohexyl radical obtained in solution at 203 K compares well with that of Ogawa and Fessenden 33 measured in the solid state at 193 K, i.e., the " collapse " to a six-line spectrum of the type found at 273 K in the solid has not occurred. The resolution of cis- and trans-forms of the radical CH,c(OH)C02Bun has been discussed in the experimental section. We thank the S.R.C. for studentships to D. G. and R. J. H. and for grants to pur- chase the spectrometer and the C.A.T. R. 0. C. Norman, Chem. SOC. Special Publication No. 24, Chap. 6. T. J. Kemp and T.J. Stone, ref. (l), Chap. 14. R. Livingston and H. Zeldes, J. Chem. Phys., 1966,44, 1245. J. K. Kochi and P. J. Krusic, ref. (l), Chap. 7. A. Hudson and H. A. Hussain, J. Chem. Sac. B, 1969, 793. A. G. Davies, D. Griller and B. P. Roberts, J.C.S. Perkin 11, 1972, 993. H. Zeldes and R. Livingston, J. Chem. Phys., 1966,45, 1946. R. Wilson, J. Chem. SOC. B, 1968, 84. R. A. Sheldon and J. K. Kochi, J. Amer. Chem. SOC., 1968,90,6688. ' R. W. Fessenden and R. H. Schuler, J. Chem. Phys., 1963,39,2147. lo J. K. Kochi, R. A. Sheldon and S. S. Lande, Tetrahedron, 1969, 25, 1197. l 2 D. Greatorex and T. J. Kemp, Trans. Faraday SOC., 1971, 67, 56. l3 D. Greatorex and T. J. Kemp, Trans. Faraday SOC., 1971, 67, 1576. l4 D. Greatorex and T. J. Kemp, J.C.S. Faraday I, 1972, 68, 121.l5 D. Greatorex, R. J. Hill, T. J. Kemp and T. J. Stone, J.C.S. Faraday 1, 1972, 68, 2059. l6 H. D. Burrows, D. Greatorex and T. J. Kemp, J. Amer. Chem. SOC., 1971,93,2539. l7 R. S. Davidson and R. Wilson, J. Chem. SOC. B, 1970, 71. l 8 A. R. Buick, T. J. Kemp, G. T. Neal and T. J. Stone, J. Chem. SOC. A, 1969, 666. l9 W. T. Dixon and R. 0. C. Norman, J. Chem. SOC., 1963, 3119. 'O A. Hudson, J. Chem. SOC. A, 1969,2513. 21 T. W. Martin, L. L. Swift and J. H. Venable, Jr., J. Chem. Phys., 1970, 52, 2138. 22 W. T. Dixon and R. 0. C. Norman, J. Chem. SOC., 1964,4850. 23 D. J. Edge, B. C. Gilbert, R. 0. C. Norman and P. R. West, J. Chem. SOC. B, 1971,189. 24 T. S. Shiga, A. Boukhors and P. Douzou, J. Phys. Chem., 1967, 71, 3559. 25 R. W. Fessenden, J. Chem. Phys., 1962,37,747.26 W. T. Dixon, R. 0. C. Norman and (in part) A. L. Buley, J. Chem. SOC., 1964, 3625. 27 A. Samuni, D. Behar and R. W. Fessenden, J. Phys. Chem., 1973,77,777. 28 A. Ledwith, P. J. Russell and L. H. Sutcliffe, Proc. Roy. SOC. A, 1973, 332, 151. 29 D. M. Allen, H. D. Burrows, A. Cox, R. J. Hill, T. J. Keinp and T. J. Stone, Chem. Comm., 30 S. Sakuraba and R. Matsushima, Bull. Chem. SOC. Japan, 1970, 43, 2359. 31 R. Matsushima and S. Sakuraba, J. Amer. Chem. SOC., 1971, 93, 5421. 32 W. H. Richardson in Oxidation in Organic Chemistry, ed. K. Wiberg (Academic Press, New 33 S. Ogawa and R. W. Fessenden, J. Chem. Phys., 1964, 41, 994. 1973, 59. York, 1965), Chap. 4. Electron Spin Resonance Detection of Radical Intermediates During Photo-oxidation by Metal Ions in Solution BY DAVID GREATOREX, RICHARD J.HILL, TERENCE J. KEMP* AND THOMAS J. STONE Department of Molecular Sciences, University of Warwick, Covehtry CV4 7AL, Warwickshire, England Received 9th July, 1973 Irradiation of fluid solutions of ceric and uranyl ions in various organic media in the temperature range 140-290 I( gives rise in many cases to well-resolved e.s.r. spectra of solvent-derived radicals, enabling characterisation of the photolytic process. Whilst UO3+ photo-oxidises predominantly, but not exclusively, by means of abstracting a hydrogen atom from carbon adjacent to an activating site such as -OH or -C02H, Ce(1V) attacks by a process of C-C fission ; for example RCOzH yields R- and R’CH20H yields, not R’CHOH as is customarily found at 77 K, but R’..The production of free radicals in the liquid phase in sufficient concentration (- pM) to enable electron spin resonance detection has depended upon two principal methods, namely (i) continuous-flow mixing of two reactants, one of which is either an exceptionally powerful oxidising agent, e.g. the Ti3+/H202 couple which behaves like OH- radical or a strong reducing agent, e.g. eFmm and (ii) continuous irradiation of a mixture of a photolabile molecule, such as H202 or di-t-butyl p e r ~ x i d e , ~ - ~ with a second molecule under study using an intense beam of u.-v. light or, alterna- tively, continuous or pulsed irradiation of a single molecule, or suitable solution of this molecule, with high energy election^.^ All of these methods depend upon mixing or irradiation processes taking place within the spectrometer cavity.Other developments include the use of triplet state sensitisers such as acetone * or benzo- phenone which absorb light to yield triplet states capable of abstracting hydrogen atoms from neighbouting solvent molecules. The demonstration, particularly by Kochi and his colleagues,1o* l 1 of the photo- lability of complexes of molecules such as alcohols and carboxylic acids with oxidising metal ions such as Pb(IV), Tl(III), Ce(Iv), Cu(II), etc., led us to attempt to characterise the free radical intermediates produced in the primary photochemical process, i.e. a one-electron transfer reaction. This proved successful, particularly with Ce(1V)‘ 2-14 and UO $+ ions,’ when dilute solutions of these in a solution of the organic molecule of interest were frozen to 77 K and then irradiated directly in the cavity : substantial concentrations of radicals derived from the organic solute (or solvent) were obtained, enabling ready characterisation.This technique depends on the immobility of the radicals once formed by photolysis, but suffers from two shortcomings, namely (i) the photolysis occurring at 77 K may not exactly parallel behaviour at ambient temperature, and (ii) the inherently broad nature of the solid-state e.s.r. absorption reduces the amount and precision of spectroscopic information, rendering character- isation equivocal or even impossible in some instances. In this paper we describe spectra obtained on irradiating liquid solutions of Ce(1v) and UO $ + ions ; the rapid rate of removal of organic radicals by processes of dimer- sation or secondary oxidation by residual oxidant being overcome in some, but by no 21 6D.GREATOREX, R . J . HILL, T . J . KEMP AND T . J . STONE 217 means all, cases by utilising both a high intensity u.-v. lamp (to boost the rate of production of radicals) and solvents (or co-solvents), retaining liquidity at ternpera- tures between 150 and 270 K. In general the photochemical pathway at temperatures of ca. 200 K does not differ from that at 77 K for oxidation by Ce(1V) or UOi+ ; however, whilst Ce(1V)- RCH20H complexes give radicals of type RCHOH at 77 K, the exclusive radical at the higher temperature is R e . A preliminary account of an application of this technique has been published.16 EXPERIMENTAL SOURCE The irradiation source was a Phillips CSX 900 W Xe-Hg point-source lamp mounted in a h i s s Ikon Xenoblock 111 unit of the type described in detail by Davidson and Wilson," which was itself maintained in a metal housing equipped with a cooling fan and coupled to a powerful extraction system to remove ozone.For some irradiations, the light output was filtered solely through continuously circulated water, but more usually a Chance- Pilkington OX7 filter was immersed in the water. SPECTROMETER The e.s.r. spectrometer has been described previously.'* An additional feature was the incorporation of a Biomac lo00 computer-of-average-transients (C.A.T.) supplied by Data Laboratories Ltd. of London to enable accumulation of weak e.s.r.signals. SAMPLES Variation of Ce(1V) concentration from to 5x M produced no change in the nature of the radical produced, but in the higher concentration range, the spectrum was reduced in intensity : in the lower concentration range the Ce(1V) was sometimes irradiated to exhaustion, and the optimum concentration was found to be 5 x M. Solutions were nitrogen-bubbled over a sinter for 15 min before being withdrawn into two 30 ml all-glass syringes. In order to run a spectrum over a suitably lengthy period, or to accumulate spectra in the C.A.T., it was found desirable to irradiate for up to 30min, when to prevent the consequent exhaustion of oxidant and the occurrence of secondary reactions, the sample was continuously replenished by flowing the reactant through the e.s.r. cell at a rate of ca.2 ml min-'. At the low temperatures employed, the sample became highly viscous and the flow was effected by driving the horizontally-mounted.syringes with a 2 r.p.m. motor (Parvalux, 35 lb in.) coupled to a four-speed planet-type gearbox constructed by Mr. C. Worland of the departmental workshop. By a suitable choice of both output gear and size of syringe, flow rates of 0.5 to 15 ml min-' could be realised. The two syringes were connected via Teflon tubing and Teflon ball-and-socket joints (to allow flexibility) to a small mixer, based on the design of Dixon and Norman,19 situated at the bottom of the flow cell. Samples consisted of a solution of the metal ion in neat solvent : for solid samples, or for liquids that froze above 250 K methanol was used as a solvent or co-solvent.Ce(1V) was normally in the form of ceric ammonium nitrate. FLOW CELL This consisted of a mixer (M), an indented cooling: column (C), a flat section (F) and an exit chamber (fig. 1) : the last three sections were jacketed in glass tubing. Although the mixer was always used (to take the contents of two syringes), its use became significant only when thermally unstable mixtures e.g., of ceric salts and carboxylic acids were to be photo- lysed. The cooling was effected by passing pre-cooled nitrogen through the outer jacketing. The cell fitted into a modified variable temperature insert (Decca Radar Ltd.), and the temper- ature of the flowing solution was measured with a thermocouple (") (Comark Ltd.). Using this apparatus it was possible to study liquid phase reactions at temperatures of as low as 140 K.21 8 E .S . R . OF RADICAL INTERMEDIATES FIG. 1 .-Low-temperature e.s.r. cell for photo-irradiations of organic liquids. M, mixer ; indented cooling column ; F, flat cell ; T, electric thermometer. C, RESULTS ALCOHOLS With ceric ammonium nitrate as oxidant, irradiated methanol solution gave no signal. This was a significant result in the sense that other simple primary alcohols gave intense e.s.r. absorptions with this oxidant and also that irradiated solid solutions of ceric salts in methanol yielded the spectrum of .CH2OH. This photostability was useful, however, in providing an excellent solvent for other molecules of interest. Uranyl perchlorate (5 x lo-, M) in neat methanol gave intense, well-resolved spectra of CH20H in the temperature range 195-21 5 K ; although the relative intensity of the centre doublet decreased as the temperature was lowered, the temperature at which the couplings from the two protons appear inequivalent 2o was never reached because of the freezing of the solution.Details of coupling constants, etc. are given in table 1. With ethanol, whilst uranyl perchlorate yielded the spectrum of CH,cHOH at 185 K (table 1), ceric ammonium nitrate gave an intense 1 : 3 : 3 : 1 quartet at nearly the same temperature with a, 2.29 mT, i.e. the methyl radical is produced. To eliminate the possibility of some secondary oxidation involving the NO,. radical, a known product of ceric ammonium nitrate photolysis,2 the experiment was repeated using ceric perchlorate as oxidant ; the production of CH3- was, however, unaffected by this change.With ceric ammonium nitrate as oxidant, propan-1-01 and butan-1-01 yielded ethyl and n-propyl radicals respectively, although the y-proton coupling in the latter radical of 0.038 mT Uranyl perchlorate gave a less intense, but well- resolved, spectrum with propan-1-01 of the CH3CH2cHOH radical (table 1) ; Essentially the same pattern emerged with higher members of the series. was unresolved.TABLE SUMMARY OF E.S.R. DATA FOR PHOTO-OXIDATION OF ALCOHOLS BY METAL IONS substrate oxidant temp./K species (couplings in mT) CH3OH CH3CH20H CH3CH2CH2CH20H Ce(1V) U(V1) CH3CH2CH2CH2CH20H Ce(1V) (CH3)2CHCH20H U(VI) PhCH2CH20H U(V1) (CH3)zCHOH various 21 5 182 185 190 180 194 273 21 0 239 264 Ce(1V) 200 CH~*(UH 2.29) U(VI) 186 (CH3)ZCOH ( ~ ~ 1 .9 3 ) Broadening of central doublet observed at lower temperatures ; b spectrum accumulated in C.A.T.220 E.S.R. OF R A D I C A L INTERMEDIATES however, with butan-1-01, U(V1) photo-oxidation produced a complex spectrum which we have not fully analysed, but which exhibits a contribution from the butan-1-01 radical. Ce(1V) photo-oxidation of pentan-1-01 gave only a very poor spectrum of nine lines analysed in terms of coupling from two protons of 2.85 mT and from two further protons of 2.18 mT and assigned to n-butyl radical. Branching of the carbon chain, or the introduction of an aryl group, introduced a diversion from the pathway of RcHOH production for (UO 9 +)* : 2-methylpropan-1 -01 (isobutanol) yielded the spectrum of (CH3),&H20H whilst 2-phenylethanol gave benzyl radical (for details see table 1).Runs with secondary alcohols gave generally much inferior spectra compared with those described above. With propan-2-01, Ce(IV) gave methyl radical at 200 K whilst at 186 K U(V1) produced (CH,),COH. Butan-2-01 and pentan-2- and -3-01s gave weak signals indicating mixtures of radicals with U(V1). ETHERS Photolysis of tetrahydrofuran containing 5 x M uranyl perchlorate at 245 K yielded six groups of septets with the groups in intensity ratios of 1 : 1 : 2 : 2 : 1 : 1. The spectrum was analysed in terms of the radical with at = 1.34 (one proton), ar = 2.82 (two protons), a: = 0.10, a: = 0.19 mT. A room temperature spectrum of this radical affords at = 1.24, a; = 2.82, a! = 0.09, a: = 0.18 mT.22 A similar photo-oxidation of 1,Zdimethoxyethane at 248 K gave a spectrum in- dicating a mixture of radicals assigned to CH,0CH2CH20CH3 with aFH, = 1.71 and a&, = 0.22 mT and also CH30cHCH20CH3 with gH = 1.73, aEHz = 0.81 and A 2-methoxyethanol solution of U(V1) was examined from 233-273 K.The major species present was -CH20CH2CH20H appearing as a triplet of triplets with the centre group relatively broadened at 233 K, but in a normal 1 : 2 : 1 distribution at 273 K (details are summarised in table 2). The coupling constants agree with those given by Shiga et aZ.24 of aCHz = 1.67, aCH, = 0.20 mT. = 0.20 mT (in conformity with the data of Norman and colleague^).,^ CARBOXYLIC ACIDS The procedure here was to photolyse 2 M solutions of the acid in methanol containing 5 x lo-, M ceric ammonium nitrate.Intense and well-resolved spectra of the radical R. were obtained from RC0,H in the cases of acetic, propionic, isobutyric, pivalic and cyclohexane carboxylic acids, and with pivalic acid, where Re = t-butyl radical, the second-order structure of each line was well-resolved, the fifth line, for example, being split into four components with relative intensities of 8 : 28 : 49 : 36 and couplings of 23, 63 and 117 pT, in fair agreement with the theoretical values 25 of the intensities 1 : 8 : 27 : 48 : 42 and the couplings of 24, 71 and 142 pT. Tri- fluoroacetic acid gave no resonance on photo-oxidation by Ce(1V). Other experi- mental data are compiled in table 3 and the solution spectrum of cyclohexyl radical is shown in fig.2 as representative of those we have dealt with. It was possible to obtain reasonably strong resonance from irradiated solutions of uranyl perchlorate in neat acetic and propionic acids at room temperature and 255 K respectively : the radicals were, respectively, CH2C02H and CH3tHCO2H (for details see table 3).substrate tetrahydro furan TABLE 2.-sUMMARY OF E.S.R. DATA FOR PHOTO-OXIDATION OF ETHERS BY METAL IONS ? E c oxidant temp. / K species (couplings in mT) U(W 245 OCH2CH2CH2CH (uH(=) 1.34, aH(B) 2 . 8 2 , U H ( , ) 0 . 1 9 ) 1 J r a Central group of lines'relatively broadened at 233 K. mTABLE 3.-sUMMARY OF E.S.R. DATA FOR PHOTO-OXIDATION OF CARBOXYLIC ACIDS AND HYDROXYACIDS AND THEIR ESTERS BY METAL IONS a substrate oxidant temp./K species (couplings in mT) CH3C02H Ce(1V) 190 CH3* ( ~ ~ 2 .2 9 ) UWI) 293 CH2C02H (QH 2.1 1) P h2C(OH)C02H CH3CH(0H)C02Bun 234 c~s-CH~C(OH)CO~CH~CH~CHCH~ (@H3 1.668, Q ~ H 0.265, 0.165) + ~~W~S-CH~C(~H)CO~CH~CH~CH~CH~ 1.656, agH 0.236, @HZ 0.1 16) d 2 Ceric ammonium nitrate photo-oxidations were performed in methanolic solution : uranyl perchlorate photo-oxidations were carried out in neat acid solution ; 6 second order couplings given io. text ; C we thank Mr. M. J. Welbourn for carrying out this expximent ; d Fessenden et aL2' give cis- CH~~(OH)CO~CH~CHJ (&I, 1.674, a& 0.223, c&, 0.151) ; trans-isomer (a&3 1.641, a& 0.192, 0.117).D. GREATOREX, R . J . HILL, T . J .KEMP AND T. J . STONE 223 , 4.0 mT , FIG. 2.-E.s.r. spectrum of cyclohexyl radical obtained by irradiating methanolic ceric ammonium nitrate containing cyclohexanecarboxylic acid at 203 K. HYDROXYACIDS Neat lactic acid at room temperature underwent photo-oxidation by U(V1) perchlorate to giveaquartet of doublets assigned to CH,t(OH)CO,Hwith a&, 1.66mT, u& 225 pT, in agreement with Norman et aZ.26 A methanolic solution M) of benzilic acid was photolysed in the presence of Fe(II1) perchlorate M) and also uranyl perchlorate to give a well-resolved spectrum of the benzophenone ketyl neutral radical, Ph2COH, with couplings listed in table 3 in good agreement with those of Wilson who obtained this radical by irradiation of benzophenone in hydrogen- donor solvents.CARBOXYLIC ESTERS Both of the esters examined presented difficulties of interpretation. Photo- oxidation of n-butyl lactate and diethyl succinate as neat liquids by U(V1) perchlorate at 234 and 258 K respectively gave spectra of reasonable quality. The spectrum from the di-ester could not be analysed, but that from n-butyl lactate (fig. 3) was interpreted as arising from a mixture of cis- and trans-conformations of the radical CH3e(OJ3)CO2CH2CH2CH2CH3 ; following an analogous interpretation by Fessenden et aZ.,27 we have been able to extract coupling constants for both confor- meric forms of the radical, and our results are in good agreement with theirs (table 3). The assignment of each member of the mixture spectrum to a particular conformation is also based on the arguments of Fessenden et aZ.27 A number of other esters were examined with U(V1) as photo-oxidant, including ethyl acetate, dimethyl succinate, ethyl lactate and methyl propionate, but no signals were observed in any of these cases.224 E .S . R . OF RADICAL INTERMEDIATES 0.4mT 'I - FIG. 3.-(u) Full e.s.r. spectrum of cis- and trans- forms of CH3c(OH)CO2Bun obtained during U(V1) photo-oxidation at 234 K ; (6) computer simulation of (a) based on coupling constants given in table 3 ; (c) third group of lines on expanded field scale. DISCUSSION PHOTOCHEMICAL MECHANISMS It is evident from the results that detectable concentrations of radicals can be realised during the photo-irradiation of solutions of Ce(IV) and U(V1) in liquid organic media using the experimental arrangement described, i.e., incorporating an intense source, low temperatures and sample renewal. However, one drawback, particularly apparent with U(VI), is that a number of substrates give no resonance for no obvious reason, thus the lower primary alcohols RCHzOH give RcHOH with U(V1) as photo-oxidant, whilst RIR,CHOH yields rather weak spectra in all cases other than propan-2-01 : again, very few esters yield any absorption with U(VI), but n-butyl lactate gives an intense, complex and well-resolved spectrum.Another generalisation which can be made is that U(V1) functions by abstracting hydrogen atoms from C-H bonds adjacent to some activating group, for example, alcohols (with the exception of 2-phenylethanol) yield hydroxyalkyl radicals, carboxy- lic acids RCH2C02H produce RCHCO,H, ethers afford -CH,OcH- species, and RCH(OH)CO,H similarly gives Rc(OH)CO,H, although Ph,C(OH)CO,H undergoes oxidative decarboxylation to give the benzophenone ketyl radical, which we also characterised optically by means of a flash photolysis experiment.16 In some instances more than one C-H site is attacked and two radicals are formed, e.g.from ethylene glycol dimethyl ether. U(V1) resembles, therefore, in its action such species as OH- and ButO- and its behaviour at temperatures between 140 and 300 K closely mirrors that we have found on photo-irradiation in rigid matrices at 77 K. One feature of note is the behaviour of (UO; +)* towards methanol. The sole radical *CH20CH2CH20CH3 + CHSOCHCH20CH3D. GREATOREX, R. J .HILL, T. J . KEMP AND T. J . STONE 225 we observe at both 77 K and 215 K is CH20H; however, Ledwith et aL2' have succeeded in spin-trapping CH30- during the photo-oxidation of neat methanol by uranyl nitrate at room temperature, a result which implies that our own observation relates to a secondary radical formed by the fast attack of CH30- upon CH30H even at 77K, (it should be pointed out that in aqueous methanol, the atbck of (UOt,+)* is upon C-H bonds, as evinced by a primary kinetic isotope effect in fluorescence quenching of (UO2,+)* by dilute, aqueous CH30H and CD30H,29 and, moreover, only CH20H radicals were spin-trapped in 2 : 1 (v/v) watermethanol by Ledwith et aL2'). With ethanol and higher alcohols, our observation of RcHOH accords well with the appreciable kinetic isotope effect observed in both the photochemical oxidation by uranyl ion of CH3CH20H and CD3CD20H 30 and the Stern-Volmer quenching constants of uranyl ion fluorescence by the two substrate^.^ Again, spin-trapping experiments indicate that U(V1) nitrate photo-oxidations of ethanol and butan-1-01 yield exclusively the hydroxyalkyl radical.28 C-C cleavage occurs with the alcohols we examined solely in the case of 2- phenylethanol (to give benzyl radical), reproducing the behaviour established in the solid state.15 This fragmentation might be concerted, viz. hv PhCH2-CH2-OH -+ PhCH2* + CH2=O + Hf + U(V) 1 W I ) or it could involve fission of a primarily formed 2-phenylethanoxy radical, PhCH2- CH20*. Ledwith et aL2' make a brief reference to spin-trapping of both PhCH2- and PhCH2CH20- in the oxidation of 2-phenylethanol by Ag+/S20,2- couple, but the latter oxidant is particularly prone to attacking 0-H bonds in primary and secondary alcohols to give alkoxyl radicals rather than hydroxyalkyl radicals.28 Ceric ion behaves in a rather unprecedented fashion in its photo-oxidation of primary alcohols.Thermal oxidation of these substrates by Ce(1V) is generally considered to proceed through a breakdown of a 1 : 1 Ce(IV)-alcohol complex involving C-H fission to give an intermediate hydroxyalkyl and our studies l2 of the photo-induced breakdown of such complexes at 77 K confirmed the production of RcHOH from RCH20H, although very minor production of a second radical displaying the relatively narrow lines of small alkyl radicals in the solid state was perceptible (fig.l(d) and (e) in ref. (12)). At temperatures of ca. 200 K, however, the exclusive radical found is that corresponding to loss of *CH20H from RCH20H, implying a pathway of C-C cleavage established for the thermal and photo-oxidation of tertiary alcohols by Ce(IV),33 viz. hv R-CH2-OH -+ Re + CH,=O + H+ + Ce(II1). 1 Ce(1V) Evidently the two photochemical pathways, ix., to give Re and RCHOH, have different energies of activation and the former, presumably of higher E,, predominates when T>200 K. In its behaviour with carboxylic acids, Ce(1V) reproduces the behaviour found at 77 K in oxidatively decarboxylating the acid RC02H hv R-C-OH -+ Re + C02 + H+ + Ce(II1) II 1 0 Ce(IV)226 E . S . R . OF RADICAL INTERMEDIATES COUPLING CONSTANTS The vast majority of the radicals we have described are commonplace and require no comment.The spectrum of cyclohexyl radical obtained in solution at 203 K compares well with that of Ogawa and Fessenden 33 measured in the solid state at 193 K, i.e., the " collapse " to a six-line spectrum of the type found at 273 K in the solid has not occurred. The resolution of cis- and trans-forms of the radical CH,c(OH)C02Bun has been discussed in the experimental section. We thank the S.R.C. for studentships to D. G. and R. J. H. and for grants to pur- chase the spectrometer and the C.A.T. R. 0. C. Norman, Chem. SOC. Special Publication No. 24, Chap. 6. T. J. Kemp and T. J. Stone, ref. (l), Chap. 14. R. Livingston and H. Zeldes, J. Chem. Phys., 1966,44, 1245. J. K. Kochi and P. J. Krusic, ref. (l), Chap. 7. A. Hudson and H. A. Hussain, J. Chem. Sac. B, 1969, 793. A. G. Davies, D. Griller and B. P. Roberts, J.C.S. Perkin 11, 1972, 993. H. Zeldes and R. Livingston, J. Chem. Phys., 1966,45, 1946. R. Wilson, J. Chem. SOC. B, 1968, 84. R. A. Sheldon and J. K. Kochi, J. Amer. Chem. SOC., 1968,90,6688. ' R. W. Fessenden and R. H. Schuler, J. Chem. Phys., 1963,39,2147. lo J. K. Kochi, R. A. Sheldon and S. S. Lande, Tetrahedron, 1969, 25, 1197. l 2 D. Greatorex and T. J. Kemp, Trans. Faraday SOC., 1971, 67, 56. l3 D. Greatorex and T. J. Kemp, Trans. Faraday SOC., 1971, 67, 1576. l4 D. Greatorex and T. J. Kemp, J.C.S. Faraday I, 1972, 68, 121. l5 D. Greatorex, R. J. Hill, T. J. Kemp and T. J. Stone, J.C.S. Faraday 1, 1972, 68, 2059. l6 H. D. Burrows, D. Greatorex and T. J. Kemp, J. Amer. Chem. SOC., 1971,93,2539. l7 R. S. Davidson and R. Wilson, J. Chem. SOC. B, 1970, 71. l 8 A. R. Buick, T. J. Kemp, G. T. Neal and T. J. Stone, J. Chem. SOC. A, 1969, 666. l9 W. T. Dixon and R. 0. C. Norman, J. Chem. SOC., 1963, 3119. 'O A. Hudson, J. Chem. SOC. A, 1969,2513. 21 T. W. Martin, L. L. Swift and J. H. Venable, Jr., J. Chem. Phys., 1970, 52, 2138. 22 W. T. Dixon and R. 0. C. Norman, J. Chem. SOC., 1964,4850. 23 D. J. Edge, B. C. Gilbert, R. 0. C. Norman and P. R. West, J. Chem. SOC. B, 1971,189. 24 T. S. Shiga, A. Boukhors and P. Douzou, J. Phys. Chem., 1967, 71, 3559. 25 R. W. Fessenden, J. Chem. Phys., 1962,37,747. 26 W. T. Dixon, R. 0. C. Norman and (in part) A. L. Buley, J. Chem. SOC., 1964, 3625. 27 A. Samuni, D. Behar and R. W. Fessenden, J. Phys. Chem., 1973,77,777. 28 A. Ledwith, P. J. Russell and L. H. Sutcliffe, Proc. Roy. SOC. A, 1973, 332, 151. 29 D. M. Allen, H. D. Burrows, A. Cox, R. J. Hill, T. J. Keinp and T. J. Stone, Chem. Comm., 30 S. Sakuraba and R. Matsushima, Bull. Chem. SOC. Japan, 1970, 43, 2359. 31 R. Matsushima and S. Sakuraba, J. Amer. Chem. SOC., 1971, 93, 5421. 32 W. H. Richardson in Oxidation in Organic Chemistry, ed. K. Wiberg (Academic Press, New 33 S. Ogawa and R. W. Fessenden, J. Chem. Phys., 1964, 41, 994. 1973, 59. York, 1965), Chap. 4.

ISSN:0300-9599    

DOI:10.1039/F19747000216

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Intermediates in the Nanosecond Pulse Radiolysis of Triphenylamine Solutions in Cyclohexane BY ERIKA ZADOR,? JOHN M. WARMAN,* LEONARD H. LUTHJENS AND ANDRIES HUMMEL Interuniversitair Reactor Instituut, Berlageweg 15, Delft, The Netherlands Received 1 ltlz July, 1973 The pulse radiolysis of solutions of triphenylamine (TPA) in cyclohexane has been investigated using nanosecond time resolution. In addition to the transient absorption at 620 nm which has been previously observed in microsecond pulse radiolysis and which is assigned to the N-phenyldihydro- carbazole (DHC) derivative, the following species have been identified : singlet TPA, fluorescence maximum 365 nm, mean lifetime 4.5 ns ; triplet TPA, absorption maximum 520 nm, mean lifetime 38 ns ; triplet N-phenyldihydrocarbazole, absorption maximum 430 nm, mean lifetime 330 ns ; the TPA+ radical cation, absorption maximum 640 nm.The rate constants for the reaction of TPA', TPA3 and DHC3 with O2 have been measured to be 1.8 x lO'O, 1.3 x 10'O and 7 x lo9 M-' s-' respectively. For a lo-' M TPA solution the yield and decay kinetics of TPA+ ( E ~ ~ ~ = 14 000 M-' cm-l) are the same as'found for the biphenyl negative ion in a M biphenyl in cyclohexane solution. The effects of NH3 and N20 on the yields of interinediates indicate ion recombination following charge scavenging by TPA to be the major soilrce of TPA1 and TPA3.,f In a previous study of the pulse radiolysis of cyclohexane solutions of triphenyl- amine (TPA), on a microsecond timescale, the predominant feature found was a strong absorption with a single maximum at 620nm.The spectral shape of this absorption and effect of scavengers indicated that the species responsible was not the radical cation of TPA and it was tentatively assigned to the triplet state of the amine. The TPA+ ion has however been observed on irradiation of TPA dissolved in electron attaching solvents 2 * It has been known for some time, from extrapolation of low temperature flash photolysis data,5 that the triplet state of TPA should be very short lived (ca. 100 ns) at room temperature. A lif'etime of 50 ns has recently been measured in nanosecond laser photolysis experiments which also showed that conversion of TPA triplet into N-phenyl-1 1 , 12-dihydrocarbazole proceeded via the intermediate formation of the triplet state of the dihydrocarbazole.The laser experiments are the latest in a lengthy series 5 * 7-13 carried out in an attempt to elucidate the mechanism of the formation of stable carbazole derivatives on photolysis of substituted diphenylamines. and in a low temperature EXPERIMENTAL The majority of the experiments reported were carried out with recrystalised triphenyl- amine (Fluka, purum) although it was later found that the unpurified reagent gave identical results. Cyclohexane (Merck, " Uvasol ") was used as received. Solutions were de-aerated by bubbling with helium for 20min. Saturated solutions of 02, N20 and NH3 (0.01,14 0.13 l5 and 0.38 M l5 respectively) were obtained by bubbling with gas for 20 min. -t On leave from the Central Research Institute for Physics of the Hungarian Academy of Sciences, Budapest 114, Hungary.$ and refer to singlet and triplet, respectively. 227228 PULSE RADIOLYSIS OF TRIPHENYLAMINE The samples were irradiated with 10 ns pulses of 3 MeV electrons from a Van de Graaff accelerator using a beam current of 1 A. The dose per pulse received by the sample, cali- brated by measuring the O.D. of the solvated electron at 3, = 578 nm in a M NaOH+ 1 M ethanol aqueous solution using &578 = 10 600 M-l cm-l l6 and G(e-) = 3.4 l7 was 1.63 x 1017 eV ~ m - ~ . Between pulses the sample in the cell (volume 1.3 cm3) was flushed with the solution under investigation using a flow system. The reproducibility of the dose per pulse as measured on an adjacent coaxial target (beam temporarily deflected) was 3 %.The analysing light source used was an Osram XBO-450 Xenon lamp which was tempor- arily pulsed l 8 (half width 0.5 ms) at approximately 20 kW (DC level 0.5 kW) during the period of measurement. The detection equipment consisted of a Bausch and Lomb grating monochromator, a 1P28 photomultiplier and a Hewlett Packard 183A oscilloscope. The rise time of the detection circuitry as measured by the Cerenkov radiation was 2 ns. The noise signal (optical bandwidth 7.5 nm at 600 nm) corresponded to approximately 0.5 % absorption. Emission signals were obtained by measuring the photomultiplier output in the absence of the analysing light beam. The emission was then corrected for (a) the photomultiplier quantum efficiency response ; (b) the self-absorption by the solute using the formula 2.303 ECZ 1 - exp (-2.303 eel) where Im is the measured intensity, I is the length of the cell (1.3 cm, assumed uniformly irradiated), c is the solute concentration, and e is the decadic extinction coefficient which was determined spectrophotometrically for the particular solution used ; (c) the Cerenkov radiation, obtained in the absence of solute which was corrected for (a) and then subtracted from the corrected emission intensity.I = RESULTS A N D DISCUSSION The absorption spectra resulting from pulse irradiation (10 ns pulse length) of a de-aerated, M solution of TPA in cyclohexane are shown in fig. 1A (full circles) for elapsed times of 0,100 and 1000 ns. Almost identical spectra reduced by a factor of three in optical density are observed for a M solution.It is apparent, from the rather complex time-dependent form of the spectrum, that several species contribute to the overall absorption. The initial problem therefore is the identification of the species involved. SPECTRAL ASSIGNMENTS On saturation of a TPA solution with O2 a much simpler spectrum, fig. 1A (crosses) is obtained, the form of which, in the region 500 to 700 nm, is independent of time. The wavelength dependence, with absorption maxima at 640 and 560 nm, is very similar to that attributed to TPAf formed by photolysis l9 (dashed line in fig. 1A) and by radiolysis of TPA dissolved in electron attaching mat rice^.^-^* 2o Assignment of this absorption to TPAf is supported by the marked reduction in the absorption at 640 nm, fig.2, produced by saturating the solution with NH3 which is a scavenger of solvent positive ions.21* 22 In contrast, saturation with N20, an electron scavenger,23 produces a much smaller reduction at 640 nm and results in a spectrum, fig. 2, almost identical to that observed for the O2 saturated solution. The decay of the absorption at 640 nm, in O2 saturated solutions, is shown in fig. 3 for M TPA. The decay kinetics are seen to be concentration dependent, which is characteristic of ion recombination in hydrocarbon l i q ~ . i d s , ~ ~ - ~ ~ thus further substantiating assignment to the TPA radical cation. Since 0, does not react with positive ions, it may be assumed that the effect produced by O2 is due only to the rapid removal of other absorbing species and that M andE.ZADOR, J . M. WARMAN, L. H. LUTHJENS AND A . HUMMEL 229 therefore the yield of the positive ion observed in the presence of O2 is the same as in a de-aerated solution of TPA. The contribution to the spectrum observed in a de- aerated solution from species other than the positive ion may therefore be obtained by subtraction of that found in the O2 saturated solution. The -resulting difference spectra at elapsed times of 0, 100 and 1000 ns are shown, for M TPA, in fig. 1B. A end 6 c of pulse L t wavelength /nm FIG. 1.-Transitory absorption spectra obtained on irradiation of a M solution of TPA in cyclohexane for elapsed times of 0, 100 and 1000 ns. Section (A), e, solution de-aerated ; +, solution oxygen saturated; dashed curve, spectrum of TPA+ from ref.(19). Section (B), 0, spectra obtained by the subtraction of the optical density found for an O2 saturated solution from that found for a de-aerated solution ; dashed curves : end of pulse, the spectrum attributed to the TPA triplet from ref. (13) ; 100 ns, the spectrum attributed to the N-phenyldihydrocarbazole triplet from ref. (6) ; 1 ps., the spectrum attributed to N-phenyldihydrocarbazole from ref. (13). The spectrum present after 1 ps (A,,, = 620 nm), fig. lB, is identical to that found by Kemp et al.’ in the microsecond pulse radiolysis of TPA-cyclohexane solutions. This absorption was attributed by the latter authors to the triplet state of TPA. However, studies of the photolysis of solutions of substituted diphenylamines 6-1 have shown that the relatively long lived absorptions in the 600 nm region in these systems originate from the 1 1,12-dihydrocarbazole derivatives (I) which are intermediates in the formation of the stable carbazole (11).230 PULSE RADIOLYSIS OF TRIPHENYLAMINE I R I (1) (11) R The spectrum previously found for N-phenyl- 1 1,12-dihydrocarbazole by photolysis of TPA solutions is shown as the dashed line in fig.1B (1 ,us elapsed time) and is in reasonable agreement with that found in the present experiments. From the decay O Q O O c XK) 600 700 wavelength/nm FIG. 2.-End of pulse spectra for a M solution of TPA in cyclohexane: 0, de-aerated; +, N20 saturated ; 0, NH3 saturated. 0.6 c FIG. 3.--Decav I I I 0.2 0.4 06 0.8 ID tltLs kinetics of TPA* at 640 nm in O2 saturated M, 0, and M, 0, TPA in cyclohexane solutions.Also, +, the decay kineti& of the biphenyl anion at 600 nm in a M biphenyl in cyclohexane solution from ref. (40).FIG. 4.-Oscilloscope traces of light absorption at the wavelengths shown for a de-aerated lo-’ M TPA-cyclohexane solution. The maximum signal for each trace corresponds to ca. 7.5 % absorption. 0 350 400 450 wavelength/nm FIG. 5.-Emission spectrum from a de-aerated lop2 M TPA-cyclohexane solution. Insert (A), oscilloscope trace of emission at 370 nm, horizontal scale 10 ns per division, vertical scale 500 mV per division. Insert (B) oscilloscope trace of Cerenkov radiation at 370 nm (cyclohexane alone), hori- zontal scale 10 ns per division, vertical scale 100 mV per division. To face page 23 1 ]E . ZADOR, J .M . WARMAN, L. H . LUTHJENS AND A . HUMMEL 231 rate of the 620 nm transient in an aerated solution a rate constant for reaction with O2 of G 107 M-1 s-1 may be estimated. This is to be compared with the much higher rate constants (> lo9 M-l s-I ) associated with deactivation of triplet states, thus supporting the assignment of this absorption to the dihydrocarbazole rather than to the triplet state of TPA. Immediately at the end of the pulse the absorption spectrum, fig. IB, has a maxi- mum at 520nm, which is the same as that attributed to the triplet state of TPA formed by photolysis,13 dashed curve fig. 1B (end of pulse). The fact that this species is present immediately following the pulse and that the rate constant for reaction with O2 is found to be approximately 1O1O M-l s-l, see below, supports assignment of this absorption, in the main, to the triplet state of TPA.After 100 ns the 520 am absorption is replaced by an absorption with A,,, = 430nm, fig. lB, as the most prominant spectral feature. The fact that the rate constant for reaction of this transient with O2 is also found to be approximately 1010 M-1 s-l , see below, suggests that it also is a triplet state. The shape of this absorption in the 400 to 500 nm region is almost identical to that observed in a recent laser photolysis study (dashed line in fig. 1B (100 ns)) by Forster and Grellmann who attributed it to the triplet state of the dihydrocarbazole. The following coase- yuences of such an assignment axe fuElled : in a de-aerated solution the growth rate of the dihydrocarbazole is the same as the decay rate of the 430 nm transient, see fig.4 ; in an air saturated solution the increase in decay rate a'. 430 nm (by a factor of five) is accompanied by an increase in the growth rate of the dihydrocarbazole with the relative yields of the two transients remaining the same as in a de-aerated solution. The emission observed during the pulse and for a few nanoseconds thereafter, fig. 5, has a spectrum with A,,, = 365nm. This is the same as the fluorescence spectrum of TPA observed by other workers l1 and indicates the formation of the lowest excited singlet state of TPA in the present system. REACTION MECHANISM A N D KINETICS IONS The presently accepted mechanism 27-30 for the formation of excited states of solute molecules in irradiated liquid cyclohexane involves initial scavenging of electrons and/or solvent positive ions, RH+, by the solute, S, followed by ion re- combination with concomitant excitation of the molecules involved.For compounds which react with both positive ions and electrons reactions (l), (2) and (3) must be considered as possible sources of excited states. S- + RH++S* + RH s-+s++s*+s. S++e-+S* Measurements of the scavenging efficiency 31 of TPA have shown electron scavenging, a- N 10 M-1,23 by this compound to be approximately ten times more efficient than positive ion scavenging, a+ N 1 M-l, thus reaction (2) may be neglected as a source of excited states. The actual calculated 29 fraction of positive ions undergoing reaction (2) is approximately 3 % for a M solution. From the above reaction efficiencies (a-values) the yields (G values) of TPA- and TPA+ for a M solution of TPA may be calculated 2 3 9 32 to be 1.35 and 1.0 respectively, indicating that approximately two-thirds of the TPA- ions recombine with a TPA+ ion.If one takes the solvation energy of positive ions in cyclohexane to be 1.6 eV,33 then both singlet (3.6 eV 34) and triplet (3.1 eV 34) state formation are energetically feasible as a232 PULSE RADIOLYSIS OF TRIPHENYLAMINE result of reactions (1) and (3) provided that the electron affinity of TPA is less than 4.8 eV for reaction (1) and less than 1.7 eV for reaction (3). At present, no estimates of the electron affinities of molecules in liquid hydrocarbons are available.However, it has been found,35 for other solutes which are both positive-ion and electron scaven- gers, that the total yield of excited states is close to the yield of anions formed. This suggests that both reactions of the type (1) and (3) are effective in excited state for- mation. Since the majority of ion pairs formed in liquid cyclohexane have an inter-ion distance shorter than the Onsager escape radius 3 6 9 37 (rc -N 300&, the kinetics of reactions (1) and (3) are complex. The very rapid decay of the correlated ion pairs ensures that, even for pulses as short as 10 ns, a considerable amount of ion recombina- tion occurs during the pulse. This is illustrated by the end-of-pulse yield of 0.56 found for P A + ions in a M solution, fig. 3, compared with the total scavenged yield of 1.0 calculated above. In order to calculate the yields (G values) of TPAf shown in fig.3, an extinction coefficient at 640 nm of 1.4 x lo4 M-l cm-l was used. This value is that which is required to give an overlap of the decay of TPAf in a M, N20 saturated TPA solution with that of TMPDf in a M, N20 sa-turated solution of TMPD under identical radiolysis conditions, using an extinction coefficient for TMPD+ at 560 nm of 1.2 x lo4 M-' ~ r n - ' . ~ ~ Since the positive ion scavenging efficiencies of TPA and TMPD have beeii found to be the same 31 and since the large excess of N20 ensures that the counter ion is the same in both cases, then the assumption that the positive ion yields are identical, which is the basis of this method of obtaining E for TPA+, would seem to be well founded.This value of E is considerably lower than that of 3.0 x lo4 M-I cm-I obtained by a chemical oxidation method.2 However, the latter measurement involved considerable experimental uncertainty, as was acknowledged by the authors, and furthermore the spectrum observed did not have the characteristic double maximum of the TPA+ ion. In addition when this value was used to calculate the yields of TPAf formed on pulse irradiation of halogenocarbon solutions, values were obtained which were consistently much lower than for the cations derived from similar solutions of the tribromo and trimethyl TPA derivatives. Although the total number of electrons and positive ions scavenged are not equal it has been calculated 39 that, at M using a 10 ns pulse, the end-of-pulse yields of negative and positive solute ions should be almost the same and should follow the same decay kinetics.Unfortunately, no absorption attributable to the TPA- ion, which has been found to have absorption maxima at 360 and 1650 nm, could be observed in the present work. Data are available, however, for the decay of the biphenyl anion in a M solution of biphenyl in cyclohexane under the same experimental conditions 40 and these are plotted as crosses in fig. 3. Since the sums of the ion mobilities in the two systems are not expected to differ appreciably good agreement between the decay curves for TPA+ and iD; would be expected and is seen to be so in fig. 3 thus further substantiating the extinction coefficient of 1.4 x lo4 M-l cm-1 for the TPA+ ion.From the decay curves in fig. 3 one can calculate that within 25 ns following the pulse 75 % of the initially formed negative ions have decayed and in the ensuing 100 ns only a further 8 % recombine. SINGLET TPA The predominant role of ion recombination in the formation of singlet TPA is indicated by the reductions of 57 % and 75 % in the emission yield brought about byE. ZADOR, J . M. WARMAN, L . H. LUTHJENS AND A . HUMMEL 233 saturation of a M TPA solution with NH3 (a positive ion scavenger) or N20 (an electron scavenger) respectively. That these reductions are due to reaction with precursors of TPA rather than reaction with TPA itself is shown by the insigni- ficant effect 41 of these scavengers on the decay rate of the emission, see fig.6A for the c t/ns FIG. 6.-Semi logarithmic decay plots for a M TPA cyclohexane solution, 0, de-aerated ; a, air saturated; 0, O2 saturated; +, NH3 saturated, (A), singlet TPA measured in emission at 370 nm (all curves are corrected for Cerenkov emission and are normalised to the same end of pulse value) (Ei), triplet TPA measured in absorption at 500 nm (corrected for underlying absorption, see text). (C), triplet DHC measured in absorption at 430nm (corrected for underlying absorption, see text). effect of NH3. From a comparison of the oscilloscope traces of the emission from a TPA solution with that of Cerenkov radiation, shown in fig. 5, it can clearly be seen that the TPAl lifetime is longer than the time resolution of the apparatus. However, because of the 2 ns fall-time of the detection circuitry, a plot of the logarithm of the emission intensity against time following a 10 ns pulse is not linear, fig.6A. A value for the true exponential decay rate of the emission may be obtained by comparison of the decay curve with simulated curves obtained by a computational procedure of convolution analysis and numerical integration over the pulse. A best fit to the present data was obtained using a decay rate of 2.2 x lo8 s-l which, together with an RC time constant of 2 ns, was the value used to compute the curve through the filled points in fig. 6A. The mean lifetime of TPAl is therefore 4.5 ns with an estimated error of k0.7 ns. This is similar to the value of 4 ns obtained for the lifetime of the singlet state of N-methyldiphenylamine in fluorescence quenchiqg experiments.234 PULSE RADIOLYSIS OF TRIPHENYLAMINE For an air saturated solution, the decay rate of the emission is not greatly changed, however saturation with 0, results in an increase in the decay rate to 4 x lo8 s-l, fig.6A. This corresponds to a rate constant for singlet deactivation by O2 of 1.8 x 1O'O M-l s-l which is of the magnitude expected for a diffusion controlled reaction. Using this rate constant for O2 deactivation, a reduction of 38 % in the end of pulse yield of emission would be expected for an O2 saturated solution. The reduction observed however, 61 %, is considerably greater than this and would indicate that 0, is in addition capable of reacting with a precursor of the singlet state.Since, as mentioned previously, O2 does not react with solvent positive ions and is a very inefficient electron scavenger 2 5 * 42 this effect cannot be due to competition with TPA for these species. It is, however, possible that TPA- is removed by electron transfer to O2 which has been found to be a very efficient reaction (k = 3 x 10'O M-l s-') for biphenyl anions.25 Alternatively, TPAl may be formed in part by reaction with the short lived 43 (0.3 ns) singlet state of cyclohexane which is also known to react with 02.44 At present, although the formation of the cyclohexane singlet has been shown 43-45 to occw in irradiated cyclohexane, the role of this species in the radiation chemistry of cyclohexane solutions awaits clarification. TRIPLET TPA Apart from a very slow growth resulting from the tail of the ion decay the forma- tion of TPA3, on ion recombination and via intersystem crossing from TPA' (effi- ciency 0.88),46 may be considered to be complete within 25 ns following the pulse.The subsequent sequence of reactions TPA3+DHC3-,DHC is illustrated by the oscilloscope traces of the absorptions of these species shown in fig. 4. Thus the decay of TPA3 and consequent formation of DHC3 is shown by traces 4a and 4b and similarly, traces 4c and 4d illustrate the growth of DHC which accompanies the decay of DHC3. An accurate quantitative measure of the rate constant for the decay of TPA3 is complicated by the occurrence of formation and decay on a similar timescale and also by the presence of underlying absorptions due to the other species present.The absorption at 500 nm, which is due mainly to TPA3, is found to reach a plateau value corresponding to approximately 25 % of the initial yield at times long compared with the first half life. The simplifying assumption has therefore been made that this long-time absorption is in fact constant from the end of the pulse and may be sub- tracted from the total in order to obtain the contribution of TPA3. In this way, the logaritbmic decay plots for TPA3 shown in fig. 6B were obtained. After the initial distortion arising from concurrent formation, the points are seen to lie on a straight line which corresponds to a mean lifetime for TPA3 of 38 ns with an estimated error of & 8 ns. This value is somewhat lower than the value of 50 ns estimated from recent laser photolysis experiments, but in view of the large error limits involved the two values may be taken to be in reasonable agreement.In an air saturated solution the mean lifetime of TPA3 is reduced to 19 ns, fig. 6B, which corresponds to a rate constant for deactivation by O2 of 1.3 x loio M-l s-l in good agreement with the value of 1.5 x 1Olo determined in the laser experiments.6 TRIPLET DHC The kinetics of the absorption at 430 nm which is due mainly to the dihydro- carbazole triplet state suffers from the same complications mentioned above for TPA3 ; therefore, the same treatment has been applied to the data and results in the decay curves for DHC3 shown in fig. 6C. The mean lifetime of 330 70 ns obtainedE . ZADOR, J .M . WARMAN, L . H . LUTHJENS AND A . HUMMEL 235 from the linear part of the decay in a de-aerated solution is somewhat lower than the value of 500 ns recently estimated.6 However, the rate constant of 7 x lo9 M-1 s-' for DHC3 deactivation by O2 obtained from the mean lifetime of 60 ns in an aerated solution is in good agreement with the value of 6 x lo9 M-l s-l calculated in the same study. GROUND STATE DHC The dihydrocarbazole absorption at 620 nm is found to decay on a much longer timescale than the other transients. The half-life of this decay, II 10 ,us, is in fact controlled by the low frequency filtering in the detection circuitry and, therefore, represents a lower limit to the lifetime of the dihydrocarbazole. A value of 24 ,us has been found for the lifetime of DHC in microsecond pulse radiolysis experiments arid a much larger value of approximately 500 ,us in flash photolysi~.~* When account is taken of the low frequency filtering of the system, the observed half-life of 4 ,us found for the 620 nm transient in an air saturated solution yields a value of approximately lo7 M-l s-l for the reaction of DHC with 02.A somewhat lower value of 2.3 x lo6 M-' s-l has been obtained ' for the reaction of the N-methyl DHC with 02. The authors thank Dr. A. Weller for bringing to their attention aspects of the photochemistry of TPA. T. J. Kemp, J. P. Roberts, G. A. Salmon and F. G. Thomson, J. Phys. Chem., 1967,71,3052 ; 1968,72,1464. H. D. Burrows, D. Greatorex and T. J. Kemp, J. Phys. Chem., 1972,76,20. ' T. Shida and W.H. Hamill, J. Chem. Phys., 1966,44,2369. D. W. Skelly and W. H. Hamill, J. Chem. Phys., 1965,43,3497. ' G. C. Terry, V. E. Uffidell and F. W. Willets, Nature, 1969,223,1050. E. W. Forster and K. H. Grellmann, Chem. Phys. Letters, 1972, 14, 536. ' C. A. Parker and N. J. Barnes, Analyst, 1957, 82, 606. E. J. Bowen and J. H. D. Eland, Proc. Chem. SOC., 1963,202. K. H. Grellmann, G. M. Sherman and H. Linschitz, J. Amer. Chem. SOC., 1963,85,1881. lo H. Linschitz and K. H. Grellmann, J. Amer. Chem. SOC., 1964,86,303. l1 H. Shizuka, Y. Takayama, J. Tanaka and T. Morita, J. Amer. Chem. SOC., 1970,92,7270. l2 E. W. Forster and K. H. Grellmann, J. Amer. Chem. SOC., 1972,94,634. l3 E. W. Forster, Doctoral mesis (University of Stuttgart, 1971). l4 Landolt Bornstein, 1962, II (2bI), 74.l5 T. Saito, K. Takahashi and S. Sato, Bull. Chem. SOC. Japan, 1968,41,2603. l6 J. Rabani, W. A. Mulac and M. S . Matheson, J. Phys. Chem., 1965,69,53. '' J. K. Thomas, Adv. Rod. Chem., ed. M. Burton and J. L. Magee, Vol. II(Wiley, N.Y., 1969), l 8 L. H. Luthjens, details to be published. l9 H. Ruppel, U. Krog and H. T. Witt, J. Electrochem. SOC., 1960, 107, 966. 2o W. H. Hamill, RadicaZ Ions, ed. E. T. Kaiser and L. Kevan (Wiley, N.Y., 1968), pp. 383,391. 21 F. Williams, J. Amer. Chem. SOC., 1964, 86, 3954. 22 K-D. Asmus, Int. J. Rad. Phys. Chem., 1971, 3,419. 23 The yield of product P formed due to the scavenging of ions by a solute S in irradiated cyclo- hexane has been found to be described by the empirical expression G(P) = Gfj+ Ggi/(l+ ' (or[S])-*) where Gfi and Ggi are the yields of free ions and geminate ions respectively and a, the " reactivity ", is a constant which is proportional to the rate constant for the scavenging reaction.For fuller details concerning the significance of a see J. M. Warman, K-D. Asmus and R. H. Schuler, Adv. Chem. Series, 1968,82,25. p. 163. 24 J. K. Thomas, K. Johnson, T. Klippert and R. Lowers, J. Chem. Phys., 1968,48,1608. 2 5 J. T. Richards and J. K. Thomas, Chem. Phys. Letters, 1971,10, 317. 26 S. J. Rzad, P. P. Infelta, J. M. Warman and R. H. Schuler, J. Chem. Phys., 1970,52, 3971. "J. K. Thomas, Ann. Rev. Phys. Chem., 1970, 21, 17.236 PULSE RADIOLYSIS OF TRIPHENYLAMINE 28 A. Singh, Radiation Res. Rev., 1972, 4, 1. 29 S. J. Rzad, J. Phys. Chem., 1972, 76, 3722. 30 J. L. Magee and J. J. Huang, J. Phys. Chem., 1972, 76, 3801. 31 E. L. Davids, J. M. Warman and A. Hummel to be published. 32 S. J. Rzad, R. H. Schuler and A. Hummel, J. Chern. Phys., 1969, 51, 1369. 33 R. A. Holroyd, J, Chem. Phys., 1972,57, 3007. 34 V. Zanker and B. Schneider, 2. phys. Chem., 1969, 68,21. 35 J. H. Baxendale and P. Wardman, Trans. Faraday SOC., 1971, 67, 2997. 36 A. Hummel, A. 0. Allen and F. H. Watson, Jr., J. Chem. Phys., 1966, 44, 3431. 37 G. R. Freeman and J. M. Fayadh, J. Chem. Phys., 1965,43, 86. 38 See ref. (2) for a review of available data. 39 P. P. Infelta and S. J. Rzad, J. Chem. Phys., 1973,58, 3775. 40 A. Hummel and L. H. Luthjens, J. Chem. Phys., 1973,59, 654. 41 With N20 a small increase in the decay rate comparable to that observed for air is observed (i.e., N- 20 %) suggesting a rate constant for singlet deactivation by NzO of approximately 3 x 10' M-l s-l. 42 G. Beck and J. K. Thomas, J. Chem. Phys., 1972,57,3649. 43 M. S. Henry and W. P. Helman, J. Chem. Phys., 1972,56, 5734. 44 F. Hirayama and S . Lipsky, Organic Scintillators, ed. D. L. Horrocks (Academic Press, N.Y., 45 P. K. Ludwig and M. M. Huque, J. Chem. Phys., 1968,49,805. 46 A. A. Lamola and G. S . Hammond, J. Chem. Phys., 1965,43,2129. 47 K. M. Bansal and R. H. Schuler, J. Phys. Chem., 1970,74, 3924. 1970).

ISSN:0300-9599    

DOI:10.1039/F19747000227

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Ellipsome tric Investigation of Black Soap Films BY DAAN DEN ENGELSEN AND GERRIT FRENS * Philips Research Laboratories, Eindhoven, The Netherlands Received 13th July, 1973 Thin soap films of sodium dodecyl sulphate and sodium dodecyl benzenesulphonate have been studied with a combination of ellipsometry and reflectometry. It is demonstrated that an optical three-layer model is superior to a one-layer model for the interpretation of the ellipsometric data. The refractive indices of the film core and the surface layers are determined, and it is found that the index for the surface is higher than that for the film core. The composition of the Newton type of film is discussed on the basis of these data. The drainage of water from the core of a newly formed soap film leads to a gradual thinning of the lamella and finally an equilibrium film is formed. With ionic surfactants there exist two distinct types of equilibrium film.One is known as the “ common ” or “ first ” black film. The other type is the “ Newton ”, “ Perrin ”, or “ second ” black film. Indeed these equilibrium films look black, because of the low reflectivity of such thin lamellae. Depending on the electrolyte concentration, the thickness of the common black films varies between 5 and 50 nrn. Appxently the forces which promote the thinning of the lamella (gravity, capillary suction, van der Wads attraction) are compensated by the repulsion of the electrical double layers at the surfaces of the film. The thinner Newton film is about 4nm thick, independent of the electrolyte concentration.Little is known 2bout the molecular structure of Newton films. Their thickness suggests that Newton films resemble bilayers of surfactant molecules. It has been found that the coverage of the surfaces with surfactant is almost equal in both types of film. The continuous thinning of a soap lamella leads to the “common” type of equilibrium film. For the formation of the Newton film it is necessary that a spot of such film nucleates and grows at the expense of thicker film. The nucleation of Newton film cz-n occur during the thinning of the lamella, but also after the formation of the common equilibrium film. Studies of the conditions for the formation of the different types of film indicate that the transition from the cornon to the Newton black film is some sort of phase transition in the film surface.This transition could be an ordering of the surfactant molecules from a two dimensional liquid in the surface of the common and thicker films into a two dimensional nematic? liquid crystal in the Newton type. The temperature dependence of the contact angles of black soap films gives a measure for the entropy change in the transition from com- mon to Newton film. Thai this term is of the same magnitude (per molecule) as that for the transition between three dimensional isotropic and nematic liquids would be an argument in favour of such a model for the transition in the equilibrium films. We have attempted to obtain experimental information concerning the structure t Nematic, not smectic. The surfactant molecules are already in a layer in the surface of a thick film.237238 ELLIPSOMETRY OF SOAP FILMS and the anisotropy of surface layers in common-and in Newton black films by combining ellipsometric and reflectometric measuring techniques. Film thicknesses are normally determined from the reflectivity of the films and interpreted in a three layer m0de1.~ Such an interpretation requires a number of assumptions which require elaborate experimentation for their verification. Therefore, film thickness data are often reported as "equivalent water thicknesses '7.5 In this paper, it is shown that a single-layer model with a constant " effective " refractive index cannot describe the ellipsometric observations on soap films. The differences with a three- layer model are analyzed, and it is shown how the refractive indices for the three-layer model may be determined from ellipsometric experiments.Furthermore, it is shown that in thin films there is a unique relationship between the ellipsometric (A) and the reflectometric (D) data which can be used to judge the quality of experiments. THEORETICAL In fig. 1 we define a three-layer system, which will serve as an optical model for a In the rectangular coordinate system the z-axis is normal to the film soap film. "I 1 I It m Lc! FIG. 1.-Optical model of a symmetrical three-layer film. surface. The inner (aqueous) layer of phase 3 is considered optically isotropic. The two surface layers (phase 2) can have a uniaxial anisotropy, the optic axis coinciding with the z-axis. It is reasonable to take n, and n, = tz, as the principal refractive indices of the surface layers.Preferential orientation of molecules in the surface should be allowed. It will not necessarily coincide with the z-axis, but there is no difference between x and y with respect to the film. Local orientation effects in the surface will therefore be averaged in time and in space with respect to x and y , though not to z. We only consider non-absorbing systems, which implies that refractive indices are real. Light polarized parallel to the plane of incidence is indicated by a subscript p , the subscript s means normal (" senkrecht ") polarization. We will, however, omit these subscripts wherever the equations for p - and s-waves are identical.D . DEN ENGELSEN AND G.FRENS 239 (a) THREE-LAYER MODEL Duyvis derived expressions for the reflection coefficient R of a symmetrical three-layer system, which may be written with pi = 2 3 ~ d ~ n ~ l - ~ cos +j is the angle of refraction in phase j ( j = 2, 3, fig. 1). A is the wavelength of the light in vacuo. Ra, Rb and Q depend on the optical constants of the three layers and on the thickness of the surface layers d2 : R, = A[rx + YIX exp( - 2iP2)IY Rb = A[rxxx + rxv exp( - 2iP2)] = - &, + rI exp( - 2iB2)], Q = A[rxrxx + exp(-WdI, A = [ 1 + rIrII exp( - 2iP2)]-l (2) (3) (4) (5) where and r are Fresnel’s two phase reflection coefficients. depend on the direction of polarization 9* l o In the case of anisotropy in nz, the phase difference PZ and the Fresnel coefficients P2p = 2&nxx*(nA - ( 6 ) Pzs = 2nd2xs;l -’ (7) (11) x s - n3 cos 43 x s + n3 cos 43 r11s = with xp = (n,2-n; sin2 +#, and xs = (ni-ni sin2 41)*, (cf.fig. 1 for nl, 41). extension of known formulae as : The transmission coefficient T for a thee-layer system is obtained by a simple Here T, = AtItxI exp( - ip2) ; Tb = AtxIItIV exp( -iP2). The products of transmission coefficients tItII and tIIltIV depend again on the anisotropy of the surface layers? The transmittivities and the reflectivities of a three-layer system are ITp12, &,I2 and I&12 respectively. By definition the ellipsometric angles @ and A for reflected and for transmitted light are respectively : tan t)r exp (iAr) = RPRl1 tan @t exp (iAJ = TpT, ’. (13) (14)240 ELLIPSOMETRY OF SOAP FILMS For a non-absorbing film At = AT, apart from trivial multiples of 180".from combining the eqn (13) and (14). This is found In the simplest case of d3 = 0 ; where rI = - rIV = r. equals With expressions for T from ref. (1 1) it is found that eqn (1 5) This result being real, it follows that Ar = A,+multiples of n. An analogous, but more tedious derivation can be given for d,>O. Since the transmittivities of black soap films-of very thin, non absorbing lamellae in general-are orders of magnitude larger than the reflectivities it is advantageous from an experimental point of view to obtain ellipsometric information in terms of At and t,bt. U 50 100 150 ' d3bm 4I = 76.38' ; h = 632.8 nm; dz = 1.5 nm; nl = 1 ; n3 = 1.333. Solid curve: nx = n, = n2 = 1.38. Dashed curve: n, = 1.30, n, = 1.38.FIG. 2.-Calculated curves of A versus d3. lT'l and ITs] are close to unity in thin films. This makes t,bt fairly insensitive for changes in the optical constants for the three layers. However, at the low reflectivities of these " black " films there is a large effect of the refractive indices on t,hr. It is therefore useful that ellipsometry in transmission (At, +J and in reflection (Ar, +r) are combined in the experiments.D. DEN ENGELSBN AND G. PRBNS 241 Fig. 2 shows changes in A for a symmetrical three-layer film when the surface layers remain unchanged and the film thickness is altered through the drainage of solution from the core of the film. A = At = Ar has been computed as a function of d3 for a few sets of refractive indices. It is found that differences in the refractive indices or in the anisotropy of the surface layers have a relatively small effect OR A.$r is a very sensitive measure for anisotropy in the surface layers when d3 is near 0 and near 345 run. These thicknesses correspond to the minima in the reflectivity of the film. The inset gives In fig. 3 it is shown how ll/r changes with d3 in this model. around these values of d3 in more detail. d 3 l m FIG. 3.-Calculated curve of & versus d3. Dash-dot curve : n, = 1.38, n, = 1.30. Other legends as in fig. 2. (b) APPROXIMATIONS FOR THIN THREE-LAYER FILMS When d2 and d3 are both small relative to Iz it is possible to obtain useful approxi- mations for the cumbersome expressions (1) and (1 2). Expanding the exponentials in eqn (1) to terms linear in Bz and f13 one finds : 9 (17) 4B2rLl- 4) + 2 P 3 C d l + 4) + 11 1(1+ &I (1 - r;J(1- r;) IRI = with (6), (7), assuming n, = 1 for a film in air, and writing r in terms of the optical constants one obtains : and n[2d2(n,2 - 1) + d3(n,2 - l)] Iz cos 41 RSI =242 ELLIPSOMETRY OF SOAP FILMS In the isotropic case n, = n, = n2 it follows from eqn (12), (18) and (19) that 2d2(nq cos2 41 - cos2 42)+d3(nz cos2 4l - cos2 #2) 2d2(nz - 1) + d3(ng - 1) IR,I IRSI (20) tan 1+9,, = - - - The approximate expression for A becomes with the same assumptions : (21) I* [ 4 4 7t sin2 +1 2(n4 - 1)2d2 + ( n i - l)2d3 Numerical comparison of these approximate expressions with the exact eqn (1)-( 1 1) shows that they may be used for the thin films (d3 < 10 nm) which are studied in this paper.We have, however, based our computations on the exact expressions for t,br, A and R. The accuracy of eqn (19) is 1 % when d2 is taken to be 1 nm ; d3 = 3 nm ; n, = n, = 1.4; n3 = 1.3 ; A = 500 nm and 41 = 75”. For this set of parameters, eqn (21) is also accurate to within 1 %. Eqn (20) is a rather poor approximation, the difference from the exact result being about 6 % for the chosen parameters. Higher order terms in b are needed for correct results. We found generally that the approxi- mation is better for IR,I than for In the approximate eqn (18), (19) and (21) it is seen that R,, R, and A are linear combinations of d2 and d3. When $ and A have been found from an ellipsometric experiment it gives no new information when R, or R, is also measured independently.However, if the reflectometry and the ellipsometry are done independently and at different angles of incidence one has a built-in check on the quality of the experi- mental data. The linearization of exp( - 2iB3) is also justified for B ~ k n , i.e., at d3 values near the interference minima at kA/2n3 cos 43 (k = 1, 2, 3, . . .). It is seen from (18) and (19) that the quantity Id&,-d&l will generally increase when the anisotropy of the outer layers increases. d& and d& are defined here as the values for d3 where IRJ = 0 and lRsl = 0 respectively (cf. fig. 3). A = A cos 41 1 d ~ , - d ~ , = 2d,[-- n2- 1 n:n,” cos2 -$ ni - I. n,2(n$ cos2 41 - cos2 43) which amounts to 0.7 nm with n, = 1.38, n, = 1.30 under the conditions of fig. 3.We have tried to determine d&-d&, measuring the reflectivity of films of approxi- mately 345 nm when d3 changed slowly due to drainage. Monitoring the reflectivity in polarized light, and determining the difference in time for the minimum in and in IRJ2 as well as the rate of thinning, we should have been able to measure ld&,-.d&l values larger than 0.4 nm. However, the results were of the order of the experimental error. This shows that the anisotropy in the surface layers of thick films is certainly small : In, - nzl < 0.03. The Brewster angle c ) ~ is the angle of incidence where IR,I = 0. From eqn (18) it follows that ngn:[2(nz- l)d,+(ng- l)d3] 2d2n:(n,” - l)+d3n~(n~ - 1) tan2 4B = - & is a complicated function of refractive indices and layer thicknesses in a three-layer model.The interpretation of Brewster angles in terms of anisotropic refractive indices is therefore not straightforward.D. DEN ENGELSEN AND G . FRENS 243 (C) SINGLE-LAYER MODELS Next we investigate the representation of a thin three-layer film as a single layer with an “ effective ” refractive index n, and a thickness d = 2d2 + d3. The optical properties of the real film are of course independent of a model. Thus, I&[, lRpl, t,k, A and +B are the same, for the single- and for the three-layer system. From the equality of lR,I one finds From the equality of A it can be derived that and from the equality of +B it follows that (nz- 1) = [2(n~-l)d2+(n~-l)d3](2d2+d3)-1. (24) (25) (nz- 1)2n,2 = [2d2(nz- 1)2n~2+d3(n~-1)2n~2](2d2+d3)-1 ngni[2d2(nz - 1)+ d 3 ( 4 - l)] 2d2ng(n; - 1) +d3n;(ni - 1) n, = The expression (24) has already been obtained by Rijnbout l 3 in a somewhat different way. In all three equations it is clear that n, is not a constant, but that it changes as a function of d3.The expressions for n, are based on the optical theory for reflection and interference of’ light in thin films. They are not equivalent to the calculated n, of a non-structured film of the same chemical composition using the Lorentz-Lorenz equation. The latter equation is based on the assumption of additivity of the refractivity Cpolariz- ability) of the mixed molecules. It does not take into account the effect of surfaces in the system, i.e., of adsorption. The single-layer and the three-layer models are not optically equivalent, but there is no difference between a three-layer and a five- or multi-layer model in this respect.9 However, the optical three-layer model is complemented with a molecular multi-layer model in these soap film systems. The parameters n2, dz of the surface layers in the optical model are a way of averaging over all the adsorption and structure in the film surface, as long as these remain unchanged during the drainage of water from the core of the am. EXPERIMENTAL Fig. 4 illustrates the experimental set up for simultaneous ellipsometry and reflectometry. Inside a closed cell C with flat, strain-free optical windows there is a soap solution. A soap film can be drawn from this solution in a flat, thin, vertical glass frame.There is one spot S on the film which is illuminated by both the horizontal beams which originate from the semitransparent mirror BS. The two beams strike the film at different angles of incidence. In this arrangement it is possible to simultaneously gather ellipsometric data-at large angle of incidence for optimum sensitivity-and monitor the thickness of the film by reflectometry at a small angle of incidence. The spot is approximately 3 cm above the level of the solution in the cell. The soap solutions contained purified sodium dodecyl sulphate (NaDS),* 0.05 % by weight, in aqueous solutions of 0.17 M NaCl and 0.1 8 M NaCl respectively. Some additional measurements were on a 0.17 % solution of commercial (BDH) sodium dodecyl benzene- sulphonate (NaDoBS) in 0.17 M NaCI.In these solutions the surfactant concentration is above the critical micelle concentration. The transition temperature below which common black films are metastable relative to Newton films, lies approximately at room temperature for each of these solutions. * Kindly given by Mr. J. A. de Feyter, van? Hofflaboratorium der Rijksuniversiteit, Utrecht.244 ELLIPSOMETRY OF SOAP FILMS The experiments were done in a thermostatted room. Temperatures were chosen in such a way that the gradual thinning of the vertical soap lamella would produce metastable common black films. The nucleation and growth of Newton film would occur approximately 10 min after the common black film had reached the point of observation S. In this way it was possible to obtain the data on both types of film without any change in the measuring system, or in the salt concentration and the temperature of the solution.The experiments were done between 23°C and 25°C. Sometimes a slight undersaturation of the atmosphere FIG. 4.-Schematic view of the experimental arrangement. L = HsNe laser of 1.5 mW power ; F = attenuation filter (50 x ) ; BS = beam splitter ; M1, Mz = mirrors ; P = polarizer (Glan- Thomson prism); K = compensator; C = cell; S = observation spot (see text); Al, A2 = analyzers ; D1, D2, D3 = light detectors. around the film was brought about by a gentle heating of the wall of the cell C. This would cause an extra thinning of the common black Hrn below its equilibrium thickness, due to the evaporation of water from the aqueous core of the lamella.This made it possible to study common black films of different thicknesses but at a nearly constant electrolyte concentration. The intensity of the reflected light was recorded during the thinning of the film and during the ellipsometric measurements on black films. This gave the reflectivity IRI" of the thin films as well as the maximum reflectivity lRmnx12 of the first interference maximum. From these the normalized reffectivity D = lR12*IRm&2 can be obtained. The ellipsometric experiments were done in transmission and in reflection, in two zones and at several angles of incidence. The angles of incidence were all close to 75', and they were measured with an accuracy better than 0.1". The settings and readings of the polarizer and the analyzer were precise to 0.01'.Zero points were determined with this precision, both on a polished Si-surface and on the actual soap films in the cell C. This was necessary since we needed accurate absolute values of t,b and A for both types of film. RESULTS AND DISCUSSION Fig. 5 gives the experimental values of t,br and A for some common- and Newton black films of NaDS. We have indicated which of the experimental points belong together in one experiment, Drawing a fresh film from the solution might conceivably have altered the position of the film relative to the beam between experiments, thereby producing slight changes in the angle of incidence. But it is certain for each individual experiment that nothing changed between the determinations of t,br and of A for the common black film and for the Newton film.D.DEN ENOELSEN AND G . FRBNS 245 We have attempted to interpret the experimental data in terms of a three-layer model. Such a model has four adaptable parameters (n2, d2, n3, d3)-0r even five if we allow unaxial anisotropy in the outer layers. There are only two independent experimental results among $, A and d, so that two or three additional assumptions are necessary. 25.25- 25.00 24.75. 2650' 24.25 3.0" L .O* A FIG. 5.-(&, A) plot of the ellipsometric results for NaDS at = 76.38'. The various symbols used to indicate individual measuring points refer to individual experiments. The drawn curve is a theoretical (&,A) curve for n2 = 1.365, d2 = 1.8 nm, n3 = 1.333, the running parameter d3 = 5.1 nm for an equilibrium common black film and 0 tun for a Newton film.The following three assumptions are made in our model : (i) the outer layers of the thick and of the common black films are isotropic. (ii) The properties n2, d2 of the surface layers in the thick and the common black films are independent of the film thickness. Differences between films are described as variations in the amount of material in the aqueous core of the film (i.e., of d3). (iii) The refractive index n3 of the core material is isotropic and equal to that of the solution from which the film was drawn. In our experiments this means n3 = 1.333 at R = 632.8 nm and T = 25°C. These assumptions make the common black film the starting point of the analysis. Comparison of the predictions of the model with the experimental results means that we must investigate whether all soap films, including Newton films, can be described with the same parameters as a common black film.The first of the assumptions in the model i3 based on our experiment concerning dip - d& and the lack of evidence to the contrary. Under the circumstances it is the simplest assumption. The second assumption is based on the experimental observa- tion that the drainage of a thick film ends in the formation of a common black film, and that the coverage of the surface of a soap film with surfactant is independent of the film thickness.246 ELLIPSOMETRY OF SOAP FILMS The bulk solution of NaDS has the assumed refractive index n3 = 1.333. Comput- ations show that the experimental data ($r, A, 0) can only be described with n3 = 1.331 k0.003 for a wide range of n2, dz.This is an indirect proof that the core of a common black film consists of essentially unchanged bulk solution. The second and the third assumption of the model serve to describe the adsorption at the film surfaces in terms of differences with the bulk material. In this model, it is possible to compute the value of n2, d2 and d3 which give the correct relation between t,br and A for the common black films. There is, of course, a whole range of combinations which give the desired result, since the separation plane of core and surface layer can be located in an arbitrary way. It is important to notice that all these solutions predict the same (t,br, A) relationship for films with changing d3.There is one rather special solution : n2 = 1.365, d2 = 1.8 nm and d3 = 5.1 nm, for the " equilibrium " common black film leads to d3 = 0 for a film with the same $r as a Newton film. The (+hr, A) curve for the three-layer model has been drawn in fig. 5 for comparison with experiment. It is found that the model provides an excellent description of $r and A for common black films while their thickness is varying by evaporation, condensation or drainage of core material. However, it is also found that the experimental observations on Newton films deviate from the predictions of the model. The discrepancy is larger than the measuring errors, the experimental $r being too large, or A too small, by more than 0.1". This implies that a Newton film is not simply a common black film with a thinner core.On the basis of our data alone it cannot be decided whether this implies a change in the anisotropy of the surface layer (n,>n,), or in the concentration of the adsorbed surfactant, or both. Fig. 6 and 7 relate the ellipsometric parameters, $r and A respectively, to the reflectometric experiment (0). Fig. 6 compares experimental data with calculated curves for three-layer models. The large effect of anisotropy on +hr is illustrated. It 9% FIG. 6.-Normalized reflectivity D versus ellipsometric angle (br for NaDS. Measured values are indicated by circles. The drawn curves are theoretical (D, $r) graphs for the following film models : A, onelayer film with constant refractive index of 1.345. Other curves refer to three-layer films with n3 = 1.333 and 0<d3<6nm; By nx = nz = n2 = 1.365, dz = 1.8 nm; C, n, = 1.339, n, = 1.320, d2 = 1.8nm; D ,nx = 1.383, n, = 1.400,d2 = 1.8nm.D .DEN ENGELSEN AND G . FRENS 247 is found that for isotropic models 12, ranges between 1.386> n2> 1.362 when d2 is varied between 1 and 2 nm. This is the same result as was obtained from the ellipso- metric experiment alone. The solution n, = 1.365 ; d2 = 1.8 nm leads again to d3 = 0 at the Newton film's Ic/r. A curve for a single-layer model with a constant D A FIG. 7.-Normalized reflectivity D against A for NaDS. Measured values are indicated by circles. Curve below A = 1.5" refers to a hypothetical one-layer film with the effective refractive index of 1.365 of the Newton film. refractive index is also shown.Such a model can not account for the experimental results, as had been anticipated by Smart and Senior.' Another refractive index for the single-layer would alter the position of the curve but not its slope, nor its curv- ature. Plotting A against D (fig. 7) gives no new information : the theory predicts that all the solutions n,, n,, n3, dz, d3 which give correct ($r, 0) curves should lead to the same (A, 0) curve. That the experimental points are in such a good agreement with the prediction gives proof of their reliability and of the internal consistency of our analysis. LORENTZ-LORENZ TREATMENT OF SURFACE LAYERS Three-layer models revert to single-layer models when d3 = 0. If the small differences between Newton films and equally thin common films relate to structure, and not to concentration of the surface layers we may ignore them in a Lorentz- Lorenz treatment of the Newton film.We have then found that a Newton film of NaDS is a single-phase lamella with a thickness 2d2 = 3.6 nm and a refractive index of 1.365. An analogous treatment of the NaDoBS system gives 2d, = 3.6 nm and n2 = 1.390+0.002. These results can then be discussed in terms of the Lorentz-Lorenz equation for a multicomponent system l4 : provided that the Newton film may be described as a single phase bulk material. V248 ELLIPSOMETRY OF SOAP FILMS is the average volume of a molecule in the film. We may restrict the summation over k components to water and the surfactant (NaDS or NaDoBS), other electrolytes being absent from thin films due to negative adsorption.x k are mole fractions, ak the polarizabilities of components k. Considering a Newton film of NaDS one obtains where V* = V/xNaDS. With an area per molecule for NaDS of 0.40 n m 2 , 1 6 aHzO = 0.15 x nm3,I7 it follows that xH20/xNaDS = 6.5k0.5. There are 6 or 7 molecules of water per molecule of surfactant in a Newton film of NaDS. We have no reliable data for the surface coverage in NaDoBS films but it is probably not very different from the value for NaDS. With 0.4 nm2 per molecule it is found again that there are 7 water molecules for 1 molecule of NaDoBS in the Newton film. It appears that there is not much of a difference in the composition of Newton films of NaDS and NaDoBS. The volume element Y* contains one molecule of surfactant and 7 water mole- cules.In NaDS, V* = 0.72 nm3, but the volume of an NaDS-molecule is approxi- mately 0.4 nm3 and 7 molecules of ordinary water would fill 0.2 nm3. This shows that the structure of a surface layer and of a Newton film is less dense than a close- packing of water and surfactant molecules. nm3 and aNaDS = 2.84 x CONCLUSIONS The ellipsometric investigation of soap films gives new information which illus- trates the necessity to describe such systems with a three-layer model. A fair descrip- tion of the surface layers in NaDS films is n, = 1.365, dz = 1.8 nm, and the core of the film, with the refractive index of the solution from which the film was drawn has thickness of approximately d3 = 5.1 nm in common black films and d3 = 0 in Newton films. There is a small change in the structure or in the composition of the surface layers when a Newton film is formed from a thick, or from a common black film. A Lorentz-Lorenz type treatment of Newton films reveals that there are approximately 7 water molecules per molecule of surfactant, both in NaDS and in NaDoBS films. And, finally, it seems important that the combination of ellipsometry and reflecto- metry provides a means to judge the quality of experimental data in optical studies of non-absorbing thin films. K. J. My&, K. Shinoda and S. Frankel, Soap Filnzs (Pergamon, London, 1959). J. M. Corkill, J. F. Goodman, D. R. Haisman and S . P. Harold, Trans. Faraday Soc., 1961,57, 821. M. N. Jones, K. J. Mysels and P. C. Scholten, Trans. Faraday Soc., 1966, 62, 1336. J. A. de Feyter, to be published. E. M. Duyvis, Thesis (Utrecht 1962). J. M. Corkill, J. F. Goodman and C. P. Ogden, Trans. Faraduy SOC., 1965,61, 583. A. Vasicek, Optics of Thin Films (North Holland, Amsterdam, 1960). H. Schopper, 2. Phys., 1952,132,146. ’ C . Smart and W . A. Senior, Trans. Faraduy Sac., 1966, 62, 3253. lo D. den Engelsen, J. Opt. SOC. Arner., 1971, 61, 1460. l 1 D. den Engelsen, J. Phys. Chem., 1972,76, 3390. l 3 J. B. Rijnbout, J. Phys. Chem., 1970,74,2001. l4 C. J. F. Bottcher, Theory of Electric Polarizufion (Elsevier, Amsterdam, 1952). l6 I. Weil, J. Phys. Chern., 1960, 70, 133. R. J. Cherry and D. Chapman, J. Mol. Biol., 1969,40,19. P. Drude, Ann. Phys., 1889, 36, 352, 865. Handbook of Chemistry and Physics (Chemical Rubber Publishing Co., 47th Edn., 1966).

ISSN:0300-9599    

DOI:10.1039/F19747000237

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Kinetics of the Thermal Isomerisation of Hexamethyldisilane BY IAIN M. T. DAVIDSON Department of Chemistry, The University, Leicester AND COLIN EABORN* AND JOHN M. SIMMIE School of Molecular Sciences, University of Sussex, Brighton BN1 9QJ Received 10th August, 1973 A kinetic investigation of the gas-phase isomerisation of hexamethyldisilane between 717 and 780 K with initial pressures of 10-125 mmHg is described, and discussed in relation to other work on the thermolysis of hexamethyldisilane. The isomerisation is a radical chain reaction of three halves order, with rate constants given by k/cm* mol-+ s-I = lO16*65*0.70 exp[ -(251 -t- 8)/RT] (activation energy in kJ mol-I). Thermolysis of hexamethyldisilane at pressures near or above atmospheric in a flow system or a sealed tube,** gives the isomeric trimethylsilyl(dimethylsily1)- methane, (I), in good yield, and we have given a preliminary account of a kinetic investigation of this is~merisation.~ Me,SiSiMe,-+ Me,SiCH,SiMe,H In the thermolysis of hexamethyldisilane at lower pressures, however, 0) is no longer the main p r o d u ~ t , ~ and is not observed at all in thermolysis at very low pressure^.^ The isomerisation has been interpreted as a radical chain reaction,l* whereas the earlier low pressure therrnolyses were believed to be non-~hain,~ enabling values of D(Me,Si-SiMe,) to be deduced directly from the kinetic results.However, a recent investigation has shown that even the low-pressure thermolysis proceeds by a chain sequence, albeit of short length, but that non-chain conditions prevail in the presence of a chain idubitor (m-~ylene).~ In consequence, previously accepted values of bond dissociation energies in hexamethyldisilane and related compounds have had to be revised ~ubstantially,~ and our results for the isomerisation of hexa- methyldisilane will be discussed in the light of these revised values.0 EXPERIMENTAL APPARATUS Kinetic experiments were conducted in an apparatus of conventional design, with a static reaction vessel consisting of a 450 cm3 quartz cylinder, from which samples could be withdrawn directly to a g.1.c. column ; details have been given.7 Quantitative analysis for reactants and products was by g.1.c. ; mass spectrometry, n.m.r. and infra-red spectroscopy were also used to identify products. MATERIALS Hexamethyldisilane was prepared by treating trimethylchlorosilane with potassium- sodium alloy in xylene, followed by fractionation, and final purification by preparative g.1.c.249250 ISOMERISATION OF HEXAMETHYLDISILANE The isomer (I) was prepared by thermolysis of hexamethyldisilane in sealed Pyrex tubes for 24 h at 723 K, with purification by preparative g.1.c. Trimethylsilane was prepared by reduction of trimethylchlorosilane with lithium aluminium hydride in dioxan and purified by trap-to-trap distillation. Tetramethylsilane and methane were obtained commercially. RESULTS Hexamethyldisilane was thermolysed between 71 7 and 780 K, with initial pressures of hexamethyldisilane ranging from 10 to 125 mmHg, equivalent to 2 x to 3 x mol ~ r n - ~ . The isomer (I), was the main product, its yield being about 17 times greater than that of the next most abundant product, trimethylsilane.A typical product analysis was : (I), 92.2 % ; trimethylsilane, 5.4 % ; tetramethylsilane, 1.6 % ; methane, 0.5 % ; and an unidentified product of long retention time, 0.3 %. Kinetic measurements were confined to the formation of (I). For a 12-fold range of initial pressures (10-125 mmHg) at 780.5 K, the variation of the ratio of [(I)]/[Me6Si2] with time indicated that the order was 1.56+0.06. In agreement with this, plots of [Me,Si,]* against time were linear up to at least 60 % conversion at 739, 759.5, and 780.5 K, and the three-halves order rate constants at 780.5 K were independent of initial pressure except for a slight fall-off at the highest pressures. Arrhenius para- meters were obtained over the temperature range with a fixed initial pressure of hexamethyldisilane of 26 mmHg ; three-halves order rate constants were given by eqn (i) (activation energy in kJ mol-I).k/cm+ mol-- s-I = 1016.65 fO.'O exp[ - (251 +_8)/RT] (9 The rate of isomerisation was unaffected by a 15-fold change in the surface to volume ratio of the reaction vessel and by added argon or benzene, but was reduced by added toluene. The effect of toluene was mainly studied in sealed tubes at 723 K ; adding an equimolar amount of toluene to the hexamethyldisilane reduced the rate of isomerisation by a factor of 10.0, but (I) was still the predominant product. A few experiments in the main apparatus at 780 K gave concordant results.DISCUSSION As previously suggested,l* the following sequence accounts for the observed Me,SiSiMe3 + 2Me3 Si. (1) (2) Me5Si2cH2+Me3SiCH2SiMe2 (3) 2Me,SiCH2SiMez + (Me3SiCH2SiMe2)2 ( 5 ) features of the thermal isomerisation of hexamethyldisilane. Me$* + Me3SiSiMe3+Me,SiH + Me,Si2cH2 Me3SiCH2SiMe2 + Me3SiSiMe3-+Me3SiCH2Si(H)Me2 + Me5Si2cH2 (4) Steady-state treatment of this scheme gives : d[(I)]/dt = (klkz/k5)+ [Me,Si,]s in agreement with the observed order of 2. The experimental activation energy E = $(El + 2E4 - E5). From the most recent work on he~amethyldisilane,~ El = 337 kJ mol-l and D(Me,Si-H) = 368 kJ mol-I 4* * ; D(Me3SiCH2-H) has been esti- mated as 406 kJ mol-l, and we assume that D(Me,Si2CH2-H) is a little smaller, say 400 kJ mol-l. Hence AH2 = D(Me5Si2CH2-H) - D(Me3Si-H) = 32 kJ mol-l, and (- 2) is an exothermic hydrogen abstraction by a substituted methyl radical for which E-2 may be estimated lo to be about 40 kJ mol-'.Thus, E2 = EA2 +AH2 = 72 kJ mol-l. Reaction (4) is similar to reaction (2), but we may assume that the silicon-bydrogen bond is somewhat weaker in (I) than in trimethylsilane, say 360 kJI . M . T . DAVIDSON, C. EABORN AND J . M . SIMMIE 25 1 mol-', whence AH4 = 40 kJ mol-1 and E4 = E-,+AH4 = 80 kJ mol-l. Radical combination reactions may be taken to have zero activation energy. Hence E = 3(337 + 2 x 80 - 0) = 248.5 kJ mol-l, in satisfactory agreement with the experimental values of 251 8 kJ mol-l. The situation with regard to A factors is less satisfactory. Al has been measured as s-l, and cannot be significantly higher on theoretical grounds lo ; and although the radicals involved in reaction (5) have a high molecular weight it is unlikely that As < 1010.5 cm3 mol-1 s-l. Hence, to reconcile the experimental results with the proposed mechanism, A4 must be 1013.3.~m3 mol-1 s-l, which ex- ceeds the conventional value for this type of reaction by about two powers of ten.An A factor as large as this has been reported for the reaction between methyl radicals and dimethyldichlorosilane,l but is now believed to be inc~rrect,~ while recent studies of chlorine abstraction by trimethylsilyl radicals from alkyl chlorides gave " normal " A factors around loll cm3 mol-1 s-l. This problem cannot be resolved until more data are available on the rate parameters for abstraction by substituted silyl radicals.The above figures may be used to calculate the relative concentrations of the various radicals in the steady-state system if k3 is known. Reaction (3) is exothermic and probably has an activation energy of about 80 kJmol-l, but to obtain a lower limit for k3 we shall take E3 = 120kJmol-l and A3 = 1013s-l. Then [Me,SiCH,$iMe,] : [Me3Sia] : [Me,Si,cH,] is > 1294 : 6.76 : 1.00 at 780 K and > 795 : 1.19 : 1.00 at 717 K. These figures justify the choice of reaction (5) as the sole termination step. Toluene, with D(PhCH,-H) = 351 kJ mol-l, has an easily-abstracted hydrogen atom, and the sequence of the isomerisation of hexamethyldisilane in toluene may be represented as follows : Me3SiSiMe3+2Me,Si- (1) (6) (71, (-7) (3) (8) 2PhcH2 + (PhCH,),.(9) Me3Si= + PhCH3 + Me3SiH + PhcH2 PhcH2 + Me,SiSiMe3 +PhCH3 + MeSSi,cH2 Me Si, cH2 + Me , SiCH, si Me, Me3SiCH,$iMe, + PhCH,-+Me,SiCH2Si(H)Me2 + PhcH, Steady-state treatment of the above scheme gives The kinetic parameters for reaction (- 7) will be similar to those for (- 2) and (- 4), and AH7 = 55 kJ mol-1 ; hence E7 = 93 kJ mol-'. Under the conditions used for the experiments with toluene, k3 was at least 57 times greater than k-,[PhCH3] at 717 K, and at least 190 times greater at 780K. Hence d[(I)]/dtw(klk,2/kg)+ [Me,Si,]3, and E = +(El +2&-Eg) = +(337+2 x 93-0) = 261.5 kJ mol-l. If this calculated activation energy is correct, the reduction in fate caused by the toluene is due to the increased activation energy (i.e.E,> E4), with the A factor remaining virtually unchanged. This in turn would imply that reaction (7), like reaction (4), has an abnormally high A factor. Although the isomer (I) was formed in the thermolysis of low pressures of hexa- methyldisilane in a nitrogen ~trearn,~ it was not the dominant product, there also being252 ISOMERISATION OF HEXAMETHYLDISILANE substantial quantities of trimethylsilane and 1,1,3,3-tetrarnethyl- 1,3-disilacyclobutane. Under these conditions the radical dissociation reaction Me,SiCH,SiMe,-+Me,Si* + Me,Si=CH, can compete with reaction (4), and the products of reaction (10) give rise to the di- silacyclobutane and the additional trimethylsilane. Addition of xylene suppresses the formation of these products because of the relatively high rate of the reaction analogous to (8), and the rate constants for isomerisation should then be the same as in our experiments with added toluene.This was found to be so, the three-halves order rate constant for isomerisation in xylene at 780 K, derived by use of data from the work described in ref. (4), being 0.11 times that calculated from eqn (i), while in our experiments with added toluene at 780 K the factor was 0.10. The small quantities of tetramethylsilane in the products may arise from the silylene elimination, (1 1). Me,SiSiMe,-+Me,Si + Me,Si: We thank the S.R.C. for a maintenance grant (to J. M. S.) and Dow Corning (Barry) Limited, for gifts of chemicals. K. Shiina and M. Kumada, J. Org. Chem., 1958,23,139. H. Sakurai, A. Hosomi and M. Kumada, Chem. Comm., 1968,930, C. Eaborn and J. M. Simmie, Chem. Comm., 1968, 1426. I. M. T. Davidson and A. V. Howard, Chem. Comm., 1973, 323. J. A. Connor, G. Finney, G. J. Leigh, R. N. Haszeldine, P. J. Robinson, R. D. Sedgwick and R. F. Simmons, Chem. Comm., 1966, 178 ; J. A. Connor, R. N. Haszeldine, G. J. Leigh and R. D. Sedgwick, J. Chem. Soc. A , 1967,768 ; I. M. T. Davidson and I. L. Stephenson, J. Chem. Soc. A , 1968,282. I. M. T. Davidson, Quart. Rev., 1971, 25, 111. R. Walsh, reported at Gas Kinetics Discussion Group Meeting, Leicester, 1972. J. A. Kerr, A. Stephens and J. C. Young, Int. J. Chem. Kinetics, 1969, 1, 339. lo S. W. Benson, Thermochemical Kinetics (Wiley, New York, 1968), p. 55 et seq. J. A. Kerr, D. H. Slater and J. C. Young, J. Chem. Suc. A, 1967, 134. l 2 P. Cadman, G. M. Tilsley and A. F. Trotman-Dickenson, J.C.S. Furuday I, 1973,69,914. ' C. Eaborn, J. M. Simmie and I. M. T. Davidson, J. Organometullic Chem., 1972, 44, 273.

ISSN:0300-9599    

DOI:10.1039/F19747000249

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Proton and Deuteron Mobility in Normal and Heavy Water Solutions of Electrolytes BY NOEL K. ROBERTS" AND HELEN L. NORTHEY Chemistry Dept., University of Tasmania, Australia 7001 Received 9th April, 1973 Proton and deuteron mobility in normal and heavy water solutions of some alkali halides (NaCI, KCl, KBr and KI) and tetra-alkylammonium bromides (where alkyl is Me, Et, Pr or Bu) is measured polarographically and the diffusion coefficients calculated from a corrected IIkovic equation. The results are analysed in terms of the effect of electrolyte concentration on the excess or Grotthuss mobility of the hydrogen and deuterium ion. The results raise some questions about the accepted mechanism for the transport of the proton and deuteron in solution. Earlier studies 1-5 on proton mobility in aqueous electrolyte solutions using polarography gave useful information about the degree of hydration of the proton and the structure of water in the presence of electrolytes.In this paper we give a more detailed analysis as a result of further work on deuteron mobility in heavy water solutions of electrolytes. EXPERIMENTAL Apparatus, purification techniques and special procedures are described elsewhere. 1-5 As was noted previously4 it is important to purify all salts thoroughly before use. For example, different samples of A.R. KCI gave different results before purification and the most marked difference was observed in the case of the tetra-alkylammonium salts. After purification the order of the diffusion currents for the hydrogen ion in the presence of the tetra-alkylammonium salts was altered from that reported previou~ly,~ fig.l(e) and (f). However, the half-wave potentials remained unchanged. Tetramethylammonium bromide (Kodak) was recrystallised three times from distilled de-ionised water and dried under reduced pressure at 110°C for 1 week. Tetraethylammonium bromide (Kodak) and tetra- propyIammonium bromide (Kodak) were recrystallised three times from super-dry ethanol and methanol respectively, and dried under reduced pressure at 110°C. Tetrabutylammon- iuin bromide (Kodak) was recrystallised five times from dry acetone, and dried under reduced pressure at 80°C. The purity of the recrystallised tetra-alkylammonium salts was determined by titration against silver nitrate using the Mohr method as percentage bromide and gave the following results, Me4NBr 100.00 %, Et4NBr 100.00 %, Pr4NBr 99.88 %, Bu4NBr 99.69 %.The alkali halides (A.R.) were recrystallised twice from distilled de- ionised water and dried under reduced pressure at 110°C for at least a week. In two previous publications 4-5 and in this work all salts were exhaustively purified as described above and the diffusion coefficients for the proton and deuteron so obtained were reproducible within 1 %. The DzO, 99.75 % by weight, was obtained from the Australian Institute of Nuclear Science and Engineering and was made 1 . 1 7 ~ mol dm3 in DCI by adding a small quantity of a solution of DCl in D20 (Stohler Isotope Chemicals). The concentration of DCl was determined by potentiometric titration with sodium tetraboraie. The solution of DCl in D20 was de-aerated with high purity nitrogen and stored in a glass dispensing vessel under nitrogen.To make up the solutions a quantity of DCl+D20 was dispensed with high purity nitrogen into a weighed glass container to which a known weight of salt had been added in a nitrogen-filled glove bag. The weight of DCl+ D20 was obtained by difference. 253254 PROTON AND DEUTERON MOBILITY I N SOLUTION A light oil was used in the constant temperature bath and while the polarogram was recorded nitrogen was run over the surface of the solution. All solutions were de-aerated before recording the polarogram of the deuterium ion. RESULTS AND DISCUSSION The corrected form of the Ilkovic equation id = 607nD*m~ct*(l i- AD&m-*t*) was used to calculate the diffusion coefficient of the deuterium ion and the limiting value for D+, Oh+, calculated as described for the hydrogen ion except that all extrapolations were performed on an aquamolality basis (i.e.moles of electrolyte per 55.51 moles of DzO). The corrected form of the Ilkovic equation for the hydrogen ion agrees with that obtained previously by Hans and Jensch. We obtained a value of 17+1 for the constant A and Hans and Jensch obtained A = 18.4+1.7.6* Strictly the value of D&+ so obtained refers to the diffusion coefficient of Df in the concentration of DCl originally added to the D 2 0 (i.e. 1.17 x mol dm-3 DCl). However Kolny and Zembura found that the concentration of H+ has no effect on the value of the diffusion coefficient in 0.1 mol dm-3 NaCl between pH 2.3 and 3.3.Also, in the diaphragm cell method the effect of the concentration of the tracer ion appears to have a negligible effect on the value of the diffusion coefficient. Stokes, Woolf and Mills reported that for the I- ion the diffusion coefficient at vanishing concentration of I- in alkali chloride solutions differed by less than 1 % from that in more concentrated solutions (-0.1 mol dm-3). Woolf lo observed a similar depend- ence for H+ in LiCl, NaCl and KC1, although there is some doubt about the applic- ability of the diaphragm cell method to the hydrogen ion? Consequently it appears legitimate to equate the experimentally obtained diffusion coefficients with the true diffusion coefficients of H+ and D+. TABLE 1 ion D 0 x lOs/cm* s-1 a D'x lW/crn2 s-1 H' 9.4+ 0.1 9.3 Df 6.2A0.1 6.5 a ref.(4) and present study ; b ref. (ll), (12) and (13). The value of Db+ is shown in table 1 together with D&+ obtained previously and Once again the good agreement between polarography and conductivity measure- the values of DE+ and Dk+ calculated from conductance data. ments validates the method used in this study. TABLE 2.-DIFFUSION COEFFICIENTS OF H+ IN 0.1 mol dnr3 SOLUTIONS OF KCl AND NaCl electrolyte conc. /mol d m - 3 0.1 KC1 0.1 KCl 0.1 KCl 0.1 NaCl 0.1 NaCl 0.1 NaCl 0.1 NaCl DHt x 1os/Cnl2 S - l 8.5k0.2 8.7k0.1 8.01 7.8k0.4 8.01 8.5+ 0.4 S.8kO.l met hod polarography polarography diaphragm cell rotating disc electrode diaphragm cell rotating disc electrode (platinum) polarography (amalgamated copper) ref.6 4 10 8 10 14 4255 It is also of interest to compare other results for the diffusion coefficient of the hydrogen in electrolyte solutions. Table 2 lists values which have been reported in 0.1 mol dm-3 solutions together with results from the present study. Only the present authors 1-5 and Woolf lo have studied hydrogen ion diffusion in more con- centrated electrolyte solutions. N . K . ROBERTS AND H . L . NORTHEY 0 - _' 0 4 5 I I aquamolality FIG.. 1.-Tracer diffusion coefficients of the proton (0) and deuteron (A) at 25°C in solutions of (a) NaCl, (b) KCI, (c) KBr, (d) KI, (e) Me4NBr, (f) Bu4NBr. 0, the self diffusion ratio for normal water at 23°C. It is clear from table 2 that values from polarography are consistently higher than those obtained from the diaphragm cell method, and that the rotating disc electrode gives both high and low values depending on the disc material.It would appear, then, that the higher figures are not peculiar to the polarographic method and this lends support to our earlier claim 4 p that our results are correct. Plots of D/D" for the hydrogen and deuterium ion against aquamolality of electrolyte were constructed for the following electrolytes, NaCl, KCl, KBr, KI, Me4NBr and Bu4NBr, fig. 1. Where available, data for the self-diffusion of water in the presence of these electrolytes were included for comparison from the work of256 PROTON AND DEUTERON M081LITY I N SOLUTION McCall and Douglas l5 and Brun et d . 1 6 As for the proton in normal water, deuteron mobility was higher in KC1 than in NaC1 solutions, greatest in KCl and least in KI.With the tetra-alkylammonium salts mobility was greatest in the tetramethyl- and least in the tetrabutyl-ammonium bromide solutions. Fig. 2 shows the results for deuteron mobility in Me,NBr, Et,NBr, Pr,NBr and Bu4NBr. aquamolalit y FIG. 2,-Tracer diffusion coeffic .. ,its of the deuteron in tetra-alkylammonium bromide solutions. 0, Me4NBr ; A, Et4NBr ; 0, PrcNBr ; V, Bu4NBr. With the alkali halides it is observed that D/D" for deuteron mobility in D20 is affected to a greater extent than proton mobility in HzO, whereas with the tetramethyl and tetrabutylammonium bromide salts the effect is almost the same up to - 1 aquamolal. Above 1 aquamolal D/DO decreases more slowly for D+ than for H+ with increasing concentration of Me,NBr while the reverse is true for Bu4NBr and the alkali halides.The anomalous behaviour of Me,NBr is evident. Furthermore the alkali halides reduce the mobility of the deuterium ion at a greater rate up to - 1 aquamolal but above this concentration the rate of change of D/D" for H+ and D+ is almost the same. In table 3 is shown the effect of concentration of supporting electrolyte on the diffusion coefficient of the hydrogen ion relative to that of the deuterium ion. Sodium chloride, and potassium chloride and bromide show much the same trend in the ratio DH + /OD + , whereas potassium iodide shows a much more rapid increase as a result of the relatively lower diffusion coefficient of the deuterium ion. The differ- ence may arise as a result of some specific interaction between the iodide ion and D20.In this connection it is worthy of note that Greyson l1 and Walrafen l2 have shown that in the case of water, structure-influencing properties are more sensitive to ion size for negative than for positive ions since the protons in water may polarise anions to an extent increasing with increasing anion size thus producing between the water and anion a bond which is less polar than the OH-0 bond of water itself and w ~ c h therefore tends to destroy structure in the surrounding water.N . K . ROBERTS A N D H . L . NORTHEY 257 TABLE 3 supporting DH+X 105/ elcctrolytc aquamoti11ity em2 s-1 NaCl 0.0 0.5 1 .o 2.0 3 .O 4.0 KCI KBr KI 0.5 1 .o 2.0 3 .O 4.0 0.5 I .o 2.0 3 .O 4.0 0.5 1 .o 2.0 3.0 4.0 Me4NBr 1.8 2.0 2.5 3 .o 3.5 Bu42\1Br 0.5 1 .o 1.5 2.0 2.5 2.9 9.3 7.1 6.0 4.9 4.1 3.4 7.3 6.7 5.9 5.5 5.1 7.1 5.6 4.0 3.2 2.6 7.1 5.7 4.0 3.1 2.5 3.8 3.3 2.5 2.1 1.9 3.9 2.1 1.4 0.95 0.64 0.49 DDI >: fos/ &? s-1 6.5 3.8 2.9 2.1 1.7 1.3 4.1 3.2 2.4 2.2 2.0 3.9 3 .O 2.0 1.4 0.98 3.7 2.7 1.7 1 .o 0.5 2.6 2.4 2.0 1.7 1.4 2.5 1.2 0.5 - - - DH+/DD+ 1.43 1.9 2.1 2.3 2.4 2.6 1.8 2.1 2.5 2.5 2.6 1.8 I .9 2.0 2.3 2.7 1.9 2.1 2.4 3.1 5.0 1.5 1.4 1.3 1.2 1.4 1.6 1.8 2.8 - - - THE GROTTHUSS COMPONENT OF HYDROGEN A N D DEUTERIUM ION DIFFUSION The discussion so far, while acknowledging the abnormal mobility of hydrogen ion diffusion, has made no attempt to separate the excess Grotthuss contribution from the total measures of the diffusion coefficients.The work on proton conductance of Conway, Bockris and Linton l 2 suggests that the Grotthuss component may be obtained by subtracting the component due to normal ionic diffusion from the values measured for the proton and deuteron diffusion coefficients. The potassium ionis of comparable radius to the hydrogen ion, so for proton diffusion in norma1 water solutions at various electrolyte concentrations, it is reasonable to say : Grouhuss contribution = (&+ - DK+)H20. 1-9258 PROTON AND DEUTERON MOBILITY I N SOLUTION There are, however, no values available for the diffusion coefficient of K+ in various electrolyte solutions, but such values are available for C1- in sodium and potassium chloride solutions,' ' and since the diffusion coefficients at infinite dilution for Kf and Cl- are little different (being 1.98 x lo5 and 2.03 x lo5 cm2 s-1 respectively), it is reasonable to substitute the Cl- values for those of K+.Then : Grotthuss contribution = (DH+ - Dcl-)H20. Grotthuss contribution = (OD+ - Dcl-)D20. Similarly for deuteron diffusion in heavy water solutions : Now the diffusion coefficient of the chloride ion in heavy water is not known, but it may be estimated using the Einstein equation : D = RT/6nqrN. which indicates that Dcc 1 /qr, where t,~ is the viscosity of the electrolyte solution and Y is the radius of the ion. The chloride ion is only slightly hydrated so there will be little error in assuming that its ionic radius is the same in D,O as in H20. Then (Dcl-)D20 in various electrolytes may be put equal to : where qH20/~D20 is known for various electrolyte solutions.Stokes and Mills l7 give values of the viscosity of potassium chloride in normal water, and those (DCl -)H20 rH2O/YD20 TABLE 4.-TOTAL DIFFUSION COEFFICIENTS AND GROTTHUSS COMPONENTS FOR HYDROGEN AND DEUTERIUM ION DIFFUSION IN ELECTROLYTE SOLUTIONS aquamolality DH+ x lo5 0.0 9.31 0.5 7.10 1 .o 5.98 2.0 4.92 3 .O 4.11 4.0 3.38 0.0 9.31 0.5 7.34 1 .o 6.68 2.0 5.94 3 .O 5.45 4.0 5.10 (DH+- DC1-)H20 (DD+- DC1-)D20 D,-~- x 105 x 105 DD+ x lo5 (DC1-)D20 x lo5 a x 10' NaCl 2.03 7.28 6.50 1.64 4.86 1.84 5.26 3.81 1.48 2.33 1.77 4.21 2.92 1.43 1.49 1.60 3.32 3.10 1.29 0.81 1.45 2.66 1.66 1.17 0.49 1.26 2.12 1.34 1.02 0.32 KCl 2.03 7.28 6.50 1.64 4.86 1.96 5.38 4.06 1.60 2.46 1.95 4.73 3.18 1.61 1.57 1.90 4.04 2.39 1.57 0.93 1.85 3.60 2.17 1.55 0.62 1.77 3.33 2.00 1.54 0.46 a ~ H ~ O / ~ D ~ O is not available for NaCl solutions ; relative viscosity for the pure liquids was used for calculations. Thus (DCl-)D20 = (DCL-)H20 x 1)H20/7DrO where 7Hz0/71D20 = 0.8072 at 25°C.for heavy water solutions were obtained from Selecki et al.18 All viscosities were calculated for an aquamolality basis. The results of these calculations are shown in table 4 which shows also the ratio of the Grotthuss component of the proton in normal water solutions to that of the deuteron in heavy water solutions of potassium chloride. This table also contains similar values for sodium chloride solutions, though another approximation was necessary to calculate these as viscosity data for heavy waterN.K . ROBERTS AND H . L . NORTHEY 259 solutions of sodium chloride were not available. Thus the ratio qH20/qD20 for the pure solvents was multiplied by (Dcl-)H20 values at the various concentrations to obtain (Dcl-)D,o values. This seems a justifiable assumption as the viscosity ratio for KCI solutions does not vary by more than 5 % from that of the pure solvents. When the ratio of the Grotthuss component for H+ in H,O solutions of electrolyte to this component for Df in D20 solutions is plotted against aquamolal concentration of' supporting electrolyte straiat lines are obtained (fig. 3). These extrapolate to the accepted value of the ratio of the diffusion coefficients at infinite dilution,12 i.e. 1.42. 7 - 6 - T 5 - 8 2 : I W 6 . ? 3 - I 0 0 : 2 3 4 aquamolalit y FIG.3.-The effect:of electrolyte concentration on the ratio of the total diffusion coefficients (curves) and the Grotthuss components (straight lines) in normal and heavy water solutions. A, KCl ; 0, NaCI. They may be contrasted with the non-linear curves obtained when the total diffusion coefficient ratios were plotted against concentration. The straight lines indicate some regularityin the reductionof the Grotthuss component in both normal and heavy water by both electrolytes. When the Grotthuss contributions alone are plotted against concentration, fig. 4, it is seen that the curves for deuterium diffusion are extremely close. This may indi- cate, once again, the importance of the anion in the reduction of deuteron mobility, though, with only two curves available, it is difficult to draw a definite conclusion.These curves also indicate that the Grotthuss contribution to the diffusion coefficient is very much greater for the proton than the deuteron. Also the Grotthuss contri- bution to Hf diffusion does not cut out at concentrations of supporting electrolyte above 2 molal, but is simply reduced less rapidly than at the low concentrations, where the Grotthuss component is 3-4 times that of normal ionic diffusion and at 3-4 mol2m P R O l O N A N D DLU I'LRON M O B l L I I Y l h SOLUTION kg-I still approximately twice t i x normal ionic: componcnt ip H,O. In contrast with thc work of Lown and Thirsk l 9 on aqueous solutions of KOH, NaOH and LiOH the Grotthuss contribution in 4 mol kg-' NaCl and KCl is significant.In D,Q solutions the Grotthuss component falls below the normal componeqt near 1 aquaivolal. e 7 - I N 6 > 6 2 9, 2 s X 4 s" I 0 4 4 + v -0 (d $ 3 s" 1 2 5 n I - - I ( j 0 I 2 3 4 aquamolaii t y FIG. 4.--The effect of electrolyte concentration on the Grotthuss component of the ditfusion coetficient in norqal and heavy water. A, KCI in HzO ; 0, NaCl in H 2 0 ; V , KCI in D20 ; 0, NaCl in P,O. MECHANISM FOR TRANSFER OF H ' A N D D I N THE PRESENCE OF ELECTROLYTES The results abwe call into question the accepted mecjxinism for the transport of H+ and D+ in H20 and D20 respectively. Currently proton mobility is considered to proceed via three steps. 1, Rotation of H 2 0 (or H,O+) ; 2, proton flip or tunnelling ; 3, randomisation of H20. At one atmosphere pressure the slow step is considered to be the rotation of the water molecule or H30T.Prima facie one wQuld expect, therdore, that electrolyks which cause water to become more fluid (i.e. reduce viscosity) and increase the self-diffusion of water e.g. KCl, KBr and KI, would make the rotational step easier and promote proton mobility. None of the electrolytes we have studied increase proton or deuteron mobility though K+ decreases it less than Na+ or Li+. However,k . K . ROBERTS AND H . t. NOkTHEY 26 1 even in the case of the cations the rate of change is incorrect e.g., the self-diffusion of water is almost constant in that region (0-1 molal) where protan mobility is changing most rapidly. The anions present a problem because they not only fail to enhance proton mobility they decrease it in the wrong order i.e.I- which produces the most fluid water and increases the self-diffusion of water decreases proton mobility to a greater extent than Br- and C1-. Thus the order for the decrease in proton and deuter- on mobility is I-> Br-> C1- whereas the order of increaing viscosity is Cl->Br-> I-. It is generally agreed that anions move by a different mechanism to cations 2o and the explanation may lie here. The tetra-alkylammonium bromides decrease proton mobility in the order one might expect i.e. in the direction of increasing viscosity so that proton and deuteron mobility are greatest in solutions of Me4NBr and least in Bu,NBr. It can be argued that the viscosity of a solution reflects both the translational and rotational movement of the water molecules and in the present discussion it is the rotational movement of the water molecule which is of importance.Fortunately it is not necessary to attempt to separate the two components in the work of McCall and Douglass from an n.m.r. study of the longitudinal relaxation time of the proton and deuteron have estimated the orientation time of H 2 0 and D20 in the pure liquids and also in many diamagnetic salt solutions. A selection of their results is given in table 5. as Hertz and Zeidler TABLE 5.-THERMALLY INDUCED ORIENTATION TIMES OF THE HzO AND DzO MOLECULE IN THE PRIMARY HYDRATION LAYER OF VARIOUS DIAMAGNETIC IONS 7 i i r ion ( 1 ) (2) Na+ 2.1 1.4 K+ 1 .o 1 .o Me4N-l- 1.6 1.6 Et4N’- - 2.1 3.1 Pr4N+ - B u ~ N + - 2.9 c1- 1 .o 1 .o Br- 0.6 0.8 I- 0.35 0.6 T? = orientation time of a H 2 0 or D20 niolecule in the primary hydration sphere; T = orient- ation time of a H 2 0 or D20 molecule in pure H 2 0 and D 2 0 respectively (-lo-“ s); ( 1 ) = results for water solutions : (2) = results for heavy water solutions.Note that : (a) 7 & ( K+)/T is placed equal to T k (C~-)/T because the longitudinal relaxation time for KCI solutions up to -4 molal equals that for pure water : (11) the ratio T+ / T depends on the value assumed for the hydration number of the ion. The lower rotational times of H,O and D20 in KBr and KI solutions compared with pure water support the previous argument based on the viscosities of the electro- lyte solutions and the self-diffusion of water in these solutions. Strictly speaking the rate determining step in proton and deuteron mobility is not the rotation of H 2 0 or D20 but the field-induced orientation of H 2 0 or D 2 0 l 2 and this may not coincide with the thermally induced orientation of water in the presence of electrolytes.The present work suggests that a measure of the thermally induced orientation of water (e.g. viscosity, self-diffusion of water or rotational orientation times) may give little or no indication of the field-induced rotztion of H,O or D 2 0 in the presence of electrolytes.”262 PROTON AND DEUTERON MOBILITY I N SOLUTION Further work is in progress on the temperature dependence of proton and deuteron mobility in the presence of electrolytes. N. K. Roberts and H. van der Woude, J .Chem. Suc. A, 1968, 940. N. K. Roberts and H. L. Northey, J. Chem. Soc. A, 1971, 2572. N. K. Roberts and H. L. Northey, J. Chem. Soc. A, 1971, 2640. N. K. Roberts and H. L. Northey, J.C.S. Furaday I , 1972, 68, 1528. N. K. Roberts and H. L. Northey, Nature, 1972, 237, 144. W. Hans and W. Jensch, Z . Elektrochem., 1952,56, 648. D. S. Turnham, J. Electroanal. Chem., 1965, 9, 440. R. H. Stokes, L. A. Woolf and R. Mills, J. Phys. Chem., 1957, 61, 1634. lo L. A. Woolf, Ph.D. Thesis (University of New England, 1958) ; J. Phys. Chem., 1960, 64, 481. R. A. Robinson and R. H. Stokes, Electrolyte Solutions (Butterworths, London, 2nd edn., 1959), p. 317. * H. Kolny and Z . Zembura, Roczniki Chem., 1971, 45, 1593. l 2 B. E. Conway, J. O’M. Bockris and Hedda Linton, J.Chem. Phys., 1956,24, 834. l 3 G. N. Lewis and F. G . Doody, J. Amer. Chem. Soc., 1933,55, 3504. l4 D. Jahna, Doctoral Thesis (University of Bonn, 1964), quoted in ref. (8). l 5 D. W. McCall and D. C . Douglas, J . Phys. Chem., 1965, 69, 2001. l 6 B. Brun, M. Servent and J. Salvinien, Compt. rend. C, 1969, 269, 1. l7 R. H. Stokes and R. Mills, Viscosity of Electrolytes and Related Properties (Pergamon, London, l 8 A. Selecki, B. Tyminski and A. Chmielewski, J. Chem. Eng. Data, 1970, 15, 127. l9 D. A. Lown and H. R. Thirsk, Trans. Furaday SOC., 1971, 67, 132. 2o H. S. Frank in Chemical Physics of Ionic Solutions, ed. B. E. Conway and R. G. Barradas (Wiley, 21 H. G . Hertz and M. D. Zeidler, Bey. Bunsmges. phys. Chem., 1963, 67, 774 and 1964, 68, 821. 22 A similar conclusion is implied in the work of H.G. Hertz and R. Klute, 2. Phys. Chem. 1965). New York, 1966), p. 63. (Frankfurt), 1970, 69, 101. Proton and Deuteron Mobility in Normal and Heavy Water Solutions of Electrolytes BY NOEL K. ROBERTS" AND HELEN L. NORTHEY Chemistry Dept., University of Tasmania, Australia 7001 Received 9th April, 1973 Proton and deuteron mobility in normal and heavy water solutions of some alkali halides (NaCI, KCl, KBr and KI) and tetra-alkylammonium bromides (where alkyl is Me, Et, Pr or Bu) is measured polarographically and the diffusion coefficients calculated from a corrected IIkovic equation. The results are analysed in terms of the effect of electrolyte concentration on the excess or Grotthuss mobility of the hydrogen and deuterium ion.The results raise some questions about the accepted mechanism for the transport of the proton and deuteron in solution. Earlier studies 1-5 on proton mobility in aqueous electrolyte solutions using polarography gave useful information about the degree of hydration of the proton and the structure of water in the presence of electrolytes. In this paper we give a more detailed analysis as a result of further work on deuteron mobility in heavy water solutions of electrolytes. EXPERIMENTAL Apparatus, purification techniques and special procedures are described elsewhere. 1-5 As was noted previously4 it is important to purify all salts thoroughly before use. For example, different samples of A.R. KCI gave different results before purification and the most marked difference was observed in the case of the tetra-alkylammonium salts.After purification the order of the diffusion currents for the hydrogen ion in the presence of the tetra-alkylammonium salts was altered from that reported previou~ly,~ fig. l(e) and (f). However, the half-wave potentials remained unchanged. Tetramethylammonium bromide (Kodak) was recrystallised three times from distilled de-ionised water and dried under reduced pressure at 110°C for 1 week. Tetraethylammonium bromide (Kodak) and tetra- propyIammonium bromide (Kodak) were recrystallised three times from super-dry ethanol and methanol respectively, and dried under reduced pressure at 110°C. Tetrabutylammon- iuin bromide (Kodak) was recrystallised five times from dry acetone, and dried under reduced pressure at 80°C.The purity of the recrystallised tetra-alkylammonium salts was determined by titration against silver nitrate using the Mohr method as percentage bromide and gave the following results, Me4NBr 100.00 %, Et4NBr 100.00 %, Pr4NBr 99.88 %, Bu4NBr 99.69 %. The alkali halides (A.R.) were recrystallised twice from distilled de- ionised water and dried under reduced pressure at 110°C for at least a week. In two previous publications 4-5 and in this work all salts were exhaustively purified as described above and the diffusion coefficients for the proton and deuteron so obtained were reproducible within 1 %. The DzO, 99.75 % by weight, was obtained from the Australian Institute of Nuclear Science and Engineering and was made 1 . 1 7 ~ mol dm3 in DCI by adding a small quantity of a solution of DCl in D20 (Stohler Isotope Chemicals).The concentration of DCl was determined by potentiometric titration with sodium tetraboraie. The solution of DCl in D20 was de-aerated with high purity nitrogen and stored in a glass dispensing vessel under nitrogen. To make up the solutions a quantity of DCl+D20 was dispensed with high purity nitrogen into a weighed glass container to which a known weight of salt had been added in a nitrogen-filled glove bag. The weight of DCl+ D20 was obtained by difference. 253254 PROTON AND DEUTERON MOBILITY I N SOLUTION A light oil was used in the constant temperature bath and while the polarogram was recorded nitrogen was run over the surface of the solution. All solutions were de-aerated before recording the polarogram of the deuterium ion.RESULTS AND DISCUSSION The corrected form of the Ilkovic equation id = 607nD*m~ct*(l i- AD&m-*t*) was used to calculate the diffusion coefficient of the deuterium ion and the limiting value for D+, Oh+, calculated as described for the hydrogen ion except that all extrapolations were performed on an aquamolality basis (i.e. moles of electrolyte per 55.51 moles of DzO). The corrected form of the Ilkovic equation for the hydrogen ion agrees with that obtained previously by Hans and Jensch. We obtained a value of 17+1 for the constant A and Hans and Jensch obtained A = 18.4+1.7.6* Strictly the value of D&+ so obtained refers to the diffusion coefficient of Df in the concentration of DCl originally added to the D 2 0 (i.e.1.17 x mol dm-3 DCl). However Kolny and Zembura found that the concentration of H+ has no effect on the value of the diffusion coefficient in 0.1 mol dm-3 NaCl between pH 2.3 and 3.3. Also, in the diaphragm cell method the effect of the concentration of the tracer ion appears to have a negligible effect on the value of the diffusion coefficient. Stokes, Woolf and Mills reported that for the I- ion the diffusion coefficient at vanishing concentration of I- in alkali chloride solutions differed by less than 1 % from that in more concentrated solutions (-0.1 mol dm-3). Woolf lo observed a similar depend- ence for H+ in LiCl, NaCl and KC1, although there is some doubt about the applic- ability of the diaphragm cell method to the hydrogen ion? Consequently it appears legitimate to equate the experimentally obtained diffusion coefficients with the true diffusion coefficients of H+ and D+.TABLE 1 ion D 0 x lOs/cm* s-1 a D'x lW/crn2 s-1 H' 9.4+ 0.1 9.3 Df 6.2A0.1 6.5 a ref. (4) and present study ; b ref. (ll), (12) and (13). The value of Db+ is shown in table 1 together with D&+ obtained previously and Once again the good agreement between polarography and conductivity measure- the values of DE+ and Dk+ calculated from conductance data. ments validates the method used in this study. TABLE 2.-DIFFUSION COEFFICIENTS OF H+ IN 0.1 mol dnr3 SOLUTIONS OF KCl AND NaCl electrolyte conc. /mol d m - 3 0.1 KC1 0.1 KCl 0.1 KCl 0.1 NaCl 0.1 NaCl 0.1 NaCl 0.1 NaCl DHt x 1os/Cnl2 S - l 8.5k0.2 8.7k0.1 8.01 7.8k0.4 8.01 8.5+ 0.4 S.8kO.l met hod polarography polarography diaphragm cell rotating disc electrode diaphragm cell rotating disc electrode (platinum) polarography (amalgamated copper) ref.6 4 10 8 10 14 4255 It is also of interest to compare other results for the diffusion coefficient of the hydrogen in electrolyte solutions. Table 2 lists values which have been reported in 0.1 mol dm-3 solutions together with results from the present study. Only the present authors 1-5 and Woolf lo have studied hydrogen ion diffusion in more con- centrated electrolyte solutions. N . K . ROBERTS AND H . L . NORTHEY 0 - _' 0 4 5 I I aquamolality FIG.. 1.-Tracer diffusion coefficients of the proton (0) and deuteron (A) at 25°C in solutions of (a) NaCl, (b) KCI, (c) KBr, (d) KI, (e) Me4NBr, (f) Bu4NBr.0, the self diffusion ratio for normal water at 23°C. It is clear from table 2 that values from polarography are consistently higher than those obtained from the diaphragm cell method, and that the rotating disc electrode gives both high and low values depending on the disc material. It would appear, then, that the higher figures are not peculiar to the polarographic method and this lends support to our earlier claim 4 p that our results are correct. Plots of D/D" for the hydrogen and deuterium ion against aquamolality of electrolyte were constructed for the following electrolytes, NaCl, KCl, KBr, KI, Me4NBr and Bu4NBr, fig. 1. Where available, data for the self-diffusion of water in the presence of these electrolytes were included for comparison from the work of256 PROTON AND DEUTERON M081LITY I N SOLUTION McCall and Douglas l5 and Brun et d .1 6 As for the proton in normal water, deuteron mobility was higher in KC1 than in NaC1 solutions, greatest in KCl and least in KI. With the tetra-alkylammonium salts mobility was greatest in the tetramethyl- and least in the tetrabutyl-ammonium bromide solutions. Fig. 2 shows the results for deuteron mobility in Me,NBr, Et,NBr, Pr,NBr and Bu4NBr. aquamolalit y FIG. 2,-Tracer diffusion coeffic .. ,its of the deuteron in tetra-alkylammonium bromide solutions. 0, Me4NBr ; A, Et4NBr ; 0, PrcNBr ; V, Bu4NBr. With the alkali halides it is observed that D/D" for deuteron mobility in D20 is affected to a greater extent than proton mobility in HzO, whereas with the tetramethyl and tetrabutylammonium bromide salts the effect is almost the same up to - 1 aquamolal.Above 1 aquamolal D/DO decreases more slowly for D+ than for H+ with increasing concentration of Me,NBr while the reverse is true for Bu4NBr and the alkali halides. The anomalous behaviour of Me,NBr is evident. Furthermore the alkali halides reduce the mobility of the deuterium ion at a greater rate up to - 1 aquamolal but above this concentration the rate of change of D/D" for H+ and D+ is almost the same. In table 3 is shown the effect of concentration of supporting electrolyte on the diffusion coefficient of the hydrogen ion relative to that of the deuterium ion. Sodium chloride, and potassium chloride and bromide show much the same trend in the ratio DH + /OD + , whereas potassium iodide shows a much more rapid increase as a result of the relatively lower diffusion coefficient of the deuterium ion.The differ- ence may arise as a result of some specific interaction between the iodide ion and D20. In this connection it is worthy of note that Greyson l1 and Walrafen l2 have shown that in the case of water, structure-influencing properties are more sensitive to ion size for negative than for positive ions since the protons in water may polarise anions to an extent increasing with increasing anion size thus producing between the water and anion a bond which is less polar than the OH-0 bond of water itself and w ~ c h therefore tends to destroy structure in the surrounding water.N . K . ROBERTS A N D H . L .NORTHEY 257 TABLE 3 supporting DH+X 105/ elcctrolytc aquamoti11ity em2 s-1 NaCl 0.0 0.5 1 .o 2.0 3 .O 4.0 KCI KBr KI 0.5 1 .o 2.0 3 .O 4.0 0.5 I .o 2.0 3 .O 4.0 0.5 1 .o 2.0 3.0 4.0 Me4NBr 1.8 2.0 2.5 3 .o 3.5 Bu42\1Br 0.5 1 .o 1.5 2.0 2.5 2.9 9.3 7.1 6.0 4.9 4.1 3.4 7.3 6.7 5.9 5.5 5.1 7.1 5.6 4.0 3.2 2.6 7.1 5.7 4.0 3.1 2.5 3.8 3.3 2.5 2.1 1.9 3.9 2.1 1.4 0.95 0.64 0.49 DDI >: fos/ &? s-1 6.5 3.8 2.9 2.1 1.7 1.3 4.1 3.2 2.4 2.2 2.0 3.9 3 .O 2.0 1.4 0.98 3.7 2.7 1.7 1 .o 0.5 2.6 2.4 2.0 1.7 1.4 2.5 1.2 0.5 - - - DH+/DD+ 1.43 1.9 2.1 2.3 2.4 2.6 1.8 2.1 2.5 2.5 2.6 1.8 I .9 2.0 2.3 2.7 1.9 2.1 2.4 3.1 5.0 1.5 1.4 1.3 1.2 1.4 1.6 1.8 2.8 - - - THE GROTTHUSS COMPONENT OF HYDROGEN A N D DEUTERIUM ION DIFFUSION The discussion so far, while acknowledging the abnormal mobility of hydrogen ion diffusion, has made no attempt to separate the excess Grotthuss contribution from the total measures of the diffusion coefficients.The work on proton conductance of Conway, Bockris and Linton l 2 suggests that the Grotthuss component may be obtained by subtracting the component due to normal ionic diffusion from the values measured for the proton and deuteron diffusion coefficients. The potassium ionis of comparable radius to the hydrogen ion, so for proton diffusion in norma1 water solutions at various electrolyte concentrations, it is reasonable to say : Grouhuss contribution = (&+ - DK+)H20. 1-9258 PROTON AND DEUTERON MOBILITY I N SOLUTION There are, however, no values available for the diffusion coefficient of K+ in various electrolyte solutions, but such values are available for C1- in sodium and potassium chloride solutions,' ' and since the diffusion coefficients at infinite dilution for Kf and Cl- are little different (being 1.98 x lo5 and 2.03 x lo5 cm2 s-1 respectively), it is reasonable to substitute the Cl- values for those of K+.Then : Grotthuss contribution = (DH+ - Dcl-)H20. Grotthuss contribution = (OD+ - Dcl-)D20. Similarly for deuteron diffusion in heavy water solutions : Now the diffusion coefficient of the chloride ion in heavy water is not known, but it may be estimated using the Einstein equation : D = RT/6nqrN. which indicates that Dcc 1 /qr, where t,~ is the viscosity of the electrolyte solution and Y is the radius of the ion. The chloride ion is only slightly hydrated so there will be little error in assuming that its ionic radius is the same in D,O as in H20.Then (Dcl-)D20 in various electrolytes may be put equal to : where qH20/~D20 is known for various electrolyte solutions. Stokes and Mills l7 give values of the viscosity of potassium chloride in normal water, and those (DCl -)H20 rH2O/YD20 TABLE 4.-TOTAL DIFFUSION COEFFICIENTS AND GROTTHUSS COMPONENTS FOR HYDROGEN AND DEUTERIUM ION DIFFUSION IN ELECTROLYTE SOLUTIONS aquamolality DH+ x lo5 0.0 9.31 0.5 7.10 1 .o 5.98 2.0 4.92 3 .O 4.11 4.0 3.38 0.0 9.31 0.5 7.34 1 .o 6.68 2.0 5.94 3 .O 5.45 4.0 5.10 (DH+- DC1-)H20 (DD+- DC1-)D20 D,-~- x 105 x 105 DD+ x lo5 (DC1-)D20 x lo5 a x 10' NaCl 2.03 7.28 6.50 1.64 4.86 1.84 5.26 3.81 1.48 2.33 1.77 4.21 2.92 1.43 1.49 1.60 3.32 3.10 1.29 0.81 1.45 2.66 1.66 1.17 0.49 1.26 2.12 1.34 1.02 0.32 KCl 2.03 7.28 6.50 1.64 4.86 1.96 5.38 4.06 1.60 2.46 1.95 4.73 3.18 1.61 1.57 1.90 4.04 2.39 1.57 0.93 1.85 3.60 2.17 1.55 0.62 1.77 3.33 2.00 1.54 0.46 a ~ H ~ O / ~ D ~ O is not available for NaCl solutions ; relative viscosity for the pure liquids was used for calculations.Thus (DCl-)D20 = (DCL-)H20 x 1)H20/7DrO where 7Hz0/71D20 = 0.8072 at 25°C. for heavy water solutions were obtained from Selecki et al.18 All viscosities were calculated for an aquamolality basis. The results of these calculations are shown in table 4 which shows also the ratio of the Grotthuss component of the proton in normal water solutions to that of the deuteron in heavy water solutions of potassium chloride. This table also contains similar values for sodium chloride solutions, though another approximation was necessary to calculate these as viscosity data for heavy waterN.K . ROBERTS AND H . L . NORTHEY 259 solutions of sodium chloride were not available. Thus the ratio qH20/qD20 for the pure solvents was multiplied by (Dcl-)H20 values at the various concentrations to obtain (Dcl-)D,o values. This seems a justifiable assumption as the viscosity ratio for KCI solutions does not vary by more than 5 % from that of the pure solvents. When the ratio of the Grotthuss component for H+ in H,O solutions of electrolyte to this component for Df in D20 solutions is plotted against aquamolal concentration of' supporting electrolyte straiat lines are obtained (fig. 3). These extrapolate to the accepted value of the ratio of the diffusion coefficients at infinite dilution,12 i.e.1.42. 7 - 6 - T 5 - 8 2 : I W 6 . ? 3 - I 0 0 : 2 3 4 aquamolalit y FIG. 3.-The effect:of electrolyte concentration on the ratio of the total diffusion coefficients (curves) and the Grotthuss components (straight lines) in normal and heavy water solutions. A, KCl ; 0, NaCI. They may be contrasted with the non-linear curves obtained when the total diffusion coefficient ratios were plotted against concentration. The straight lines indicate some regularityin the reductionof the Grotthuss component in both normal and heavy water by both electrolytes. When the Grotthuss contributions alone are plotted against concentration, fig. 4, it is seen that the curves for deuterium diffusion are extremely close.This may indi- cate, once again, the importance of the anion in the reduction of deuteron mobility, though, with only two curves available, it is difficult to draw a definite conclusion. These curves also indicate that the Grotthuss contribution to the diffusion coefficient is very much greater for the proton than the deuteron. Also the Grotthuss contri- bution to Hf diffusion does not cut out at concentrations of supporting electrolyte above 2 molal, but is simply reduced less rapidly than at the low concentrations, where the Grotthuss component is 3-4 times that of normal ionic diffusion and at 3-4 mol2m P R O l O N A N D DLU I'LRON M O B l L I I Y l h SOLUTION kg-I still approximately twice t i x normal ionic: componcnt ip H,O.In contrast with thc work of Lown and Thirsk l 9 on aqueous solutions of KOH, NaOH and LiOH the Grotthuss contribution in 4 mol kg-' NaCl and KCl is significant. In D,Q solutions the Grotthuss component falls below the normal componeqt near 1 aquaivolal. e 7 - I N 6 > 6 2 9, 2 s X 4 s" I 0 4 4 + v -0 (d $ 3 s" 1 2 5 n I - - I ( j 0 I 2 3 4 aquamolaii t y FIG. 4.--The effect of electrolyte concentration on the Grotthuss component of the ditfusion coetficient in norqal and heavy water. A, KCI in HzO ; 0, NaCl in H 2 0 ; V , KCI in D20 ; 0, NaCl in P,O. MECHANISM FOR TRANSFER OF H ' A N D D I N THE PRESENCE OF ELECTROLYTES The results abwe call into question the accepted mecjxinism for the transport of H+ and D+ in H20 and D20 respectively. Currently proton mobility is considered to proceed via three steps.1, Rotation of H 2 0 (or H,O+) ; 2, proton flip or tunnelling ; 3, randomisation of H20. At one atmosphere pressure the slow step is considered to be the rotation of the water molecule or H30T. Prima facie one wQuld expect, therdore, that electrolyks which cause water to become more fluid (i.e. reduce viscosity) and increase the self-diffusion of water e.g. KCl, KBr and KI, would make the rotational step easier and promote proton mobility. None of the electrolytes we have studied increase proton or deuteron mobility though K+ decreases it less than Na+ or Li+. However,k . K . ROBERTS AND H . t. NOkTHEY 26 1 even in the case of the cations the rate of change is incorrect e.g., the self-diffusion of water is almost constant in that region (0-1 molal) where protan mobility is changing most rapidly.The anions present a problem because they not only fail to enhance proton mobility they decrease it in the wrong order i.e. I- which produces the most fluid water and increases the self-diffusion of water decreases proton mobility to a greater extent than Br- and C1-. Thus the order for the decrease in proton and deuter- on mobility is I-> Br-> C1- whereas the order of increaing viscosity is Cl->Br-> I-. It is generally agreed that anions move by a different mechanism to cations 2o and the explanation may lie here. The tetra-alkylammonium bromides decrease proton mobility in the order one might expect i.e. in the direction of increasing viscosity so that proton and deuteron mobility are greatest in solutions of Me4NBr and least in Bu,NBr.It can be argued that the viscosity of a solution reflects both the translational and rotational movement of the water molecules and in the present discussion it is the rotational movement of the water molecule which is of importance. Fortunately it is not necessary to attempt to separate the two components in the work of McCall and Douglass from an n.m.r. study of the longitudinal relaxation time of the proton and deuteron have estimated the orientation time of H 2 0 and D20 in the pure liquids and also in many diamagnetic salt solutions. A selection of their results is given in table 5. as Hertz and Zeidler TABLE 5.-THERMALLY INDUCED ORIENTATION TIMES OF THE HzO AND DzO MOLECULE IN THE PRIMARY HYDRATION LAYER OF VARIOUS DIAMAGNETIC IONS 7 i i r ion ( 1 ) (2) Na+ 2.1 1.4 K+ 1 .o 1 .o Me4N-l- 1.6 1.6 Et4N’- - 2.1 3.1 Pr4N+ - B u ~ N + - 2.9 c1- 1 .o 1 .o Br- 0.6 0.8 I- 0.35 0.6 T? = orientation time of a H 2 0 or D20 niolecule in the primary hydration sphere; T = orient- ation time of a H 2 0 or D20 molecule in pure H 2 0 and D 2 0 respectively (-lo-“ s); ( 1 ) = results for water solutions : (2) = results for heavy water solutions.Note that : (a) 7 & ( K+)/T is placed equal to T k (C~-)/T because the longitudinal relaxation time for KCI solutions up to -4 molal equals that for pure water : (11) the ratio T+ / T depends on the value assumed for the hydration number of the ion. The lower rotational times of H,O and D20 in KBr and KI solutions compared with pure water support the previous argument based on the viscosities of the electro- lyte solutions and the self-diffusion of water in these solutions.Strictly speaking the rate determining step in proton and deuteron mobility is not the rotation of H 2 0 or D20 but the field-induced orientation of H 2 0 or D 2 0 l 2 and this may not coincide with the thermally induced orientation of water in the presence of electrolytes. The present work suggests that a measure of the thermally induced orientation of water (e.g. viscosity, self-diffusion of water or rotational orientation times) may give little or no indication of the field-induced rotztion of H,O or D 2 0 in the presence of electrolytes.”262 PROTON AND DEUTERON MOBILITY I N SOLUTION Further work is in progress on the temperature dependence of proton and deuteron mobility in the presence of electrolytes. N. K. Roberts and H. van der Woude, J . Chem. Suc. A, 1968, 940. N. K. Roberts and H. L. Northey, J. Chem. Soc. A, 1971, 2572. N. K. Roberts and H. L. Northey, J. Chem. Soc. A, 1971, 2640. N. K. Roberts and H. L. Northey, J.C.S. Furaday I , 1972, 68, 1528. N. K. Roberts and H. L. Northey, Nature, 1972, 237, 144. W. Hans and W. Jensch, Z . Elektrochem., 1952,56, 648. D. S. Turnham, J. Electroanal. Chem., 1965, 9, 440. R. H. Stokes, L. A. Woolf and R. Mills, J. Phys. Chem., 1957, 61, 1634. lo L. A. Woolf, Ph.D. Thesis (University of New England, 1958) ; J. Phys. Chem., 1960, 64, 481. R. A. Robinson and R. H. Stokes, Electrolyte Solutions (Butterworths, London, 2nd edn., 1959), p. 317. * H. Kolny and Z . Zembura, Roczniki Chem., 1971, 45, 1593. l 2 B. E. Conway, J. O’M. Bockris and Hedda Linton, J. Chem. Phys., 1956,24, 834. l 3 G. N. Lewis and F. G . Doody, J. Amer. Chem. Soc., 1933,55, 3504. l4 D. Jahna, Doctoral Thesis (University of Bonn, 1964), quoted in ref. (8). l 5 D. W. McCall and D. C . Douglas, J . Phys. Chem., 1965, 69, 2001. l 6 B. Brun, M. Servent and J. Salvinien, Compt. rend. C, 1969, 269, 1. l7 R. H. Stokes and R. Mills, Viscosity of Electrolytes and Related Properties (Pergamon, London, l 8 A. Selecki, B. Tyminski and A. Chmielewski, J. Chem. Eng. Data, 1970, 15, 127. l9 D. A. Lown and H. R. Thirsk, Trans. Furaday SOC., 1971, 67, 132. 2o H. S. Frank in Chemical Physics of Ionic Solutions, ed. B. E. Conway and R. G. Barradas (Wiley, 21 H. G . Hertz and M. D. Zeidler, Bey. Bunsmges. phys. Chem., 1963, 67, 774 and 1964, 68, 821. 22 A similar conclusion is implied in the work of H. G. Hertz and R. Klute, 2. Phys. Chem. 1965). New York, 1966), p. 63. (Frankfurt), 1970, 69, 101.

ISSN:0300-9599    

DOI:10.1039/F19747000253

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Electron Spin Resonance Studies of Isotopically Labelled Oxygen Species Adsorbed on Supported Molybdenum B Y MICHEL CHE,? ANTHONY J. TENCH* Chemistry Division, A .E.R.E., Harwell CLAUDE NACCACHE AND Institut de Recherches sur la Catalyse, C.N.R.S. 69 10-Villeurbanne, France Received 30th April, 1973 The molecular ion (O,)s has been observed on the surface of the molybdenum/SiOz and molyb- denum/Al2O3 systems by e.s.r. Using oxygen isotopically enriched in "0, two different hyperfine splittings were observed in the Mo/Si system corresponding to A, = 85 and 72 G respectively ; the A, and AZz values of the tensor were too small to be resolved. The observed values for Axx are shown to arise from an unequal interaction of the unpaired electron with two oxygen nuclei in the same molecular ion.This result is contrary to the previous work for (O& on other oxides where the oxygen nuclei have been found to be equivalent ; the difference is attributed either to a more covalent bonding with the surface ions or to adsorption at an unusual site at the surface. Thisis thought to be a sensitive function of the surface environment since the oxygen nuclei appear to be nearly equivalent for the Mo/AI system although the g-tensors for both systems are almost identical. A number of papers have recently appeared on oxidation reactions using supported catalysts and some attention has been paid to molybdenum impregnated on a variety of ~upports.~'~ It has been shown that the valence state of surface molybdenum ions is important in determining how the hydrocarbon adsorbs on the surface and therefore the catalyst selectivity.6 The other important aspect is the state of the adsorbed oxygen and this was first studied in some detail by Dufaux et aL7 who concluded that 0; was formed on the surface.More recently Kazansky et aL2 extended this work and showed that (O-)s could also be stabilized on the surface after N20 adsorption at 100°C. The (O-)s species was shown to be more reactive than (O;)s in agreement with the results on Mg0.8* However, the (O-)s species has not been detected after the adsorption of 0, and in this work the adsorption of O2 isotopically enriched with 170 was studied to confirm that the ( 0 ~ ) ~ species was formed on such active oxidation catalysts and to see whether the structure of the adsorbed species was similar to that found on other oxides.The slight departure of the lowest component of the g-tensor from the free elec- tron g value has been recently the subject of some attention.14 Kazansky et all5 showed that, on V/Si02, the hyperfine tensor of (l6O$), could be explained by some covalency of the (OT)s bond with the transition metal ion through a partial transfer of both the unpaired electron and the pair of electrons on the oxygen to the vacant orbitals of the ion. The associated lowest component of the g-tensor was 2.004. t permanent address : Institut de Recherches sur la Catalyse. 263264 OXYGEN ON SUPPORTED MOLYBDENUM The model proposed was a molecular oxygen radical ion adsorbed with the inter- nuclear axis parallel to the surface with the electron equally delocalised on to the two oxygen nuclei.One would expect in such a case a (170;), spectrum similar to that obtained with Mgd.'O Unfortunately, it has not been possible to check ?.he validity of such a model using oxygen enriched with 1 7 0 owing to the complication arising from the nuclear spin of 51V(Z = 7/2) and the rather poor resolution of the e.s.r. spectrum obtained.16 More recently, it has been shown l 4 that, on CeO,/SiO,, the lowest component of the g-tensor was 2.0109, which was tentatively explained by the influence of thef-orbital electrons of cerium. In this case, the (170;), spectrum showed that the electron was equally delocalized on the two oxygen nuclei. It was also interesting to investigate the case of supported molybdenum where the lowest component of the g-tensor was -2.0048 and covalent bonding might be expected. This work has extended the previous studies of isotopically labelled adsorbed oxygen to the supported molybdenum system and confirms earlier work 17* that the (170;), radical ion has an unequal didtfibution of the unpaired electron between the two oxygen nuclei.This suggests that a different type of bonding may exist on some oxides or that the radicdl ion may be adsorbed at an unusual site at the surface. This type of study is relevant not only to catalysis where heterogeneous oxidation reactions ate used extensively for the production of chemicals from crude oil but also in biochemistry where modefs of oxygen adsorptlon on oxides lo have been success- fuHy applied to the case of enzymically generated superoxide ions l9 providing final confirmattion that the one-electron reduetion of oxygen can occtrr in biological systems.EXPERIMENTAL The supported Mo containing samples were prepared by impregnation of appropriate supports by aqueous (NH4),Mo04 sloutions. The catalysts were dried at 400 K and calcined in an oxygen atmosphere at 800 K for twenty-four hours ; the following solids were obtained (% by weight) : 6.3 % Mo/y-A1203 (referred to as Mo/A1203) and 2.3 % Mo/SiO, (referred to as Mo/Si02). The oxide MOO, was prepared by decomposition at 800 K of ammonium molybdate. The specific surface areas were measured using the B.E.T. method af nitrogen ddsorption at 77 K after heat tteatment at 850 k under a vacuum of 1W6 Tort and were found to be : 170 mz g-I, 450 m2 g-' and 1 mz g-' respectively.The physical prapenies of these solids have been repotted in a previous paMr.' Before oxygen adsurptiofi, the samples were reduced in silica rubes at 800 K either under hydrogen (Mo/AI20 j ) or undet a v a c m bf 1k6Torr (Mo/SiO,). The gas pressute was measured using a thermistor gauge and adsorpfions were carried out in EE conventional volumetric apparatus. Oxygeh was adsorbed at pressure of to lo-' Torr at 77 K using 1 6 0 2 and isotopically enriched O2 (58 atom % 170) on various samples. The e.s.r. spectra were obtained using a Varian V4502 instrument at 9.3 GHz (X band) or at 35 GHz (Q band) with 100 kHz field modulation. Measurements were carried out at 77 and 300K and the magnetic field splittings were calibrated using a proton resonance probe together with a Marconi TFl417A/TF2400 electronic counter to measure the ptaton resonance frequency.Measurements of g factors were made by reference to DPPH with a g factbt of 2.0036. The values of the g-tetisor components were measured at the ttirhing points as described in ref. (20), and theabsoluteaccuracyisbelieved to bewithin + 0.0010. The values are quoted to four decimal places for comparison with the powder spectra of (0;') in earlier papers. The spin concentratians were measured by comparison witli a Cr3+ standard in MgO using the LABCOM system.21 The e.s.r. spectra illustrated in the figures are first derivative curves and the magnetic field increases from left to right ; the modulation has been adjusted to resolve the maximum detail, unless otherwise stated.M.CHE, A . J . TENCH AND C . NACCACHE 265 RESULTS Thermal treatrneht under vacuum reduces the surface molybdenum ions and leads to the appearance of a broad asymmetric e.s.r. signal which has been assigned to pentavakrft molybdenum as described in earlier work.’ MO/SiO, SYSTEM The adsorption of 1602 at 77 K on the reduced samples leads to a debrease of the molybdenum e.s.r. signal and the appearanee of a new e.s.r. signal (fig. I). The line shape of the new signal i s characteristic of species located in orthorhombic crystalline 10 gauss FIG. 1 .-Oxygen-16 adsorbed on Mo/SiO, at 77 K. 160 170-, I--.’ 170160- ~ L - U I 7 0 170- 1 1 I I I I FIG. 2.-Oxygeh-17 adsorbed on Mo/SiOz at 77 K.266 OXYGEN ON SUPPORTED MOLYBDENUM field.The principal g values of the triplet are g1 = 2.0176, 9 2 = 2.0098 and g 3 = 2.0042. Similar experiments were repeated with isotopically enriched oxygen and the new e.s.r. spectrum exhibited a complex hyperfine structure (fig. 2). It has been suggested in earlier work that this g-tensor corresponds to the 0; radical ion. These ( 0 ~ ) ~ e.s.r. signals are broadened by admission of gaseous oxygen but can be restored if oxygen is pumped off. The (O;)s species is stable under oxygen at 77 K or 293 K. The profile of the hyperfine structure lines does not change upon variation of the microwave power or of the temperature and there is no indication of isotopic exchange with the oxide lattice ions either at 77 K or at 293 K.These results indicate that only one type of species is responsible for the observed spectra after adsorption of I6O2 or of the 1 7 0 enriched oxygen. FIG. 3.-Oxygen-16 adsorbed on Mo/AI,O, at 77 K. 50gauss FIG. 4.-Oxygen-17 adsorbed on Mo/A1203 at 77 K.M. CHE, A . J . TENCH AND C . NACCACHE 267 Mo/A120, SYSTEM Similar experiments repeated with the Mo/A1203 catalysts lead after adsorption at 77 K of 1 6 0 2 or isotopically enriched oxygen to the e.s.r. spectra of fig. 3 and 4 respectively. These 0; e.s.r. signals behave as those described in the previous section. The principal g values are : g1 = 2.0170, g2 = 2.0103 and g 3 = 2.0045. MOO, POWDER No new e.s.r. signal appears upon admission of oxygen at 77 K on reduced MOO,. QUANTITATIVE ANALYSIS Spin concentrations were obtained using an on-line laboratory computing system (LABCOM) 2 1 ; the spectra were recorded in digital form with a resolution of 0.1 % along the magnetic field axis.Double integration of the spectra could be rapidly carried out with the operator choosing the required field range and calculating the base line from the position of the experimental line before and after the derivative signal. A number of integrations could be performed in quick succession covering various field ranges on the same spectrum to check for convergence of the double integral. Before oxygen adsorption (spectrum A) the Mo5+ spin concentration (A) was obtained by direct double integration of the e.s.r. signal using the on-line computer and compared with a Cr3+ standard in MgO of known concentration.After oxygen adsorption (spectrum B) using the same procedure the sum ( T ) of the concentraion of ( 0 ~ ) ~ (C,) and the unreacted Mo5+ (C,) was measured. Subtraction of spectrum B, whose intensity could be modified by a scaling factorf, from spectrum A was carried out on the computer for several values of thef; the correct choice offleads to the disappearance of the Mo5+ signal, and the e.s.r. signal remaining corresponds to the concentration of (O;)s. Relations Cz =fA and C, = T-fA were used to obtain the concentrations. In all cases, the Mo5+ e.s.r. signal of the reduced samples decreases in intensity on adsorption of 0,. Typical quantitative analysis gave : 1.34 x 1017 spins of Mo5+ after reduction, and 0.98 x lo1' spins of Mo5+ after admission of oxygen at 77 K with the formation of 0.03 x lo1' spins of 0; radical ions. These figures apply only to Mo/A120, and Mo/SiO, since no 0; could be detected on the sample of MOO,.DISCUSSION THE g-TENSOR AND THE ADSORPTION SITE Several forms of adsorbed oxygen, such as (O-)s, ( 0 ~ ) ~ and (O.;),, have been reported and from the g and hyperfine values available in the literature 22-24 it is evi- dent that (O-)s and (O;)s are not responsible for the e.s.r. signals observed after oxygen adsorption on reduced Mo/Si02 and Mo/Al,03. This leaves (O;)s as the most probable species ; it has the correct form of the g-tensor and there are several possible adsorption sites such as 0-, A13+, Si4+ and Mo6+, on the supported molyb- denum catalysts. If (O;)s is stabilised on oxide lattice ions, the unpaired electron is probably no longer in the orbital usually defined as 71: (by comparison with the case of positively charged adsorption sites) but in the orbital perpendicular to it, i.e., 71:.The resulting spectrum should resemble the normal (O;)s adsorbed on a positively charged site and it is not possible from its powder spectrum to assign a direction to each of the g-tensor components. The uncertainty may be resolved from the hyperfine tensor268 OXYGEN ON SUPPORTED MOLYBDENUM data, but, as has been discussed previou~ly,'~ even if the largest hyperfine splitting occurs along the y direction (with the previous labelling), the stabilisation of (O;), on lattice oxide ions was unlikely to happen on chemical grounds. Also, it has been shown that the g-tensor component with the highest value (g,,, where z i s the direction of the internuclear axis) exhibits a departure (Ag,,) from the free electron g value, proportional to the charge of the ion on which (O;), is adsorbed.25 From this argu- ment, one would expect a gzz value somewhere around 2.077 to 2.051 for (O;)s on a lattice oxygen ion from comparison with values obtained for doubly positively charged adsorption sites (MgO and ZnO respectively).26 From electrostatic reasons, it seems more likely that the ( 0 ~ ) ~ radical ion adsorbs at a positively charged site.For A13+, g,, has been found to be 2.034 27 and for Si4'-, 2.024.28 Therefore the observed g,, value (2.017) seems to favour rather a Mo6+ adsorption site in agreement with earlier r e s ~ l t s . ~ The results we report here differ Somewhat from those reported by Kazansky et ~ 1 .~ who observed two types of Mo5+ in square pyramidal and tetrahedral arrangements respectively, only the latter leading to 0; ions on adsorption of oxygen. The experi- mental conditions we chose were such as to obtain only the pyramidal type of moIyb- denum ions, and our results show that these ions also lead to (O;), ions upon adsorp- tion of oxygen. If we assume that only one type of reaction occurs at a time : M O ~ + + ~ ~ - - + M O ~ + + ~ ; (1) Mo5++ 02+Mo6+ + 0; (2) then the quantitative results we report here show that there are no Mo4+ ions formed after thermal reduction. This is in agreement with the work reported earlier where the reduced molybdenum ions were produced by a milder thermal treatment and disappeared upon oxygen adsorption.If a more drastic reduction procedure is used, then the reduced molybdenum ions are formed deeper in the bulk and are no longer available to further oxygen adsorption, which explains the incomplete removal of the reduced molybdenum ion e.s.r. signal. The quantitative results show that reaction (2) cannot account for the figures obtained. It is probable that species such as (0;-), are responsible for the discrepancy between the number of 0; formed and the number of Mo5+ being reoxidised to Mo6+ upon adsorption of oxygen. Finally, we believe that (OF), is not observed on reduced MOO, bulk oxide because of the low specific surface area. This is also found for some low specific surface area reduced oxides where ( 0 ~ ) ~ ions are known to be produced on adsorption of oxygen on the supported oxides.THE HYPERFINE TENSOR TWO EQUIVALENT OXYGEN NUCLEI In the systems reported which have two equivalent oxygen nuclei the e.s.r. spectrum eonsists of two sets with 6 and 11 hyperfine lines and each set is centrcd on a single g component (in general the lowest, g3). The presence of several sites with slightly different 9, factor but identical hyperfine constants can be distinguished as all the lines reflect the distribution of the g factor. Each situation where not only 9, but also the associated hyperfine constants vary from site to site needs to be treated as a separate case and more experimental data are required for an unambiguous analysis.For "0; on Mo/Si02 and Mo/Al,O, there was no change of the profile of the hyperfine structure on variation of the microwave power or the temperature indicating only onk predominate species and one site in agreement with earlier recult$ l 8 o n Mo/SiO,.IM. CHE, A . 3 . ’TLNCH A N D C . NACCACHE 269 TWO INEQUIVALENT OXYGEN NUCLEI Some possible configuratioqs for inequivalent oxygen nuclei in the 0; ion are summarized in fig. 5. The presence of a configuration similar to case 1 can be ruled out since the corresponding spectrum for two sites would appear as a series of doublets with a splitting (A1-B2)rn, providing A , - B2 .g A , and B2 ; where A and B ale the two splitting constants of the two nuclei, rn, the magnetic quantum number varying from + 5/2 to - 5/2 and 1 and 2 refer to different surface sites. This analysi$ leads to the conclusion that for two sites the separation of the doublet will increase away from the ceptre of the spectrum.For two inequivalent nuclei in the same ( 0 ~ ) ~ ion (case 2, fig. 5) the spectruni should appear to be sets of 1, 2, 3, 4, 5 and 6 lines for m, = 5, 4, 3, 2, 1 and 0 res- pectively with a sepqration of ( A , - B , ) centred on m,(A1 +B,)/2 ; where A , and B1 9 ( A - B,). In general these lines will not be resolved and the overall result will be that the integer lines should appear to be increasing in width towards the centre of the spc’ctrum, i.e. tire exact reverse of that predicted when two sites are irivolved. By analysis of the e.s.r. spectra in fig. 2 and 4 using this method on the half integer lines we obtain A = 72 G and B = 85 G for Mo/SiO, and A = 77 G and B = 80 G for Mo/A1,0,.The absence of doubling and the increase in the line width of the integer lines towards the centre of the spectrum correspond to case 2 of fig. 5 and confirm that the (O;)s ion is present in only one form on these oxides. Although the 0 n777 @f Sites 1,2 + 6) 0” /7/ /’ ’//’ Sites 1.2 0; with one labelled nucleus Site I Site 2 Cam 1 : cquivalcnt nuclei, different sites I ’ / I I I ‘ \ I I I I I 0; with both nuclei I Case 2: equivalent sites, different nuclei labelled FIG. 5.-Configuration for inequivalent oxygen nuclei in the (Oils ion. pattern of the line intensity is different for the half iuteger lines in Mo/A1,03 and Mo/SiO,, this does not seem to be explained by a dfference in population of sites because lines corresponding to m, = 5 and to the case of two different sites were not detected in the far wings of the spectrum.The comparison of Lunsford’s results and those given in this paper and earlier ‘’ concern only the Mo/Si02 system. The slight discrepancy between Lunsford’s values 1 G is only just outside the estimated experimental accuracy and the differewe in the splitting between the oxygen nuclei is 13 G in both cases. of 69 and 82f 1 G and our values of 72 and 85270 OXYGEN ON SUPPORTED MOLYBDENUM Since the 0; species is adsorbed at a Mo6+ site and the negative charge probably moves closer to the surface leaving the unpaired electron concentrated more on the oxygen further from the adsorption site on the surface, the larger hyperfine splitting is therefore likely to be associated with the oxygen further from the Mo6+ ion.The fact that the two oxygen nuclei are inequivalent within the same (O;)s radical ion suggests that there is some covalency involved, a situation comparable with that of (CHJ3COO* radical obtained in solution 29 where values of Aiso = 21.8 and 16.4 G for the isotropic interaction were assigned to the terminal and inner oxygen respect- ively. We have used these two values (both expected to be negative 14) together with the values obtained for A , in the case of Mo/SiO, to construct the two following axial tensors (-85, +9.8, +9.8) and (-72, + 11.4, + 11.4) with Aiso = (AII +2A,)/3. The two dipolar traceless tensors of the form (BII, B,, B,) are then found to be (-63.2, 31.6, 31.6) and (-55.6,27.8,27.8) for nuclei B and A (fig.5) respectively with Bit = A - Aiso and B, = A , - Aiso (case A). It is also possible to use another set of hyperfine tensors assuming A,, and A,, to be zero (case B). In fact, the spectra obtained in Q-band lead to the same g and A tensors for Mo/SiO,. It is noteworthy that the splittings between consecutive hyperfine lines for each sextet due to 1 7 0 1 6 0 - and 1 6 0 1 7 0 - are equal within the experimental error and that the centres of the hyper- fine patterns for each value of m, are the same either in X or in Q band spectra. This means 30 that the second order corrections are negligible, i.e., A,, and A,, are small compared with Axx. The spin densities obtained from these values are summarized in table 1 using two sets of values for A .and Bo that are currently available. TABLE SP SPIN DENSITIES FOR ON SUPPORTED Mo total spin density (Or), outer oxygen inner oxygen spin case A case B case A case B case A case B densities (4 (6) (4 (6) (4 (6) (4 (6) (4 (6) (4 (6) Mo/SiOz 2s 0.013 0.012 0.017 0.015 0,010 0.009 0.014 0.013 2PZ 0.61 0.53 0.55 0.47 0.54 0.46 0.47 0.40 total 0.62 0.54 0.57 0.49 0.55 0.47 0.48 0.41 1.17 1.01 1.05 0.90 M0lA1203 2s 0.013 0.012 0.106 0.014 0.010 0.009 0.016 0.014 2PZ 0.565 0.486 0.516 0.444 0.59 0.51 0.50 0.43 total 0.58 0.50 0.53 0.46 0.60 0.52 0.52 0.44 1.18 1.02 1.05 0.90 (a) using A0 = 1655 and SO = 51.5 G 3 1 ; (6) using A0 = 1890 and Bo = 59.9 G.32 The total spin density obtained by summing the population of the 2s and 2p orbitals for both oxygen nuclei should be exactly unity. In practice, considerable variation is observed and values of about 1.17 and 1.01 are obtained for both systems from the data in table 1 using values (a) and (b) respectively.The value of 1.17 is rather large although the assumption of axial symmetry of the hyperfine tensor is probably not correct when compared with the orthorhombic g-tensor and the value of 1.01 fits very well. For comparison, values of 1.05 and 0.90 are obtained if it is assumed that A,, = A,, = 0, although this is thought to be unlikely. In some instances,l8 it has been the practice where the derived total spin density is less than unity, to attribute the difference to a measure of the extent of delocalisation on to the orbitals of the surface metal ion.This delocalisation would be expected to be larger in the case where covalent rather than ionic bonding is involved and would be expected to be significant for 0; on molybdenum where the oxygen nuclei are not equivalent. In fact, of the values calculated, only two are < 1 .O and this emphasises that argumentsM . CHE, A . J . TENCH AND C . NACCACHE 27 1 based on the summation of orbital populations can only be used as a guide at present and cannot be used to give an indication of the type of bonding with the surface in the absence of better theoretical models. Improvement is required both in the theor- etical values for A . and Bo and a better understanding of the use of single electron rather than the full many-electron analysis in the description of the orbitals containing the unpaired electron.Also, the problem of spin polarisation effects which may induce negative spin densities has been ignored. It is striking that the g-tensor does not vary significantly with the support (SO2, y-Al,O,) in contrast with the A-tensor. This shows that the g-tensor may be used to some degree for identifying the nature of the paramagnetic species but does not give much information on the structure, and it is here that the A-tensor proves to be a much more powerful tool. The difference in the hyperfine constants (13 G for Mo/ SO2, 3 G for Mo/A1203) cannot arise just from the molybdenum ions since it depends upon the support; nor does it reflect the properties of the support alone, since for example on the CeOJSiO, l4 system the 0, radical ion exhibits nearly equivalent oxygens (difference : <4 G).We conclude that this important inequivalency of the oxygen nuclei is due to a particular association of molybdenum and silica. The inequivalency of the nuclei may reflect the covalency of the ion sitting at an angle to an essentially flat surface or the surface geometry of a particular association of molyb- denum and silica. In this latter case an ionic ( 0 ~ ) ~ could be adsorbed parallel on a zigzag surface, It is important to understand how the ( 0 ~ ) ~ is adsorbed on the surface. 0)- / 0 (0 0 / \ ' / \ Si Mo Si and due to steric hindrance one oxygen is closer to the adsorption site of Mo6+. This situation has not been discussed previously but must be considered as a serious possibility.Clearly, the e.s.r. of the reduced molybdenum system needs to be reinvestigated in view of the present results, as it probably might give the reason for the non-equivalency of the oxygen nuclei. It is also interesting to note that Mo/SiO, catalysts present a better selectivity than Mo/Al,03 ones.33 C . Naccache acknowledges the facilities made available at A.E.R.E. Harwell for some of this work. M. Che thanks the C.N.R.S. and the Royal Society, London, for financial support under the European Science Exchange Programme, and acknow- ledges the award of a Van't Hoff Foundation Fellowship during 1972-73. M. Dufaux, M. Che and C. Naccache, J . Cliern. Phys., 1970, 67, 527. V. M. Vorotyntsev, V. A. Shvets and V.B. Kazansky, Kinetika i Katalyz, 1971, 12, 1249. 0. V. Kryslov, G. B. Pariiskii and K. N. Spiridonov, J. Catalysis, 1971, 23, 301. V. A. Shvets and V. B. Kazansky, J. Catalysis, 1972, 25, 123. M. Cathala and J. E. Germain, Bull. Soc. Chem. France, 1970, 4114. M. Dufaux, M. Che and C. Naccache, Conipt. Rend., l969,268C, 2255. C . Naccache and M. Che, Preprint No. 104, Vth Int. Congress Catalysis, Palm Beach, U.S.A., 1972. A. J . Tench, T. Lawson and J. F. J. Kibblewhite, J.C.S. Faraday I, 1972, 68, 1169. ' K. N. Spiridonov, G. B. Pariiskii and 0. V. Kryslov, Zzv. Akad. Nuuk S.S.S.R., 1971,11, 2161. l o A. J. Tench and P. Holroyd, Chem. Comm., 1968,471. '' A. J. Tench and T. Lawson, Chem. Phys. Letters, 1971, 8, 177. l 2 C. Naccache, P. Meriaudeau, M. Che and A.J. Tench, Trans. Faraduy Suc., 1971, 67, 506.272 OXYGEN ON SUPPORTED MOLYBDENUM l 3 P. Meriaudeau, C. Naccache and A. J. Tench, J. Catalysis, 1971, 21, 208. l4 M. Che, J. F. J. Kibblewhite, A. J. Tench, M. Dufaux and C. Naccache, J.C.S. Faraduy I, 1973,69,857. V. M. Vorotyatsev, V. A. Shvets, G. M. Zhidomirov and V B Kazansky, Kinetika i Katalyz, 1971, 12,1020, l6 M. Che, C. Naccache and A. J. Tench, unpubliskd results. ’’ M. Che and A. J. Tench, Chem. Phys. Letters, 1973, 18, 199. l 9 R. C. Bray, F. M. Picks and D. Samuel, Eur. 3. Biochem., 1970, IS, 352. 2o F. K. Kneubuhi, J. Chem. Phys., 1960,33, 1074. 22 N. B. Wong and J. H. Lunsford, J. Chem. Phys., 1971,55, 3007. 23 N. B. Wong and J. H. Lunsford, J. dhem. Phys., 1972, 56, 2664. 24 A. J. Tench, J.C.S.Faraa2z.v I, 1972, 68, 1181. ’’ P. H. Kasaj, J. Chem. Phys., 1965,43, 3322. ” A. A. Gezalav, G. M. Zhabrova, V. V. Nikisha, G. B. Pariiskii and Z. N. Spirjdonov. Kineriku 28 3. Vedrine, G. Dafmai and B. Imelik, Colloque Ampere XV, North Holland, Amsterdam, 1969, 2 9 K. Adamjc, K. U. Ingold and J. R. Morton, J. Anier. Chem. SOC., 1970, 92, 922. ’O J. C. Vqdrine and C. Naccache, J. Phys. Chem., in the press. 31 J. R. Morton, J. R. Rowlands and D. H. Whiffen, Nat. Phys. Lab. Circular No. BPR 13. 32 C. M. Hurd and P. Coodin, J. Phys. Chem. Solids, 1967, 28, 523. 33 M. Tosa-Brussin, Thesis (Lyon, 1971). Y. Ben Taarit and J. R. Lunsford, J. Phys. Cltern., 1973,77,780. M. Deane, C. Kenwood and A. J. Tench, 1972, A.E.R.E. R.7020. J. H. Lunsford and J. P. Jayne, J.Chem. Phys., 1966, 44, 1487. i Katalyz, 1968, 9, 462. 304. Electron Spin Resonance Studies of Isotopically Labelled Oxygen Species Adsorbed on Supported Molybdenum B Y MICHEL CHE,? ANTHONY J. TENCH* Chemistry Division, A .E.R.E., Harwell CLAUDE NACCACHE AND Institut de Recherches sur la Catalyse, C.N.R.S. 69 10-Villeurbanne, France Received 30th April, 1973 The molecular ion (O,)s has been observed on the surface of the molybdenum/SiOz and molyb- denum/Al2O3 systems by e.s.r. Using oxygen isotopically enriched in "0, two different hyperfine splittings were observed in the Mo/Si system corresponding to A, = 85 and 72 G respectively ; the A, and AZz values of the tensor were too small to be resolved. The observed values for Axx are shown to arise from an unequal interaction of the unpaired electron with two oxygen nuclei in the same molecular ion.This result is contrary to the previous work for (O& on other oxides where the oxygen nuclei have been found to be equivalent ; the difference is attributed either to a more covalent bonding with the surface ions or to adsorption at an unusual site at the surface. Thisis thought to be a sensitive function of the surface environment since the oxygen nuclei appear to be nearly equivalent for the Mo/AI system although the g-tensors for both systems are almost identical. A number of papers have recently appeared on oxidation reactions using supported catalysts and some attention has been paid to molybdenum impregnated on a variety of ~upports.~'~ It has been shown that the valence state of surface molybdenum ions is important in determining how the hydrocarbon adsorbs on the surface and therefore the catalyst selectivity.6 The other important aspect is the state of the adsorbed oxygen and this was first studied in some detail by Dufaux et aL7 who concluded that 0; was formed on the surface. More recently Kazansky et aL2 extended this work and showed that (O-)s could also be stabilized on the surface after N20 adsorption at 100°C.The (O-)s species was shown to be more reactive than (O;)s in agreement with the results on Mg0.8* However, the (O-)s species has not been detected after the adsorption of 0, and in this work the adsorption of O2 isotopically enriched with 170 was studied to confirm that the ( 0 ~ ) ~ species was formed on such active oxidation catalysts and to see whether the structure of the adsorbed species was similar to that found on other oxides.The slight departure of the lowest component of the g-tensor from the free elec- tron g value has been recently the subject of some attention.14 Kazansky et all5 showed that, on V/Si02, the hyperfine tensor of (l6O$), could be explained by some covalency of the (OT)s bond with the transition metal ion through a partial transfer of both the unpaired electron and the pair of electrons on the oxygen to the vacant orbitals of the ion. The associated lowest component of the g-tensor was 2.004. t permanent address : Institut de Recherches sur la Catalyse. 263264 OXYGEN ON SUPPORTED MOLYBDENUM The model proposed was a molecular oxygen radical ion adsorbed with the inter- nuclear axis parallel to the surface with the electron equally delocalised on to the two oxygen nuclei.One would expect in such a case a (170;), spectrum similar to that obtained with Mgd.'O Unfortunately, it has not been possible to check ?.he validity of such a model using oxygen enriched with 1 7 0 owing to the complication arising from the nuclear spin of 51V(Z = 7/2) and the rather poor resolution of the e.s.r. spectrum obtained.16 More recently, it has been shown l 4 that, on CeO,/SiO,, the lowest component of the g-tensor was 2.0109, which was tentatively explained by the influence of thef-orbital electrons of cerium. In this case, the (170;), spectrum showed that the electron was equally delocalized on the two oxygen nuclei.It was also interesting to investigate the case of supported molybdenum where the lowest component of the g-tensor was -2.0048 and covalent bonding might be expected. This work has extended the previous studies of isotopically labelled adsorbed oxygen to the supported molybdenum system and confirms earlier work 17* that the (170;), radical ion has an unequal didtfibution of the unpaired electron between the two oxygen nuclei. This suggests that a different type of bonding may exist on some oxides or that the radicdl ion may be adsorbed at an unusual site at the surface. This type of study is relevant not only to catalysis where heterogeneous oxidation reactions ate used extensively for the production of chemicals from crude oil but also in biochemistry where modefs of oxygen adsorptlon on oxides lo have been success- fuHy applied to the case of enzymically generated superoxide ions l9 providing final confirmattion that the one-electron reduetion of oxygen can occtrr in biological systems.EXPERIMENTAL The supported Mo containing samples were prepared by impregnation of appropriate supports by aqueous (NH4),Mo04 sloutions. The catalysts were dried at 400 K and calcined in an oxygen atmosphere at 800 K for twenty-four hours ; the following solids were obtained (% by weight) : 6.3 % Mo/y-A1203 (referred to as Mo/A1203) and 2.3 % Mo/SiO, (referred to as Mo/Si02). The oxide MOO, was prepared by decomposition at 800 K of ammonium molybdate. The specific surface areas were measured using the B.E.T. method af nitrogen ddsorption at 77 K after heat tteatment at 850 k under a vacuum of 1W6 Tort and were found to be : 170 mz g-I, 450 m2 g-' and 1 mz g-' respectively.The physical prapenies of these solids have been repotted in a previous paMr.' Before oxygen adsurptiofi, the samples were reduced in silica rubes at 800 K either under hydrogen (Mo/AI20 j ) or undet a v a c m bf 1k6Torr (Mo/SiO,). The gas pressute was measured using a thermistor gauge and adsorpfions were carried out in EE conventional volumetric apparatus. Oxygeh was adsorbed at pressure of to lo-' Torr at 77 K using 1 6 0 2 and isotopically enriched O2 (58 atom % 170) on various samples. The e.s.r. spectra were obtained using a Varian V4502 instrument at 9.3 GHz (X band) or at 35 GHz (Q band) with 100 kHz field modulation.Measurements were carried out at 77 and 300K and the magnetic field splittings were calibrated using a proton resonance probe together with a Marconi TFl417A/TF2400 electronic counter to measure the ptaton resonance frequency. Measurements of g factors were made by reference to DPPH with a g factbt of 2.0036. The values of the g-tetisor components were measured at the ttirhing points as described in ref. (20), and theabsoluteaccuracyisbelieved to bewithin + 0.0010. The values are quoted to four decimal places for comparison with the powder spectra of (0;') in earlier papers. The spin concentratians were measured by comparison witli a Cr3+ standard in MgO using the LABCOM system.21 The e.s.r. spectra illustrated in the figures are first derivative curves and the magnetic field increases from left to right ; the modulation has been adjusted to resolve the maximum detail, unless otherwise stated.M.CHE, A . J . TENCH AND C . NACCACHE 265 RESULTS Thermal treatrneht under vacuum reduces the surface molybdenum ions and leads to the appearance of a broad asymmetric e.s.r. signal which has been assigned to pentavakrft molybdenum as described in earlier work.’ MO/SiO, SYSTEM The adsorption of 1602 at 77 K on the reduced samples leads to a debrease of the molybdenum e.s.r. signal and the appearanee of a new e.s.r. signal (fig. I). The line shape of the new signal i s characteristic of species located in orthorhombic crystalline 10 gauss FIG. 1 .-Oxygen-16 adsorbed on Mo/SiO, at 77 K. 160 170-, I--.’ 170160- ~ L - U I 7 0 170- 1 1 I I I I FIG.2.-Oxygeh-17 adsorbed on Mo/SiOz at 77 K.266 OXYGEN ON SUPPORTED MOLYBDENUM field. The principal g values of the triplet are g1 = 2.0176, 9 2 = 2.0098 and g 3 = 2.0042. Similar experiments were repeated with isotopically enriched oxygen and the new e.s.r. spectrum exhibited a complex hyperfine structure (fig. 2). It has been suggested in earlier work that this g-tensor corresponds to the 0; radical ion. These ( 0 ~ ) ~ e.s.r. signals are broadened by admission of gaseous oxygen but can be restored if oxygen is pumped off. The (O;)s species is stable under oxygen at 77 K or 293 K. The profile of the hyperfine structure lines does not change upon variation of the microwave power or of the temperature and there is no indication of isotopic exchange with the oxide lattice ions either at 77 K or at 293 K. These results indicate that only one type of species is responsible for the observed spectra after adsorption of I6O2 or of the 1 7 0 enriched oxygen.FIG. 3.-Oxygen-16 adsorbed on Mo/AI,O, at 77 K. 50gauss FIG. 4.-Oxygen-17 adsorbed on Mo/A1203 at 77 K.M. CHE, A . J . TENCH AND C . NACCACHE 267 Mo/A120, SYSTEM Similar experiments repeated with the Mo/A1203 catalysts lead after adsorption at 77 K of 1 6 0 2 or isotopically enriched oxygen to the e.s.r. spectra of fig. 3 and 4 respectively. These 0; e.s.r. signals behave as those described in the previous section. The principal g values are : g1 = 2.0170, g2 = 2.0103 and g 3 = 2.0045. MOO, POWDER No new e.s.r. signal appears upon admission of oxygen at 77 K on reduced MOO,.QUANTITATIVE ANALYSIS Spin concentrations were obtained using an on-line laboratory computing system (LABCOM) 2 1 ; the spectra were recorded in digital form with a resolution of 0.1 % along the magnetic field axis. Double integration of the spectra could be rapidly carried out with the operator choosing the required field range and calculating the base line from the position of the experimental line before and after the derivative signal. A number of integrations could be performed in quick succession covering various field ranges on the same spectrum to check for convergence of the double integral. Before oxygen adsorption (spectrum A) the Mo5+ spin concentration (A) was obtained by direct double integration of the e.s.r.signal using the on-line computer and compared with a Cr3+ standard in MgO of known concentration. After oxygen adsorption (spectrum B) using the same procedure the sum ( T ) of the concentraion of ( 0 ~ ) ~ (C,) and the unreacted Mo5+ (C,) was measured. Subtraction of spectrum B, whose intensity could be modified by a scaling factorf, from spectrum A was carried out on the computer for several values of thef; the correct choice offleads to the disappearance of the Mo5+ signal, and the e.s.r. signal remaining corresponds to the concentration of (O;)s. Relations Cz =fA and C, = T-fA were used to obtain the concentrations. In all cases, the Mo5+ e.s.r. signal of the reduced samples decreases in intensity on adsorption of 0,. Typical quantitative analysis gave : 1.34 x 1017 spins of Mo5+ after reduction, and 0.98 x lo1' spins of Mo5+ after admission of oxygen at 77 K with the formation of 0.03 x lo1' spins of 0; radical ions.These figures apply only to Mo/A120, and Mo/SiO, since no 0; could be detected on the sample of MOO,. DISCUSSION THE g-TENSOR AND THE ADSORPTION SITE Several forms of adsorbed oxygen, such as (O-)s, ( 0 ~ ) ~ and (O.;),, have been reported and from the g and hyperfine values available in the literature 22-24 it is evi- dent that (O-)s and (O;)s are not responsible for the e.s.r. signals observed after oxygen adsorption on reduced Mo/Si02 and Mo/Al,03. This leaves (O;)s as the most probable species ; it has the correct form of the g-tensor and there are several possible adsorption sites such as 0-, A13+, Si4+ and Mo6+, on the supported molyb- denum catalysts.If (O;)s is stabilised on oxide lattice ions, the unpaired electron is probably no longer in the orbital usually defined as 71: (by comparison with the case of positively charged adsorption sites) but in the orbital perpendicular to it, i.e., 71:. The resulting spectrum should resemble the normal (O;)s adsorbed on a positively charged site and it is not possible from its powder spectrum to assign a direction to each of the g-tensor components. The uncertainty may be resolved from the hyperfine tensor268 OXYGEN ON SUPPORTED MOLYBDENUM data, but, as has been discussed previou~ly,'~ even if the largest hyperfine splitting occurs along the y direction (with the previous labelling), the stabilisation of (O;), on lattice oxide ions was unlikely to happen on chemical grounds.Also, it has been shown that the g-tensor component with the highest value (g,,, where z i s the direction of the internuclear axis) exhibits a departure (Ag,,) from the free electron g value, proportional to the charge of the ion on which (O;), is adsorbed.25 From this argu- ment, one would expect a gzz value somewhere around 2.077 to 2.051 for (O;)s on a lattice oxygen ion from comparison with values obtained for doubly positively charged adsorption sites (MgO and ZnO respectively).26 From electrostatic reasons, it seems more likely that the ( 0 ~ ) ~ radical ion adsorbs at a positively charged site. For A13+, g,, has been found to be 2.034 27 and for Si4'-, 2.024.28 Therefore the observed g,, value (2.017) seems to favour rather a Mo6+ adsorption site in agreement with earlier r e s ~ l t s .~ The results we report here differ Somewhat from those reported by Kazansky et ~ 1 . ~ who observed two types of Mo5+ in square pyramidal and tetrahedral arrangements respectively, only the latter leading to 0; ions on adsorption of oxygen. The experi- mental conditions we chose were such as to obtain only the pyramidal type of moIyb- denum ions, and our results show that these ions also lead to (O;), ions upon adsorp- tion of oxygen. If we assume that only one type of reaction occurs at a time : M O ~ + + ~ ~ - - + M O ~ + + ~ ; (1) Mo5++ 02+Mo6+ + 0; (2) then the quantitative results we report here show that there are no Mo4+ ions formed after thermal reduction.This is in agreement with the work reported earlier where the reduced molybdenum ions were produced by a milder thermal treatment and disappeared upon oxygen adsorption. If a more drastic reduction procedure is used, then the reduced molybdenum ions are formed deeper in the bulk and are no longer available to further oxygen adsorption, which explains the incomplete removal of the reduced molybdenum ion e.s.r. signal. The quantitative results show that reaction (2) cannot account for the figures obtained. It is probable that species such as (0;-), are responsible for the discrepancy between the number of 0; formed and the number of Mo5+ being reoxidised to Mo6+ upon adsorption of oxygen. Finally, we believe that (OF), is not observed on reduced MOO, bulk oxide because of the low specific surface area.This is also found for some low specific surface area reduced oxides where ( 0 ~ ) ~ ions are known to be produced on adsorption of oxygen on the supported oxides. THE HYPERFINE TENSOR TWO EQUIVALENT OXYGEN NUCLEI In the systems reported which have two equivalent oxygen nuclei the e.s.r. spectrum eonsists of two sets with 6 and 11 hyperfine lines and each set is centrcd on a single g component (in general the lowest, g3). The presence of several sites with slightly different 9, factor but identical hyperfine constants can be distinguished as all the lines reflect the distribution of the g factor. Each situation where not only 9, but also the associated hyperfine constants vary from site to site needs to be treated as a separate case and more experimental data are required for an unambiguous analysis.For "0; on Mo/Si02 and Mo/Al,O, there was no change of the profile of the hyperfine structure on variation of the microwave power or the temperature indicating only onk predominate species and one site in agreement with earlier recult$ l 8 o n Mo/SiO,.IM. CHE, A . 3 . ’TLNCH A N D C . NACCACHE 269 TWO INEQUIVALENT OXYGEN NUCLEI Some possible configuratioqs for inequivalent oxygen nuclei in the 0; ion are summarized in fig. 5. The presence of a configuration similar to case 1 can be ruled out since the corresponding spectrum for two sites would appear as a series of doublets with a splitting (A1-B2)rn, providing A , - B2 .g A , and B2 ; where A and B ale the two splitting constants of the two nuclei, rn, the magnetic quantum number varying from + 5/2 to - 5/2 and 1 and 2 refer to different surface sites.This analysi$ leads to the conclusion that for two sites the separation of the doublet will increase away from the ceptre of the spectrum. For two inequivalent nuclei in the same ( 0 ~ ) ~ ion (case 2, fig. 5) the spectruni should appear to be sets of 1, 2, 3, 4, 5 and 6 lines for m, = 5, 4, 3, 2, 1 and 0 res- pectively with a sepqration of ( A , - B , ) centred on m,(A1 +B,)/2 ; where A , and B1 9 ( A - B,). In general these lines will not be resolved and the overall result will be that the integer lines should appear to be increasing in width towards the centre of the spc’ctrum, i.e. tire exact reverse of that predicted when two sites are irivolved.By analysis of the e.s.r. spectra in fig. 2 and 4 using this method on the half integer lines we obtain A = 72 G and B = 85 G for Mo/SiO, and A = 77 G and B = 80 G for Mo/A1,0,. The absence of doubling and the increase in the line width of the integer lines towards the centre of the spectrum correspond to case 2 of fig. 5 and confirm that the (O;)s ion is present in only one form on these oxides. Although the 0 n777 @f Sites 1,2 + 6) 0” /7/ /’ ’//’ Sites 1.2 0; with one labelled nucleus Site I Site 2 Cam 1 : cquivalcnt nuclei, different sites I ’ / I I I ‘ \ I I I I I 0; with both nuclei I Case 2: equivalent sites, different nuclei labelled FIG. 5.-Configuration for inequivalent oxygen nuclei in the (Oils ion. pattern of the line intensity is different for the half iuteger lines in Mo/A1,03 and Mo/SiO,, this does not seem to be explained by a dfference in population of sites because lines corresponding to m, = 5 and to the case of two different sites were not detected in the far wings of the spectrum.The comparison of Lunsford’s results and those given in this paper and earlier ‘’ concern only the Mo/Si02 system. The slight discrepancy between Lunsford’s values 1 G is only just outside the estimated experimental accuracy and the differewe in the splitting between the oxygen nuclei is 13 G in both cases. of 69 and 82f 1 G and our values of 72 and 85270 OXYGEN ON SUPPORTED MOLYBDENUM Since the 0; species is adsorbed at a Mo6+ site and the negative charge probably moves closer to the surface leaving the unpaired electron concentrated more on the oxygen further from the adsorption site on the surface, the larger hyperfine splitting is therefore likely to be associated with the oxygen further from the Mo6+ ion.The fact that the two oxygen nuclei are inequivalent within the same (O;)s radical ion suggests that there is some covalency involved, a situation comparable with that of (CHJ3COO* radical obtained in solution 29 where values of Aiso = 21.8 and 16.4 G for the isotropic interaction were assigned to the terminal and inner oxygen respect- ively. We have used these two values (both expected to be negative 14) together with the values obtained for A , in the case of Mo/SiO, to construct the two following axial tensors (-85, +9.8, +9.8) and (-72, + 11.4, + 11.4) with Aiso = (AII +2A,)/3.The two dipolar traceless tensors of the form (BII, B,, B,) are then found to be (-63.2, 31.6, 31.6) and (-55.6,27.8,27.8) for nuclei B and A (fig. 5) respectively with Bit = A - Aiso and B, = A , - Aiso (case A). It is also possible to use another set of hyperfine tensors assuming A,, and A,, to be zero (case B). In fact, the spectra obtained in Q-band lead to the same g and A tensors for Mo/SiO,. It is noteworthy that the splittings between consecutive hyperfine lines for each sextet due to 1 7 0 1 6 0 - and 1 6 0 1 7 0 - are equal within the experimental error and that the centres of the hyper- fine patterns for each value of m, are the same either in X or in Q band spectra. This means 30 that the second order corrections are negligible, i.e., A,, and A,, are small compared with Axx.The spin densities obtained from these values are summarized in table 1 using two sets of values for A . and Bo that are currently available. TABLE SP SPIN DENSITIES FOR ON SUPPORTED Mo total spin density (Or), outer oxygen inner oxygen spin case A case B case A case B case A case B densities (4 (6) (4 (6) (4 (6) (4 (6) (4 (6) (4 (6) Mo/SiOz 2s 0.013 0.012 0.017 0.015 0,010 0.009 0.014 0.013 2PZ 0.61 0.53 0.55 0.47 0.54 0.46 0.47 0.40 total 0.62 0.54 0.57 0.49 0.55 0.47 0.48 0.41 1.17 1.01 1.05 0.90 M0lA1203 2s 0.013 0.012 0.106 0.014 0.010 0.009 0.016 0.014 2PZ 0.565 0.486 0.516 0.444 0.59 0.51 0.50 0.43 total 0.58 0.50 0.53 0.46 0.60 0.52 0.52 0.44 1.18 1.02 1.05 0.90 (a) using A0 = 1655 and SO = 51.5 G 3 1 ; (6) using A0 = 1890 and Bo = 59.9 G.32 The total spin density obtained by summing the population of the 2s and 2p orbitals for both oxygen nuclei should be exactly unity.In practice, considerable variation is observed and values of about 1.17 and 1.01 are obtained for both systems from the data in table 1 using values (a) and (b) respectively. The value of 1.17 is rather large although the assumption of axial symmetry of the hyperfine tensor is probably not correct when compared with the orthorhombic g-tensor and the value of 1.01 fits very well. For comparison, values of 1.05 and 0.90 are obtained if it is assumed that A,, = A,, = 0, although this is thought to be unlikely. In some instances,l8 it has been the practice where the derived total spin density is less than unity, to attribute the difference to a measure of the extent of delocalisation on to the orbitals of the surface metal ion.This delocalisation would be expected to be larger in the case where covalent rather than ionic bonding is involved and would be expected to be significant for 0; on molybdenum where the oxygen nuclei are not equivalent. In fact, of the values calculated, only two are < 1 .O and this emphasises that argumentsM . CHE, A . J . TENCH AND C . NACCACHE 27 1 based on the summation of orbital populations can only be used as a guide at present and cannot be used to give an indication of the type of bonding with the surface in the absence of better theoretical models. Improvement is required both in the theor- etical values for A .and Bo and a better understanding of the use of single electron rather than the full many-electron analysis in the description of the orbitals containing the unpaired electron. Also, the problem of spin polarisation effects which may induce negative spin densities has been ignored. It is striking that the g-tensor does not vary significantly with the support (SO2, y-Al,O,) in contrast with the A-tensor. This shows that the g-tensor may be used to some degree for identifying the nature of the paramagnetic species but does not give much information on the structure, and it is here that the A-tensor proves to be a much more powerful tool. The difference in the hyperfine constants (13 G for Mo/ SO2, 3 G for Mo/A1203) cannot arise just from the molybdenum ions since it depends upon the support; nor does it reflect the properties of the support alone, since for example on the CeOJSiO, l4 system the 0, radical ion exhibits nearly equivalent oxygens (difference : <4 G).We conclude that this important inequivalency of the oxygen nuclei is due to a particular association of molybdenum and silica. The inequivalency of the nuclei may reflect the covalency of the ion sitting at an angle to an essentially flat surface or the surface geometry of a particular association of molyb- denum and silica. In this latter case an ionic ( 0 ~ ) ~ could be adsorbed parallel on a zigzag surface, It is important to understand how the ( 0 ~ ) ~ is adsorbed on the surface. 0)- / 0 (0 0 / \ ' / \ Si Mo Si and due to steric hindrance one oxygen is closer to the adsorption site of Mo6+.This situation has not been discussed previously but must be considered as a serious possibility. Clearly, the e.s.r. of the reduced molybdenum system needs to be reinvestigated in view of the present results, as it probably might give the reason for the non-equivalency of the oxygen nuclei. It is also interesting to note that Mo/SiO, catalysts present a better selectivity than Mo/Al,03 ones.33 C . Naccache acknowledges the facilities made available at A.E.R.E. Harwell for some of this work. M. Che thanks the C.N.R.S. and the Royal Society, London, for financial support under the European Science Exchange Programme, and acknow- ledges the award of a Van't Hoff Foundation Fellowship during 1972-73. M. Dufaux, M. Che and C. Naccache, J . Cliern. Phys., 1970, 67, 527. V. M. Vorotyntsev, V. A. Shvets and V. B. Kazansky, Kinetika i Katalyz, 1971, 12, 1249. 0. V. Kryslov, G. B. Pariiskii and K. N. Spiridonov, J. Catalysis, 1971, 23, 301. V. A. Shvets and V. B. Kazansky, J. Catalysis, 1972, 25, 123. M. Cathala and J. E. Germain, Bull. Soc. Chem. France, 1970, 4114. M. Dufaux, M. Che and C. Naccache, Conipt. Rend., l969,268C, 2255. C . Naccache and M. Che, Preprint No. 104, Vth Int. Congress Catalysis, Palm Beach, U.S.A., 1972. A. J . Tench, T. Lawson and J. F. J. Kibblewhite, J.C.S. Faraday I, 1972, 68, 1169. ' K. N. Spiridonov, G. B. Pariiskii and 0. V. Kryslov, Zzv. Akad. Nuuk S.S.S.R., 1971,11, 2161. l o A. J. Tench and P. Holroyd, Chem. Comm., 1968,471. '' A. J. Tench and T. Lawson, Chem. Phys. Letters, 1971, 8, 177. l 2 C. Naccache, P. Meriaudeau, M. Che and A. J. Tench, Trans. Faraduy Suc., 1971, 67, 506.272 OXYGEN ON SUPPORTED MOLYBDENUM l 3 P. Meriaudeau, C. Naccache and A. J. Tench, J. Catalysis, 1971, 21, 208. l4 M. Che, J. F. J. Kibblewhite, A. J. Tench, M. Dufaux and C. Naccache, J.C.S. Faraduy I, 1973,69,857. V. M. Vorotyatsev, V. A. Shvets, G. M. Zhidomirov and V B Kazansky, Kinetika i Katalyz, 1971, 12,1020, l6 M. Che, C. Naccache and A. J. Tench, unpubliskd results. ’’ M. Che and A. J. Tench, Chem. Phys. Letters, 1973, 18, 199. l 9 R. C. Bray, F. M. Picks and D. Samuel, Eur. 3. Biochem., 1970, IS, 352. 2o F. K. Kneubuhi, J. Chem. Phys., 1960,33, 1074. 22 N. B. Wong and J. H. Lunsford, J. Chem. Phys., 1971,55, 3007. 23 N. B. Wong and J. H. Lunsford, J. dhem. Phys., 1972, 56, 2664. 24 A. J. Tench, J.C.S. Faraa2z.v I, 1972, 68, 1181. ’’ P. H. Kasaj, J. Chem. Phys., 1965,43, 3322. ” A. A. Gezalav, G. M. Zhabrova, V. V. Nikisha, G. B. Pariiskii and Z. N. Spirjdonov. Kineriku 28 3. Vedrine, G. Dafmai and B. Imelik, Colloque Ampere XV, North Holland, Amsterdam, 1969, 2 9 K. Adamjc, K. U. Ingold and J. R. Morton, J. Anier. Chem. SOC., 1970, 92, 922. ’O J. C. Vqdrine and C. Naccache, J. Phys. Chem., in the press. 31 J. R. Morton, J. R. Rowlands and D. H. Whiffen, Nat. Phys. Lab. Circular No. BPR 13. 32 C. M. Hurd and P. Coodin, J. Phys. Chem. Solids, 1967, 28, 523. 33 M. Tosa-Brussin, Thesis (Lyon, 1971). Y. Ben Taarit and J. R. Lunsford, J. Phys. Cltern., 1973,77,780. M. Deane, C. Kenwood and A. J. Tench, 1972, A.E.R.E. R.7020. J. H. Lunsford and J. P. Jayne, J. Chem. Phys., 1966, 44, 1487. i Katalyz, 1968, 9, 462. 304.

ISSN:0300-9599    

DOI:10.1039/F19747000263

出版商:RSC    

年代:1974

数据来源: RSC

摘要:

Electrical Properties of Molten T1,S BY Y. NAKAMUAA, K. hhTSUMIJRAt AND M. SHIM011 * Department of Chemistry, Faculty of Science, Hokkaido University, Sapporo, Japan Received 24th May, 1973 The electrical conductivity and thermoelectric power of TlzS have been measured as a function of temnerature. Non-stoichiometric and impurity doping effects have been investigated. The results for molten TlzS containing excess metal atoms are discussed with relation to Mott's theory for the strong scattering process. It has been reported that molten T1-Te,1-3 and TI-Se systems exhibit a sharp minimum in the electrical conductivity-composition curve at the stoicluometric composition T12Te or T1,Se and that the absolute thermoelectric power changes its sign around this composition. In this paper we report the experimental results on the electrical conductivity and thermoelectric power of T1,S in the solid and liquid states.The effects of non-stoichiometric doping, as well as those of impurity doping, have also been investigated and compared with the results for molten TI,Te and T1,Se. EXPERIMENTAL The apparatus and experimental procedures for measurements of the electrical conduct- ivity and thermoelectric power were similar to those described elsewhere.2 Since tungsten 300 400 5 00 600 temperaturerc FIG. 1.-Electrical conductivity of pure TlzS as a function of temperature. A, d.c. four-probe method ; 0, a.c. four-probe method ; 0, Velikanov and Tert~kh.~ t present address : Toshiba Electric Co., Ltd., Kawasaki, Japan. 273274 ELECTRICAL PROPERTIES OF MOLTEN TIZS electrodes corrode with samples containing excess sulphur at high temperatures, graphite rods in which a tungsten wire was embedded, were used as electrodes.The electrical conductivity was measured both by the d.c. and ax. four-probe methods. Results from these two methods coincided with each other within the range of experimental error of 1 % (fig. 1). The thermoelectric power measurements were made using a supplementary heater placed under one end of the cell. A correction was made for the absolute thermoelectric power of the tungsten electrode^.^ Electrolysis of the stoichiometric T12S was carried out in the conductivity cell by using a constant current electrolyzer and the variation of conduct- ivity during and after the electrolysis was recorded. The purity of TI metal was 99.95 %.The reagent grade sulphur was purified by repeated recrystallization from CS2 solutions. Samples were prepared in a sealed evacuated ampoule by melting carefully weighed amounts of the constituent elements. RESULTS The experimental results for the conductivity, 0, of pure T1,S are shown in fig. 1 as a function of temperature. The values of 0 increase with increasing temperature both in the solid and liquid states. The conductivity increases from 0.48 to 2.3 ohm-' cm-I at the melting point (448+ I T ) . The present values of CT are in good agreement with those determined by Velikanov and T e r t ~ k h . ~ The value reported by Joffe and Regel,6 1 x ohm-' cm-' just above the melting point, is too small 2 1 4 I rl I .c b M 0 E 1 - 0 I 1 I I 1 1 I 1 TL 2.0 "lo \ 0-5010- 0.3 '10 S O \ i d I i q u i d T 1 2.0 OIO 1.0 "lo I .I I . 3 1.5 I .7 103 K I T FIG. 2.-EIectricaI conductivity of doped T12S as a function of reciprocal temperature.Y . NAKAMURA, K . MATSUMURA AND M. SHIMOJI 800 400 4 0 \ % w -400 -800 I I s o l i d I iqui d T l 2.0'10 -A& 1.5'10 o O O / o 1.0 "0 2 O 0 . 5 " o u "u n u 0 0.3 " l o 0.2°10 T I 1 . 5 ' / o 2.0 "lo 0 0.5 '10 0 v> A 0 il +-v a A - D 0 0 -0 n 300 400 500 600 temperature/ "C FIG. 3.-Absolute thermoelectric power of doped TlzS as a function of temperature. I50 I00 rl I 4 I c b E --- 50 1.0 Y 0.5 0 275 excess S excess T1 (atom %) 4.-ElectricaI conductivity of molten TI,S at 600°C. 0, present Tertykh5; ---,yfromeqn(l) witha Z 3.0A; - - - - Y 9 work; U, Velikanov from eqn (3).and276 ELECTRICAL PROPERTIES OF MOLTEN T12S and appears to be erroneous. The value of the absolute thermoelectric power, S, for pure molten T1,S is negative and its absolute value decreases slightly with increasing temperature, as shown in fig. 3. The non-stoichiometric effect is investigated by adding the component elements excessively to molten stoichiometric T1,S; the data for Q and S are given in fig. 2 and 3, and their composition isotherms at 600°C are shown in fig. 4 and 5. The conductivity values increase almost linearly with the amount of excess TI, whereas the effect of excess sulphur is much less significant. The absolute thermoelectric power changes sign at the stoichiometric composition : positive for the S-rich region and negative for the TI-rich region.This result is similar to those of molten T1,Te and T12Se.1-3 -100 - 4 ;L! \ ?i h -300 -. -500 ,- I 2 2 I 0 1 excess S excess TI (atom ‘x) Fiti. 5.-Absolutc thcrmoclcctric power of molten TlzS at GOO’C. As seen in fig. 2, the (log 0, 1/T) plots for the samples containing excess T1 exhibit a knee at a temperature which depends on the concentration of excess T1. Similar conductivity curves obtained in the T1,Se +T1 system were related to the “ saturation effect ” in our previous paper,2 but the results can now well be explained from the phase diagram redetermined by Kanda et al.’ : these knees correspond exactly to the temperature of liquid-liquid phase separation. From the analogous point of view we can determine the phase diagram ,of the T1- S system in the vicinity of the composi- tion TlzS (or X,, = 0.6667) by using the present conductivity data.Thc results given in table 1 are somewhat diffcrent from Hansen’s diagram,* as in the TI-Se system. The impurity elements chosen here were Ag, Cd and In which belong to the Vth period. The increase in Q is approximately proportionzi to the concentration of excess electrons which can be given by the concentration of excess metal multiplied by its valence. The assigned valences are the same as those found in molten Tl,Se.2* The values of S of the impurity-doped T1,S were similar to thosc of molten T1,S containing excess T1, characterized by a large negative value and a slight temperature dependence. Fig. 6 shows the effect of impurity doping on Q of molten T1,S.Y.NAKAMUkA, K . MATSOMARA AND M. SHIMOJI 277 Since the value of 0 for pure molten T1,S is as low as 2.3 ohm-l ctn-' at the melting point, a contribution from ionic conduction was expected. In order to detect this effect the molten T1,S was electrolyzed and the change in CJ after the electrolysis was measured. The increase in conductivity, ACT, was found to be roughly propor- tional to the amount of electricity passed through the melt. When the electrolyzing current was turned off, the enhanced conductivity decreased with time, and a more rapid recovery of the resistivity was observed if the polarity of electrolysis was changed. I50 190 'F - u I c b - E \ 50 I / I 0 l z 100 zi.yi FIG. 6.--Electrical conductivity of molten TI,S doped with impurity elements (600°C).0, T1; n , A g ; 0 , C d ; C7,In. If we assunie that the increase in Q is mainly due to the accuinulation of T1 atoms in the capillary part of the cell, the average concentration of the excess T1 in the capillary can be calculated from the observed Ao. We can then estimate roughly the transport number of the ioiiic species frojn the current efficiency.'O The estimated current cfficiency at 600°C was -2 x Belashchenko et d . l l have also shown, from the e.1n.f. nieasurenients of a galvanic cell with concentration gradient, that the ionic contribution to the total conductivity in molten T1,S is only of the order of I %. In view of these results we neglect the ionic contribution for further discussion in the present temperature range.The density of molten Tl,S, determined by pyknometry, is expressed by the following equation : p / g ~ n 1 - ~ = 8.970- 18.66 x T/"C. DISCUSSION In the table due to Allgaier," molten TI,S belongs to the Range C, characterized by a low value of Q (< lo2 ohm-' cm-l), a positive da/dT and a negative dlSl/dT. The observed conductivity of pure molten T1,S is far below the value of Molt's278 criterion for non-localized conduction, - 100 ohm-l cm-l.12 In the T1-rich samples the conductivity shows a weaker temperature dependence (fig. 2). This suggests that the excess metal atom donates its valence electron to the electron gas in the conduction band of molten “ T1,S ”. The conduction electrons are degenerate at sufficiently large concentrations of excess metal (e.g., excess T12 1.5 atom %).We also note that the enthalpies of mixing show a deep minimum at the composition corres- ponding to T12S.15 This indicates that the molten T1-S system may be regarded as consisting of two pseudo-binary mixtures: Tl-T1,S and Tl,S-S, similar to the molten T1-Te system.16 If the electrical properties of the TI-T1,S mixtures are treated with the usual conductivity mechanism for a heavily doped n-type semi- conductor (Le., “ T1,S ” is an intrinsic semiconductor), the mean free path is calculated from the formula L = 3n2Fia/e2(3n2n)5 with the excess electron concentration, n = ziNi, where zi and Ni are the valence and the concentration of the dissolved metallic element. The values of L calculated in this way are small compared with the average interatomic distance a at metal concentrations of interest ; for example, L = 2.3 A at 2 atom % T1 and 2.0 A at 1.5 atom % T1.Therefore, it appears that the electron behaviour in the conduction band deviates considerably from the free electron state. We attempt to explain the electrical properties of these mixtures using a modified version of Mott’s theory for the strong scattering process, which is analogous to the cases of concentrated metal-ammonia or metal-amine solutions. l4 According to Mott l 2 the expression for 0 resulting from strong scattering takes the form ELECTRICAL PROPERTIES OF MOLTEN Tl,S Q = S,e2ag2/12n3h, (1) where SF is the Fermi surface area and g is the ratio of the density of states at the Fermi level to that of the free electron case.* Though eqn (1) was originally derived for all the valence electrons in a system, we apply it to the excess electrons in the doped samples, by assuming that the values of S, is given by S, = 4n(3n2n)3, with the excess electron concentration.By taking a E 3.0 A, which is the average of the sum of atomic and ionic radii of TI and S , we have estimated the value of g from the conductivity data at 600°C. The results are shown in fig. 4 by a broken line. The value of g tends to unity at high excess metal concentrations. The expression for the thermoelectric power is given by E=EF if the conductivity is determined by electrons with energies in the neighbourhood of the Fermi energy EF. Assuming an invariant density of states, we can write where N,(E,) is the density of states (per unit vohme) for free electrons at the Fermi level.The values of (d In o/dE)E=EF and (d In a/dn) can be determined from the observed results for the thermoelectric power and conductivity respectively. Thus we can estimate the value of g as a function of excess metal concentration by use of eqn * An identical expression to eqn (1) was used in the analysis of electrical conductivity of extrinsic semiconductors (N. F. Mott, Ado. Phys., 1972,21,875), in which the concentration of excess electrons is so high that a metal-insulator transition of the Mott-Hubbard type may occur. The average donor-donor distance was taken for the mean free path in eqn (1) for the extrinsic semiconductors whose conductivity measurements were made at very low temperatures.Y .NAKAMURA, K . MATSUMURA A N D M . SHIMOJI 219 (3). The result is given in fig. 4 by a dotted line, which is in reasonable agreement with that deduced from the conductivity data. The value of (dlnaldn) decreases with temperature, and the temperature coefficient of the thermoelectric power, dlSl/dT, is estimated to be very small, in accordance with observation (fig. 3). The assumption that the density of states is not dependent greatly on composition should be valid for fairly small deviations from the stoichiometric composition. A similar analysis has been made for amorphous Mg-Bi alloys by Ferrier and HerreIl.l* TABLE 1.-PHASE DIAGRAM OF TI-S SYSTEM I N THE VICINITY OF T12S (XT~ = 0.6667) XTI temp. /"C remark 0.6732 574 phase separation 0.6716 539 phase separation 0.6700 496 phase separation 0.6667 448 melting point 0.6601 446 liquidus 0.6536 441 liquidus 0.6349 41 8 liquidus Near the stoichiometric composition, where carrier concentrations are small, the density of states is expected to be reduced to a large extent from the free electron value, and electron states may be localized.For such regions the strong scattering approach may no longer be adequate. The electrical conduction is due either to hopping of electrons with energies within -kT of the Fermi energy or to band conduction by electrons excited to energies beyond the mobility shoulder. For the band conduction the electrical conductivity and thermoelectric power can be expressed as arid CT = cr0 exp( - AE/kT) = C T ~ exp(y/k) exp( - Eo/kT), S = (k/e)(AE/kT + 1) = (k/e)(Eo/T -y/k + l), (4) (5) where AE is the activation energy for conduction, which can be written AE = Eo - y T, taking account of temperature dependence.If we apply eqn (5) to the present data for pure Tl,S, we obtain Eo = 0.57eV and y = 6 . 7 ~ 10-4eVK-1, and from eqn (4) we have Eo = 0.53 eV and ts0 = 6.3 ohm-' cm-l with the aid of y obtained from eqn (5). A fairly good agreement of the values of Eo from eqn (4) and (5) may indicate that the band conduction approach is not in serious conflict with the observed results. The applicability of similar expressions to the hopping conduction has been discussed by Ichikawa and Thompson.16 In the sulphur-rich side, the excess sulphur may dissolve by the reaction S +Tlf+S2-+Tl3+, and conduction will be possible by a hopping mechanism such as Tl+ +T13++T13+ +Tl+.(6) Using the conductivity data together with the concentration of hopping electrons estimated from eqn (6), we have for the mobility, p z 0.1 cm2 V-l s-l. This value of p is well in the range of hopping conduction, though further quantitative arguments are not possible for the moment.286 ELECTRICAL PROPERTIES OF MOLTEN T1,S I M. CWtr and C. E. Mallan; Phys. Rev., 1966, 144, 642. Y. Nadtamura and M. Shimoji, fiuns. Fwhduy Soc., 1969; 65; 1509. J. E. Enderby and C. J. Simmons, Phil. Mug., 1969, 20,125. N. Cusack and P. Kendall, Proc. Phys. Soc., 1 58 72 898. A. A. Velikanov and V. A. Tertykh, Russ. J. jhji. d e r n . . 1969,43, 1444. ti A. F. hfk and A. R. Regel; Progr. Semicond., f960,4, 23g.F. A. Kanda, R. C. Faxon and R. Grant, Shy$. Chem. Liquids, 1968,1, 61. M. Hansen, Constitution of Binary Alloys (McGraw-Hill, New York, 1958). The valence of TI is taken as + 1 to be fitted to the data. The value of + 3 for TI in T1,Se reported in ref. (2) should be corrected to + 1. l o 0. Kubaehewski and K. Reinark, Z. Elektrdchem., 1948,52,75. D. K. Belashchenko, I. A. Madidson and F. L. Kmopelko, Ukr. Fiz. Zhur., 1967, 12, 66. I N. F. Mott and E. A. Davis, Electronic Processes in Noncrystalline Materials (Clarendon Press, Oxford, 1971). l 3 J. V. Acrivos and N. F. Mott, Phil. Mug., 1971, 24, 15. l4 Y. Nakamura, M. Yamamoto and M. Shimoji, Proc. 2nd itd. Conf. Prop. Liquid Metals, Tokyo, 1972 (Taylor and Ftanck; London, 1979, p. 385. l 5 T. Maekawa, T.Yakokawa and K. Niwa, J. Chm. Thermodyhdrtfics, 1971, 3, 707. l 6 Y. Nakamura and M. Shimoji, 2’Fans. Furuday Soc., 1971, 67, 1270. l 8 R. P. Ferrier and D. .I. Herrell, Phil. Mug., 1969, 19, 853. K. Ichikawa and J. C. Thomp$on, Phil. Mug., 1972, 26, 483. Electrical Properties of Molten T1,S BY Y. NAKAMUAA, K. hhTSUMIJRAt AND M. SHIM011 * Department of Chemistry, Faculty of Science, Hokkaido University, Sapporo, Japan Received 24th May, 1973 The electrical conductivity and thermoelectric power of TlzS have been measured as a function of temnerature. Non-stoichiometric and impurity doping effects have been investigated. The results for molten TlzS containing excess metal atoms are discussed with relation to Mott's theory for the strong scattering process.It has been reported that molten T1-Te,1-3 and TI-Se systems exhibit a sharp minimum in the electrical conductivity-composition curve at the stoicluometric composition T12Te or T1,Se and that the absolute thermoelectric power changes its sign around this composition. In this paper we report the experimental results on the electrical conductivity and thermoelectric power of T1,S in the solid and liquid states. The effects of non-stoichiometric doping, as well as those of impurity doping, have also been investigated and compared with the results for molten TI,Te and T1,Se. EXPERIMENTAL The apparatus and experimental procedures for measurements of the electrical conduct- ivity and thermoelectric power were similar to those described elsewhere.2 Since tungsten 300 400 5 00 600 temperaturerc FIG.1.-Electrical conductivity of pure TlzS as a function of temperature. A, d.c. four-probe method ; 0, a.c. four-probe method ; 0, Velikanov and Tert~kh.~ t present address : Toshiba Electric Co., Ltd., Kawasaki, Japan. 273274 ELECTRICAL PROPERTIES OF MOLTEN TIZS electrodes corrode with samples containing excess sulphur at high temperatures, graphite rods in which a tungsten wire was embedded, were used as electrodes. The electrical conductivity was measured both by the d.c. and ax. four-probe methods. Results from these two methods coincided with each other within the range of experimental error of 1 % (fig. 1). The thermoelectric power measurements were made using a supplementary heater placed under one end of the cell. A correction was made for the absolute thermoelectric power of the tungsten electrode^.^ Electrolysis of the stoichiometric T12S was carried out in the conductivity cell by using a constant current electrolyzer and the variation of conduct- ivity during and after the electrolysis was recorded.The purity of TI metal was 99.95 %. The reagent grade sulphur was purified by repeated recrystallization from CS2 solutions. Samples were prepared in a sealed evacuated ampoule by melting carefully weighed amounts of the constituent elements. RESULTS The experimental results for the conductivity, 0, of pure T1,S are shown in fig. 1 as a function of temperature. The values of 0 increase with increasing temperature both in the solid and liquid states. The conductivity increases from 0.48 to 2.3 ohm-' cm-I at the melting point (448+ I T ) .The present values of CT are in good agreement with those determined by Velikanov and T e r t ~ k h . ~ The value reported by Joffe and Regel,6 1 x ohm-' cm-' just above the melting point, is too small 2 1 4 I rl I .c b M 0 E 1 - 0 I 1 I I 1 1 I 1 TL 2.0 "lo \ 0-5010- 0.3 '10 S O \ i d I i q u i d T 1 2.0 OIO 1.0 "lo I . I I . 3 1.5 I .7 103 K I T FIG. 2.-EIectricaI conductivity of doped T12S as a function of reciprocal temperature.Y . NAKAMURA, K . MATSUMURA AND M. SHIMOJI 800 400 4 0 \ % w -400 -800 I I s o l i d I iqui d T l 2.0'10 -A& 1.5'10 o O O / o 1.0 "0 2 O 0 . 5 " o u "u n u 0 0.3 " l o 0.2°10 T I 1 . 5 ' / o 2.0 "lo 0 0.5 '10 0 v> A 0 il +-v a A - D 0 0 -0 n 300 400 500 600 temperature/ "C FIG.3.-Absolute thermoelectric power of doped TlzS as a function of temperature. I50 I00 rl I 4 I c b E --- 50 1.0 Y 0.5 0 275 excess S excess T1 (atom %) 4.-ElectricaI conductivity of molten TI,S at 600°C. 0, present Tertykh5; ---,yfromeqn(l) witha Z 3.0A; - - - - Y 9 work; U, Velikanov from eqn (3). and276 ELECTRICAL PROPERTIES OF MOLTEN T12S and appears to be erroneous. The value of the absolute thermoelectric power, S, for pure molten T1,S is negative and its absolute value decreases slightly with increasing temperature, as shown in fig. 3. The non-stoichiometric effect is investigated by adding the component elements excessively to molten stoichiometric T1,S; the data for Q and S are given in fig. 2 and 3, and their composition isotherms at 600°C are shown in fig.4 and 5. The conductivity values increase almost linearly with the amount of excess TI, whereas the effect of excess sulphur is much less significant. The absolute thermoelectric power changes sign at the stoichiometric composition : positive for the S-rich region and negative for the TI-rich region. This result is similar to those of molten T1,Te and T12Se.1-3 -100 - 4 ;L! \ ?i h -300 -. -500 ,- I 2 2 I 0 1 excess S excess TI (atom ‘x) Fiti. 5.-Absolutc thcrmoclcctric power of molten TlzS at GOO’C. As seen in fig. 2, the (log 0, 1/T) plots for the samples containing excess T1 exhibit a knee at a temperature which depends on the concentration of excess T1. Similar conductivity curves obtained in the T1,Se +T1 system were related to the “ saturation effect ” in our previous paper,2 but the results can now well be explained from the phase diagram redetermined by Kanda et al.’ : these knees correspond exactly to the temperature of liquid-liquid phase separation.From the analogous point of view we can determine the phase diagram ,of the T1- S system in the vicinity of the composi- tion TlzS (or X,, = 0.6667) by using the present conductivity data. Thc results given in table 1 are somewhat diffcrent from Hansen’s diagram,* as in the TI-Se system. The impurity elements chosen here were Ag, Cd and In which belong to the Vth period. The increase in Q is approximately proportionzi to the concentration of excess electrons which can be given by the concentration of excess metal multiplied by its valence. The assigned valences are the same as those found in molten Tl,Se.2* The values of S of the impurity-doped T1,S were similar to thosc of molten T1,S containing excess T1, characterized by a large negative value and a slight temperature dependence.Fig. 6 shows the effect of impurity doping on Q of molten T1,S.Y. NAKAMUkA, K . MATSOMARA AND M. SHIMOJI 277 Since the value of 0 for pure molten T1,S is as low as 2.3 ohm-l ctn-' at the melting point, a contribution from ionic conduction was expected. In order to detect this effect the molten T1,S was electrolyzed and the change in CJ after the electrolysis was measured. The increase in conductivity, ACT, was found to be roughly propor- tional to the amount of electricity passed through the melt. When the electrolyzing current was turned off, the enhanced conductivity decreased with time, and a more rapid recovery of the resistivity was observed if the polarity of electrolysis was changed.I50 190 'F - u I c b - E \ 50 I / I 0 l z 100 zi.yi FIG. 6.--Electrical conductivity of molten TI,S doped with impurity elements (600°C). 0, T1; n , A g ; 0 , C d ; C7,In. If we assunie that the increase in Q is mainly due to the accuinulation of T1 atoms in the capillary part of the cell, the average concentration of the excess T1 in the capillary can be calculated from the observed Ao. We can then estimate roughly the transport number of the ioiiic species frojn the current efficiency.'O The estimated current cfficiency at 600°C was -2 x Belashchenko et d . l l have also shown, from the e.1n.f.nieasurenients of a galvanic cell with concentration gradient, that the ionic contribution to the total conductivity in molten T1,S is only of the order of I %. In view of these results we neglect the ionic contribution for further discussion in the present temperature range. The density of molten Tl,S, determined by pyknometry, is expressed by the following equation : p / g ~ n 1 - ~ = 8.970- 18.66 x T/"C. DISCUSSION In the table due to Allgaier," molten TI,S belongs to the Range C, characterized by a low value of Q (< lo2 ohm-' cm-l), a positive da/dT and a negative dlSl/dT. The observed conductivity of pure molten T1,S is far below the value of Molt's278 criterion for non-localized conduction, - 100 ohm-l cm-l.12 In the T1-rich samples the conductivity shows a weaker temperature dependence (fig.2). This suggests that the excess metal atom donates its valence electron to the electron gas in the conduction band of molten “ T1,S ”. The conduction electrons are degenerate at sufficiently large concentrations of excess metal (e.g., excess T12 1.5 atom %). We also note that the enthalpies of mixing show a deep minimum at the composition corres- ponding to T12S.15 This indicates that the molten T1-S system may be regarded as consisting of two pseudo-binary mixtures: Tl-T1,S and Tl,S-S, similar to the molten T1-Te system.16 If the electrical properties of the TI-T1,S mixtures are treated with the usual conductivity mechanism for a heavily doped n-type semi- conductor (Le., “ T1,S ” is an intrinsic semiconductor), the mean free path is calculated from the formula L = 3n2Fia/e2(3n2n)5 with the excess electron concentration, n = ziNi, where zi and Ni are the valence and the concentration of the dissolved metallic element. The values of L calculated in this way are small compared with the average interatomic distance a at metal concentrations of interest ; for example, L = 2.3 A at 2 atom % T1 and 2.0 A at 1.5 atom % T1.Therefore, it appears that the electron behaviour in the conduction band deviates considerably from the free electron state. We attempt to explain the electrical properties of these mixtures using a modified version of Mott’s theory for the strong scattering process, which is analogous to the cases of concentrated metal-ammonia or metal-amine solutions.l4 According to Mott l 2 the expression for 0 resulting from strong scattering takes the form ELECTRICAL PROPERTIES OF MOLTEN Tl,S Q = S,e2ag2/12n3h, (1) where SF is the Fermi surface area and g is the ratio of the density of states at the Fermi level to that of the free electron case.* Though eqn (1) was originally derived for all the valence electrons in a system, we apply it to the excess electrons in the doped samples, by assuming that the values of S, is given by S, = 4n(3n2n)3, with the excess electron concentration. By taking a E 3.0 A, which is the average of the sum of atomic and ionic radii of TI and S , we have estimated the value of g from the conductivity data at 600°C. The results are shown in fig. 4 by a broken line. The value of g tends to unity at high excess metal concentrations.The expression for the thermoelectric power is given by E=EF if the conductivity is determined by electrons with energies in the neighbourhood of the Fermi energy EF. Assuming an invariant density of states, we can write where N,(E,) is the density of states (per unit vohme) for free electrons at the Fermi level. The values of (d In o/dE)E=EF and (d In a/dn) can be determined from the observed results for the thermoelectric power and conductivity respectively. Thus we can estimate the value of g as a function of excess metal concentration by use of eqn * An identical expression to eqn (1) was used in the analysis of electrical conductivity of extrinsic semiconductors (N. F. Mott, Ado. Phys., 1972,21,875), in which the concentration of excess electrons is so high that a metal-insulator transition of the Mott-Hubbard type may occur.The average donor-donor distance was taken for the mean free path in eqn (1) for the extrinsic semiconductors whose conductivity measurements were made at very low temperatures.Y . NAKAMURA, K . MATSUMURA A N D M . SHIMOJI 219 (3). The result is given in fig. 4 by a dotted line, which is in reasonable agreement with that deduced from the conductivity data. The value of (dlnaldn) decreases with temperature, and the temperature coefficient of the thermoelectric power, dlSl/dT, is estimated to be very small, in accordance with observation (fig. 3). The assumption that the density of states is not dependent greatly on composition should be valid for fairly small deviations from the stoichiometric composition.A similar analysis has been made for amorphous Mg-Bi alloys by Ferrier and HerreIl.l* TABLE 1.-PHASE DIAGRAM OF TI-S SYSTEM I N THE VICINITY OF T12S (XT~ = 0.6667) XTI temp. /"C remark 0.6732 574 phase separation 0.6716 539 phase separation 0.6700 496 phase separation 0.6667 448 melting point 0.6601 446 liquidus 0.6536 441 liquidus 0.6349 41 8 liquidus Near the stoichiometric composition, where carrier concentrations are small, the density of states is expected to be reduced to a large extent from the free electron value, and electron states may be localized. For such regions the strong scattering approach may no longer be adequate. The electrical conduction is due either to hopping of electrons with energies within -kT of the Fermi energy or to band conduction by electrons excited to energies beyond the mobility shoulder.For the band conduction the electrical conductivity and thermoelectric power can be expressed as arid CT = cr0 exp( - AE/kT) = C T ~ exp(y/k) exp( - Eo/kT), S = (k/e)(AE/kT + 1) = (k/e)(Eo/T -y/k + l), (4) (5) where AE is the activation energy for conduction, which can be written AE = Eo - y T, taking account of temperature dependence. If we apply eqn (5) to the present data for pure Tl,S, we obtain Eo = 0.57eV and y = 6 . 7 ~ 10-4eVK-1, and from eqn (4) we have Eo = 0.53 eV and ts0 = 6.3 ohm-' cm-l with the aid of y obtained from eqn (5). A fairly good agreement of the values of Eo from eqn (4) and (5) may indicate that the band conduction approach is not in serious conflict with the observed results.The applicability of similar expressions to the hopping conduction has been discussed by Ichikawa and Thompson.16 In the sulphur-rich side, the excess sulphur may dissolve by the reaction S +Tlf+S2-+Tl3+, and conduction will be possible by a hopping mechanism such as Tl+ +T13++T13+ +Tl+. (6) Using the conductivity data together with the concentration of hopping electrons estimated from eqn (6), we have for the mobility, p z 0.1 cm2 V-l s-l. This value of p is well in the range of hopping conduction, though further quantitative arguments are not possible for the moment.286 ELECTRICAL PROPERTIES OF MOLTEN T1,S I M. CWtr and C. E. Mallan; Phys. Rev., 1966, 144, 642. Y. Nadtamura and M. Shimoji, fiuns. Fwhduy Soc., 1969; 65; 1509. J. E. Enderby and C. J. Simmons, Phil. Mug., 1969, 20,125. N. Cusack and P. Kendall, Proc. Phys. Soc., 1 58 72 898. A. A. Velikanov and V. A. Tertykh, Russ. J. jhji. d e r n . . 1969,43, 1444. ti A. F. hfk and A. R. Regel; Progr. Semicond., f960,4, 23g. F. A. Kanda, R. C. Faxon and R. Grant, Shy$. Chem. Liquids, 1968,1, 61. M. Hansen, Constitution of Binary Alloys (McGraw-Hill, New York, 1958). The valence of TI is taken as + 1 to be fitted to the data. The value of + 3 for TI in T1,Se reported in ref. (2) should be corrected to + 1. l o 0. Kubaehewski and K. Reinark, Z. Elektrdchem., 1948,52,75. D. K. Belashchenko, I. A. Madidson and F. L. Kmopelko, Ukr. Fiz. Zhur., 1967, 12, 66. I N. F. Mott and E. A. Davis, Electronic Processes in Noncrystalline Materials (Clarendon Press, Oxford, 1971). l 3 J. V. Acrivos and N. F. Mott, Phil. Mug., 1971, 24, 15. l4 Y. Nakamura, M. Yamamoto and M. Shimoji, Proc. 2nd itd. Conf. Prop. Liquid Metals, Tokyo, 1972 (Taylor and Ftanck; London, 1979, p. 385. l 5 T. Maekawa, T. Yakokawa and K. Niwa, J. Chm. Thermodyhdrtfics, 1971, 3, 707. l 6 Y. Nakamura and M. Shimoji, 2’Fans. Furuday Soc., 1971, 67, 1270. l 8 R. P. Ferrier and D. .I. Herrell, Phil. Mug., 1969, 19, 853. K. Ichikawa and J. C. Thomp$on, Phil. Mug., 1972, 26, 483.

ISSN:0300-9599    

DOI:10.1039/F19747000273

出版商:RSC    

年代:1974

数据来源: RSC