摘要:

J . Chem. SOC., Furaday Trans. I , 1985, 81, 105-112 p-Doping of (CH)x to the Metallic Regime with Gaseous Oxygen Application to Oxygen Fuel-cell-type Electrodes BY ROBERT J. MAMMONE AND ALAN G. MACDIARMID* Department of Chemistry, University of Pennsylvania, Philadelphia, Pennsylvania 19104, U.S.A. Received 29th March, 1984 Truns-polyacetylene, (CH),, can be p-doped (oxidized) to the metallic regime by gaseous oxygen in the presence of an aqueous solution of a non-oxidizing acid such as HBF,. The oxygen oxidizes the (CH), to (CHY+)Z while the HBF, supplies a stable fluoroborate counter-anion to give a species such as [CHY+A;], ( y = 0.02; A- = fluoroborate). The doping process is consistent with the reduction potentials of (CH), and 0, in acid solution. In the absence of the acid, oxygen reacts irreversibly with the semiconducting (CH), to destroy its conductivity.The [CHy+A;], formed above may be connected to a lead anode in a HBF, electrolyte whereupon it is reduced to (CH), with concomitant production of an electric current. If oxygen is bubbled over the (CH), during this process it is continuously electrochemically reduced by the lead as rapidly as it is oxidized chemically by the oxygen. The (CH), therefore acts as a catalyst electrode: a ‘fuel-cell’ electrode for the room-temperature reduction of 0, to H,O at atmospheric pressure. The overall reaction does not occur to any significant extent in the absence of the (CH),. It is well established that the conductivity of parent polyacetylene, (CH),, the prototype conducting polymer, or that of its p- or n-doped forms is rapidly and irreversibly destroyed on exposure to oxygen or air.’.A more detailed understanding of the oxidation (p-doping) process by which the conductivity of cis- or trans-(CH), can be increased through the semiconducting to the metallic regime3 has now enabled us to use gaseous oxygen to ‘ dope ’ (CH), under appropriate experimental conditions to the metallic regime. It has been reported previously that when cis-(CH), or trans-(CH), is exposed to oxygen or air for > ca. 1 h irreversible oxidative degradation of the material 0ccurs.l However, during the first few minutes of exposure the conductivity increases by one to two orders of magnitude [from ca. to 10F7 R-l cm-l for cis-(CH),] in a reversible process, uiz.During this time the 0,-doping reaction may be reversed by pumping off the oxygen. It is believed the material probably consists of a delocalized polycarbonium ion in combination with an 0;- i0n.l This system is intrinsically unstable in view of the strongly oxidizing nature of the anion. It may be regarded as the transition state for the irreversible oxidation of the (CH), t o an insulating material containing carbonyl and other oxygen-containing groups. The degradation by oxygen of (CH), p-doped to the metallic regime, e.g. [<=H0.07+(13)0.0,],,2 presumably occurs in a similar manner, the oxygen oxidizing the material still further to give [CH(0.07+y)+ (I 3);. ,, ( 0,)E72],, which then spontaneously decomposes. It is not surprising that reduced (n-doped) polyacetylene, e.g. 105106 0, DOPING OF (CH), [Na$.l(CH)o.l-],, reacts instantly and irreversibly with air in view of the strongly reducing nature of the (CHOal-), polycarbanion. This results in complete loss of conductivity.EXPERIMENTAL Polyacetylene film was synthesized as previously de~cribed.~ For the 0,-doping studies, a piece of cis-rich (CH), (ca. 85% cis isomer) was thermally isomerized to the more stable trans isomer by heating at 160 "C for 20 min6 This piece was cut into several small pieces (ca. 5 x 20 x 0.1 mm; ca. 4 mg) and several large pieces (ca. 20 x 30 x 0.1 mm; ca. 20 mg) which were later sent for elemental analysis.' All the pieces were placed in a 48% (7.4 mol dm-3) aqueous solution of the strong non-oxidizing acid HBF, (Aldrich Co.) in two polyethylene containers. The solutions were previously deaerated by bubbling argon through them for 2 h.Oxygen was then bubbled continuously through one container so that the samples of film were constantly floating and moving throughout the solution. After a selected time interval one piece of the film was removed and dried in the vacuum system for 2 h by dynamic pumping and its conductivity (4-probe) was measured. Passage of oxygen was continued for an additional period of time and another sample of film was removed and treated in a similar manner. This process was repeated until all pieces of film had been examined. The solution in the other container was covered with an argon atmosphere and was used as a control. Samples of film were removed at intervals and treated as described above. For the 0,-electrode studies, cis-rich (CH), (ca.85 % cis isomer) was first electrochemically isomerized in the following manner. A strip of (CH), film, to the top of which a platinum wire had been attached by mechanical pressed contact, was placed in a solution of 0.5 mol dm-3 [Bu,NJ+[BF,]- in methylene chloride under a dry, nitrogen atmosphere. A platinum counter- electrode (1.5 cm2) was placed in the solution at a distance of ca. 5 cm from the (CH), electrode. Electrochemical oxidation (p-doping) of the (CH), to [CH0-06+(BF,);~o,], was accomplished by having it serve as the anode. It was held at a constant potential of +9.0 V with respect to the Pt counter-electrode. Oxidation of the (CH), was discontinued after a set amount of charge, corresponding to 6 mol% oxidation, had been passed.The (Bu,N)+ ion was reduced at the cathode, with the production of hydrocarbons and amines.* This step took ca. 15 min. The film was then washed with methylene chloride and dried in U ~ C U O with constant pumping. Oxidation as described above results in spontaneous isomerization of the cis isomer to the trans is~mer.~ A strip of the [CHo.os+(BF,);.o,], (ca. 10 x 10 x 0.1 mm; ca. 5.1 mg), to the top of which a platinum wire had been attached by mechanical pressed contact, was then placed in one leg of a polyethylene H-cell containing a solution of 48% aqueous HBF,. The platinum wire attached to the film and the area of the film touching the platinum was covered with molten paraffin wax. The latter step was done to insure that the platinum current collector was isolated from the electrolyte solution.A lead counter-electrode (ca. 10 x 40 x 0.13 mm) was placed in the solution in the other leg of the H-cell at a distance of ca. 5 cm from the [CHo.06+(BF,);.06], electrode. Electrochemical reduction of the [CHo.06+(BF,);.o,], was accomplished by having it serve as the cathode by holding it at a constant potential of 0.01 V with respect to the lead counter-electrode for 1 h. The net electrochemical reaction which occurred is expressed by (2) [CHo-06+(BF4);~06], + 0.03xPb + (CH), + 0.03xPb(BF4),. The trans-(CH), prepared using the above method gave better results in the 0,-electrode studies than those obtained from a cell constructed using either cis-(CH), or trans-(CH), obtained by thermal isomerization of cis-(CH), in vacua This may be related to the fact that trace amounts of dopant remained in the trans-(CH), after the above electrochemical treatment, which would tend to make it more hydrophillic and therefore give higher currents with aqueous electrolytes.Current against time curves were recorded continuously for the above 1 h electrochemical reduction of the [CHo-06+(BF,);o,], and for all subsequent studies. Oxygen was then permitted to bubble for 1 h over the (CH),, still held at a constant applied potential of 0.01 V us the lead counter-electrode. This resulted in the continuous chemical oxidation and electrochemical reduction of the partly oxidized polyacetylene electrode. The oxygen stream was then turned off and the current was recorded as a function of time.These two steps were repeated severalR. J. MAMMONE AND A. G. MACDIARMID I07 -4t X -5t I 1 I 1 I I I 1 1 2 3 4 5 6 7 time/days Fig. 1. Conductivity (4-probe) of trans-(CH), film as a function of time when immersed in 48 % (7.4 mol dm-3) aqueous HBF, and exposed either to oxygen (upper curve) or argon (lower curve): (a) 0, one sample treated for 2 h (dried for 2 h); (b) ., one sample treated for 2 h under argon (dried for 2 h); (c) 0, five samples treated for varying lengths of time in the same container (dried for 2 h); ( d ) 0, five samples treated for varying lengths of time in the same container but HBF, replaced by a fresh HBF, solution every 48 h (dried for 2 h); ( e ) 0, five samples treated for varying lengths of time under argon in the same container (dried for 2 h); cf) x , conductivity of trans-(CH),, (2.3 x R-l cm-l); ( g ) A, one sample (ca. 20 x 30 x 0.1 mm) treated for 24 h (dried for 24 h); (h) A, one sample (ca. 20 x 30 x 0.1 mm) treated for 7 days (dried for 24 h); (i) V, one sample (ca. 20 x 30 x 0.1 mm; 18.1 mg) treated for 24 h under argon (dried for 24 h, 18.5 mg).times. In order to confirm that the effect being observed was due to the polyacetylene and not to the current collector, the exposed portion of the polyacetylene was cut off parallel to the current collector (see fig. 1). A thin cross-section (ca. 0.1 mm) of film was still exposed to the solution. The experiment was repeated once utilizing this configuration. Constant potential oxidation and reduction (for electrochemical doping and 0,-electrode studies) was carried out utilizing a Princeton Applied Research (PAR) model 363 potentiostat- galvanostat in conjunction with an Electro-Synthesis Co.(ESC) model 630 digital coulometer. Voltages and currents (for conductivity studies) were measured with a Keithley 169 or 177 microvolt digital multimeter. Current against time curves (for 0,-electrode studies) were recorded with a Houston Instruments series D500 strip-chart recorder. RESULTS AND DISCUSSION The relationship between the conductivity of the (CH), samples and the length of their exposure to their environments is given in fig. 1 . The initial conductivity of the trans-(CH), was 2.3 x R-l cm-l. The conductivity of the film exposed to oxygen increased by ca.lo5 during the first day to the beginning of the metallic regime (3 f 2 R-l cm-l) and remained constant within experimental error at this value for the following 7 days. The increase in conductivity of the (CH), in the control108 0, DOPING OF (CH), Table 1. Analytical data for doped trans-(CH), films" C H B F sample composition found calc. found calc. found calc. found calc. ~cH[BF2(oH)d0.0035(0H)0.0025~Z 89.34 89.96 7.73 7.62 0.56 0.28 0.97 1.00 {CH[BF2(0H)~Io.0560)~C 68.32 68.02 6.75 6.35 3.55 3.43 12.38 12.05 ~ C H ~ B F 3 0 H ~ o . o o ~ ~ ~ B F 2 ~ 0 H ~ 2 ~ o . o ~ 2 3 ~ ~ d 83.17 82.71 7.18 7.15 1.47 1.34 6.36 5.45 ~ C H ~ B F ~ ~ o . o o ~ ~ ~ B F 3 0 H ~ 0 . 0 ~ ~ 0 ~ 0 H ~ o . o ~ ~ ~ ~ e 76.19 75.96 6.50 6.87 1.33 1.32 8.01 7.95 a Since oxygen analysis of compounds of this type presents certain difficulties, the amount of oxygen was deduced, as is common practice, by assuming that the difference between 100 and the sum of the percentage elemental composition of all other elements represents the percentage of oxygen by weight in the compound.Oxygen determined by difference: found 9.00, calc. 10.15. Oxygen determined by difference: found 7.97, calc. 7.90. Oxygen determined by difference: found 1.40, calc. 1.14. Oxygen determined by difference: found 1.82, calc. 3.35. experiment was significantly less, reaching a maximum of only ca. 2 x 0-l cm-l. Elemental analysis [fig. 1 (i); table 1, sample 13 of a piece of (CH), from the control experiment under argon shows that the HBF, used in these studies undergoes negligible reaction with (CH), under the experimental conditions employed.The (much smaller) increase in conductivity which does occur is probably related to the presence of minute traces of oxygen remaining in the HBF, solution since, upon doping, the conductivity of (CH), increases very rapidly at first, even at extremely low doping levels.1o As can be seen from the legend to fig. 1, the data given in the graphs were obtained from several different experiments carried out under slightly different conditions. In one of the 7-day experiments the same HBF, solution was used throughout the experiment. In the other 7-day experiment a fresh solution was used every 2 days. The overall reaction which occurs can be expressed by and is consistent with a mechanism involving first the oxidation of (CH), to (CHu+), by the oxygen as given by reaction (1) followed by reaction of the [cHu+(o,);~~], so formed with the HBF,, viz.(4) Since the 0;- ion is a hard base it is expected to react preferentially with the hard acid H+ (or its aquated form) rather than with the soft acid (CHu+),. We have shown in a separate study that a ca. 0.1 mol dm-3 solution of H,O, in 48 % HBF, will oxidize trans-(CH), to the metallic regime (a = 12 0-l cm-l; table 1, sample 2) during ca. 30 min as given by (5) [CHY+(O,>~~,], + xyHBF, + [CHu+(BF,)J, + +syH,O,. 2(CH), +xyH,O, + 2xyHBF4 + 2[CHg+(BF,);], + 2xyH20. A combination of the processes given by reactions (l), (4) and ( 5 ) results in the net reaction represented by reaction (3). In the above equations the anion has been represented, for convenience, as (BF,)-.However, it is known that (BF4)- tends to undergo pH-dependent hydrolysis1' to give species such as [BF,(OH)]-, [BF,(OH),]-, etc. Hence it is not surprising to find that elemental analyses for samples of the oxygen-doped film [fig. 1 (g) and (h); table 1, samples 3 and 41 indicate compositions such as e.g. {CHo~018+(BF30H)~~oo,7-R. J. MAMMONE AND A. G. MACDIARMID tivities of 2.9 and 1.4 Sz-l cm-l obtained for these ca. 2 mol% oxidized materials fall in the same range as that observed for similar levels of doping of trans-(CH), with iodine (ca. 0.2 i2-l cm-l)12 and AsF, (ca. 10 R-l cm-l).13 An elemental analysis obtained after exposing a sample of (CH), to oxygen continuously for 7 days in the HBF, solution [fig.1 ( A ) ; table 1, sample 41 indicated the presence of additional oxygen. This is believed to be caused by partial hydrolysis of the (CHy+), ion to form [CH(OH),], groups which occurred to a greater extent during this more lengthy experiment. We have recently determined thermodynamic (equilibrium) reduction potentials for polyacetylene in a number of different oxidation states3 Since oxidation of (CH), occurs in a continuous manner, rather than in a series of distinct integral steps, as for example in the electrochemical intercalation of graphite with (HSO,)- etc.,14 we define the reduction potential of a polyacetylene couple for a given level of oxidation of the polyacetylene as that potential such that the application of an infinitesimally small potential greater or smaller than that of the couple will result in the removal or addition, respectively, of an infinitesimally small amount of negative charge, ax e-, from or to the c ~ u p l e .~ Values of the reduction potentials for polyacetylene couples determined in non-aqueous solution us the Li+, Li couple may be arbitrarily converted to values in aqueous solution us the standard hydrogen electrode (SHE) by the subtraction of 3.05 V. In the non-aqueous solvents employed for studying the polyacetylene couples the error introduced by this approach is not greater than ca. 0.1 V.3 Hence the reduction potentials can be used to rationalize qualitatively the expected thermodynamically permitted reactions. Two representative reduction potentials for the polyacetylene system3 109 [BF,(OH)&.o1,3), and [ ~ ~ o ~ O 1 g ~ ~ ~ ~ ~ ~ ~ .o o ~ ~ ~ ~ F ~ ~ H ~ , , , , , ( O H ~ o The conduc- (CH(y+')t), + ax e- + (CH,), [ y = 0; neutral (CH),]* - 0.7 V us SHE +0.48 V us SHE (CH(o.o2+a)+), + ax e- (CH0.02+ 1, when compared to the standard oxygen reduction potential 0, + 4H+ + 4e- + 2H,O + 1.23 V us SHE show qualitatively that (CH), should be oxidized by oxygen in the presence of acid even more extensively when the hydrogen-ion concentration is greater than unity. This is consistent with reaction (3). Similarly the reduction potential of the couple H,O, + 2H+ + 2e- 2H,O is consistent with reaction (5). The observation that the (CH), is not oxidized to the extent expected from the + 1.23 V reduction potential of the O,/H,O couple may be due to an increase in activation energy for the oxidation process as oxidation proceeds to higher levels and/or to partial hydrolysis of the more highly oxidized polyacetylene [see for example fig.l(h); table 1, sample 41. The effect of gaseous oxygen on the current produced by the 0, electrode is shown in fig. 2. The initial current, given in the decreasing left-hand curve, is the initial electrochemical reduction of the [CHo.06+(BF4)~.06]Z described previously. When the film had been almost completely reduced to (CH), the oxygen stream was turned on and the current was recorded as given in fig. 2. After a period of 1 h the oxygen stream was turned off and the current was again recorded. The oxygen stream was then turned on and off several times as shown in fig.2. The amount of charge released during the + 1.77 V us SHE * y refers to the degree of oxidation or reduction of the polyacetylene; it is a positive number for p-doped (oxidized) polyacetylene and a negative number for n-doped (reduced) polyacetylene.110 0, DOPING OF (CH), 2 0, on time Fig. 2. Change in current produced by a (CH),IO,IHBF,(aq)lPb cell when oxygen stream bubbling over the (CH), electrode is turned on and off. Table 2. Amount of charge released and currents observed for a (CH),IO,IHBF,(aq)lPb cell ~ trial amount of chargea/C currentb/mA 1 1.802 (0.0668) 0.530 2 1.745 (0.0647) 0.483 3 1.625 (0.0602) 0.460 4 1.543 (0.0572) 0.458 a Amount of charge released during 1 h exposure to oxygen. The number in parentheses is Current observed after 1 h the number of electrons released per CH unit present in the film.exposure to oxygen. Exposed area of (CH), film is 1.1 cm2. time oxygen was bubbled over the (CH), as well as the current observed after 1 h exposure to oxygen is given in table 2 for each individual oxygen treatment. Note that a total of 6.715 C was released during the time oxygen was bubbled over the (CH),, This corresponds to the passage of 0.249 electrons per CH unit in the (CH),. A significant current flowed only when oxygen was bubbled over the (CH), electrode. When the oxygen stream was stopped, the current immediately dropped from ca. 0.5 mA levelling off at ca. 0.1 mA after 1 h. After four trials the film was severed from the platinum wire by means of a razor blade, by cutting through the wax-covered portion of the film as shown by the dotted line on the electrode configuration given in the inset to fig.2. When oxygen was bubbled over what then remained of the (CH), electrode only a minute increase in current was observed (see fig. 2), thus proving that the (CH), film, not the platinum current collector, was responsible for the observed phenomenon. In another experiment, after turning the oxygen stream off, argon was bubbled through the solution to remove dissolved oxygen. The current fell to an even lower value (ca. 0.03 mA crn-,). Experiments are now in progress to determine the optimum acid strength to be used in the electrolyte in order to obtain maximum currents, voltages, recyclability etc. In these preliminary experiments, in which no attempt has been made to optimize conditions for maximum performance, a steady-state short-circuit current, Isc, of ca.0.4 mA cm-2 of (CH), is obtained simply by bubbling oxygen over the surface of the (CH), film immersed in the electrolyte. Since the bulk density of the film is ca. 0.4 g ~ m - ~ and the density of the ca. 200 A fibrils comprising the film is ca. 1.2 g ~ m - ~ the material is therefore approximately two-thirds void space. Experiments are now in progress to force the oxygen through the porous film. Such a modification in design is expected to increase the current to significantly higher levels. That higher currentsR. J . MAMMONE AND A. G . MACDIARMID 111 should be obtainable is evidenced by the observation that if oxygen is bubbled over the film for 1 h when it is not connected to the lead anode so as to permit the interior portions of the film to become oxidized to a similar level to the outer surface, a short-circuit current of ca.10 mA cm-, is obtained. The open-circuit voltage, Voc, of the cell after 1 h exposure to oxygen, immediately before the short-circuit current measurements, was ca. 0.74 V.I5 The overall reaction which occurs can be expressed by (CH), Pb++02+2HBF,- Pb(BF,),+H,O. If a piece of [CH0.02+(BF,),o,], film and a strip of lead are placed in a 48% aqueous HBF, solution nothing happens. The overvoltage for H, evolution at lead is such that lead does not dissolve spontaneously to any significant extent. If, however, these two electrodes are now connected via an external wire the lead dissolves, liberating electrons : Pb -+ Pb2+ + 2e-.(7) The electrons flow through the wire and are taken up by the (CH0.02+), ion (CH0.02+), +O.O2x e- -+ (CH), 0.OlxPb + [CH0.02+(BF,),o,], + (CH), + O.OlxPb(BF,), (8) (9) resulting in the net electrochemical reduction reaction which regenerates the (CH),. The lead, which acts as the reducing agent or ‘fuel’, is converted to Pb(BF,),. The important point to note is that the (CH), can be reconverted back to [CH0.02S(BF,);.o,], in 48% aqueous HBF, by bubbling oxygen over it while it is immersed in the acid solution as given by reaction (3). Hence, if oxygen is constantly bubbled over the polyacetylene electrode it is possible continuously to oxidize chemi- cally the polyacetylene to [CH0-02+(BF,),o,], as rapidly as it is reduced electro- chemically according to reaction (9).Hence, neither the chemical composition nor the total mass of the p-doped polyacetylene electrode changes during the reaction at steady state, i.e. the p-doped polyacetylene acts as a ‘catalyst electrode’ permitting reaction (6) to occur. The (CH), is therefore acting as an electrocatalytic ‘ fuel-cell-type ’ electrode for oxygen, the oxidizing agent being elemental oxygen, the ‘ fuel ’ being lead and HBF,. It is clear that the operation of the (CH), electrode in this system is completely different from that of a battery electrode during either charge or discharge conditions. In a battery the chemical composition and/or the total mass of the electrode changes. It should be stressed that the potential of the oxygen catalyst electrode is determined by the value of y in the steady-state composition of the [CHU+(BF,);!, comprising the electrode.The value ofy will not be affected by the mechanism by which the (CH), is converted to [CHg+(BF,);],, i.e. whether it is by a four-electron reduction of 0, to 20,- (to produce H,O) as given by reaction (3) or whether it is by two two-electron reduction steps as given first by reaction (4) (to produce H202) followed by reaction (5). The role of (CH), in acting as a catalyst electrode for the net reduction of 0, is completely different from the catalytic role of a material such as platinum. In the latter case the potential of the electrode will depend critically on the degree to which four-electron or two-electron reduction steps are involved. It should also be stressed that lead has been used in this study of the electrocatalytic properties of a (CH),/O, electrode purely as a convenient counter- and reference- electrode.It should not be implied that lead should be considered as a ‘fuel’ in any112 0, DOPING OF (CH), practical oxygen fuel-cell system ! Studies are in progress to investigate other ‘ fuels ’, particularly organic compounds, as a replacement for lead, especially at a (CH), electrocatalytic counter-electrode. A secondary, very important observation arising from this study is that, contrary to general belief, p-doped polyacetylene in the metallic conducting regime can be stable in aqueous (acid) solution. We believe that this stability arises from the delocalization of the positive charge on the polymer chain over ca.15 CH units in a positive soliton,l6 which makes the material less susceptible to nucleophilic attack than if the charge were localized on one CH unit. In conclusion, these studies show that, contrary to previous belief, gaseous oxygen, rather than destroying the conductivity of (CH),, can be used to dope it to the metallic regime. This is accomplished by replacing the 0;- anion in [CH~+(O,)~~,], by a fluoroborate anion which does not react with the (CHY+), cation to destroy its conductivity as does the 0;- ion. Furthermore the reaction is in qualitative agreement with the reduction potentials of (CH), and oxygen in acid solution. Although it is far too early to make any realistic predictions, these unexpected catalytic properties of a conducting organic polymer suggest that (CH), and possibly other conducting polymers might be useful either in fuel-cell-type processes for the generation of electricity using oxygen as an oxidizing agent, or for the facile oxidation of certain organic or inorganic compounds to give useful materials not readily synthesized by other methods.This work was supported by the U.S. Office of Naval Research and by the Defense Advanced Research Projects Agency through a grant monitored by the US. Office of Naval Research. We thank one of the referees for helpful comments. J. M. Pochan, H. W. Gibson and F. C. Bailey, J. Polym. Sci., Polym. Lett. Ed., 1980, 18, 447; J. C. Chien, J. Polym. Sci., Polym. Lett. Ed., 1981, 19, 249; J.M. Pochan, D. F. Pochan, M. Rom- melmann and H. W. Gibson, Macromolecules, 1981, 14, 110; J. M. Pochan, H. W. Gibson and J. Harbour, Polymer, 1982,23,439; H. W. Gibson and J. M. Pochan, Macromolecules, 1982,15,242. A. G. MacDiarmid, Y. W. Park, A. J. Heeger and M. A. Druy, J. Chem. Phys., 1980,73,946. A. G. MacDiarmid, R. J. Mammone, J. R. Krawczyk and S. J. Porter, Mol. Cryst. Liq. Cryst., 1984, 105,89; R. J. Mammone and A. G. MacDiarmid, Synth. Met., 1984,9, 143. T-C. Chung, A. Feldblum, A. J. Heeger and A. G. MacDiarmid, J. Chem. Phys., 1981,74, 5504. H. Shirakawa and S. Ikeda, Polym. J., 1971, 2, 231; H. Shirakawa, T. Ito and S. Ikeda, Polym. J., 1973,4,460; T. Ito, H. Shirakawa and S. Ikeda, J. Polym. Sci., Polym. Chem. Ed., 1974,12, 1 1 ; 1975, 13, 1943. €3. Shirakawa, T. Ito and S. Ikeda, Makromol. Chem., 1978,179, 1565; B. R. Weinberger, H. Ehren- freund, A. Pron, A. J. Heeger and A. G. MacDiarmid, J . Chem. Phys., 1980, 72, 4729; P. Bernier, C. Linenya, M. Rolland and M. Aldissi, J. Phys. Lett., 1981, 42, L295. ’ Analysis performed by Schwartzkopf Microanalytical Laboratory, Woodside, New York 1 1 377, U.S.A. J. E. Dubois, A. Monvernay and P. C. Locaze, Electrochim. Acta, 1970, 15, 315. T-C. Chung, A. G. MacDiarmid, A. Feldblum and A. J. Heeger, J. Polym. Sci., Polym. Lett. Ed., 1982, 20, 427. lo A. G. MacDiarmid and A. J. Heeger, Synth. Met., 1979/80, 1, 101. l1 D. W. A. Sharp, Adv. Fluorine Chem., 1961, 1, 68. l2 C. K. Chiang, Y. W. Park, A. J. Heeger, H. Shirakawa, E. J. Louis and A. G. MacDiarmid, J. Chem. l3 C. K. Chiang, S. C. Gau, C. R. Fincher Jr, Y. W. Park, A. G. MacDiarmid and A. J. Heeger, Appl. l4 A. Metrot and J. E. Fischer, Synth. Met., 1981, 3, 201; J. 0. Besenhard, E. Wudy, H. Mohwald, J. J. Nickl, W. Biberacher and W. Foag, Synth. Met., 1983, 7, 185; see also Intercalation Chemistry, ed. M. S . Whittingham and A. J. Jacobsen (Academic Press, New York, 1982). Phys., 1978, 69, 5089. Phys. Lett., 1978, 33, 18. l5 J. R. Krawczyk and A. G. MacDiarmid, unpublished results (1 983). l6 W. P. Su, J. R. Schreiffer and A. J. Heeger, Phys. Rev. Lett., 1979,42, 1698; Phys. Rev. B, 1980,22, 2099. (PAPER 4/5 12)

ISSN:0300-9599    

DOI:10.1039/F19858100105

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. SOC., Furaday Trans. I, 1985, 81, 1 13-1 16 Carbon Dioxide-mediated Decomposition of Hydrogen Peroxide in Alkaline Solutionsf- BY LASZLO J. CSANYI* AND ZOLTAN M. GALBACS Institute of Inorganic and Analytical Chemistry, A. Jozsef University, P.O. Box 440, 6701 Szeged, Hungary Received 2nd April, 1984 Rapid decomposition of hydrogen peroxide in aerated alkaline solution is induced by carbon dioxide when transition-metal impurities are transformed from their polynuclear oxohydroxo complexes into less aggregated carbonato complexes. In connection with the base-induced decomposition of hydrogen peroxide, an interesting observation has recently been described by Navarro et al.' They found that the rate of decomposition was enhanced when a continuous stream of CO, or air was passed through alkaline hydrogen peroxide solutions.In contrast to the effect of CO,, carbonate ions proved completely inactive. Further, it was observed that: 'The pH of unbuffered solutions containing H,O, initially adjusted to 12 decreased, as the peroxide was decomposing, at a higher rate than that of similarly treated solutions without H,O, in which the pH decrease was only due to CO, contained in the air being passed through the solutions. ' In the presence of CO, the rate of decomposition varied according to a maximum curve as a function of the pH, with a maximum between pH 11.5 and 11.7, just as in the absence of CO,. This interesting observation was explained by Navarro et al.' in terms of the formation of peroxycarbonic acid : CO, + H,O, + H,CO, H,CO, + HO; -+ 0, + H,O + HCO, .(1) (2) which interacts with HO, anions, the following oxidation-reduction reaction taking place : Since the hydrogen carbonate ion is a much weaker base than OH-, it follows that the pH of the solution decreases as the decomposition progresses. This pH change is in contrast to the change observed during the autodecomposition of alkaline hydrogen peroxide, where the pH increases because of the removal of the slightly acidic H,O, molecules (pK, 11.7). The mechanism advanced by Navarro et u1.l cannot be supported by direct experimental evidence and can be questioned for the following reasons. (i) Peroxy derivatives of carbonic acid have been described many times, but nothing exact is known about their compositions. At present it cannot be stated definitely whether they are true peroxy compounds containing -0OH group(s) or whether they contain active oxygen in the form of hydrogen peroxide of crystallization. In a recent paper Carrondo et al., reported that in a solid peroxycarbonate the H,O, was hydrogen-bonded and was released instantly when the compound was dissolved in t A comment on the paper of the same title by Navarro et u/.* 5 113 1 A K I114 C02-MEDIATED DECOMPOSITION OF H2 Table 1.Stoichiometry of the C0,-mediated decomposition of hydrogen peroxide according to the data of fig. 4 of ref. ( 1 ) time 0, evolved CO, introduced" AQ /min / 1 O6 mol dm-3 / lo6 mol dm-3 ACO, 0 0.0 5 28.95 10 65.30 20 87.80 30 96.10 0.0 6.13 12.25 24.52 36.77 - 4.7 5.3 3.6 2.6 a The average CO, content of the air was taken as 0.03 vol% and the flow rate was 0.1 dm3 min-l.water. However, the presence of the -0OH group activated by an acyl radical is of vital importance for the proposed redox reaction [step (2)]. (ii) The pK, values of aliphatic peroxycarboxylic acids are in the range 6 9 . If such a value is also valid for peroxycarbonic acid (if this exists), then the pH maximum of decomposition must be found in the range 8.8-10.3 rather than at pH 11.7 if the proposed mechanism operates. s - ~ ) , ~ but it becomes much faster in the presence of hydroxide ions (kCO,+OH- = 7.7 x lo3 dm3 mol-1 s - ~ ) . ~ By analogy, it is expected that perhydration with H20, (if it takes place) will also be a slow process, and consequently step (1) cannot contribute to the enhancement of the rate of decomposition of hydrogen peroxide.In contrast, the perhydration would occur quickly (if at all) with the aid of the strong nucleophile HO; and the mechanism should therefore be corrected as follows: (iii) The hydration of CO, is a slow process (kCOz+HzO = 3.0 x CO, + HO; + HCO; HCO, + H,02 -+ 0, + H 2 0 + HCO;. (iv) According to the mechanism advanced by Navarro et aL,l the stoichiometry of the C0,-mediated decomposition is A02/AC02 = 1 : 1. However, the data obtained by these authors (see table 1) do not support this: up to a hydrogen peroxide conversion of ca. 85% the ratio varies between 5 and 2.5. It follows that another explanation is required to interpret the interesting obser- vations of Navarro et aZ.l A simpler and more probable interpretation can be offered if it is assumed that the solution investigated is somehow contaminated or that the activities of catalysing impurities are increased when CO, is bubbled through alkaline hydrogen peroxide solutions.In the cited paper nothing was said about the means of introducing air into the solution. If the air was not thoroughly filtered before passage through the solution (by suction or pumping) then airborne particles (among them metal oxides) could contaminate the system considerably, with resulting enhancement of the rate of decomposition. On the other hand, it was mentioned that chemicals of reagent grade were used without further purification. In conjunction with this, rate coefficients of 6.8 x and 1.2 x dm3 mol-1 s-l can be calculated from the experimental data in fig. 2 of ref.(1) [N, and 0, atmospheres and (presumably) room temperature]. If we compare these values with the (5.6 & 0.6) x lo-' dm3 mol-l s-l determined recently in 0,-saturated solution at 303 K with purified reagents and the use of a sequesteringL. J. CSANYI AND z. M. GALBACS 115 Table 2. Rate of dioxygen evolutiona CO, injected rate /cm3 PH /lo7 mol dm-3 s-l (1) in the absence of iron(m) 0.00 11.74 1.83 0.80 11.67 1.83 1.60 11.65 2.06 2.30 11.60 1.54 0.00 11.92 4.99 0.75 11.87 9.62 1.54 11.82 10.26 2.16 11.79 8.98 0.00 12.22 5.50 0.56 12.20 21.15 1.15 12.10 23.99 1.69 12.00 38.49 (2) in the presence of 9.52 x mol dmP3 iron(II1) (3) in the presence of 47.60 x mol dmP3 iron(II1) a Conditions: 25.0 cm3 1.0 mol dm-3 NaOH (purified with magnesium hydroxide5 and partially converted into carbonate with ultrapure CO,) + 15.0 cm3 triply distilled water+2.0 cm3 0.516 mol dm-3 H,O,, 303 K.The volume of 0, evolved was measured at constant pressure with a simple automatic apparatus described elsewhere.6 agent in an attempt to remove the last traces of metal imp~rities,~ it can be concluded that the reagents used by Navarro et al. contained catalysing impurities in consider- able quantities. From the position of the rate maximum (pH ca. 11.7) it can be assumed that iron(m) was, as usual, the main ~ontaminant.~ To check the influence of impurities, the rate of decomposition of hydrogen peroxide was measured using purified reagent solutions both in the absence and in the presence of small quantities of CO, of ultrapure grade.Measurements were also made with reaction mixtures deliberately contaminated with iron(II1). The injected CO, diffused into the solution quickly and almost quantitatively and did not cause any change in the rate of decomposition in the absence of catalyst; however, it resulted in a rate increase when iron(Ir1) was present (table 2). The rate enhancement was observed not only during the initial dosing of CO,, as postulated by Navarro et aI.,l but afterwards, thereby pointing to a permanent change in the system. This rate enhancement can be explained as follows. At higher pH transition-metal ions, among them iron(III), form polynuclear oxohydroxo complexes, the catalytic activities of which depend strongly on the degree of aggregation.When CO,, an acidic reagent, is continuously introduced into alkaline hydrogen peroxide solution, disaggregation of the polynuclear metal complexes and metal hydrogen carbonate complex formation may set in, and as a consequence the rate of decomposition of hydrogen peroxide is increased. Such a disaggregation cannot be caused by carbonate ions, as found by Navarro et a/.' Another explanation may be considered here. Sodium hydroxide preparations always contain various amounts of magnesium ions, which form colloidal magnesium hydroxide with a highly active surface when dissolved in water. The colloid helps to reduce the catalytic activities of heavy-metal impurities pre~ent.~ Howevei , when CO, 5-2116 C02-MEDIATED DECOMPOSITION OF H2 is introduced into the solution, Mg(OH), is transformed into its hydrogen carbonate and dissolves, and at the same time its inhibitory effect is eliminated. Further, we refer to the experimental observations7? that the catalytic efficiencies of some transition-metal ions such as manganese, cobalt, nickel and copper are increased by many orders of magnitude when their aqua complexes are transformed into their carbonato complexes. J. A. Navarro, M. A. De La Rosa, M. Roncel and F. F. De La Rosa, J. Chem. SOC., Faraduy Trans. I , 1984,80, 249. M. A. A. F. de C. T. Carrondo, W. P. Griffith, D. P. Jones and A. C. Skapski, J. Chem. SOC., Dalton Trans., 1977, 2323. D. A. Palmer and R. VanEldik, Chem. Rev., 1983,83, 651. L. J. Csanyi, L. Nagy, Z. M. Galbacs and I. Horvath, 2. Phys. Chem. (Frankfurt am Main), in press. Z . M. Galbacs and L. J. Csanyi, J. Chem. SOC., Dalton Trans., 1983, 2353. L. J. Csanyi, Z. M. Galbacs and L. Nagy, J. Chem. SOC., Dalton Trans., 1982, 237. A. Ya. Sychev, V. J. Isac and Dao Van Lan, Zh. Fiz. Khim., 1977, 51, 363. A. Ya. Sychev, V. J. Isac and Dao Van Lan, Zh. Fiz. Khim., 1978, 52, 107. (PAPER 4/540)

ISSN:0300-9599    

DOI:10.1039/F19858100113

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1985, 81, 117-125 Studies of Electrical Double-layer Interactions in Concentrated Silica Sols by Small-angle Neutron Scattering BY JEFF PENFOLD AND JOHN D. F. RAMSAY* Neutron Division, Rutherford Appleton Laboratory, Chilton, Oxfordshire and Chemistry Division, AERE, Harwell, Oxfordshire Receiued 5th April, 1984 Effects of ionic strength, volume fraction and pH on the interparticle interactions in concentrated silica sols of different particle size have been investigated by small-angle neutron scattering. The results have been modelled using the solution of the mean spherical approximation (MSA) of Hayter and Penfold and Hansen and Hayter. In this model the pair potential comprises a hard core and a soft tail, due to the screened coulombic repulsion between the electrical double layers surrounding the spherical particles. Closed analytic solutions of the structure factor, S(Q), for this potential have been used in the fits to the experimental scattering data to obtain the surface charge and potential of the silica particles.Although the surface charge so obtained rises with increases in the pH of the sols, as expected, values are markedly less than the total surface charge density of silica as previously reported. There has recently been considerable interest in the determination of the effective pair potential resulting from the double-layer interaction between colloids in concen- trated dispersions. Particular progress has been made from the analysis of small-angle neutron scattering (SANS) from monodispersed latex particle^^-^ and surfactant micelles4 by modelling the behaviour of the dispersions as a one-component fluid of colloidal particle^.^ This approach, which has also been applied to dilute dispersions, has been based on established liquid-state models which involve either computer- simulation6? ’ or integral-equation methods.8> The latter approach, using the solution of the mean spherical approximation (MSA) as developed by Hayter et al.9*10 has recently been applied to experimental SANS data obtained with silica Results of the MSA analysis are described more fully in the present paper, which is an extension of our previous structural investigations12 and results obtained by modelling the systems using a hard-sphere (HS) p0tentia1.l~ An important feature which is demonstrated here is the effect of the pH of the silica sols on the small-angle scattering behaviour and the corresponding effective surface charge obtained using the MSA model.EXPERIMENTAL MATERIALS Concentrated silica sols of different particle size (Ludox SM, HS and TM) were obtained commercially (E. I. du Pont de Nemours & Co.) from designated sample batches (S2, S3 and S4) which have been studied previously.l29 l 3 These stock sols were dialysed repeatedly against dilute NaNO, solutions of different ionic strength and controlled pH as already described. To reduce the background incoherent neutron scattering from water, the sols were prepared in deuterium oxide ( > 99 D,O). Particle-size distributions, determined by transmission electron 117118 DOUBLE-LAYER INTERACTIONS IN SILICA SOLS microscopy,12 showed discrete, almost spherical, particles having mean diameters, d,, of 12, 17 and 30 nm for S2, S3 and S4, respectively (standard deviation o z 0.2).SMALL-ANGLE NEUTRON SCATTERING Measurements were made as previo~slyl~ at a wavelength, A, of 10 A on sol samples contained in silica cells (path length 2 mm) using a multidetector instrument14 installed in the PLUTO reactor at AERE, Harwell. Data were analysed using standard programs to normalise counter efficiencies and to correct for sample self-absorption and incoherent background. Absolute scattered intensities, expressed as macroscopic cross-sections, [dZ/di2],,,, were obtained using a light-water standard. MICROELECTROPHORESIS Measurements of electrophoretic mobility, u, were made on dilute (< g cmP3) S3 sols with a laser-electrophoresis instrument of similar design to that described15 previously.The corresponding zeta potentials, [, were calculated using the Huckel equation (viz. u = 3&[/2q; for KR < 1). DATA TREATMENT AND ANALYSIS NEUTRON SCATTERING The coherent macroscopic scattering cross-section for a concentrated colloidal dispersion of identical particles is given by16 where Q is the scattering vector, defined as for a scattering angle 28 and wavelength A, pp and ps are, respectively, the mean scattering length densities of the particles and solvent, & is the volume of each particle, np is the particle number density and P(Q) is the particle form factor, which for spheres of radius R is given by 3[sin (QR) - QR cos (QR)] Q3R3 P(Q) = ( (3) The structure factor, S(Q), is determined by the nature of the particle interaction potential; for non-interacting systems S(Q) = 1 .The spatial distribution of the particles as a function of the mean interparticle separation, r, is given by the particle pair-distribution function g(r) and is related to S(Q) by the Fourier transform as already illustrated12 for the silica sols considered here. DATA FITTING The MSA model gives a closed analytic solution to S(Q) as a function of several parameters of the system defining the interparticle potential : (volume fraction, d p , charge, 0, and screening, K ) . This function is calculated as has been described previo~sly,~~ lo and the product KP(Q)S(Q), where K contains the terms remaining in eqn (l), is fitted to the experimental data.The form of the potential, U(r), between the spherical particles is defined by a hard-J. PENFOLD AND J. D . F. RAMSAY 119 core potential and the standard Coulombic potential due to the mutual interaction of their electric double layers :17 U(r) = co, r < 2R U(r) = 4 n ~ , ER*I,Y; exp [ - ~ ( ~ - 2 R ) l / r , r > 2R where I,Y, is the surface potential, E is the dielectric constant of the solvent, E , is the permittivity of vacuum and IC is the Debye-Hiickel inverse screening length, which is given by where ci is the ionic strength of the solution and N is Avogadro’s number. The surface charge on the particle zp is related to yo by the approximation y / , = Zp/47tEEo R( 1 + KR).(8) By inputting K and &,, experimental data are fitted by least-squares refinement to an arbitrary intensity scale given by the calculated product of KP(Q)S(Q) to obtain 0, the surface charge density. The factor, YSCAL, by which the absolute theory must be multiplied to match the absolutely scaled data, is then calculated. In view of systematic errors and uncertainties in cross-sections, fits are considered satisfactory1* if this ratio is within 30% of unity. Here we have taken psio2 and pDZ0 as 3.49 x loplo cmP2 and 6.34 x cm-*, respectively, and [dC/dR],2, = 0.514 cm-l sr-l, and have obtained values of YSCAL which were consistently in the range ca. 0.8-> 0.9. RESULTS AND DISCUSSION EFFECTS OF SOL CONCENTRATION AND IONIC STRENGTH The dependence of scattered neutron intensity, I@), on momentum transfer, Q, for S3 silica sols, covering a range of concentration, which have been dialysed against electrolytes of two different ionic strengths (lop4 and 5 x lop3 mol dmP3) but similar pH (ca.8), are shown in fig. 1 and 2, respectively. The development of the maxima in I(Q) is caused by interference effects and indicates that the particles are not arranged at random but have some short-range ordering due to interparticle repulsion. Thus the movement of the maxima to higher values of Q with increasing sol concentration reflects a reduction in the equilibrium separation distance, rg(r)max, between the particles as already discussed.l* Values of rg(r)max in table 1 are those derived by Fourier-transforming [cf.eqn (4)] the fitted S(Q) obtained with the MSA model. It would seem from the similarities in scattering behaviour shown in fig. 1 and 2 that S(Q), and hence the interparticle potential, is almost unaffected by changes in the ionic strength of the dialysing electrolyte, cf. This at first sight would seem surprising because it is well known17 that the screening of the electrostatic interaction between colloidal particles is controlled by the ionic strength of the supporting electrolyte, ci. In the present situation, however, cf and ci may differ significantly due to the Donnan effect, which results from the establishment of osmotic equilibrium across the dialysing membrane, as has recently been discussed.3 Thus if we consider the colloidal particles as macroions of charge zp, then to preserve electroneutrality there will be an additional compensating counter-ion contribution to the bulk - electrolyte such that where zi is the counter-ion charge.Nc, zi 2 Ncf zi + np zp (9)120 DOUBLE-LAYER INTERACTIONS IN SILICA SOLS 100 3 I I Fig. 1. Small-angle neutron scattering of S3 sols of different concentrations dialysed against mol dm-3 sodium nitrate solution: 0, 0.081; 0, 0.175 and 0, 0.384 g ~ m - ~ ; continuous lines are the theoretical fits to the data. 0 0.025 0 0.025 0 0.025 0.05 Q1A-l Fig. 2. Small-angle neutron scattering of S3 sols of different concentrations dialysed against 5 x mol dm-3sodiumnitratesolution: .,0.137; .,0.266and +,0.550 g ~m-~;continuous lines are the theoretical fits to the data. It can be seen that the difference between ci and cf will increase as the particle number concentration, np, increases, and will hence be particularly marked where the particles are of small size and have a high volume fraction, as occurs here.Thus it can be shown that for a colloidal dispersion containing particles with radius R, volume fraction q$,J. PENFOLD AND J. D. F. RAMSAY 121 Table 1. Results from fitting the MSA model to SANS data obtained on silica sols of different concentrations, pH ca. 8 sol conc./g cmP3 cflmol dm-3 c;/mol dm-3 2R/nm a/pC cmP2 cy,/mV ~ ~ ( ~ ) ~ ~ ~ / n s 3 0.08 1 ca. loP4 5 x 20.0 0.346 28.2 45 s 3 0.175 ca. lop4 1.5 x 19.2 0.452 27.7 33 s 3 0.384 ca. lop4 4 . 6 ~ loP3 17.9 0.613 25.2 24 s 3 0.137 5 x 7 x 19.7 1.01 28.0 35 s 3 0.266 5 x 10-3 9 x lop3 18.6 0.765 26.0 27 s 3 0.550 5 x 1 .4 ~ 10P 16.7 0.662 18.5 21 with an effective surface density of charge, a, corresponding to the diffuse part of the electrical double layer, we have (10) where 4, = (4/3)nR"n,. Thus for a 1 : 1 electrolyte the effective counter-ion concentration, c; = [ci/( 1 -&)I, within the interstitial solvent volume between particles (rather than the overall dispersion volume) which will determine the screening parameter, IC, is then given by np zp = 3q4, a/ R c; = (cf + 3.12& all?)/( 1 - f$p) where, expressed in dimensionless units, we have c;/mol dmP3, R / A and a/pC cm-2. In fig. I and 2 the fits to the experimental data make allowance for an increase in K resulting from the additional counter-ion concentration, c;.Values of the fitted parameters are given in table 1. These show that R is slightly larger, although in satisfactory accord with that determined by electron microscopy, and that the effective surface potential, yo, is approximately half the zeta potential as measured by electrophoresis (see fig. 4 later) on much more dilute S3 sols under similar pH conditions. Values of the effective surface charge, a, are furthermore markedly lower than those obtained by conductimetric titration19 ( > 2 pC cm-2). It is also evident that values of c( for the two sets of sols, despite having been dialysed at markedly different values of ci, are similar. This would thus explain the similar scattering behaviour observed at comparable sol concentrations and the apparent insensitivity of the interaction potential to the ionic strength of the supporting electrolyte.The latter feature will be particularly evident as the particle size is decreased and thus may explain why maxima in S(Q) for silica and ceria sols, containing exceptionally small particles ( R < 100 A), do not increase markedly with &,13 as is observed2 with latex dispersions with a considerably greater particle diameter (2 300 A). Nevertheless there is evidence that S(Q) does indeed become less sensitive to K with increasing sol concentrations. This is demonstrated in fig. 3, which shows scattering data for a sol of high concentration (dialysed mol dmp3) and fits using the MSA model with fixed values of K corresponding to ionic strengths of lop4, and lop2 mol dm-3.Thus it is evident that all the fittings are similar in the region of the maximum in I(Q) and only show marked discrepancies in the low-Q limit, which is outside the range of measurement with the present instrument. Furthermore, using unrealistically high and low values of K , fits can be achieved by relatively marginal changes in the fitted surface charge. This would suggest that at high volume fractions the form of S(Q) is much more sensitive to a than K .122 DOUBLE-LAYER INTERACTIONS IN SILICA SOLS Fig. 3. Theoretical fits to scattering data I(Q) for silica sol S3 (0.38 g ~ r n - ~ ) dialysed against NaNO,; c; = and mol drn-3. K is fixed and corresponds to ionic strengths of mol drn-3 for (a), (6) and (c) giving a/pC cm-2 of 0.42, 0.44 and 0.83 respectively.Fig. 4. 50 - - I n - -80 E E - > E . - -40 I I I I 0 10 --. ; - 5 : 3 7 25 2 4 6 8 PH Dependence of electrophoretic mobility, u, and zeta potential, [, on pH (< g ~ r n - ~ ) silica sol S3 (“a+] z lop4 mol dmp3). I n I m --. 5 3 0. I lo 10 for a dilute EFFECTS OF pH It is well known20 that the surface of silica has a negative charge which increases gradually with rising pH, as is demonstrated in fig. 4 by the electrophoresis results obtained with dilute S3 sols. Changes in the scattering behaviour of concentrated S3 sols shown in fig. 5 reflect the difference in surface charge which results from dialysing at pH values of 7.2 and 4.5. Thus the sharper peak obtained at the higher pH indicates a stronger repulsive interaction, which arises from a greater effective surface charge, as is confirmed by the 0 values obtained from the fits given in table 2.The marked effect of surface charge on S(Q) is also demonstrated (fig. 6) by the scattering of S2 sols. Because these have a smaller particle size it is possible, with the same accessible range at small Q, to examine effects at lower volume fraction. It is then evident that differences in ionic strengths of the dialysing electrolyte produce significant changes in the S(Q) of sols of similar pH, which leads to a correspondingJ. PENFOLD AND J. D. F. RAMSAY I23 100 - I L. n ... I f 2 50 5 , w =3 ( b ) 1 I 0.03 0 0.03 0.06 QiA Fig. 5. I(Q) plotted against Q for S3 sols of similar concentrations (ca. 0.19 g cmP3) but different pH, ( a ) 7.2 and (b) 4.5, both dialysed against loP4 mol dmP3 sodium nitrate.-A 0.03 Fig. 6. Small-angle scattering of S2 sols of different pH dialysed against sodium nitrate of concentration (a) 5 x loP3 and (b) loP4 mol dmP3. Concentrations/g and pH of sols are, respectively: 0, 0.129 and 10.0, 0, 0.157 and 5.6, 17, 0.092 and 7.5 and ., 0.097 and 4.1.124 DOUBLE-LAYER INTERACTIONS IN SILICA SOLS Table 2. Results from MSA model fits to SANS data obtained on silica sols of different pH sol conc./g ci/mol dm-3 pH c;/mol dm-3 2R/nm o/pC cmP2 y/,/mV s3 0.197 ca. lop4 4.5 6 x 18.7 0.181 13.7 s 3 0.182 ca. 7.2 1 . 2 ~ 18.9 0.352 22.9 s 2 0.092 ca. 7.5 1 x 10-3 14.8 0.357 21.4 s2 0.097 ca. lop4 4.1 5~ 10-4 15.3 0.177 11.9 s 2 0.05 1 5~ 10-3 10.0 6 x 15.2 0.886 32.8 s2 0.129 5~ 10-3 10.0 9~ 10-3 13.9 1.32 40.9 s 4 0.314 ca.5.0 1 x 10-3 27.3 0.226 18.1 difference in the fitted 0 value (cf. table 2). However, the most striking feature is the complete absence of a maximum in I(Q) in fig. 6(a), which demonstrates that the interparticle potential is very weak at pH 5 5 owing to the low charge and short screening distance, K . CONCLUSION Measurements of interaction potential between silica-sol particles obtained from the MSA model reported here clearly reflect the effects of changing pH on the zeta potential as determined independently by laser electrophoresis. Zeta potentials are nevertheless significantly larger, a feature which may arise because measurements are of necessity restricted to much less concentrated sols. The considerably smaller effective surface charges obtained using the MSA model and the total surface charge as measured by conductimetric titration can be ascribed to the adsorption of counter-ions (Na+) at the silica surface.This feature, which has already been noted to a lesser extent with latex particles,l9 may arise because the silica particles have a diffuse surface with a high adsorption capacity for counter-ions.21* 22 This diffuseness may also be responsible for the slight, but significant, increase in the value of R, obtained from the MSA fits as the sol concentration is decreased compared with the particle size determined by electron microscopy. Another possible reason for the discrepancy between surface potentials obtained from model fits and other experimental measurements which cannot be overlooked concerns the assumptions inherent in the MSA model, in particular the use of the Debye-Huckel potential to describe double-layer interactions in concentrated dispersions.Consequently while the application of the Hansen-Hayter MSA form for S(Q) is a useful approach for displaying trends in potential parameters as shown here, it may have limitations in providing absolute values. Further verification of the model may be obtained in the future, if SANS measurements are extended to a much lower Q than accessible here, since it is now well known from liquid theory that details of the interaction potential are contained in S(Q) particularly in this range and that the form of S(Q) at higher Q represents the ‘softness’ of the repulsive core of the pair potential. We are indebted to Mr L.Benest for making electrophoretic mobility measurements and gratefully acknowledge helpful discussions with Prof. R. H. Ottewill and Dr J. B. Hayter.J. PENFOLD AND J. D. F. RAMSAY 125 K. Alexander, D. J. Cebula, J. W. Goodwin, R. H. Ottewill and A. Parentich, Colloids SurJ, 1983, 7, 233. D. J. Cebula, J. W. Goodwin, G. C. Jeffrey, R. H. Ottewill, A. Parentich and R. A. kchardson, Faraday Discuss. Chem. Soc., 1983, 76, 37. B. Beresford-Smith and D. Y. C . Chan, Faraday Discuss. Chem. Soc., 1983, 76, 65. J. B. Hayter, Faraday Discuss. Chem. Soc., 1983, 76, 7. W. van Megen and I. Snook, J. Chem. Phys., 1977,66, 813. ’ E. Dickinson, Faraday Discuss. Chem. Soc., 1978, 65, 127. D. W. Shaeffer, J. Chem. Ph-ys., 1977, 66, 3980. J. B. Hayter and J. Penfold, Mol. Phys., 1981, 42, 109. l o J. P. Hansen and J. B. Hayter, Mol. Phys., 1982, 46, 651. J. D. F. Ramsay, Faraday Discuss. Chem. Soc., 1983, 76, 108. l 2 J. D. F. Ramsay and B. 0. Booth, J . Chem. Soc., Faraday Trans. 1, 1983, 79, 173. l 3 J. D. F. Ramsay, R. G. Avery and L. Benest, Faraday Discuss. Chem. Soc., 1983, 76, 52. l 4 D. I. Page, Atomic Energy Res. Estab. Rep. (AERE-R 9878, 1980). l 5 A. W. Preece and N. P. Luckman, Phys. Med. Biol., 1981, 26, 11. l 6 B. Jacrot, Rep. Prog. Phys., 1976, 39, 91 1. * J. B. Hayter and J. Penfold, J . Chem. Soc., Faraday Trans. I , 1981, 77, 1851. E. J. W. Verwey and J. Th. G. Overbeek, Theory of the Stability of Lyophobic Colloids (Elsevier, Amsterdam, 1948). l 8 J. B. Hayter and J. Penfold, Colloid Pofm. Sci., 1983, 261, 1022. l9 G. H. Bolt, J . Phys. Chem., 1957, 61, 1166. *O R. J. Hunter, Zeta Potential in Colloid Science (Academic Press, London, 1981). 21 J. Lyklema, J . Electrounal. Chem., 1968, 18, 341. 22 J. W. Perram, J. Chem. Soc.. Faraday Trans. 2, 1973, 69, 993. (PAPER 4/561)

ISSN:0300-9599    

DOI:10.1039/F19858100117

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1985, 81, 127-136 1 7 0 Nuclear Magnetic Resonance Study of the Rotational Motion of the Sulphate Ion in Aqueous Solution B Y YUICH MASUDA, MITSURU SANOt AND HIDEO YAMATERA* Department of Chemistry, Faculty of Science, Nagoya University, Nagoya 464, Japan Received 17th April, 1984 Linewidths of 1 7 0 n.m.r. spectra of the sulphate ion in D,O solutions have been measured at temperatures ranging from 4 to 70 "C. The dependence of the linewidth on concentration has been investigated in various alkali-metal and tetra-alkylammonium sulphate solutions. The electric-field gradient at the oxygen site of an isolated sulphate ion has been calculated using the ab initio molecular-orbital method. This gave a field gradient of 1.52 atomic unit for the component along the 0-S bond.The field gradient at the sulphate oxygen caused by a nearby single positive charge has also been calculated. This calculation showed that the fluctuation of the field gradient caused by the movement of the surrounding ions and water molecules is of secondary importance in its effect on the 1 7 0 linewidth of the sulphate ion in aqueous solutions. The rotational correlation times of the sulphate ion, z, in solution have been obtained from the measured linewidths and the calculated field gradient (1.52 a.u.). A plot of the rotational correlation time at infinite dilution, z(O), against the ratio of viscosity to temperature (q/7') curved downward at low temperatures. z(0) values at various temperatures are well represented by the Arrhenius relation with an activation energy of 13 kJ mol-l for rotation of the sulphate ion.The rotational correlation time of the sulphate ion increased with increasing salt concentration and showed a good correlation with the decrease in mobility of water molecules in the hydration sphere of the cation contained in the solution. An investigation of the motions of ions in solution gives useful information for understanding the dynamic structure of electrolyte solutions, particularly the dynamic features of solvation and ion-ion interactions. The translational motion of various ions has been investigated by measuring diffusion coefficients1 and electric conductivities.2* A number of theoretical approaches to the translational motion of ions have also been p r ~ p o s e d .~ However, the rotational motion of ions has been studied less extensively.: Studies of the rotational motion of common inorganic ions are limited to a few cases, i.e. those of the nitrate5* and thiocyanate6 ions by Raman scattering5 and depolarized Rayleigh light scattering.6 In aqueous solutions containing these ions, the effects of solvation and ion-ion interaction on the rotational motion of the ions are complicated, and no quantitative relationship has been found between these interactions and the rotational motion of the ions. Nevertheless the rotational motion of common inorganic ions in solution must be elucidated in order to understand the dynamic properties of electrolyte solutions. This paper is concerned with the rotational motion of the sulphate ion in solution. t Present address : Laboratory of Inorganic Chemistry, College of General Education, Nagoya IJniversity .1 One of the reasons is the limitation in the experimental technique. For example, the methods of Raman scattering and depolarized Rayleigh scattering cannot be applied to the rotational motion of isotropic ions such as sulphate and perchlorate ions. 127128 1 7 0 N.M.R. STUDY OF THE SULPHATE ION This is one of the simplest and most highly symmetrical ions whose rotational motion can be observed. The measurement of the 170 nuclear relaxation or the 170 linewidth is a useful technique for investigating the rotational motion of the sulphate ion in solution. The magnetic relaxation of a nucleus with I > 1/2 is usually caused by the interaction of its electric quadrupole moment with the electric-field gradient at the nuclear site.The linewidth of the 170 signal, which is equal to (z T,)-l (T, being the spin-spin relaxation time), is determined by the magnitude of the electric-field gradient at the 170 nucleus and the relaxation time of the gradient's fluctuation. At the extreme-narrowing limit, where T, = & (& being the spin-lattice relaxation time), the linewidth (vlj2) of the 170 signal of the sulphate ion is related to its rotational correlation time (z) as follows:7 where I, a and e2qQ/h are the spin of 1 7 0 ( I = 5/2), asymmetry parameter and so-called quadrupole coupling constant, respectively. The parameters eQ and eq are the quadrupole moment of the 170 nucleus (Q = -0.0265 x cm2)8 and the electric- field gradient at the 170 nucleus along the 0-S direction, respectively.In aqueous sulphate solutions other ions and water molecules exist near the sulphate ion and their movement produces fluctuations in the electric-field gradient at the oxygen nucleus of the sulphate ion. This field fluctuation can be the origin of the relaxation of the 170 nucleus. Then additional terms representing the effects of the surrounding ions and solvent molecules have to be added to eqn (1). The quadrupole relaxation of the nucleus of a monoatomic ion such as the halide and alkali-metal ions is known to be mostly caused by the field fluctuation due to the motion of surrounding ions and solvent molecules, and a detailed formulation for such a quadrupole relaxation has been given by Hertz et aL9910 Although ions and solvent molecules surrounding the nucleus in question generally affect the linewidth (or &) of the nucleus in a complicated manner, their effect is of secondary importance in the 170 relaxation of the sulphate ion; the field gradient at the 1 7 0 nucleus of the sulphate ion is mainly caused by the 0-S bond, and the contribution of the surrounding ions and water molecules to the field gradient is less important.This was confirmed by rough estimates of the effect of the ions and solvent molecules on the linewidth (or q). The field gradient at the oxygen site in the presence of ions or solvent molecules was calculated using an ab initio molecular-orbital method which had been reported by Engstrom et a/." in the analysis of the quadrupole relaxation of the nuclei of the chloride, lithium and sodium ions in aqueous solutions. Nuclear magnetic relaxation experiments were carried out with relatively dilute aqueous sulphate solutions [(3 x 10-1)-10-2 mol dm-3] containing various univalent cations to obtain information on solvation and ion-ion interactions. EXPERIMENT MATERIALS The sulphate salts enriched in 170 were prepared as follows. A weighed portion of "0-enriched water (35 atom %, Prochem Co.Ltd) was added to an equivalent amount of solid sulphur trioxide in a dry nitrogen stream. The sulphuric acid produced was neutralized by 2 mol dm-3 sodium hydroxide with cooling. This solution was dried under reduced pressure at room temperature and the Na,S60,l70 produced was recrystallized from water.Addition of calcium chloride to a portion of the solution of the 170-enriched sodium sulphate gave "0-enriched calcium sulphate which was converted into various alkali-metal sulphates by an ion-exchangeY. MASUDA, M. SANO AND H. YAMATERA 129 technique. To another portion of the 170-enriched sodium sulphate solution silver nitrate was added in excess to result in the formation of 170-enriched silver sulphate, from which a series of tetra-alkylammonium sulphates were obtained by double decomposition with tetra- a1 kylammonium chlorides. 1 7 0 N.M.R. MEASUREMENTS The 1 7 0 n.m.r. spectra were obtained on a Jeol Fx-60 Fourier-transform spectrometer equipped with a special probe for multinuclear measurements operating at 8.1 MHz.The measurements were performed with a 9.5 mm 0.d. spherical sample tube. The temperature was maintained at 28 f 0 . 3 "C during the measurement unless otherwise stated. The sample solutions for measuring the n.m.r. spectra were prepared as follows. Each sulphate was dissolved in D,O and dried; this procedure was repeated. A spectrum of 4000 Hz width and 4096 data points was used to acquire free-induction decay (f.i.d.). The repetition time of the pulse was 0.1 s. The window function used in Fourier transformation was selected so as to have a negligible effect on the 1 7 0 linewidth. A good spectrum was obtained after accumulating ca. lo6 free- induction-decay signals when a 0.02 mol dm-3 sulphate solution was used. (The content of 1 7 0 was 6% of the total sulphate oxygen atoms.) Several sample solutions were measured five times. The linewidths were reproduced to & 5 % .In order to confirm the relation vl,, = (nT1)-l, the 7; values were also measured by the inversion-recovery method12 for several sample solutions of concentrations >0.1 mo1/(55.5 mol D,O). The observed values for each sample solution satisfied the relation. and ab initio MOLECULAR-ORBITAL CALCULATION Calculations of the LCAO MO SCF type were performed using the MIDI Gaussian-type function sets given by Huzinaga et al.l3 as the basis sets. For sulphur a 3d orbital with an orbital exponent of 0.46 was added to the basis set. The electric-field gradient at the oxygen nucleus was calculated for an isolated sulphate ion (S-0 = 1.488 A14) and for a sulphate ion with a nearby single positive charge using the program package JAMOL~ of Kashiwagi et al.l5 RESULTS AND DISCUSSION DETERMINATION OF THE ROTATIONAL CORRELATION TIME OF THE SULPHATE ION IN AQUEOUS SOLUTION Nuclear quadrupole interaction is the dominant mechanism in the relaxation of the l7O nucleus ( I = 5/2). At the extreme-narrowing limit the linewidth of the 170 n.m.r. signal of the 'isolated' sulphate ion is expressed by eqn (1). The electric-field gradient at the oxygen nucleus of the isolated sulphate ion [eg in eqn (l)] was calculated by an ah initio molecular-orbital method. The calculated electric-field gradient along the 0-S direction was 1.52 a.u.,$ which gave 9.4 MHz (a = 0) for the quadrupole coupling constant (e2qQ) of 170 in the isolated sulphate ion.This value is slighly larger than that in sulphones and approximately equal to that in sulphoxides;17 it is also larger than that in the phosphate ion.18 The surrounding ions and water molecules also cause the field gradient at the 170 nucleus of the sulphate ion in aqueous solution. We estimated these contributions to the 1 7 0 linewidth using the ab initio molecular-orbital method. Here we outline our method of estimating the effect of a neighbouring univalent cation and that of the surrounding water molecules on the 170 linewidth. Details are given in the Appendix. The molecular-orbital calculation was carried out for a system containing a sulphate t Although the magnitude of electric-field gradient at the sulphate-oxygen site in crystals can be determined by measuring the nuclear quadrupole resonance, it does not necessarily represent the magnitude in solutions.'fi $ 1 a.u.= 9.7175 x 10" V mPd.130 1 7 0 N.M.R. STUDY OF THE SULPHATE ION A\ Ob Ob Ob Ob Ob ( A) (B) Fig. 1. Possible configurations of the system consisting of a sulphate ion and a single positive charge, assumed for ab initio molecular-orbital calculations. ion and a cation,? and the interaction between the sulphate ion and the univalent cation was assumed to be purely electrostatic, i.e. the univalent cation was regarded as a single positive point charge. The same assumption has been found to be reasonable in estimating the quadrupole relaxation of halide and alkali-metal ions.ll The field gradient at the oxygen nucleus was calculated for the system consisting of a sulphate ion and a single positive charge with configurations as shown in fig.1. Table 1 lists the results, which show that the increase in the 170 linewidth caused by the presence of a nearby cation is ca. 3% for rop = 1.8 A (the distance between the sulphate oxygen and a lithium ion in contact with each other). This increase is smaller than the experimental error in the measurement of the 170 1inewidth.S Next, we consider the effect of hydration on the field gradient at the sulphate-oxygen nucleus. We consider only the electrostatic interaction due to fractional positive charges on the hydrogen atoms of the nearest water molecules, because the contri- butions to the 170 linewidth of the negative charge on the oxygen atom of the nearest water molecule and of the electric dipoles of more distant water molecules are much smaller.ll The electrostatic effect of a water molecule on the field gradient at the sulphate oxygen was then estimated to be smaller than that of a nearby cation; no appreciable change can be expected in the 1 7 0 1inewidth.s The discussion described above leads to the conclusion that the value of eq at the 170 nucleus of the sulphate ion in solution can be approximated by that for an isolated sulphate ion, 1.52 a.u.In the following discussion we use this eq value to calculate the rotational correlation time from the observed v1/2 value uia eqn (1). THE ROTATIONAL CORRELATION TIME OF THE SULPHATE ION IN D,O AT INFINITE DILUTION The linewidth of the 170 resonance at infinite dilution, V~,~(O), was determined by extrapolating the linewidths of K2S60,170 solutions at various concentrations to zero concentration at each temperature.With this value of V ~ , ~ ( O ) , eqn (1) gives the t The interactions of a sulphate ion with more than one cation and with another sulphate ion are disregarded in the present study, where the n.m.r. measurements were carried out with relatively dilute solutions. 1 The increase in the field gradient at the oxygen site caused by a nearby charge (rap = 1.8 A) was also calculated using the Sternheimer shielding constant ( y m ) for oxygen in Mg0.19 The increase in the linewidth predicted from this calculation is ca. 2%. This is even smaller than the increase estimated in the text. 5 The quantum-mechanical effect of the hydrogen bond on the field gradient at the sulphate-oxygen nucleus can also be ignored, because it has been shown for the chloride (35Cl), sodium (23Na) and lithium ('Li) ions that the quantum-mechanical effect is smaller than the electrostatic one and that both effects show opposite signs to each other at normal hydrogen-bond distances."Y. MASUDA, M.SANO AND H. YAMATERA 131 Table 1. Calculated values of eq [in eqn (I)] and ego, [in eqn (A I)] for various configurations of the SO!-- + e system 1.6 1.7 1.8 2.0 2.2 0" 1.6 1.7 1.8 2.0 2.2 me Oa 0, Oa Ob Oa Ob Oa 0, Oa Ob 'a, b Oa Ob Oa Ob Oa 0, O a Oa 'a, b Ob Ob configuration (A)" 1.61 0.0 1.52 0.1 1 1.57 0.0 1.52 0.09 1.55 0.0 1.52 0.09 1.54 0.0 1.52 0.05 1.52 0.0 1.52 0.03 1.52 0.0 configuration (B)" 1.57 0.02 1.52 0.03 1.55 0.02 1.52 0.01 1.54 0.0 1 1.52 0.02 1.53 0.0 1 1.52 0.02 1.52 0.01 1.52 0.0 1 1.52 0.0 0.08 0.06 0.04 0.02 0.01 - - - - - - 0.10 0.08 0.06 0.03 0.02 __ - - - - 0.0 0.0 0.0 0.0 0.0 0.80 0.84 0.86 0.86 0.86 " See fig.1. The main axis coincides with the 0-S vector. Asymmetry parameter. The This case corresponds to an isolated sulphate ion, where main axis is taken along rap. eq = ego,. rotational correlation time at infinite dilution, z(0). We first consider the rotational correlation time on the basis of the hydrodynamic model; the rotational correlation time often follows a semi-empirical relation : z O z = C Q / k T + z 0 ( 2 ) where q is the viscosity of the solution, I/ is the molecular volume, k is Boltzmann's constant, T is the absolute temperature, C is an experimentally determined dimen- sionless parameter which is concerned with the shape of the rotating molecule and the hydrodynamic boundary conditions and z, is the zero-viscosity intercept.Eqn (2) shows that z depends primarily on q / T ; the other parameters are not appreciably dependent on temperature. We previously reported that the temperature dependences of the rotational correlation times of [Co(en),I3+ (en = ethylendiamine)21 and of [M(phen),ln+ and [M(bpy),ln+ (phen = 1,lO-phenanthroline, bpy = 2,2'- bipyridine, M = RuI1 or C O I I I ) ~ ~ were well represented by eqn (2) at infinite dilution, and that this was attributed to the large size of the complex ions (as compared with the solvent molecules) and to weak ion-solvent interactions.In fig. 2 the rotational correlation time of the sulphate ion at infinite dilution, z(O), is plotted against q / T . The plot shows a non-linear relation at large q/T. This132 200 E =; 100 c P) - a 0 1 7 0 N.M.R. STUDY OF THE SULPHATE ION I I I I 2 4 6 8 8.0 6.0 v) a 4.0 2 0, 2 .o 0 (V/T)/ 1 0 -3 CP K -1 Fig. 2. Plot of the linewidth of the 1 7 0 signal (or the rotational correlation time) at infinite dilution against q/T. The broken line indicates the tangent at 28 "C (1 CP = kg m-l s-l). indicates that the increase in microviscosity around the sulphate ion with decreasing temperature is smaller than the corresponding increase in the viscosity of the bulk water, especially at low temperatures. In other words, the parameter C in eqn (2) decreases with decreasing temperature.Thus the hydrodynamic model, in which the solvent is regarded as a continuous medium, is not adequate. This is in contrast to previously reported results for the rotational motion of metal-complex We now consider the rotational correlation time of the hydrated sulphate ion from another point of view. We assume that the rotational motion of the sulphate ion is accompanied by the breaking of hydrogen bonds with neighbouring water molecules. If a mean activation energy, E,, can be defined for the rotation of the sulphate ion in aqueous solution, then the rotational correlation time is expressed by z = A exp(E,/RT) 22 where A is a constant, or In z = E,/RT+ In A. (3) Experimental values of In z are plotted against 1/T in fig. 3.The plot shows a better linearity than that of fig. 2, although a slight upward deviation is observed at large 1 / T. By assuming a linear relationship between In z and 1 / T we obtained 13 kJ mol-1 t for the activation energy. The activation energies for various phases of dynamic behaviour of water, such as viscous and rotational motion,24 increase with decreasing temperature. On the other hand, the activation energy for the rotational correlation time of the sulphate ion is almost independent of temperature in the range 4-70 "C. This indicates that the rotational motion of the sulphate ion is mainly influenced by short-range interactions with water molecules rather than the long-range interactions which are reflected by the properties of bulk water. This reasoning is consistent with the fact that the temperature dependence of z does not follow eqn (2), which is based on the hydrodynamic model.t This is slightly smaller than the value of 14 kJ m o P for the rotation of the C, axis of the nitrate ion5-26 .O -26.5 x s u C w -27.0 Y. MASUDA, M. SANO AND H. YAMATERA 133 A 1 I I i 3.0 3.2 3.4 3.6 Gt lo3 KIT Fig. 3. Plot of In [z(O)] against 1/T. ROTATIONAL MOTION OF THE SULPHATE ION IN D20 AT VARIOUS SALT CONCENTRATIONS The rotational correlation times of the sulphate ion for alkali-metal and tetra- alkylammonium sulphates in aqueous solution were obtained from the observed v1/2 values according to eqn (1). These results are shown in fig. 4. z for the sulphate ion increased with increasing concentration of the solution for lithium, sodium and tetra-alkylammonium sulphates, while it showed no change or a slight decrease for solutions of potassium, rubidium, caesium and ammonium sulphates.The cations of the former group are structure-making, while those of the latter group are structure- breaking. The tetrapropyl- and tetrabutyl-ammonium ions showed a large increase in z. These changes in z for the sulphate ion are probably caused by direct interaction, or by interaction through the water of hydration, with the cation. For a quantitative treatment of the changes in z we defined a parameter &(SO:-) by the following equation, assuming a linear relationship between z and the sulphate salt concentration : (4) where z/z(O) is the ratio of the rotational correlation time of the sulphate ion at a given concentration of sulphate salt to that at infinite dilution and cM,so4 is the concentration of the sulphate (in mol per 55.6 mol D,O).B,(SO:-) should reflect the dynamic properties of the cation itself or of its water of hydration. Hertz and coworkers estimated the rotational correlation times of water molecules in the hydration spheres of various cations.25 Fig. 5 shows a plot of z(H20, M+)/z(H,O), the ratio of the rotational correlation time of water molecules in the hydration sphere of the cation to that of pure water, against B,(SOi-). A linear relationship is observed. z for the sulphate ion increases in the presence of a cation having less mobile water molecules in its hydration sphere. This indicates that the rotational motion of the sulphate ion is subject to a greater restriction when the ion interacts with the less mobile water molecules bound to the cation.This view is consistent with the band profile of the Raman spectra of the sulphate2s in the presence z/z(O) = 1 + Br(SO:-) c M 2 s 0 4134 1 7 0 N.M.R. STUDY OF THE S U L P H A ~ ION 3 . 8 3.6 3 . 4 vl . a 3.2 3.0 L I / 0.1 0.2 0.3 ~ ~ ~ ~ , / ( r n o l per 55.6 mol D20) Fig. 4. Relationship between the rotational correlation time of the sulphate ion and the concentration of the sulphates of various univalent cations: 0, Pr,N+; e, , Bu,N+; 0, Et,N+; A, Li+; a, Na+; V, Me,N+; A, K+; 0, NHZ; 0, Rb+; 0, Cs+. 3 .O h 0, 2.0 z 0- s 1.0 \ - + E v 0 a Cs* (8 I Pr4N'(20) a 0 Bu~N' (25) a Li'(6) E t4N*(15) a Na*(6) Me4N* (10) I 1 1 1 -0.25 0 0.25 0.50 0.75 1.00 B, (SO:-) Fig.5. Plot of z(H,O, M+)/z(H,O) against B,(SOi-). The number in parentheses indicates the hydration number of each cation.25 of the lithium ion, where the v,(A) band has a shoulder on the higher-frequency side; this has been attributed to the sulphate ion interacting with water molecules under restraint in the coordination sphere of the lithium ion. Both the dependence of z(0) on temperature described previously and the dependence of z on the choice of cation described above show that the dynamic properties of theY. MASUDA, M. SANO AND H. YAMATERA 135 water molecules in contact with the sulphate ion, rather than those in the bulk, have an important effect on the rotational motion of the sulphate ion. The computations were carried out on a HITAC M-200H computer at the Computer Centre of the Institute for Molecular Science.APPENDIX CHANGE IN THE FIELD GRADIENT AT THE SULPHATE-OXYGEN NUCLEUS CAUSED BY A NEIGHBOURING CATION AND ITS EFFECT ON THE 1 7 0 LINEWIDTH In order to estimate the effect of a nearby univalent cation on the I7O linewidth of the sulphate ion the field gradient is divided into two parts: (i) the field gradient at the oxygen nucleus of an isolated sulphate ion and (ii) that at the same oxygen nucleus caused by a neighbouring cation. Part (ii) can be obtained by subtracting the field-gradient tensor for the isolated sulphate ion from that for the sulphate ion with a univalent cation nearby. We consider the following two extreme cases of dynamic correlation between the sulphate ion and the cation.CASE 1 This is the extreme of strong correlation between the motion of the sulphate ion and that of the cation. The configuration of the system is maintained for a time interval much longer than the rotational correlation time of the sulphate ion. Then the time constant for the fluctuation of the field gradient of part (ii) is equal to that of part (i), i.e. it is equal to the rotational correlation time of the sulphate ion. Consequently the fluctuation of the total field gradient, which is equal to the sum of parts (i) and (ii), depends only on the rotational correlation time of the sulphate ion (7). Then the linewidth of the 1 7 0 signal is given by eqn (1) with the eq value at the l70 nucleus in the presence of a cation and the rotational correlation time, 7, of the sulphate ion under the experimental conditions.CASE 2 This is the extreme of weak correlation. The motion of the sulphate ion is independent of that of the cation. Then the cross-correlation between the field gradients of parts (i) and (ii) need not be considered because their fluctuations are independent of each other; the decay of the cross-correlation function is infinitely fast. Then the vljZ value of the sulphate-oxygen nucleus with a neighbouring cation is expressed by where eqos and eq,, are the components of the field gradients of parts (i) and (ii), respectively, along their main axes, a‘ and a” are the asymmetry parameters and to, is the correlation time of the fluctuation of the field gradient of part (ii), which is smaller than 7.The actual dynamic features of a system consisting of a sulphate ion and a nearby univalent cation in solution are intermediate between the two extreme cases. For these two extreme cases we estimated the effect of the change in the field gradient at the sulphate-oxygen nucleus on the l7O linewidth; the field gradient at the oxygen site of the sulphate ion with a nearby single positive charge was calculated by the ab initio molecular-orbital method for two representative configurations with various distances between the sulphate-oxygen atom and the charge (fig. 1). The results of the calculations are listed in table 1 . The eq value calculated for 0, (see fig. 1) increased steeply with a decrease in the distance between the sulphate oxygen and the single positive charge (rcIp).Therefore the effect of a cation on the field gradient at the oxygen nucleus is largest when the sulphate ion is in contact with the smallest ion, i.e. that of lithium, for which rOp is 1.8-2.2 A.zi The increases in eq at rOp = 1.8 A were 2 and 1 % for configurations (A) and (B), respectively (table 1). For case 1 of the dynamic features of this system the contribution of this increase in eq to the “0 linewidth is estimated to be ca. 3 and 2”4, respectively, for configurations (A) and (B), considering that vljZ is proportional to the square of eq [see eqn (l)]. For case 2 increases of < 2% in the I7O linewidth are expected from eqn (A 1 ) with the eq,,, values for rop = 1.8 A. The contribution of hydration of the sulphate ion to the l70 linewidth was estimated in a136 1 7 0 N.M.R.STUDY OF THE SULPHATE ION manner similar to that of a nearby cation. Here we considered only the positive charges on the hydrogen atoms of the nearest water molecule. The distance between the sulphate-oxygen nucleus and the hydrogen atom of the water molecule is ca. 1.7 A28 (the hydrogen-bond distance) or slightly larger; this is smaller than that between the sulphate oxygen and the lithium ion in contact with each other. On the other hand, the hydrogen atoms of a water molecule have only fractional charges; e.g. +0.2e is assigned to the hydrogen atoms in the ST229 and BNS30 models. The effect of electrostatic interactions on the field gradient is proportional to the magnitude of the nearby charge and inversely proportional to the cube of the Thus a water molecule has a smaller effect than a nearby cation,t the effect of the lower charge overcoming that of the smaller distance.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 2a 29 30 31 32 R. A. Robinson and R. H. Stokes, Electrolyte Solutions (Butterworths, London, 1959). R. L. Kay, in Water, A Comprehensive Treatise, ed. F. Franks (Plenum, New York, 1973), vol. 3, chap. 4. N. Takisawa, J. Osugi and M. Nakahara, J. Phys. Chem., 1981,85, 3582; J. Chem. Phys., 1982,77, 4717; M. Nakahara, T. Torok, N. Takisawa and J. Osugi, J. Chem. Phys., 1982,76, 5145. J. Hubbard and J. Onsager, J . Chem. Phys., 1977, 67, 4850; J. Hubbard, J. Chem. Phys., 1978, 68, 1649; R. Zwanzig, J. Chem. Phys., 1963, 38, 1 103; 1970,52, 3625.D. James and R. L. Forst, Faraday Discuss. Chem. SOC., 1977, 64, 48; T. Kato, J. Umemura and T. Takenaka, Mol. Phys., 1978,36, 621. M. Whittle and J. H. R. Clarke, Mol. Phys., 1981, 44, 1435. A. Abragam, The Principles of Nuclear Magnetism (Oxford University Press, London, 1961), chap. 8. J. S. M. Harrey, Proc. R. SOC. London, Ser. A, 1965, 285, 581. H. G. Hertz, Ber. Bunsenges. Phys. Chem., 1973,77, 531; 688. H. G. Hertz, M. Holz, R. Klute, G. Stalidis and H. Versmold, Ber. Bunsenges. Phys. Chem., 1974,78, 24; H. G. Hertz, M. Holz, G. Keller, H. Versmold and C. Yoon, Ber. Bunsenges. Phys. Chem., 1974, 78, 493. S. Engstrom and B. Jonsson, Mol. Phys., 1981, 43, 1235; S. Engstrom, B. Jonsson and G. Jonsson, J. Magn. Reson., 1982, 50, 1.R. L. Vold, J. S. Waugh, M. P. Klein and D. E. Phelps, J. Chem. Phys., 1968,48, 3831. H. Tatewaki and S. Huzinaga, J. Comput. Chem., 1980,1,205; Y. Sakai, T. Tatewaki and S. Huzinaga, J. Comput. Chem., 1981, 2, 100; 108. H. Johansen, Theoret. Chim. Acta., 1974, 32, 273 and references therein. H. Kashiwagi, T. Takada, E. Miyoshi and S. Obara, Program JAMOL~ (Program Library, Computer Centre of the Institute for Molecular Science). H. G. Herz, Progr. Nutl. Magn. Reson. Spectrosc., 1967, 3, 192. C. P. Chang and T. L. Brown, J. Am. Chem. SOC., 1980, 102, 6418. R. Blinc, J. Seliger, R. Osredkar and T. Prelesnic, Chem. Phys. Lett., 1973, 23, 486. G. Bums and E. G. Wikner, Phys. Rev., 1961, 121, 155. P. A. Madden, Annu. Rev. Phys. Chem., 1980,31, 523. Y . Masuda and H. Yamatera, J. Phys. Chem., 1983,87, 5339. Y. Masuda and H. Yamatera, J. Phys. Chem., in press. D. Eisenberg and W. Kaumann, The Structure and Properties of Water (Oxford University Press, London, 1969), chap. 4. K. Krynicki, Physica, 1966, 32, 167. G. Engel and H. G. Hertz, Ber. Bunsenges. Phys. Chem., 1968,72,808; H. G . Hertz and M. D. Zeidler, Ber. Bunsenges. Phys. Chem., 1964, 68, 821. H. Nomura, S. Koda and Y. Miyahara, in Water and Metal Cations in Biological Systems, ed. B. Pullman and K. Yagi (Japan Scientific Societies Press, Tokyo, 1980), p. 31. A. C. Larson and L. Helmholtz, J. Chem. Phys., 1954, 22,2049; T. L~rland and J. Krogh-Moe, Acta Chem. Scand., 1957 11, 565. H. W. Ruben, D. H. Templeton, R. D. Rosenstein and I. Olovosson, J. Am. Chem. Soc., 1961, 83, 821. F. H. Stillinger and A. Rahman, J. Chem. Phys., 1974, 60, 1545. A. Ben-Naim and F. H. Stillinger, in Water and Aqueous Solutions, ed. R. A. Hone (Wiley, New York, 1971), chap. 8. E. A. C. Lucken, Nuclear Quadrupole Coupling Constant (Academic Press, New York, 1969), chap. 5. H. G. Hertz, in Water, A Comprehensive Treatise, ed. F. Franks (Plenum Press, New York, 1973), vol. 3, chap. 7. (PAPER 4/638) t The hydration number of an oxygen atom of the sulphate ion is about three.32 The total effect of these water molecules is not simply additive: the effects cancel each other.

ISSN:0300-9599    

DOI:10.1039/F19858100127

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1985, 81, 137-141 Characterization of Silica-supported Vanadium Species Interactions with Methanol and Ammonia Adsorbates Studied by Electron Spin-Echo Modulation Spectrometry BY MYSORE NARAYANA, CHAKRAVARTHULA S. NARASIMHAN AND LARRY KEVAN* Department of Chemistry, University of Houston, Houston, Texas 77004, U.S.A. Receiued 18th April, 1984 The interactions of deuterons in CH,OD and CD,OH adsorbates on V,O,/SiO, catalysts have been analysed by electron spin-echo modulation (e.s.e.m.) spectrometry. The best fit to the experimental data indicates that the tetravalent vanadium interacts with three methanol molecules with V-DO,, = 0.27 and V-D,C,, = 0.35 nm in the first coordination sphere and with two farther methanol molecules at V-D,C,, = 0.45 nm.Fast Fourier transform of the time domain e.s.e.m. data for samples with adsorbed ND, clearly shows a strong interaction between vanadium and nitrogen. The active sites in supported vanadium catalysts indicated in reactions such as the oxidation of hydrocarbons, reduction of NO with NH, etc. have been hypothesized to be the surface V=O species.l A number of attempts have been made to characterize the coordinatively unsaturated vanadium species on these supported cataly~ts,l-~ but very few quantitative results have been obtained. Van Reijen and Cossee5 proposed that the initial calcined V,O,/SiO, system contains (vo4)4- units which change into (VO),+ in square-pyramidal coordination on interaction with NH, or H,O. More recently, Hanke et aL6T9 suggested on the basis of chemical analyses and optical measurements that tetravalent vanadium in the calcined V,O,/SiO, system is initially present in tetrahedral or square-pyramidal coordination which transforms into an octahedrally coordinated vanadium species on interaction with water, methanol or ammonia.In recent years the relatively new technique of electron spin-echo modulation (e.s.e.m.) spectroscopy has been used quite successfully to obtain information on short-range order in disordered systerns.l0-l3 Using this technique Cu2+ in Cu2+/Si0, samples was shown14 to interact with two molecules of water, methanol or ammonia. We have also shown15 that in V,O,/SiO, the vanadium coordinates to two water molecules and also interacts with a closer proton. In this paper we extend this study to methanol and ammonia adsorbates on the V,O,/SiO, system.Three molecules of methanol are found to interact with vanadium while with ammonia as adsorbate a very strong interaction with the nitrogen nucleus is seen. EXPERIMENTAL Silica (Alfa Chemicals, 400 m2 g-I) was washed with 1 mol dm-, HC1 to remove impurity Fe3+ which otherwise was found to reduce the phase memory time of the electron spin-echo signals through spin-spin interactions with vanadium species. Aqueous solutions of NH,VO, were used to prepare the V,O,/SiO, system through impregnation followed by calcination in air at 700-800 K. The vanadium content in these samples was 0.25-1.0 wt%. Samples for adsorption of methanol or ammonia were prepared as follows. 137138 SILICA-SUPPORTED VANADIUM SPECIES (a) CH,OD or CD,OH (Stohler Isotopes) were vacuum distilled over activated 3A zeolite.Calcined samples of V,O,/SiO, were dehydrated in vacuum at 773 K for 5 h to a residual pressure of Torr. Then the samples were reduced at 773 K for 30 min in 50 Torr of H, (99.995 % ). The hydrogen was pumped off at the same temperature and the samples were cooled to ambient temperature before exposing them to the saturated vapour pressure of CH,OD or CD,OH for 2 h. After brief evacuation the samples were sealed at 77 K. (6) ND, (Stohler Isotopes) was purified by vacuum distillation. The calcined samples of V,O,/SiO, were heated at 773 K in 250 Torr oxygen for ca. 3 h after dehydration in vacuum at 773 K for 5 h. This prevents the formation of V3+ since ammonia itself acts as a reducing agent.After cooling to room temperature the samples were exposed to 50 Torr of ND, for 1 h. It is critical to pump off the excess of ND, yet not to desorb the coordinated ND,. Thus the V4+ e.s.r. signal is monitored after pumping until there is a slight decrease in intensity and the samples are sealed at 77 K. If an excess of ND, remains, a weak V3+ e.s.r. signal at 77 K is observed to develop after several hours of storage at room temperature and the characteristic vanadyl ion e.s.r. spectrum disappears. Electron spin resonance spectra were recorded on a Varian E-4 e.s.r. spectrometer at room temperature and at 77 K. The e.s.e.m. spectra were recorded at 4.2 K with a home-built spectrometer. l3 RESULTS AND DISCUSSION The e.s.r.spectra for samples of V205/Si02 with adsorbed methanol were very similar to those with adsorbed D20 and to other spectra of V02+ reported in the literat~re.~-~ The spectra show g and A axial anisotropy with a 51V nuclear spin of 7/2. With adsorbed ammonia the only difference in the e.s.r. spectra was the width of the high-field hyperfine lines. They were considerably broader than the corresponding lines with H20 or CH,OH adsorbates, presumably because of unresolved super- hyperfine interaction with the nitrogen nucleus (14N, Z = 1). The theory, methods of analyses and applications of electron spin-echo modulations have been described in detail.1°-13 When a paramagnetic spin system interacts with a suitable sequence of resonant microwave pulses, microwave echoes are generated.Since the nearby magnetic nuclei are also excited by the same pulse sequence the echoes are modulated by the characteristic Larmor frequencies of the interacting nuclei. These modulations can be analysed to yield the number and distance of such interacting nuclei and also the isotropic hyperfine coupling if any small delocalization of the unpaired spin onto those nuclei is present.12 Fig. 1 shows the experimental and calculated two-pulse electron spin-echo spectra of V205/Si02 with adsorbed CH,OD. The vanadium content in these samples was 0.5 wt% . The best fit to the experimental data is obtained for three interacting deuterons at 0.27 nm with an isotropic coupling of 0.4 MHz. An independent determination of the magnitude of the isotropic coupling was obtained by fast Fourier transform (f.F.t.) of the three-pulse e.s.e.m.time-domain data into the frequency domain.12 Fig. 2 shows one such frequency spectrum for the same sample. Two lines are seen, one with a peak at 2.13 MHz and a second poorly resolved line at 2.52 MHz. The free precession frequency, v,, of the deuterium nucleus at the magnetic field corresponding to this spectrum is 2.39 MHz. From ENDOR theory it is known that in the presence of a small hyperfine interaction two peaks would be seen at v,fa/2 MHz. Thus the two peaks observed correspond to an isotropic coupling of 0.4 MHz. The two peaks for such a coupling at the observed magnetic field should be at 2.19 and 2.59 MHz, but the differences with the observed values are within experimental error.The genuineness of the experimental peaks was confirmed by transforming the electron spin-echo spectra recorded at different magnetic fields for which the peaks shifted correspondingly.M. NARAYANA, C. S. NARASIMHAN AND L. KEVAN 139 10 8 .G 6 c, 5 0 4 * e 0 .- 2 0 1 2 time/ps Fig. 1. Calculated (-) and experimental (---) two-pulse e.s.e.m. spectra of 0.5 wt% V,O,/SiO, with adsorbed CH,OD. The additional structure seen between z = 0.25 and 0.5 ps are proton modulations from the methyl group. The decay function used for the best fit was g(z) = exp (3.0 - 2.0652 + 0 . 0 8 ~ ~ ) . n = 3 , Y = 0.27 nm and aiso = 0.4 MHz. I 1 I I I I I I I 1 10 01 0 5 frequency /MHz Fig. 2. Fast Fourier transform of an experimental three-pulse e.s.e.m. spectrum of 0.5 wt% V,O,/SiO, with adsorbed CH,OD.The data tapering parameter is 10%. The peaks indicated at 2.13 and 2.52 MHz are assigned to v(D)+iaiso. H = 3660 G and z = 0.25 ,us. Analysis of two- and three-pulse echo spectra obtained for the samples V,O,/SiO, with adsorbed CD,OH indicate that vanadium interacts with two types of deuterons, nine at 0.35 nm and six more at 0.45 nm (see fig. 3). Because of shorter phase memory times these farther methanol molecules cannot be detected for samples with adsorbed C€-€,OD. Assuming that the molecular dipole of the methanol molecule is oriented away from the metal ion and along the V-0 bond, the nine deuterons at 0.35 nm correspond to a V-0 bond length of 0.19 nm and a V-HO distance of 0.26 nm. These distances compare very well with the data obtained for the samples with140 SILICA-SUPPORTED VANADIUM SPECIES I 1 I I I I 1 I I I I \I '- I I I I 1 1 I 1 I I 1 2 time/ps Fig. 3.Calculated (-) and experimental (---) two-pulse e.s.e.m. spectra of 0.5 wt% V205/Si02 with adsorbed CD,OH. The decay function used for the best fit was g(z) = exp (3.1 1 - 1.9752 + 0 . 1 3 5 ~ ~ ) . n, = 9, rl = 0.35 nm and n2 = 6, r2 = 0.45 nm. 2 .3 a 5 0 0 5 10 15 20 frequency/MHz Fig. 4. Fast Fourier transform of an experimental two-pulse e.s.e.m. spectrum of 0.5 wt% V205/Si02 with adsorbed ND,. The data tapering parameter is 15%. The three frequencies attributed to the nitrogen interaction are indicated at 3.25, 5.55 and 7.0 MHz. One deuterium frequency is indicated at 2.06 MHz along with the associated double frequency at 4.20 MHz.H = 3235 G. adsorbed CH,OD with V-0 = 0.20 and V-DO = 0.27 nm. Thus the two sets of data are consistent with a vanadium interaction with three methanol molecules in its first coordination sphere. Note that this contrasts with two methanol adsorbate molecules around Cu2+ in Cu/Si02.14 The electron spin-echo spectra for the samples with adsorbed ammonia turned out to be very complicated, unlike the case of Cu/Si0,.lgu For the vanadium samples strong nitrogen modulations could be seen in both two-pulse and three-pulse electronM. NARAYANA, C. S. NARASIMHAN AND L. KEVAN 141 spin-echo spectra. Fig. 4 shows the f.F.t. spectrum of a two-pu1see.s.e.m. time-domain decay curve of V,O,/SiO, with adsorbed ND,. By comparison with the data available for samples with adsorbed NH,, the v(D) and 2v(D) frequencies at 2.06 and 4.20 MHz can be identified.The other frequencies of 3.25, 5.55 and 7.0 MHz are assigned to nitrogen; the free nitrogen nuclear frequency at this magnetic field is 0.995 MHz. To interpret these 14N frequencies more experimental data are necessary. Because of the complication of combined nitrogen and deuterium modulation it is more difficult to count the number of ammonia molecules coordinated to the vanadium ion. However, note that the same 14N frequencies have been observed16 for samples of MoO,/SiO, with adsorbed NH, and ND,. Belokopytov et al.17 studied the interaction of ammonia with the oxidized surfaces of V,05 and MOO, by infrared spectroscopy. They concluded that ammonia adsorbs via a strong coordinative bond to the metal species (V5+ or Mo6+) on these surfaces anti that a weak NH; species also forms, through interaction of ammonia with the surface hydroxyls.This NH,+ species was found to be easily removable by brief evacuation even at room temperature. CONCLUSIONS Specifically deuterated methanols were adsorbed on activated and reduced V,O,/SiO, samples to study the coordination sphere of the catalytically active vanadium species. Analysis of the time-domain electron spin-echo data and the corresponding frequency spectra obtained by fast Fourier transformation indicate that vanadium interacts with three methanol molecules in the first coordination sphere with V-OM, z 0.2, V-DO,, z 0.27 and V-D3CMe z 0.35 nm. It also seems to interact weakly with two more molecules of methanol farther away with V-D3CMe2 z 0.45 nm.The experiments with adsorbed ammonia showed strong nitrogen interaction and no definitive analysis was made owing to quadrupole complications. This research was supported by the National Science Foundation, the Robert A. Welch Foundation and the Energy Laboratory of the University of Houston. I K. Mori, A. Miyamoto, T. Ui and Y. Murakami, J. Chem. SOC., Chem. Commun., 1982, 260. K. Tarama, S. Yoshida, S. Ishida and H. Kakioka, Bull. Chem. SOC. Jpn, 1968, 41, 2840. K. Hirota, Y. Kera and S. Teratani, J. Phys. Chem., 1968, 72, 3133. A. Miyamoto, Y. Yamazaki, T. Hattori, M. Inamata and Y. Murakami, J. Catal., 1982, 74, 144. L. L. Van Reijen and P. Cosse, Discuss. Faraday SOC., 1966, 41, 277. W. Hanke, K. Heise, H-G. Jerschkewitz, G. Lischke, G. Ohlmann and B. Parlitz, Z. Anorg. Allg. Chem., 1978,438, 176. V. A. Shvets, M. F. Saricheu and V. B. Kazansky, J. Catal., 1968, 11, 378. V. M. Vorotyntsev, V. A. Shvets and V. B. Kazansky, Kinet. Katal., 1971, 12, 678. W. Hanke, B. Bienet and H-G. Jerschkewitz, 2. Anorg. Allg. Chem., 1975, 414, 109. lo W. B. Mims, J. Peisach and J. L. Davis, J. Chem. Phys., 1977, 66, 5536. ' I L. Kevan, in Time Domain Electron Spin Resonance, ed. L. Kevan and R. N. Schwartz (Wiley- l 2 P. A. Narayana and L. Kevan, Magn. Reson. Rev., 1983, 7, 239. l3 T. Ichikawa, L. Kevan and P. A. Narayana, J. Phys. Chem., 1979,83, 3378. l4 (a) T. Ichikawa, H. Yoshida and L. Kevan, J. Chem. Phys., 1981, 75, 2485; (b) T. Ichikawa, Interscience, New York, 1979), chap. 8. H. Yoshida and L. Kevan, J. Phys. Chem., 1982, 86, 881. M. Narayana, C. S. Narasimhan and L. Kevan, J. Catal., 1983, 79, 237. l6 M. Narayana, R. Zhan and L. Kevan, unpublished results. Yu. V. Belokopytov, K. M. Kholyavenko and S. V. Gerei, J. Catal., 1979, 60, 1. (PAPER 4/64])

ISSN:0300-9599    

DOI:10.1039/F19858100137

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1985, 81, 143-159 Electron Transfer and Dimerization of Viologen Radicals on Colloidal TiO, BY ENRICO BORGARELLO Institute of Physical Chemistry, Ecole Polytechnique Federale, CH- 10 15, Lausanne, Switzerland AND EZIO PELIZZETTI* Institute of Analytical Chemistry, The University of Turin, Turin, Italy AND WILLIAM A. MULAC AND DAN MEISEL* Chemistry Division, Argonne National Laboratory, Argonne, Illinois 60439, U.S.A. Received 25th April, 1984 The reaction of several viplogen radical cations with colloidal TiO, particles of 70 A radius has been studied in detail. The electron-transfer reaction from methyl viologen radicals (MV+) to the TiO, particles is controlled by the rate of the heterogeneous electron-transfer step. Protonation of the reduced particle follows the electron-transfer reaction in a temporally well separated reaction.Analysis of the subsequent equilibrium stage, but before protonation occurs, allows an estimate of the charge carrier density in the colloid. The effect of added Pt, either as a separate colloid or by photodeposition on the TiO, particle, has also been studied. Kinetic analysis indicates that a mixture of Pt and TiO, colloids results in adsorption of the Pt particle on the TiO, colloid. However, the adsorbed Pt colloid reacts independently of the TiO,.particle. Heptyl viologen radicals (HV+) catalytically dimerize in the presence of TiO, particles in a heterogeneous reaction between an absorbed HV+ and a dissolved radical. The competition between dimerization and hydrogen evolution can be directed towards the latter reaction either by reducing the pH or by loading the TiO, particles with Pt.The unsymmetric C,,MV+ viologen radicals aggregate even in the absence of any colloid. At low pH values and high Pt loadings they can, however, be directed towards hydrogen evolution. The use of TiO, in photocatalytic water cleavage1 is presently stimulating intensive investigations. When used in colloidal form the role of the semiconductor particle in these systems is envisioned as the photosensitizer itself upon band-gap irradiation2 and as a catalyst or a support for other catalytic components such as Pt and RuO,.~. Further studies on the photosensitization of oxide semiconductors have demonstrated the feasibility of direct charge injection of electrons from solution species into the conduction band of the semiconductor.j* The kinetics of photoinduced charge injection from the conduction band of colloidal TiO, into 1,l’-dimethyl-4,4’- bipyridinium (MV2+) have been recently studied’ and thermodynamic as well as kinetic parameters were analysed.8 The flat-band potential of the colloidal TiO, has been shown to be more negative than that of rutile electrodes, presumably owing to structural differences.The rate-determining step in the electron-ejection process has been shown to be the heterogeneous electron-transfer step to the solution acceptor.8 Hydroxide groups at the particle surface provide for the pH dependence of the flat-band potential, Efb, of the colloidal TiO,. Fig. 1 represents the pH dependence of the three systems of present interest, namely: EMV2+,MV+ = -0.445 V, EH+,iH2 = 143144 REACTION OF VIOLOGEN RADICALS WITH TiO, 445 400 > E \ 7 200 I 20 0 0 2 4 5.5 6 PH Fig.1. pH dependence of the reduction potentials of the couples MV2+/MVf (a) and H+/$H (b) and the flat-band potential (c) [taken from ref. (7)] of the colloidal TiO, particle of R = 200 A on pH. - 0.059 pH V and Efb = -0.12 - 0.059 pH V for a particle of 200 A radius.' Under standard conditions charge injection from MV+ into the conduction band of the semiconductor can be seen to be thermodynamically favourable at pH < 5.5. Under similar conditions but at higher pH, charge injection of conduction-band electrons (populated electrochemically or by band-gap irradiation) into MV2+ in the solution is favoured.The kinetics of the charge injection will, however, be determined by the amount of overlap between the state density functions of the donor/acceptor couples. Once steady-state conditions are obtained, the catalytic conversion of 2MV+ radicals into H, molecules will proceed at equal rates of anodic MV+ oxidation and cathodic H+ reduction. Detailed analysis of the electrochemical catalytic rates has been previously reported and applied to the conversion of MV+ to H, on metallic colloid^.^^ lo In the present report we investigate the kinetics of electron transfer from MV+ to colloidal TiO, in the low pH range: MV+ + (TiO,), f MV2+ + (TiO,);. (1) MV+ radicals were produced by pulse radiolysis and their decay in the presence of TiO, was followed by kinetic spectrophotometry. The rate of the protonation reaction (2) was followed conductometrically. Similar techniques have been previously employed to follow these two steps on gold and platinum colloids.1° That the net reaction in the presence of TiO, is hydrogen evolution (TiO,); + H+ =$ Had + (TiO,), MV+ + H+ e MV2+ + iH2 (3) has been demonstrated previously.6 We have also attempted to probe the effect of Pt deposition on the semiconductor catalyst.Furthermore, in view of the broad interestE. BORGARELLO, E. PELIZZETTI, W. A. MULAC AND D . MEISEL 145 Table 1. Physical properties and conversion factors for the TiO, colloid used in the present studya ~ ~~ property symbol and formula units 3w 4nN6R3 [TiO,], = ___ = 2.72 x lo-’ w q = [TiO,]/[TiO,], = 4.6 x lo4 S = 47rNR2 = 3.71 x 10“ mol (particles) dm-3 molecules per particle cm2 mol-l (particles) concentration of particles aggregation no.molar surface area molar particles volume 47rN 3 V = - R3 = 8.65 x lo5 cm3 mol-l (particles) effective molecular TiO, radius r = (:Ed)li( - = 1.95 A R-2r f = 1 -(R) = 0.158 fraction of surface molecules a Assuming the following values: molecular weight M = 80 g mol-l; density from light-scattering experiments. 6 = 4.25 g cmP3; average radius of the particles R = 70 w is the total TiO, concentration in g dm-3. N = Avogadro’s number. in using amphiphilic viologens to achieve charge separationll we have also studied the same reactions for the l,l’-diheptyl-4,4’-bipyridinium (HV2+) and 1 -tetradecyl- l’-methyl-4,4’-bipyridinium (CI4MV2+) analogue^.^ EXPERIMENTAL MATERIALS TiCl, (Agfa, Ventron), MV2+ and HV2+ (Aldrich), H,PtCl, (Fluka) and all other materials were of highest purity commercially available and were used as received, unless otherwise specified.The unsymmetric viologen, Cl,MV2+, was synthesized as the di-iodide salt according to the literature procedure12 and was then exchanged to the dichloride form on an ion-exchange resin. Water used was triply distilled. Deaeration was achieved by bubbling Ar through the solutions using the syringe technique. PREPARATION AND CHARACTERIZATION OF THE CATALYSTS Colloidal TiO, was prepared by hydrolysis of TiC1, in a water + ice mixture according to a procedure described e1~ewhere.l~ The colloid was then dialysed and the final pH at the end of the dialysis adjusted to 1.5.X-ray diffraction studies performed on the TiO, colloidal particles showed that the material is mainly amorphous. The particle size of the presently prepared batch was determined by photon correlation spectroscopy as previously described14* l5 using a Chromatix light-scattering instrument. Application of the Stokes-Einstein equation yields a hydrodynamic radius of R , = 70 A. The TiO, content of the stock solution was determined by atomic absorption spectroscopy (Pye Unicam-SP 191 spectrophotometer) to be 15 g dm-3 (0.187 mol dm-3) in the presently studied batch. The stock solution of colloidal TiO, was then diluted to the desired concentration and the pH adjusted by addition of HCI. Based on these measured quantities, the concentration of particles and their surface area as well as several other physical properties could be calculated.The results of these calculations are compiled in table 1. Two different techniques for loading the TiO, particles with Pt were employed. In the first method a solution of H,PtCl, was reduced by refluxing with a citrate solution as previously described6* l6 to produce colloidal Pt. No stabilizer was used in the preparation of the colloidal Pt and excesses of electrolytes were removed by a mixed-bed ion-exchange resin (to conductances 6 F A R 1146 REACTION OF VIOLOGEN RADICALS WITH TiO, of 5 pS). This solution of ultrafine Pt particles (R = 16 A) was then either used in irradiated samples as a blank or mixed with the TiO, colloid described above.Since the isoelectric point of the TiO, colloid is 3.713 while that of the Pt sol is 2.3,6 it is crucial that the pH be preserved at 2.5-3.0 during this stage of mixing. The second method for Pt loading of the TiO, particles used photoplatinization. A solution containing both the colloidal TiO, and H,PtCl, was irradiated in a Pyrex flask using a 450 W Xe lamp. Complete reduction of the chloroplatinic acid was verified both by the disappearance of its absorption at 420 nm and by the initiation of H, evolution at the end of its reduction. Presumably reduction of the chloroplatinic acid occurs by scavenging the electrons promoted into the conduction band of the semiconductor colloid. Hole scavenging by water simultaneously produces 0,. This procedure avoids the need for addition of external sacrificial electron donors previously described.l7 None of the colloids described above presented any coagulation problems when MV2+ was used under the experimental conditions presently employed. However, when HV2+ was used as the radical source, precipitation of the Pt-loaded TiO, did occur. Furthermore, perchlorate ions precipitated HV2+ ions while sulphate ions precipitated the colloidal TiO,. We therefore used HCl in the present study to control the acidity of the solutions. RADIATION PROCEDURE The pulse-radiolysis technique was used to produce the viologen radical cations (V+) and to follow their decay kinetics. Electron pulses of 4-40 ns width, producing 4-125 pmol dm-3 of total radical concentration per pulse, were utilized to produce the viologen radicals through the following well known sequence of reactions: (4) ( 5 ) (6) (7) (8) H,O vv\h) e& 'OH, H', H,, H,O,, H+ e;,., + V2+ -, V+ e& + H+ + H' 'OH(H') + (CH,),CHOH + (CH,),COH + H,O(H,) (CH,),COH + V2+ + (CH3),C0 + V+ + H+.Ideally all the radicals are thus converted into the viologen radicals. However, at high HCl concentrations formation of C1; radicals diminished the yield of V+ radicals. We found, however, no effect of these radicals on the much slower decay kinetics of the V+ radicals. The rate of proton consumption was determined using the pulse-radiolysis-conductivity technique, Since equivalent amounts of protons and V+ are produced [reactions (4)-(8)] a transient pH change is to be expected. However, owing to reaction (3), no permanent pH change occurs, even at the highest doses used.RESULTS MV2+/Ti0, SYSTEM The rate of MV+ disappearance in the presence of colloidal TiO, was measured under a variety of conditions. In fig. 2 we show the rate of MV+ decay following a string of pulses (2 pulses per second; 10 ns pulse width; ca. 25 pmol dm-3 MV+ produced per pulse). The radical can be seen to decay to an equilibrium level which is dependent on the total dose (and thus on [MV+]) produced in this system. The decay rate, however, was found to be independent of the total dose. This equilibrium level further decays on a slower timescale, in a process which presumably corresponds to reaction (2). In fig. 2 MV+ decays in a first-order process with k, = 43 4 s-l following each of the 6 pulses shown.The same first-order rate constant was obtained for the solution described in fig. 2 when pulsed with 4 or 40 ns single pulses (6.0 or 106 pmol dmA3 radicals per pulse). This independence of dose is in clear contrast toE. BORGARELLO, E. PELIZZETTI, W. A. MULAC AND D . MEISEL 147 0 . 5 0 0 ~ 3 -0 100 0 ti s Fig. 2. Rate of decay of MV+ observed at 605 nm following a string of 10 ns pulses at 2 pulses per s. Experimental conditions: [MV2+] = mol dmP3; pH 2 (HCl); 0.17 mol dm-3 propan- 2-01; Ar saturated; [TiO,] = 5 g dm-3. the strong dependence of MV+ decay on its concentration when catalysed by colloidal gold or platinum.1° Another deviation from catalysis on metallic colloids is the dependence on [TiO,]. While in the former case the decay rate depends on the catalyst concentration with a second-order dependence, close to first-order dependence is observed in the present case (fig.3). The observed rate constant at pH 2.0 and lop3 moldmP3 MV2+ is calculated from fig. 3 to be (5.9k0.5) x 10, dm3 mol-l s-l in terms of total TiO, concentration. From the aggregation number (table 1) one can calculate 2.7 x lo7 dm3 mo1-I s-' in terms of particle concentration, and from the particle molar surface area a heterogeneous rate constant of 7.3 x Only a slight dependence of the rate constant for MV+ decay on [MV2+] could be observed. Thus kobs = 6.5k0.3 s-l is obtained at [MV2+] = 0.5 mmol dmP3, while kobs = 7.3 1 s-l is obtained at [MV2+] = 2 mmol dm-3 (pH 2, [TiO,] = 1 g dm-3). The dependence of this rate constant on the solution pH is more complex.At pH 3.0 and 2.0, the observed rate constants are 5.9+ 1 and 6.7+ 1 s-l, respectively ([MV2+] = mol dm-3, [TiO,] = 1 g dm-3). However, at pH 1.0 MV+ decays with The equilibrium situation and its dependence on the same variable parameters were also evaluated. The equilibrium concentration of MV+ which remains in solution, [MV+Ieq, was found to depend on all of these variables, i.e. on [MV2+], [MV+],, [TiO,] and pH. The dependence is indicative of equilibrium (1) being achieved, and these results are analysed accordingly. The value of [MV+],, is calculated from the absorbance at the end of the reaction using A,/& (E,,,, = 1.15 x lo4 dm3 mol-1 cm-l) while the amount that has reacted with TiO, is calculated from ( A , - A , ) / & .The results are shown in fig. 4 as a plot of the ratio [MV2+]/[MV+],, against [TiO,]/[TiO;]. Since the same units were used in the calculation of the latter ratio, this ratio then represents the reciprocal of the fraction of TiO, molecules to which electrons were added. At the highest charge density we therefore charge ca. 1 % of the TiO, molecules cm s-l is obtained. kobs = 1 16 9 S-'. 6-2148 REACTION OF VIOLOGEN RADICALS WITH TiO, S/10-6 cm2 dm-3 2 4 I I [Ti02 1 /mmol dm-3 Fig. 3. Dependence of the decay rate constant of MV+ on [TiO,] : 0 , 2 ns pulse; .,40 ns pulse; other experimental conditions as in fig. 2. with electrons. If these were indeed to behave as electrons in the conduction band, one would calculate charge-carrier densities of ca.2 x 1020 ~ m - ~ . As would be expected for equilibrium, a straight line passing through the origin is obtained in fig. 4. The slope of this line, denoted K', yields K' = 0.138 at pH 2.0. The same result is obtained when either [MV2+], or [TiO,], is held constant. However, the equilibrium constant does change, as expected, with the pH of the electrolyte solution (fig. 5). The results in fig. 5 adhere to the equation logK'= a-pH (9) where a is a constant. The rationalization of this pseudo-homogeneous behaviour will be discussed below. PROTON CONSUMPTION IN THE MV2+/Ti0, SYSTEM The rate of proton consumption [reaction (2)] in this system was measured using the conductivity-detection technique. Owing to the experimental limitations of this method and of the chemical system, only a small pH region is available for these studies.Below pH 2.5 the high level of electrolytes renders the signal-to-noise ratioE. BORGARELLO, E. PELIZZETTI, W. A. MULAC AND D. MEISEL 149 lOOr 1 [TiOz] /[Tioil Fig. 4. Dependence of the ratio [Ox]/[Red] for the MV2+/MV+ couple on the same ratio for the TiOi/TiO; couple at pH 2: 0, [MV2+], = mol dm-3, constant; 0, [TiO,], = 1 g dmP3, constant. too low to be measured. On the other hand, raising the pH too close to the isoelectric point of the colloid (3.5) reduces the signal level owing to the buffering effect of the colloid. An example of the conductivity results is given in fig. 6 . In fig. 6 ( a ) the rate of H+ consumption is measured at pH 2.6 in the presence of 5 g dm-3 TiO,.The measured initial conductivity change corresponds to 14 pmol dm-3 H+ production in the sequence of reactions (4)-(8). The calculated yield of H+ is, however, 19 pmol dm-3. The discrepancy is explained by the buffering effect of the colloid. Conductivity changes due to reduction of MV2+ to MV+ [reactions (5) and ( S ) ] are negligible in comparison with those resulting from H+ production [reactions (4) and (S)]. The decay in fig. 6 ( a ) is exponential and yields kobs = 5.5 s-l. Under the same conditions the decay of MV+ measured spectrophotometrically yields kobs = 45 s-l (e.g. fig. 3 at a lower pH). We therefore find a mismatch of approximately an order of magnitude between the rates of reactions (1) and (2) under these experimental conditions. MV2+/Ti0,/Pt SYSTEM Addition of colloidal Pt (or photoplatinization of TiO,) results in a dramatic effect on both the equilibrium concentration of MV+ and on its rate of disappearance.This rate is substantially increased (compared with the rate in the absence of Pt), while the equilibrium level of MV+ at the end of the reaction is reduced practically to zero in the acidic pH range presently investigated. Details of this system will be described150 I0.C I .o k" 0. I REACTION OF VIOLOGEN RADICALS WITH TiO, 0.0 I PH Fig. 5. Dependence of the equilibrium constant on pH obtained as described in fig. 4. separately, but some conclusions from the preliminary experiments can already be drawn. Generally speaking metal-like behaviour is observed once colloidal Pt is added to the system.All the dependences on reactant concentrations and dose resemble those observed for colloidal metals1* rather than those presently observed for colloidal TiO,. A comparison between the two methods of Pt deposition described in the Experimental section is particularly instructive. A comparative compilation of some of these experiments is given in table 2, showing that deposition of Pt by photoreduction is greatly superior to addition of the same amount of colloidal Pt separately reduced by citrate [cf. experiments (c) and (d) in table 21. In fact, addition of the TiO, colloid to a Pt colloid prepared by citrate reduction decreases the catalytic activity of the same Pt colloid in the absence of the TiO, by approximately an order of magnitude [cf. experiments (b) and (c) in table 21.The rate of H+ consumption was also checked in a solution in which the TiO, particles were photoplatinized with 0.01 g dm-3 Pt [pH 3.2,0.5 g dm-3 TiO,; fig. 6(b)]. The decrease in the initial conductivity change (although the same dose as deposited in the solution as in the previous system) as well as its decay below the prepulse conductivity level are indicative of the stronger buffering effect of the colloid in this system. Indeed, photodeposition of Pt on TiO, was found to decrease the isoelectric point to ca. 2. 18.13 The protons disappear in this experiment with a half-life of 4.6 msE. BORGARELLO, E. PELIZZETTI, W. A. MULAC AND D. MEISEL 1 Eu 1.2 : 1 o.'t. 0 . . . . . . . . . . . . . . . . . . . - . . . I 1 I I I 0 0.4 0. a 1.2 I .6 tl s u I--- 151 .. . . _..., .. - ...... ........ " ..- -I.-. ............ . . . . - - - I..... . . . . . ........... - ..-- ._L . . . . . . . . . . I I I I I I I 1 I tlms -0.2 Fig. 6. Rate of consumption of protons measured by the conductivity technique: (a) 5 x low4 mol dm-3 MV2+, pH 2.6, 5 gdm-3 TiO,; (b) 5 x mol dmb3 MV2+, pH 3.2, 0.5 g dm-3 TiO, and 0.01 g dm-3 Pt deposition by photoplatinization; all solutions contain 0.17 mol dm-3 propan-2-01 and are Ar saturated. and exhibit the same dose dependence previously observed for the catalytic reactions of MV+ on colloidal Pt.lO* HV2+/Ti0, SYSTEM Several new features, absent in the MV2+/Ti0, system, could be observed when the decay of HV+ was followed in the presence of TiO,. Fig. 7 shows the disappearance152 REACTION OF VIOLOGEN RADICALS WITH TiO, Table 2.Effect of Pt and method of its deposition on the decay of MVfa ITi0,l [Ptl [MV+Ic [MV+IeqC /g dm-3 /g dm-3 t t / s /pmol dmV3 /pmol dm-3 method of Pt preparation (a) 0.5 - 0.1 1 5.8 1.9 - (6) - 0.01 3.9x 10-5 6.1 0.03 reduction with citrate (c) 0.5 0.01 3.3 x 10-4 6.3 0.2 1 reduction with citrate ( d ) 0.5 0.01 1 . 7 ~ 6.2 0.06 photoreduction (e) 0.5 0.01 4 . 6 ~ 19 0.05 photoreductiond a All solutions contain mol dm-3 MV2+ at pH 2 unless otherwise stated. First half-life of MV+ decay [or proton consumption in experiment (e)]. [MV+], and [MV+Ieq are the amount of MV+ initially produced by the pulse and the amount remaining at the end of the reaction, respectively. Concentrations below 0.5 pmol dm-3 are highly inaccurate and should be considered only semi-quantitatively.Results in this experiment were measured conductometrically at 5 x lop4 mol dm-3 MV2+ and pH 3.2. of HV+ in the presence of 0.5 g dm-3 TiO, mol dm-3 HV2+, pH 3.2) at various doses. Following an induction period the absorption of the HV+ radical at 605 nm is seen to decay at a dose-dependent rate. On the other hand, when followed at 530 nm an increase in absorption of the product can be seen [fig. 7(d)]. The rate of formation of this absorption at 530 nm exactly matches the decay at 605 nm [cf. fig. 7 ( d ) and (e)]. The absorption spectrum of the product at the end of this process is clearly that of a dimer of viologen radicals (see fig. 8). The TiO, particles evidently catalyse the dimerization process of this rather hydrophobic viologen radical.No detailed kinetic analysis is intended here. Nevertheless, the rate of the dimerization reaction can be seen in fig. 7 to increase linearly with the dose. The rate of dimer formation (and monomer disappearance) fits exponential kinetics better than second-order kinetics under the experimental conditions of fig. 7 . mol dm-3 (pH 3.2, 0.5 g dm-3 TiO,) dimer formation is then followed by its disappearance. On the other hand, when the pH is decreased from 3.2 to 2.0 or lower no dimer formation could be observed. At these lower pH values the same features of the MV+ reaction described above were observed. However, the absolute rate constant is smaller by a factor of ca. 20 than for the MV2+ system under otherwise identical conditions.Photodeposition of Pt on the TiO, particles also diverts the dimerization reaction towards reduction and eventually to H, formation. As mentioned in the Experimental section, platinization of the TiO, considerably reduces the stability of the TiO, colloid in the presence of HV2+. Only qualitative observations on freshly prepared sols can therefore be described. At g dm-3 Pt concomitant competitive decay of HV+ to dimerize and to form H, can be seen (see fig. 9). The dimer, however, further decays at a slower rate, presumably further to produce H,. At 0.01 gdmF3 Pt no dimer formation could be observed. When the initial concentration of HV2+ is decreased to 3 x C,,MV2+/Ti02 SYSTEM A few experiments were carried out with the unsymmetric viologen radical C,,MV+.Under all experimental conditions presently employed, no micellization of the parent molecule could occur.11 Nevertheless, the radical cation could dimerize and further proceed to form larger aggregates. Indeed, the radical cation decays even in the absence of the catalysts. Since no increase in the absorption at 530nm could beE. BORGARELLO, E. PELIZZETTI, W. A. MULAC AND D . MEISEL 153 00201 ' ' ' ' ' ' ' I ' J 0 t - . _.. 0 600 I 0 2 0 O 0 L I 2 0.100 I 1 2 I200 F' r i I I 02001 I I 2 t l s Fig. 7. Decay of HV+ following pulses of various doses. mol dmh3 HV2+, pH 3.2,0.5 g dm-3 TiO, in 0.1 7 mol dmP3 propan-2-01. Pulse width: (a) 2, (b) 4, ( c ) 10, ( d ) and (e) 20 and (f) 40 ns. All followed at 605 nm except ( d ) at 530 nm. observed, the dimer probably proceeds to form these aggregates. The kinetics of its decay are rather complex.Qualitatively the decay could be described as two processes well separated in time when the initial concentration of the radical is high. A fast process of T ~ , ~ z 0.5 ms and a much slower process in the second timescale are observed.154 REACTION OF VIOLOGEN RADICALS WITH TiO, 0.4 - 8 0.3- 5 e - s1 -53 0.2 - 0.1 - 1 1 0' I I I 1 I t I I I 460 500 540 580 620 660 wavelength/ nm Fig. 8. Absorption spectra of HV+ taken ca. 15 ps after the pulse (a) and of its dimer formed catalytically in the presence of TiO, (b). Experimental conditions: [HVe+] = low3 mol dm-3, pH 3.2, 0.5 g dm-3 TiO,, 0.17 mol dm-3 propan-2-01. 0.4 00 I I I 1 I I I I I I 1 I . 0.000 -0.100 0 10 tl S Fig.9. Decay of HV+ and formation of its dimer in photoplatinized TiO, system. Experimental conditions: [HV2+] = mol dm-3, pH 3.2, 0.5 g dm-3 TiO,, g dm-3 Pt, A = 530 nm. Addition of colloidal TiO, (up to 2.5 x lo-, mol dm-3 total TiO,) has little effect on these decay processes {[C,,MV2+] = (0.5-5) x mol dm-3, pH 3 or 2). Neither does photodeposition of Pt on the TiO, colloid ( 5 x g dmb3 Pt) have any observable effect, under otherwise similar experimental conditions. However, when the pH is reduced to 1 , the C,,MV+ radical decays at a much faster single-exponential process. {At [TiO,] = 6.2 x mol dm-3, [C,,MV2+] = 0.5 x mol dm-3, 5 x g dm-3 Pt and pH 1 the observed rate constant for the radical decay isE. BORGARELLO, E. PELIZZETTI, W. A. MULAC AND D .MEISEL 155 (3.8 kO.3) x lo4 s-l.} Apparently, under the above-mentioned conditions at pH 2 2 the catalytic reaction (1) cannot compete kinetically with dimerization and aggregation of the radical. Furthermore, the added thermodynamic stabilization of the radicals due to these aggregation processes prevents the system from proceeding towards the hydrogen-evolution reaction even when Pt is added. However, at pH < 1 the H+/H, redox potential is high enough to make the latter reaction quite efficient. DISCUSSION EQUILIBRIUM IN THE MV+/TiO, SYSTEM The first question to address is the nature of the equilibrium observed in the experiments such as those shown in fig. 2. Many indications in the Results section point to reaction (1) as being the equilibrium observed rather than reaction (3).Comparison of the conductivity results with the spectrophotometric ones shows temporal separation between the charging reaction (1) and the protonation reaction (2). It can also be shown that the amount of MV+ remaining at the end of its decay exceeds appreciably the amount calculated from the known redox potentials of the MV2+/MV+ and H+/fH2 couples under all experimental conditions of the present study. Indeed when Pt is used, either as the sole catalyst or on TiO, carriers, the reaction is strongly shifted towards the H, side. The dependence of [MV+],, on dose and [TiO,] also leads to the same conclusion, namely that equilibrium (1) is achieved before an appreciable amount proceeds towards reaction (2). The slower decay of [MV+Ieq was observed to parallel the proton consumption rate and was not investigated in detail in the present report.However, only when the rate of these two reactions is the same can the particle be considered to operate under truly catalytic conditions. The charge density of electrons in the particle following the electron-injection reaction is quite high. From the amount of MV+ reacting in reaction (1) and from the particle radius we calculate 1019-1020 electron cmP3. Since these electrons stay in the particle for an appreciable period of time, the possibility that these are now electrons in the conduction band should be considered. As has been previously found, practically no band bending can be expected initially in TiO, particles of the sizes presently used.22 The width of the depletion layer cannot be bigger than the particle radius.The extent of band bending, V,, above the flat-band potential can be calculated from the Schottky relation V, = enr2/2&&, (10) where the dielectric constant is E = 173, E , is the permittivity of free space, e is the electronic charge, n is the density of charge carriers and the depletion layer width is equated to r = 70 Only at relatively high levels of impurity of 1019 cmP3 would V, then amount to 25 mV, namely to kT at room temperature. Note also that below a density of charge carriers of l O l s cmP3 most of the particles will contain < 1 excess electron. Therefore, although the impurity level is difficult to estimate, band bending may be completely neglected when the particles equilibrate initially with the electrolyte.However, once equilibrium ( I ) is achieved, a large number of electrons have been transferred to the particle. These would initially have flattened any residual depletion layer and finally will produce an accumulation layer.23 This bending reversal from a slight depletion layer to an accumulation layer is expected to create a barrier to further electron transfer from a reducing electrolyte (which, however, was not observed in156 REACTION OF VIOLOGEN RADICALS WITH TiO, the kinetic experiments, see below). On the other hand, it is conceivable that following high-intensity excitation of such colloidal TiO, and hole injection to water,7 a sufficient excess of electrons would be created to afford such an inversion.This would then facilitate electron transfer to acceptors in the electrolyte. Pinning of the Fermi energy by surface states will not change this general argument. Injection of electrons into the particles via reaction ( 1 ) will raise the Fermi level relative to its original position, Efb, by an amount AEf = Ef - E f b which can be approximated by AEf =-kTln l+- . ( 3 Since, as discussed above, the initial density of charge carriers, no, is small compared with An, AEf z - k T l n ( g ) . When equilibrium (1) is achieved, Ef = EMV~+/MV+; using eqn ( I I b) one obtains which may be converted to the linear form where K = 1 0 ( ~ r ~ - E ~ ~ ~ ~ + ~ ~ ~ + ) / 0 ~ " 5 g Eqn (1 2 b) provides the rationale for the linear dependence observed in fig. 4 since An (in cmP3) is equal to [TiO,]/([TiO,]/q) ( V / N ) where q, V and N are defined in table 1.Thus the slope of the line in fig. 4, denoted K' in the Results section, yields K' = KVno/qN Furthermore, since K is dependent on pH through Efb = E;b-0.059pH, where E& is Efb at pH 0, it can be shown that a in eqn (9) is given by and the linear dependence of K' on pH is expected as observed in fig. 5. It is generally difficult to estimate the density of impurities, no, in the colloidal particle. However, Etb has been estimated by Gratzel and coworkers to be -0.12 V.7 Using the value for E& and the particle parameters given in table I , we can estimate the impurity density from the K' values or a (fig. 4 and 5). The value thus obtained is no = (1.1 k0.5) x lo1* ~ m - ~ . This is consistent with the assumption An > no leading to eqn ( I 1 a) and substantiates the argument for the lack of a depletion layer in the particles [eqn (lo)].The linear dependence of fig. 5 indicates that changes in the solution composition (e.g. pH) have little effect on the doping concentration in the colloid after its preparation. The Fermi level is raised by ca. 130 mV at the highest charge densities injected into the particles in the present study. This change could, however, be increased by increasing the dose/[particle] ratio or by using stronger reducing radicals. The effect of the former ratio on the Fermi level is similar to the effect of light intensity recently observed by Bard and The nature of the product, TiO;, cannot be determined by the experiments described above.However, Henglein has recently suggested a TiO, species obtained upon electron injection from (CH,),COH radicals to colloidal TiO,.,O With theE. BORGARELLO, E. PELIZZETTI, W. A. MULAC AND D. MEISEL 157 extinction coefficient given for this species (ca. 5 x lo3 dm3 mol-1 cm-l at 600 nm compared with 1.2 x lo4 dm3 mol-1 cm-l for MV+) the latter could not be the same species as that obtained in the present study. In the absence of a highly exoergic disproportionation reaction in the MV+ system it is not surprising that no dissolution of the colloid20 has been observed in the present study. This may also explain the lack of catalytic hydrogen formation in the propan-2-01 system, while the much more weakly reducing radical MV+ has been shown to evolve H, in this system.6 KINETICS OF ELECTRON TRANSFER The rate constant for electron transfer from MV+ to the TiO, particles is much slower than the diffusion-controlled limit.Thus from fig. 3 and table 1 a value of k I= (2.7k0.3) x lo7 dm3 mol-1 (particles) s-' is obtained, which is more than lo3 slower than the diffusion-controlled rate constant. Clearly, the rate-controlling step is the heterogeneous electron-transfer step. Interestingly, this rate constant is inde- pendent of dose, i.e. of the amount of charge transferred to the particle. This may indicate either two compensating effects (raising the Fermi level and decreasing the net positive change on the particle) or a reaction with a localized surface state. The fact that the same rate constant is obtained at pH 2 and 3 indicates that the rate constant is independent of AE in this pH region.The acceleration of the reaction at pH 1 is probably a result of fast protonation [reaction (2)], which at this pH is expected to follow immediately the electron-transfer step. However, the possibility that some TiO, is dissolved to yield a reactive hydroxide at high acidities cannot be EFFECT OF Pt DEPOSITION Addition of the TiO, particles to a solution containing colloidal Pt reduces the decay rate of MV+ as compared to that in the absence of TiO, (table 2). As previously observed, the reaction of MV+ with colloidal Pt is diffusion limited.1° The reduction in decay rate upon addition of TiO, is an indication of the adsorption of the Pt particles on the TiO, colloids.Such an adsorption will reduce the effective surface area available for MV+ reaction. The ratio of the decay rates in these two experiments [(b) and (c) in table 21 indicates that a surprisingly low fraction of Pt surface is available for the reaction once the Pt particle is adsorbed on the TiO,. In fact, since the electron-transfer reaction to the Pt particle is diffusion controlled, the increase in particle radius (from 16 A for Pt to ca. 70 A for the TiO, particles) should increase the decay rate by approximately a factor of four for the same surface area. The order of'magnitude decrease in the decay rate indicates, therefore, that < 10% of the original surface area is available for the reaction upon adsorption. The slowness of the electron-transfer reaction from MV' to the TiO, particles [experiment (a), table 21 indicates that the reaction on the platinized catalyst occurs directly by collision with the Pt island on the colloid.No fast transmission of electrons through the TiO, portion is possible in this system. As discussed above, owing to the small size of the colloidal particle, no significant Schottky barrier can be created at the Pt/TiO, junction. The effect of addition of Pt colloids to the TiO, colloid also provides information on the possible adsorption of MV+ on TiO, colloids prior to the electron-transfer step of' reaction (1). Since the ratio of concentrations of TiO, particles to Pt particles in experiment (c) of table 2 is ca. 3 : 1, one can exclude fast adsorption of MV+ (to yield an adsorbed radical which has a similar absorption spectrum to the free radical) on the TiO, particles.If this were the case, not more than 30% of the initially produced MV+ would have decayed in the process with Pt. However, a fast adsorption4esorption equilibrium of MV+ on TiO, could produce a decrease in the decay rate as observed in experiment (c) of table 2, even if the Pt particle is not adsorbed on the TiO, particles.158 REACTION OF VIOLOGEN RADICALS WITH TiO, Indeed, MV+ radicals have been observed to adsorb on an indium oxide transparent electrode and to have an absorption spectrum similar to that of the free radicaL21 Photodeposition of Pt by band-gap irradiation of the TiO, colloid produces a catalyst which is superior to the Pt colloid itself. Two factors may now contribute to the increase in the diffusion-limited decay rate: the increase in the particle radius and perhaps an increase in the effective surface area.The effect, however, is too small for this result to be analysed further. ADSORPTION AND DIMERIZATION OF v+ RADICALS An added complication to the system is the possible dimerization or further aggregation when hydrophobic viologen radicals (HV+ or C,,MV+) are produced. This aggregation is particularly severe in the C,,MV+ system. Aggregation will compete kinetically with the hydrogen evolution reaction and will furthermore reduce the free energy available for the latter reaction. The catalysed dimerization reaction in the HV2+/Ti0, system shows a short (ca. 0.1 s) induction period (fig. 7). We attribute this period to the establishment of an adsorption-desorption equilibrium : HV+ + (TiO,), (HV+)ads.(13) Since no spectral changes occur during the induction period, either only a small fraction of HV+ is adsorbed or (HV+)ads has the same spectrum as free HV+. The former seems to be the case, as is shown below. Once equilibrium (13) is achieved, a dimerization reaction occurs on a slower timescale. The dependence of the half-life of this reaction on dose indicates that this is a heterogeneous dimerization-desorption reaction : (14) In such a mechanism an exponential formation rate is expected to be first order in [HV+] provided [(HV+),,,] < [HV+]. Regardless of whether reaction (14) is a heterogeneous reaction, as depicted, or a surface reaction between two adsorbed molecules, the mere fact of the existence of HV+ in the adsorbed state for an appreciable period of time further indicates the slowness of the heterogeneous charge-transfer reaction.Low concentrations of photoplatinization do not completely prevent the dimer- ization reaction. Under the experimental conditions of fig. 9, only ca. 40 Pt atoms are produced per TiO,. particle if homogeneously distributed among the TiO, particles. Under such conditions the Pt islands on the TiO, particles seem to operate competitively independent of the TiO, particles themselves. Only when the coverage of TiO, by Pt reaches ca. 10% of the surface molecules will the hydrogen evolution reaction completely dominate. (HV+),,, + HV+ --+ (HV+),. The dedicated assistance of the ANL linac team and of P. Walsh and R.Clarke is gratefully acknowledged. E. P. and D. M. gratefully acknowledge NATO support under grant no. 1780. E. B. is grateful to the Swiss National Foundation and to Ciba-Geigy, Switzerland for their support. This work was performed under the auspices of the Office of Basic Energy Sciences, Division of Chemical Science, US-DOE, under contract no. W-3 1-109-ENG-38.E. BORGARELLO, E. PELIZZETTI, W. A. MULAC AND D . MEISEL 159 (a) E. C. Dutoit, F. Cardon and V. P. Gomes, Ber. Bunsenges. Phys. Chem., 1976, 80, 1285; (b) A. Fujishima, T. Inoue, T. Watanabe and K. Honda, Chem. Lett., 1978, 357; (c) A. K . Ghosh and H. P. Maruska, Electrochem. Soc., 1977, 124, 1516; ( d ) A. J. Bard, J. Photochem., 1979, 10, 59; (e) V. Guruswamy and J.O M . Bockris, Solar Energy Mater., 1979, 1, 141; (f) M. S. Wrighton, D. S. Ginley, P. T. Wolczanski, D. L. Morse and A. Liwz, Proc. Natl Acad. Sci. USA, 1975,72,2858; (g) M. V. Rao, K. Rajeshwar, V. R. Pai Verneker and J. Du Bow, J. Phys. Chem., 1980, 84, 1987; (h) F. T. Wagner and G. A. Somorjai, Nature (London), 1980, 285, 559; (i) G. N . Schrauzer and T. D. Guth, J. Am. Chem. Soc., 1977,99, 7189; (5) H. Von Damrue and W. K. Hall, J. Am. Chem. Soc., 1979, 101, 4373. A. Fujishima and K. Honda, Nature (London), 1972, 238. J. Kiwi, E. Borgarello, E. Pelizzetti, M. Visca and M. Gratzel, Angew. Chem., Znt. Ed. Engl., 1980,19, 647. E. Borgarello, J. Kiwi, E. Pelizzetti, M. Visca and M. Gratzel, J. Am. Chem. Soc., 1981, 103, 6324. E. Borgarello, J. Kiwi, E. Pelizzetti, M. Visca and M. Gratzel, Nature (London), 1981, 289, 158. D. Duonghong, E. Borgarello and M. Gratzel, J. Am. Chem. SOC., 1981, 103,4783. M. Gratzel and A. Frank, J. Phys. Chem., 1982, 86, 2964. (a) A. Henglein and J. Lilie, J. Am. Chem. Soc., 1981, 103, 1059; (b) D. S. Miller, A. J. Bard, G. McLendon and J. Ferguson, J. Am. Chem. Soc., 1981,103,5336. D. S. Miller and G. McLendon, J. Am. Chem. Soc., 1981, 103, 6791. l o (a) D. Meisel, W . A. Mulac and M. S. Matheson, J. Phys. Chem., 1981,85, 179. (b) M. S. Matheson, P. C. Lee, D. Meisel and E. Pelizzetti, J. Phys. Chem., 1983, 87, 394. l 1 P. A. Brugger, P. P Infelta, A. M. Braun and M. Gratzel, J. Am. Chem. Soc., 1981, 103, 320. l 2 M. Kreig, M. P. Pileni, A. M. Braun and M. Gratzel, J. Colloid Interface Sci., 1981, 83, 209. l 3 J. Moser and M. Gratzel, Helv. Chim. Acta, 1982, 65, 1436. l 4 M. Corti and V. Degiorgo, Ann. Phys. (Paris), 1978, 3, 303. l5 (a) J. Kiwi and M. Gratzel, J. Am. Chem. Soc., 1979, 101, 7214; (b) K. Monnserrat, M. Gratzel and P. Tundo, J. Am. Chem. Soc., 1980, 102, 3689. l6 J. Turkevich, K. Aika, L. L. Bau, I. Okura and S. Namba, J. Res. Inst. Catal., Hokkaido Univ., 1976, 24, 54. l 7 B. Krautler and A. J. Bard, J. Am. Chem. SOC., 1978, 100,4318. l a S. R. Morrison, The Chemical Physics of Surfaces (Plenum Press, New York, 1977), chap. 2, p. 25. l9 M. Ward, J. White and A. Bard, J. Am. Chem. Soc., 1983, 105, 27. 2o A. Henglein, Ber. Bunsenges. Phys. Chem., 1982, 86, 241. 21 E. Steckhan and T. Kuwana, Ber. Bunsenges. Phys. Chem., 1971, 78, 253. 22 A. Henglein, in Photochemical Conversion and Storage of Solar Energy, Part A , ed. J. Rabani 23 S. Morrison, The Chemical Aspects of Surfaces (Plenum Press, New York, 1977), p. 37. 24 C. F. Baes and R. E. Mesmer, The Hydrolysis of Cations (John Wiley, New York, 1976), pp. 147-152. ’ D. Duonghong, J. Ramsden and M. Gratzel, J. Am. Chem. Soc., 1982, 104, 2977. (Weizmann Science Press, Jerusalem, 1982), p. 1 15. (PAPER 4/670)

ISSN:0300-9599    

DOI:10.1039/F19858100143

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1985, 81, 161-173 Retention Volumes and Retention Times in Binary Chromatography Determination of Equilibrium Properties BY BRYAN A. BUFFHAM,* GEOFFREY MASON AND GANAPATI D. YADAV~ Department of Chemical Engineering, Loughborough University of Technology, Loughborough, Leicestershire LE 1 1 3TU Receiwd 25th April, 1984 A change in the composition (accompanied perhaps by a change of flow rate) of a binary gas mixture flowing into a column which can adsorb both species gives rise to a chromatographic transient. Observed at the outlet, this is seen as the concentration varying sigmoidally with time; there is also a flow-rate fluctuation. A theory has been developed which relates these transients to the phase-equilibrium properties. A retention time is defined for the concentration transient and another for the flow transient; the slopes of the equilibrium curves are shown to be explicit functions of these retention times.Experiments to determine the slopes at infinite dilution of the equilibrium curves for the simultaneous adsorption of argon and nitrogen on Linde 5A molecular sieve by measuring both the retention times are reported. The passage of a chromatographic pulse or front through a column depends on the nature of the equilibrium between the phases and on the rates of those processes which tend to bring the phases to equilibrium. Many physico-chemical properties may be determined by the examination of chromatograms.' The time taken for a transient to pass through a column, the retention time, depends mainly on equilibrium properties while the shape of the chromatogram is determined by the rate processes.If the retention time is properly the rule by which the retention time is calculated from the chromatogram is independent of the mechanism of the rate processes, so that the retention time depends only on the equilibrium properties and is independent of the mass-transfer rates and the degree of departure from equilibrium. Within a column, in the neighbourhood of a transient, the velocity of the mobile phase depends on position relative to the pulse or front. For example, if a single sorbable species is being transported by a non-sorbed carrier, the mobile phase velocity is greater at positions where there is sorbed material: the more material sorbed the greater the velocity.Bosanquet and Morgan4 called this phenomenon the sorption eflect. The sorption effect is small at high dilution but affects the passage of large samples and of perturbations of finite concentrations. Our main purpose here is to present a theory for the retention time in binary gas chromatography that is valid whatever the nature of the transport processes. Our theory relates the slopes of phase-equilibrium curves to the composition transient and the sorption effect, and is expressed in terms of natural chromatographic variables. We also report some preliminary experiments which apply the theory to simultaneous measurements of the composition transient and the sorption effect. f Present address : Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario N2L 3G1.Canada. 161162 EQUILIBRIUM PROPERTIES IN BINARY CHROMATOGRAPHY 1.0 -XZ 0 I J P - CZ 0 0 P CI - I I 0 1.0 XI - Fig. 1. Slopes of the equilibrium curves for an adsorbent in equilibrium with a binary gas of molar concentrations c1 and c, (mole fractions XI and X,); q1 and q2 are the molar concentrations in the solid. MODEL-SPECIFIC RETENTION EXPRESSIONS Chromatography is said to be ideal when the velocity profile is flat, the phases are locally equilibrated and axial mixing is absent. When fronts are sharp in ideal chromatography there is no difficulty in identifying the retentior time. Peterson and Helfferi~h,~ Conder and PurnelP and others have taken the sorption effect into account in versions of the theory of ideal chromatography.For binary finite-concentration ideal chromatography with both components sorbable by the stationary phase, Peterson’s and Helfferich’s result for the retention volume, i.e. the volume flowing in the retention time, is where V is the column volume of which a fraction E is void, XI and X , are the mole fractions of species 1 and 2 and ql, q2 and c,, c, are the molar concentrations in the163 stationary and mobile phases, respectively, of the indicated species. It is assumed in the derivation of eqn (1) that the fluid is of constant molar density p ; this is the case for gases at low pressure and ideal solutions of components of the same molar volume. Fig. 1 illustrates the notation and displays the slopes of the equilibrium curves. Recently Ruthven and Kumar' have shown that eqn (1) is still valid in the presence of axial dispersion in the mobile phase.An outline of their proof is of interest because it shows that their analysis could be extended to include other types of (chromatographic) non-ideality. Their starting point is a material balance for species i in a stream in piston flow with interstitial velocity u at position z undergoing axial dispersion with dispersivity D : B. A. BUFFHAM, G. MASON AND G. D . YADAV A linearized equation for the composition transient can then be developed as follows: (i) write eqn (2) for substances 1 and 2 and substitute x+x and z - x for Xl and X,, where and x are initial equilibrium values and x is a deviation variable; (ii) multiply the resulting equations by x and G, respectively, and then subtract one equation from the other; (iii) discard the term xau/az; (iv) substitute (dq,/dci)p@x/at) for d q i / a t .The linear equation so obtained for the deviation, x, of X I is where (4) Eqn (3) is a convective dispersion equation with an effective interstitial velocity of The corresponding retention volume is given by eqn (1) and the transit time is t , = - &+(1-&) x p + q L v ( 2, 31 where L is the column length. Ruthven and Kumar7 identify the retention time for a step response by using the point of inflexion of the chromatogram. The sorption effect is present in eqn (2) in the non-linear term involving the interstitial velocity. Linearization could proceed in the same way as above if the mass-transfer and mixing phenomena were represented differently, which means that there are other models for which eqn (6) is valid.A difficulty in the use of eqn (6) for the determination of equilibrium properties is that it involves the slopes of both equilibrium curves. To overcome this problem, Ruthven and Kumar' used a computational scheme in which K and the slopes are represented by polynomials. We shall show that another way of overcoming the problem is to measure the flow-rate transient (sorption effect) in addition to the composition transient. MODEL-INDEPENDENT RETENTION EXPRESSIONS The usual procedure adopted in devising theories of chromatography is to incorporate terms representing local rates of mass transfer into differential material balances to obtain partial differential equations.In the case of ideal chromatography164 EQUILIBRIUM PROPERTIES IN BINARY CHROMATOGRAPHY time. t Fig. 2. Change in the molar outflow rate of species i, N , ( t ) , in response to a change of the molar inflow rate of i from Mi to Mi + M t . Note that this figure refers to a single column. the equations obtained are relatively simple because axial mixing is absent and transverse equilibration is instantaneous. The theory of multicomponent ideal chromatography is available.x, starts with material balances and avoids the need to specify mass-transfer rates. It shows that retention times defined in terms of arithmetic means depend on conditions before and long after the transient, but not on the rate processes. I t is imagined that the column is initially in a steady state and that the feed is changed and the column then evolves to a new steady state.The material balances represent the change in the inventory of each species in the column in response to this change. In suitable cases the steady states are equilibrium states in which the column is in equilibrium internally and with the feed. Suppose that the molar rate at which species i enters the column is changed at t = 0 from Mi to Mi+ M I . in response the outlet rate N,(t) changes as shown in fig. 2. In the model-independent t h e ~ r y , ~ the retention time for the ith species is defined as In contrast, model-independent chromatography theory'. It has a value corresponding to a position on the sloping part of the curve in fig. 2, This retention time is related to H t , the change in the total molar holdup of species i in the column as the system progresses from the initial to the final steady state, by3 ti = H t / M : .(8) The sorption effect is automatically included in this analysis: N , ( t ) depends on the flow rate in addition to the composition of the effluent stream. Pulse perturbations can be treated in the same way.3 For rapid injections small enough for the response to be linear, the response per unit of extra i added is and eqn (7) again gives the retention time ti. A drawback of impuse injection is that the concentrations in the forward part of the column might lead to non-linear behaviour.B. A. BUFFHAM, G. MASON AND G. D . YADAV 165 I .o 0 0 0 Fig. 3. Perturbation variables (a) for the composition transient and (b) for the total molar outflow rate.Note that x is the composition perturbation for species 1 and that this figure refers to a single column. BINARY CHROMATOGRAPHY MEASURABLE VARIABLES A typical gas chromatograph is provided with a pair of matched columns and a detector whose output signal depends on the difference between the streams leaving the columns. For binary gas streams that differ slightly in composition, the signal is proportional to the difference in mole fraction. Usually a soap-bubble meter is provided to measure the volumetric flow rate which, for a gas at low pressure and constant temperature, is proportional to the molar flow rate. This suggests analysis of chromatographic transients in terms of the departures of mole fractions and molar flow rates from initially established values.We shall assume that the columns are isothermal and that the pressure drop is low enough for the steady states to be equilibrium states.166 EQUILIBRIUM PROPERTIES IN BINARY CHROMATOGRAPHY RETENTION VOLUMES Suppose that in the initial equilibrium state the total molar flow rate through the column is W and the mole fractions of the mobile phase are z and X. In terms of perturbation variables n(t) and x(t), defined to be zero at t = 0 (see fig. 3), (10) (1 1) For small perturbations, substituting eqn (10) and (1 1) into eqn (7) yields, ignoring 4 0 x ( 0 , W) = w + Wl [xl) + 401 N,(O = "" + 401 LG - x(0l. Eqn (12) and (13) may be solved to give the following expressions for the integrals: The M t ti terms can be related to holdups by using eqn (8): M: ti = Hz = EV&i+(l - E ) V6qi (16) where 6ci and Sqi are the changes in the mobile- and stationary-phase molar concen- trations as the column evolves from the initial to the final equilibrium state.For the case where the mobile phase has a constant molar density 6C' = -6c,. (1 7) Moreover, 6ql/6cl and 6q2/6c, will be the slopes of the equilibrium curves if the perturbation is not too large, and eqn (14) and (15) become: We call the retention uolume for the composition transient and Vn the retention volume for the $ow transient. Thought of as a retention-volume relation, eqn (18) is the same as eqn (1). What is new is first that the relation holds irrespective of the nature of the mass-transfer mechanism and secondly that the integral provides the definitive rule for evaluating V, from an x against t curve.Eqn (19) is new. It is important because it allows dq,/dc, and dq,/dc, to be calculated directly and unambiguously from the measurable quantities V, and V,. As a practical matter, it will usually be convenient to assign subscript 1 to the more strongly sorbed species in order to make V, positive. When the adsorption properties of the two species are alike, dq,/dc, = dc,/dc, and there is no sorption effect.B. A. BUFFHAM, G. MASON AND G. D . YADAV 167 If the mobile phase is an ideal solution of components with molar densities p1 and p2, eqn (17) is replaced by dc, = -(pl/p2) dc,; if the mobile phase is not an ideal solution, partial molar volumes are involved.Counterparts of the later equations may be derived but are more complex. RETENTION TIMES As N" = pQ", where Qo is the volumetric flow rate corresponding to IV, and dc, = pdX,, it follows that the above relations for retention voEumes may be translated into expressions for retention times: This pair of equations yields the slopes of the equilibrium curves: dq, - 7, + K zn -&V/Q" dc1 (1 -&I v/Qo dc2 (l-&)V/Qo * dq, - z x - X zn-&V/QO Eqn (22) and (23) give dql/dc, and dq2/dc2 in terms of z, and z,, which are related to measurable quantities via eqn (20) and (21). Chromatographs are often provided with integrators. By using integration by parts, the retention times may be related to areas on records of how x and n vary with time. Define (fig.3) x*(t) = x(OO)-x(t) (24) and n*(t) = n(Co)-n(t). (25) Then 1 " O z =-1 x* dt z, = - jan*dt. SXl 0 N " w 0 (27) The integrals in eqn (26) and (27) represent the areas between the x against t and n against t curves and their respective asymptotes for large values o f t (see fig. 3). It is usually convenient to think of unpacked parts of the columns, connecting tubing etc., as dead volume rather than as contributing to E . If a column contains dead volume V,, VD/Qo must be deducted from the value of z, calculated from a chromatogram by using eqn (20) or (26). So far our analysis has considered a single column. We shall now show how dual-column chromatograph construction can be exploited. MATCHED COLUMNS For small perturbations, the response will be (mathematically) linear : a stimulus of twice the strength results in a response that is twice as great.For perturbations that differ only in sign, the responses will be of the opposite sense. Now imagine a chromatograph with precisely matched columns being fed with streams of slightly different compositions at exactly the same (constant) molar rates.168 EQUILIBRIUM PROPERTIES IN BINARY CHROMATOGRAPHY Fig. 4. Schematic diagram of the apparatus. Rotating the valve through 90" interchanges the flows. C1 and C2, columns; K, katharometer; NVI and NV2, needle valves; P, deflecting- diaphragm differential-pressure transducer; R 1 and R2, capillary-flow restrictors; RV, four-port reversing valve. These flows have been maintained for so long that steady conditions prevail. At the beginning of an experiment the feed streams are interchanged and the columns experience opposite step changes in their feed composition.The composition responses seen on the two sides of the detector (e.g. katharometer) will be equal and opposite. As the detector measures the difference in composition between the two streams, the indicated response will be twice the response that would be observed for a step change in the composition of a single stream. Similarly, if devices are fitted to measure the effluent rates, the difference in their readings will be twice the variation in flow from one of the columns. Thus x(t) and n(t) [or x*(t) and n*(t)] are directly measurable by using a two-stream chromatograph equipped with a differential flow meter in addition to the usual differential composition detector.DEMONSTRATION EXPERIMENTS We report here some preliminary experiments designed to demonstrate the feasibility of measuring composition and flow (sorption effect) transients simultaneously, and of using these measurements to determine the slopes of equilibrium curves. A standard (Pye model 104) katharometer chromatograph was the basis of the apparatus. It was modified to allow the feed streams to be interchanged quickly and the difference in the outlet flow rates to be measured. Capillary flow restrictors (70 cm x 0.5 mm diameter) were fitted to the column outlets. A differential pressure transducer (Mercurey, 10 mm H,O) connected between the restrictor inlets gave an indication that depended on the difference in flow rates.A four-port Valco valve was fitted to the column entrances to permit the feed streams to be interchanged (see fig. The chromatograph was fitted with 4 mm diameter columns, each packed with 13.2 g of Linde 5A molecular-sieve adsorbent, giving a packed length of 1.5 m. The particle size was in the range 0.5-0.7 mm. Between 310 and 370 K the temperature could be controlled to within 1 K. Below 3 10 K, control was erratic. Our experiments were performed at 313 K. Argon + nitrogen mixtures were chosen for study because their isotherms are nearly straight: the values of dqJdc, and dq,/dc, are nearly constant at a given temperature.' This means that isotherm curvature is not a source of non-linearity and that step size is relatively unimportant. Important values of dq/dc are those at infinite dilution (see fig.5). In order to study, say, nitrogen at high dilution in argon, pure argon was fed into one column and nitrogen containing a little argon into the other. The gases, 4).B. A. BUFFHAM, G. MASON AND G. D . YADAV - c2 P 0 z .5 c! 2 E 0 .r( c c 5 8 3 E 2 9 2 d 0 0 P c , - molar concentration in gas 169 Fig. 5. Slopes of the equilibrium curves at infinite dilution. are the Henry’s-law constants. regulated at a pressure of 0.5 bar above atmospheric pressure, passed to the reversing valve via fine-control needle valves. Most of the pressure drop was at the needle valves so that the feed flow rates were virtually constant, as assumed in the theory, and the column pressure was not substantially different from atmospheric. Experiments were performed in the way already suggested: the columns were allowed to come to the steady state with no composition or flow-rate drift. Then the streams were interchanged by operating the Valco valve.Fig. 6 and 7 show typical results. The sigmoidal composition response is just what would be expected. The flow response returns to the original baseline because in the final steady state the flows are the same as they were in the initial steady state. Although the input flows are maintained constant, the output flows vary. Each column experiences a step change in feed composition and so has either an excess or deficit of adsorbed gas in relation to the final condition. Consequently the outlet flows differ from their respective inlet flows and each other during the transient.The composition and differential flow scales were calibrated by the injection of known volumes of gas into one stream with the flow of both streams otherwise steady. Flow-rate changes of a few percent did not affect the katharometer reading and the influence of composition on viscosity is not enough to affect the differential flow measurements seriously. Flow rates in the steady state were measured in the usual way with a soap-bubble meter. The areas represented by the integrals in eqn (26) and (27) were reproducible within a few percent, composition-difference reproducibility being somewhat better than170 EQUILIBRIUM PROPERTIES IN BINARY CHROMATOGRAPHY 0 100 200 300 tls Fig. 6. Composition and flow transients for a step change with = 0, i.e.nitrogen at high dilution in argon. flow-difference reproducibility. Very similar results were obtained with composition step changes of about 0.015, 0.02 and 0.025 mole fraction (see table 1). Retention times were calculated from eqn (26) and (27). The appropriate forms of eqn (22) and (23) at infinite dilution are: (a) for nitrogen in argon, for which = 0 and = 1, and (b) for argon in nitrogen, for which = 1 and = 0, The slopes of the equilibrium curves at infinite dilution were found by using these171 B. A. BUFFHAM, G. MASON AND G. D. YADAV * 0 2 0 40 60 80 100 120 t l s Fig. 7. Composition and flow transients for a step change with = 1, i.e argon at high dilution in nitrogen. Table 1. Experimentally determined retention times for nitrogen (1) + argon (2) mixtures passing through a Linde 5A molecular-sieve column retention time volumetric composition, step size, composition, flow, flow rate, XI 6x1 % I S % / S Qo/cm3 s-' 0.00 0.01 1 128.9 64.1 0.622 0.00 0.017 131.0 63.7 0.624 0.00 0.026 129.9 66.7 0.624 1 .oo 0.016 57.3 57.2 0.634 1 .oo 0.021 56.8 53.7 0.634 1 .oo 0.026 58.6 55.1 0.633 forms of eqn (22) and (23) (using V = 18.85 cm3, E = 0.376 and making allowance for dead volume of 1.5 cm3 per column).The results are reported in table 2. DISCUSSION Our experiments demonstrate that it is feasible to make the measurements we suggest. In particular it is possible to measure the retention time for the flow transient.172 EQUILIBRIUM PROPERTIES IN BINARY CHROMATOGRAPHY Table 2. Limiting slopes of the equilibrium curves for the simultaneous adsorption of nitrogen (1) and argon (2) on Linde 5A molecular sieve this work, Ruthven and Kumar, 313 K 304 K dc, 0.00 6.3 2.8 10.3 3.7 1 .oo 5.5 2.5 11.5 2.6 In addition to our own results we give in table 2 estimates based on the data of Ruthven and Kumar.' These authors give q against c curves determined by their chromato- graphic method (see above) for mixtures of nitrogen and argon adsorbed on 5A sieve.They express q as molecules per cavity and c as partial pressure in Torr. In making our estimates from their data we have used Ruthven's value of 1 molecule per cavity = 0.45 mmol g-l of Linde pellet1* and taken the pellet density to be 1.12 g ~ m - ~ . Our findings are consistent with those of Ruthven and Kumar when it is borne in mind that their lower temperature would give a greater degree of adsorption and consequently steeper slopes.Judging from table 1, the reproducibility of our retention times is ca. 2-3 % for composition and ca. 5% for flow. This is encouraging in a preliminary study. It might be thought that the close values of z, and z, when XI = 1 (see table 1) would render the calculation of the limiting slopes prone to error. This is not the case, as can be seen from eqn (22a and b). Note also that z, and z, are proportional to V/Q" [see eqn (20) and (21)], so that dq,/dc, and dq,/dc, are insensitive to V/Qo. (In principle, thermodynamic properties cannot depend on V and Qo.) Eqn (22) and (23) only become sensitive to experimental error when the gases are hardly adsorbed and there is a large void volume.This was not a problem with the experiments reported here, as can be seen by comparing our value of only 11.4 s-l for EV/Q" with the values of z, and z, in table 1. We are not certain of the reason for the oscillations at the start of the flow transient (see fig. 6 and 7). Perhaps it is due to imperfect matching of the columns. The integrals were evaluated literally and no attempt was made to 'correct' for the oscillation. In any case, uncertainty about the early part of the flow transient would not greatly affect the value of t,. CONCLUSIONS Expressions have been derived that relate equilibrium properties to the sort of measurements that are natural in chromatography. These relations are independent of the nature of the mass-transfer and mixing processes involved in chromatography.It has been assumed that the perturbations involved are small enough for the response to be linear. This is a constraint imposed for convenience in the mathematical analysis and is not equivalent to assuming that the column is always near equilibrium. This assumption and others (the nature of the mobile phase, no volume change of stationary phase) can be removed at the expense of complicating the analysis. A further refinement would be to consider the detectors to respond simultaneously to changes in both flow and composition. Such problems are best treated on an ad hoe basis by methods parallel to those given here.B. A. BUFFHAM, G. MASON AND G. D. YADAV 173 The retention expressions for the composition transient are the same as those reported by previous workers5~ for particular models. The importance of also introducing the flow transient is that it permits the derivation of explicit expressions for the slopes of the equilibrium curves. Preliminary experiments to measure the flow transient simultaneously with the composition transient were carried out successfully for the adsorption of mixtures of nitrogen and argon on Linde 5A molecular sieve. The slopes of the equili- brium curves were calculated from these results. The experiments were performed by D. I. Smith and B. R. Tookey. G.D.Y. gratefully acknowledges a Leverhulme Research Fellowship. J. R. Conder and C. L. Young, Physicochemical Measurement by Gas Chromatography (Wiley, Chichester, 1979). B. A. Buffham, Proc. R. SOC. London, Ser. A , 1973, 333, 89. B. A. Buffham, Proc. R. SOC. London, Ser. A , 1978, 364, 445. C. H. Bosanquet and G. 0. Morgan, in Vapour-phase Chromatography, ed. D. M. Desty (Butter- worths, London, 1957). D. L. Peterson and F. Helfferich, J . Phys. Chem., 1965, 69, 1283. J. R. Conder and J. H. Purnell, Trans. Faraday Soc., 1968,64, 3100. D. M. Ruthven and R. Kumar, Ind. Eng. Chem. Fundam., 1980, 19, 27. * H-K. Rhee, R. Aris and N. R. Amundson, Philos. Trans. R. SOC. London, Ser. A , 1970, 267, 419. F. Helfferich and G. Klein, Multicomponent Chromatography (Marcel Dekker, New York, 1970). lo D. M. Ruthven, AIChE J., 1976, 22, 753.

ISSN:0300-9599    

DOI:10.1039/F19858100161

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1985, 81, 175-183 Hydrogenation of Buta- 1,3-diene on Supported Metal Catalysts in Aqueous Solution Part 1 .-Differences in the Catalytic Action of A120,-supported Pt, Pd, Rh and Ru Catalysts BY KATSUAKI SHIMAZU* AND HIDEAKI KITA Department of Chemistry, Faculty of Science, Hokkaido University, Sapporo 060, Japan Received 30th April, 1984 Deuteration of buta-1,3-diene on Al,O,-supported Pt, Pd, Rh and Ru catalysts in 0.5 mol dm-3 D,SO, has been examined to see if there are differences in their catalytic action. The results are discussed on the basis of a simple common scheme in which butenes are formed via half-hydrogenated species and deuterium uptake proceeds by the reversible occurrence of two steps, namely (i) the formation of the half-hydrogenated species and (ii) the ionization of adsorbed hydrogen [x(a) + Xf + e-, X = H or D].Simulation of the distribution of deuterium in the products leads to the conclusions that for the former step the rate follows the order Pd > Pt % Rh > Ru and for the latter step the rate follows the order Pt % Pd > Rh > Ru. H/D exchange is fastest on Pt and slowest on Ru. The hydrogenation of lower unsaturated hydrocarbons has been studied on metal catalysts (electrodes) in aqueous solution, e.g. ethylene,' propene,2 butenes,, buta- 1 ,3-diene,,-' buta- 1 ,2-diene8 and but-2-yne9 on Pt and buta- 1,3-diene on Pd.lo7 l1 These studies confirm the usefulness of hydrogenation in aqueous solution (i.e. in an electrochemical system) for elucidating the reaction mechanism for the following reasons.(i) The potential of the catalyst during the reaction reflects the activity of the adsorbed hydrogen as calculated from the Nernst equation for a reversible elementary step, H(a) e H+ +e-. (ii) Its adsorption state and amount are readily measured in situ by using a potential-sweep method. (iii) The above reversible step increases the fraction of adsorbed deuterium atoms,f,, in the total adsorbed hydrogen (H and D) atoms in D20 solution more than in gas-phase D, and hence highly deuterated products are formed. Phillipson and Burwell12 used a deuterium-containing organic solvent but the effect onfD was not as large as in D,O solution. No other electrochemical work has been reported on the hydrogenation of buta-1,3-diene, but very many studies in the gas phase for various transition metal catalysts such as Pt,13-16 Pd,13-159 l7 Rh,l3? 1 5 9 l6 Ru,l3? Cols-zO and Nil8, 20* 21 have been reported. The two main reaction mechanisms reported are those which include adsorbed buta- 1,3-diene and half-hydrogenated species with delocalized or localized 7t orbitals.The former agrees with the reaction mechanism we proposed. All of these mechanisms are of the associated type where two adsorbed hydrogen atoms are added stepwise. Hence we regard the associated mechanism as being generally acceptable. In the present paper the deuteration of buta-1,3-diene was performed on A120,- supported Pt, Pd, Rh and Ru catalysts in 0.5 mol dm-, D2S0, solution in order to examine the catalytic differences of the dispersed metals.(i) and (iii) above hold in the present case, but (ii) does not because of the insulating nature of A120,. Instead 175176 HYDROGENATION OF BUTADIENE ON Pt, Pd, Rh AND Ru Table 1. Catalysts used in this work and their preparation conditions preparation conditions amount of catalyst metal (wt%) method metal salt calcination reduction in H, Pt/Al,O, 0.50 impregnation H,PtCl, 723 K, 3 h in 0, 573 K, 3 h Pd/Al,O, 0.48 adsorption PdCI,-HCl 603 K, 5 h in air 723 K, 4 h Rh/Al,O, 0.50 adsorption RhCl, 603 K, 5 h in air 723 K, 3 h Ru/Al,O, 0.38 adsorption RuC1, 473 K, 5 h in air 623 K, 2 h Table 2. Dispersion of metal and deuteration of buta-1,3-diene in 0.5 mol dmP3 D,SO, ~~ ~ dispersion products (%) DNb of from from but- trans- cis-but- cis-but- catalyst CO ads.t.p.d. kDa $,,/mV butane 1-ene but-2-ene 2-ene 2-ene Pt/A1,0, 0.61" 1.38 1.8 140f30 47 32 11 10 5.1 Pd/A1,0, 0.67d 0.87 3.5 225+5 2 57 32 10 2.3 Ru/Al',O, 0.04d 0.09 3.8 210+25 10 52 19 20 1.9 Rh/A1,0, 0.75" 0.88 2.9 220+20 11 45 25 19 2.1 a Units: lop2, mol min-' mmHg-' metal-atom-'. DN = &Xi, where Xi is the fraction of [,Hi] species. Catalyst reduced at 723 K. Catalyst reduced at 573 K. of the potential-sweep method, we used the temperature-programmed desorption (t.p.d.) of hydrogen from the catalysts. The adsorption state of hydrogen and the H/Pt stoichiometry obtained by electrochemical measurements have been shown22 to be very similar to those for the gas/solid interface. Hence, the results of the t.p.d. measurements are valid for the present system.f, may vary with the metal because the value off, depends on the reversibility of the formation of the half-hydrogenated species and the ionization of adsorbed hydrogen.EXPERIMENTAL CATALYST All catalysts used in this study were reference catalysts supplied from the Catalysis Society of Japan and are listed in table 1 together with preparation conditions. The catalysts were prepared as follows.23 Each metal component was loaded from an aqueous solution of its salt onto a support oxide by an impregnation or adsorption method. After drying, the catalyst was calcined at the desired temperature for several hours and finally reduced in H, gas for 2-4 h. The y-Al,O, support has a B.E.T. surface area of 174 mz g-' and a pore volume of 0.66 mm3 g-l. The dispersion of metal was measured by the adsorption of C02, and typical values are listed in table 2.TEMPERATURE-PROGRAMMED DESORPTION T.p.d. of hydrogen was carried out using home-made apparatus equipped with a gas chromatograph without a column. A U-tube containing small amount of 3A molecular sieve was placed between the sampling tube and a thermal conductivity detector to remove moisture.K . SHIMAZU AND H. KITA 177 The carrier gas, commercial Ar (4N purity), was further purified by 13X molecular sieves, Cu and Ni gauze catalysts and another molecular sieve. A sample of catalyst (ca. 200 mg) was first heated in air and then in a stream of Ar at the highest temperature in the pretreatment procedure, T,,,, for 1 h. It was reduced in flowing H, at 523 K for 1 h and then the flowing gas was switched to Ar.To desorb the hydrogen adsorbed in the reduction treatment, the catalyst was heated at T,,, for 1&20 min. The catalyst was then cooled to room temperature under vacuum and exposed to 1 atm of H, for 30 min. All desorption runs were carried out at a rate of 8 K min-' and at an Ar flow rate of 50 mm3 min-l. REACTION PROCEDURE The catalyst (100 mg), after heating in an H, stream at 523 K for 1 h, was suspended in 0.5 mol dmP3 D,SO, and re-reduced by bubbling D, (Showa Denko, 7N purity, 99.97 D% , P,) = 15-50 mmHg, 1 mmHg = 133.3 N m-,) through the solution. The reaction gas was a mixture of buta-1,3-diene (4N purity, PR = 14 mmHg), D, (P,, = 28 mmHg) and He (7N purity) at a total pressure of 1 atm. The reaction was conducted by circulating the reaction gas through the solution at 293+ 1 K.A blank test confirmed that no reaction occurred without the catalyst. The potential of the powdered catalyst was monitored using a gold probe electrode during the pretreatment by D, and the reaction. The catalyst potential is determined by a charge transfer which takes place on the direct contact of a Pt particle on the support with the gold probe electrode. All the potentials were referred to the reversible hydrogen electrode in the same solution (RHE). ANALYSIS Analyses of products were conducted gas chromatographically and mass spectrometerically, as described ear lie^.^ RESULTS T.P.D. SPECTRA We first describe temperature-programmed desorption spectra of hydrogen pre- adsorbed at room temperature in order to investigate the surface state of the catalysts.The spectra for Pt, Pd and Rh catalysts [fig. I (a)-(c)] have a main peak at 363-383 K and another smaller peak (or shoulder) at 473-573 K. The relative intensity of these two peaks is dependent on the metal. On Ru/A1,0, [fig. l(d)], only one small broad peak was observed. The appearance of the two peaks clearly shows the presence of two kinds of adsorbed hydrogen species. The peak area gives the amount of adsorbed hydrogen and the dispersion of the metal. The latter is compared with the dispersion obtained by the CO adsorption method in table 2. Except for Pt/Al,O,, they are in good agreement with each other if one takes into account the difference in the pretreatment temperature and the uncertainty in the stoichiometry for CO adsorption.The dispersion in excess of unity for Pt/AI,O, is reproducible and may be attributed to hydrogen spillover or to multi-hydrogen bonding on a Pt atom. In the following discussion, the dispersions obtained from t.p.d. spectra are used because the CO adsorption was carried out on the catalysts reduced at a different temperature from the pretreatment temperature. REACTION RATE AND DISTRIBUTION OF PRODUCTS Under the present conditions, the reaction rate for the hydrogenation of buta- 1,3-diene in aqueous solution is first order in the partial pressure of hydrogen on Pt4 and Pd1° catalysts. This is also the case on Rh and Ru catalysts, as confirmed in this study over a pressure range from 15 to 50 mmHg. First-order rate constants per exposed (surface) metal atom, kI,, are listed in table 2.Except for Pt/AI,O, these are very similar, independent of the metal. The value of kD for Pt/A1,0, is approximately one-half that of the others. 7 FAR 1178 HYDROGENATION OF BUTADIENE ON Pt, Pd, Rh AND Ru (4 1 300 400 500 600 700 800 Fig. 1. T.p.d. of hydrogen preadsorbed at room temperature on Al,O,-supported ( a ) Pt, (6) Pd, (c) Rh and ( d ) Ru catalysts. Heating rate, 8 K min-l; Ar flow rate, 50 mm3 min-l. The potential of the catalyst at 5 min, #oc (us. RHE), is also listed in table 2. The reproducibility was not very good, but the averaged values are close to those obtained on pure metal electrode^.^^^^ All of them are much more positive than the value expected from the Nernst equation for the hydrogen electrode reaction, indicating the deficient supply of hydrogen to the catalyst surface.Among the catalysts used here, Pt/Al,O, exhibits the lowest potential, indicating the least effect of the adsorbed buta- 1,3-diene. Distribution of the products is largely dependent on the metal (table 2). In particular, the selectivity of butenes, S, defined as T l K amount of butenes amount of butenes and amount of butane S = varies from metal to metal. The metals can be placed in the following sequence of decreasing selectivity: Pd > Rh z Ru > Pt, which confirms the order observed in the gas phase.13 ISOTOPIC DISTRIBUTION IN BUTENES The distribution of deuterium in the butenes is shown in fig. 2. The present results for Pt/Al,O, reproduce the previous ones for the same catalyst.6 In this case, each butene takes up many deuterium atoms. In particular, the average deuterium number (DN = cp, iXi, where Xi is the fraction of [ 2 H i ] species) of cis-but-2-ene is 5.1.The distributions, especially trans-but-2-ene, show a saw-toothed shape. The reaction mechanism on Pt/Al,O,, has already been discussed in detail.6 The distributions on Pd/A1,0, substantially reproduce those on Pd foil electrode under similar conditions.1° Little effect of the dispersion of Pd is observed. The distributions show a common shape with a maximum at C2HJ species for all the butenes.K. SHIMAZU AND H. KITA 179 60 40 h < 20 0 0 2 4 6 8 number 0 2 4 6 8 of D atoms Fig. 2. Distributions of deuterium in but-1-ene (O), trans-but-2-ene (A) and cis-but-2-ene (0) formed on the deuteration of buta-1,3-diene by D, on Al,O,-supported (a) Pt, (b) Pd, (c) Rh and ( d ) Ru catalysts in 0.5 mol dmP3 D,SO,.Conversions are 9.0, 6.9, 13.8 and 1.4% for Pt, Pd, Rh and Ru, respectively. Similar distributions are obtained on Rh/Al,O, and Ru/A1,0,, but narrowed with a maximum at [,H,] species, which is sharpest for Ru/Al,O,. The fractions of [,H,] and highly deuterated species become very small on Ru/Al,O,. The average deuterium number falls in the sequence (table 2): Pt % Pd > Rh > Ru. The average deuterium number is a measure of the amount of H/D exchange between the adsorbed intermediates and the adsorbed hydrogen. HYDROGENATION BY H, IN 0.5 mol dmP3 D,SO, Buta-1,3-diene was hydrogenated by H, in 0.5 mol dmP3 D,SO, instead of D, to estimate the reversibility of the step: X(a) g X+ + e- (X = H or D).If the reversibility is sufficiently large, the same distributions of deuterium are expected for reactions with both D, and H,. If not, the deuterium uptake should be suppressed to a large extent. The results are shown for the distribution of deuterium in but-1-ene in fig. 3. The former case approximately holds on Pt, while on the other metals the distributions are entirely different and the fraction drops sharply from [,H,] to [,HI], [,H,] and [2H,]. 'The amount of highly deuterated species is extremely small. DISCUSSION The present discussion is based on scheme 1 for partial hydrogenation. Both reactants dissolved in the solution, C,H,(b) and H,(b), migrate to and then are adsorbed on the catalyst surface.Adsorbed buta-l,3-diene reacts with an adsorbed hydrogen atom to form the half-hydrogenated species, C,H,(a) [step (2)]. The 1-2180 8C 60 n 40 %- 2c 0 HYDROGENATION OF BUTADIENE ON Pt, Pd, Rh AND Ru 0 A I! 0 2 4 6 8 0 2 4 6 8 0 2 4 6 6 0 2 4 6 8 number of D atoms Fig. 3. Distributions of deuterium in but-1-ene formed on the hydrogenation of buta-l,3-diene by H, (0) and by D, (A) in 0.5 mol dmP3 D,SO, on Al,O,-supported (a) Pt, (b) Pd, (c) Rh and ( d ) Ru catalysts, and the distributions calculated by using parameters in table 3 [(- -- -) and (---)I. The observed distributions for the reaction with D, are the same as those in fig. 2. Conversions in the reaction by H, are 5.9, 18.1, 5.4 and 0.6';/, for Pt, Pd, Rh and Ru, respectively.reversible occurrence of this step causes multiple H/D exchange under the present conditions. The second addition of a hydrogen atom to the half-hydrogenated species gives butenes [step (3)]. The butenes will be hydrogenated to butane in a similar manner, but this is not dealt with in the following discussion. Surface hydrogen atoms can exchange with hydrogen ions in the solution through step (4), which is always in equilibrium because no current flows under our experimental conditions. The surface reaction of scheme 1 is generally accepted for the hydrogenation of various olefins on metal catalysts. Although there is some discrepancy in deducing the structures of the reaction intermediates, it is not essential in the following discussion because we will discuss the pattern of the distribution of deuterium in terms of the reaction rates of step (4) and the reverse of step (2).DIFFUSION OF HYDROGEN GAS IN THE SOLUTION The rate constants independent of the metal are ca. 2.8 x mol min-l mmHg-l metal-atom-', as observed on Pt/graphite where the rate-determining step is concluded to be the diffusion of hydrogen gas. Thus, the same conclusion will hold for the catalysts used in this work. This is further supported by the potential of the catalysts being more positive than the equilibrium value for the hydrogen electrode reaction. The above rate constant for Pt/A1,0, was calculated using the implicit assumption that all the adsorbed hydrogen is related to the hydrogenation reaction. However,K.SHIMAZU AND H. KITA 181 Scheme 1. Partial hydrogenation of buta- 1,3-diene. one-half of the adsorbed hydrogen on the Pt electrode has been shown to be inactive for the hydrogenation of buta-1 ,3-diene.5 If this situation holds for Pt/A1,0,, then kIl takes the same value as those for the other catalysts. This explains the discrepancy in k,, values (table 2) between Pt and the other catalysts, for which hydrogen diffusion is rate controlling. Further discussion will require more detailed information concerning the adsorption state of hydrogen on these catalysts. Since the rate constant for the present catalysts is one order of magnitude smaller than the value of 2.5 x mol min-l mmHg-l metal-atomp1 for unsupported metal electrode^,^^ lo there certainly exist secondary effects due to the dispersion, support and/or porous structure. These may influence electronic properties and positions of the Pt particles, resulting in a decrease in the number of effective sites for the H, diffusion. However, the reaction mechanisms on supported and unsupported catalysts are taken to be the same.3 The problems of the secondary effect are now under investigation.HYDROGEN-ATOM SCRAMBLING, STEP (2) In order to discuss the rate of step (2) quantitatively, a simulation of the observed distribution of deuterium has been carried out. Only step (2) is responsible and its forward and reverse rates are denoted as uf and u,, respectively: V f 1) C,H,(a) G C,H,(a) --+ C,H,(a) step (2) step (3). rr Under steady-state conditions the difference, uf -cr, is equal to the net reaction rate u.In the calculation, uf/u and u,/u are taken as parameters. Another parameter is the fraction of deuterium in the adsorbed hydrogen, f,,. We assume that all hydrogens in each molecule are kinetically equivalent and neglect any isotopic effect. The distribution of deuterium is calculated for a given set of parameters by using the mass-balance equation, as already r e p ~ r t e d . ~ The set of parameters which gives the best fit was sought using a trial-and-error method. The simulation showed us that no single set of parameters can reproduce the results on Pt and Pd. For the hydrogenation of buta-1,3-diene on Pt/Al,O,, two reaction paths are assumed on the basis of our previous results.6 One path [path (l)] is responsible for the production of highly deuterated species up to [2H,] and includes H/D exchange which is so fast that the distribution of deuterium is approximated by a random distribution among eight hydrogen atoms.The other path [path ( 2 ) ] is for less deuterated species up to [2H6] with a moderate rate of H/D exchange. In a182 HYDROGENATION OF BUTADIENE ON Pt, Pd, Rh AND Ru Table 3. Parameters for the simulation of the distribution of deuterium fraction of maximum no. of catalyst path each path hydrogen atoms z+/u Z I , / L ~ fn(D,) fn(H2) fD(H2)/flj(D,) Pt/AI,O, 1 0.21 2 0.79 Pd/Al,O, 1 0.62 2 0.38 Rh/A1203 - 1 .oo Ru/AI,O, - 1 .oo 6 4.6 random) 3.6 0.89 0.60 0.67 8 random ] 0.32 0.10 0.32 8 6 1.0 0.0 8 3.4 2.4 0.61 0.15 0.24 8 1.6 0.6 0.86 0.13 0.15 strict sense this two-path mechanism is only a first approximation since the reaction mechanism consists of four reaction paths where two, four, six or eight hydrogen atoms are exchangeable.6 However, simplification into two paths reduces the number of parameters.A new parameter needs to be introduced to give the relative fraction of the two paths. For both paths, the samef,, is assumed. The two-path approximation is also adopted for Pd/Al,O, since the reaction mechanism has been concluded to be the same as that for the Pt-catalysed reaction.1° The best sets of the parameters are listed in table 3, wheref,, is expressed byf,(D,) orf,(H,) depending on whether the hydrogenation is by D, or H,, respectively. These parameters reproduce each experimental distribution satisfactorily, as shown in fig.3. Path (1) (random distribution) occupies a major fraction on Pd/A1,0, but a minor fraction on Pt/Al,O,. Thef,(D,) values for Pt/Al,O, and Ru/Al,O, are very high and close to 0.9 while that for Pd/Al,O, is extremely low. The f,(D,) value is concluded to be metal dependent, as expected. The reverse rate of step (2), i.e. u,, varies with the metal. A zero u, value for path (2) with Pd/Al,O, means simple addition of a hydrogen atom. A separate calculation shows that random distribution occurs for u,/u > 200. Taking into account this and the fraction of each path, the sequence for decreasing the reverse rate of step (2) is Pd > Pt 9 Rh > Ru. EQUILIBRIUM STEP FOR THE IONIZATION OF ADSORBED HYDROGEN, STEP (4) The relative rate of step (4) to the surface reaction can be estimated by the ratio fD(H2)/fD(D2), which is expected to be unity if the rate is sufficiently fast compared with that of the surface reaction and zero if it is very slow.f,(H,) is determined so that the best fit is obtained when the other parameters are fixed at the same values as those for hydrogenation by D,.Calculated distributions are shown in fig. 3 and exhibit a good fit. The ratiofD(H2)/fr,(D2) for the respective metals (table 3) decreases in the following order: Pt B Pd > Rh > Ru. The above analysis leads to the following explanation for the observed distributions of deuterium on the catalysts. On Pt/Al,O,, fast hydrogen-atom scrambling occurs via path (2), under the conditions thatf, is kept high through the fast equilibrium step (4).As a result, a large number of deuterium atoms are taken up into the products. In the case of Pd/A1,0,, the hydrogen-atom scrambling is fast, but the rate of step (4) is very slow. These cause a very low f, value and hence small deuterium uptake, as shown by high fraction of [,H0] and [,H,] species. For Ru-catalysed hydrogenation by D,, a highf, value is obtained although the rate of step (4) is the least for the metals used here. This is due to the extremely slow reverse rate of step (2). The most important difference between the electrochemical and gaseous systems is the existence of step (4). Through this step,f, is raised to a higher value, as exemplifiedK. SHIMAZU AND H. KITA 183 by the values off,(H,)/f,(D,) which inevitably become zero in the gaseous system.On Pd, Rh and Ru/Al,O,, however, the increase off,, i.e. fn(H,), is so small that the extent of deuterium uptake is slightly increased. The mechanistic study in the electrochemical system is most effective for Pt-catalysed reaction and it is desirable in the other cases to increasef,, by selecting appropriate reaction conditions, e.g. a large ratio of PD/PR, as realized in the Pd-catalysed deuteration of buta- 1 ,3-diene.11 However, high H/D exchange would not be expected for the Ru-catalysed reaction even if step (4) could raisef, sufficiently during the reaction, because hydrogen-atom scrambling is very slow on Ru/A1,0,. K. Fujikawa, H. Kita and K. Miyahara, J. Chem. Soc., Faraday Trans. 1, 1973,69,481; K. Fujikawa, A. Katayama and H.Kita, J. Chem. Soc., Faraday Trans. 1, 1974, 70, 1 ; K. Fujikawa, H. Kita, K. Miyahara and S. Sato, J. Chem. Soc., Faraday Trans. I , 1975,71. 1573; K. Fujikawa and H. Kita, J. Chem. Soc., Faraday Trans. I , 1979, 75, 2638; K. Fujikawa, H. Kita and S . Sato, J. Chem. Soc., Faraday Trans. 1, 1981, 77, 3055. 2 K. Shimazu and H. Kita, J. Catal.. 1984.86, 129; H. Kita, H. Ito, K. Fujikawa and H. Kano, Denki Kagaku, 1974,42, 408. H. Kita, K. Shimazu, Y. Kakuno and A. Katayama-Aramata, J. Catal., 1982, 74, 323. H. Kita, N. Kubota and K. Shimazu, Electrochim. Acta, 1981, 26, 1185. K. Shimazu and H. Kita, Electrochim. Acta, 1979, 24, 1085. K. Shimazu and H. Kita, Bull. Chem. SOC. Jpn, 1984, 57, 1731. N. Kubota, T. Masui and H. Kita, Electrochim. Acta, 1983, 28, 1663. H. Nakajima and H. Kita, J. Chem. Soc., Faraday Trans. I , 1983, 79, 1027. H. Kita and H. Nakajima, J. Chem. Soc., Faraday Trans. I , 1981, 77, 2105. lo K. Shimazu and H. Kita, J . Catal., 1983, 83, 393. l 1 K. Shimazu and H. Kita, J . Catal., 1983, 83, 407. l2 J. J. Phillipson and R. L. Burwell Jr, J. Am. Chem. Soc., 1970, 92, 6125. I 3 G. C. Bond, G. Webb, P. B. Wells and J. M. Winterbottom, J. Chem. Soc., 1965, 3218. l4 P. B. Wells and A. J. Bates, J. Chem. Soc., 1968, 3064. l5 A. J. Bates, Z. K. Leszczynski, J. J. Phillipson, P. B. Wells and G. R. Wilson, J. Chem. Soc., 1970, l6 P. B. Wells, J. Catal., 1978, 52, 498. I’ E. F. Meyer and R. L. Burwell Jr, J. Am. Chem. SOC., 1963, 85, 2881. l8 J. J. Phillipson, P. B. Wells and G. R. Wilson, J. Chem. Soc., 1969, 1351. I s B. J. Joice, J. J. Rooney, P. B. Wells and G. R. Wilson, Discuss. Faraday SOC., 1966, 41, 223. 2o M. George, R. B. Moyes, D. Ramanarao and P. B. Wells, J. Catal., 1978, 52, 486. 21 Y. Okamoto, K. Fukino, T. Imanaka and S. Teranishi, J. Catal., 1982, 74, 173. 22 W. Vogel, J. Lundquist, P. Ross and P. Stonehart, Electrochim. Acta, 1975, 20, 79; J. Bett, K. Kino- 23 Committee on Reference Catalyst, Catalysis Society of Japan, Reports of 4th Meeting of Reference 2435. shita, K. Routsis and P. Stonehart, J. Catal., 1973, 29, 160. Catalyst, 1982. (PAPER 4/697)

ISSN:0300-9599    

DOI:10.1039/F19858100175

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1985, 81, 185-192 Linear Isotherms in Multicomponent Adsorption onto Silica Gel from Organic Solvents BY ZHENGUO ZHAO AND TIREN Gu* Department of Chemistry, Peking University, Peking, China Received 1st May, 1984 The linearity and independence of adsorption isotherms of multiple trace components in the presence of a dominant adsorbate, previously found for activated carbon in water and in organic solvents, are also found for silica gel in organic solvents. The systems studied are as follows: ( 1 ) n-butyl alcohol, n-hexyl alcohol and n-octyl alcohol in cyclohexane solutions (0.1 and 0.3 mol dmP3) of ethyl alcohol and (2) n-amyl acetate, cyclohexanone and n-heptyl alcohol in carbon tetrachloride solutions (0.1 and 0.3 mol dm-3) of methyl ethyl ketone.Isotherms are reported for one, two and three minor components. The results can be explained in terms of a Langmuir-type mixed-adsorption equation. Linear adsorption isotherms of trace components in the presence of dominant solutes have been reported for activated carbon in water by Greenbank and Manes1 and by Gu and Manes;2 the latter study showed the adsorption isotherms of multiple trace components to be not only linear but also mutually independent. More recently, Gu and mane^^$^ extended these earlier studies to organic solvents. These results suggest the application of adsorptive displacement to the detection and determination of low loadings of strongly held multiple adsorbates on activated carbon as well as the use of activated carbon to remove multiple impurities from a crude product or from waste water where one component predominates.As an example Gu and Manes4 have reported the use of adsorptive-displacement methods in the determination of small amounts of polychlorinated biphenols (the model compound used was 4,4’- dichlorobiphenol) on activated carbon. These studies have suggested that the linearity and independence of trace adsorption isotherms under the condition that the dominant component is not much more weakly displacing than any of the trace components may well be a widespread phenomenon. In this study we extend these earlier studies to another adsorbent, silica gel. From a theoretical viewpoint, the Polanyi-based approach of Greenbank and Manes1 leads to the expectation that multiple trace adsorbates should exhibit linear and independent adsorption isotherms in the presence of a dominant adsorbate, provided only that the dominant adsorbate is not of distinctly lower displacing power than any of the trace adsorbates.Gu and Manes2 have shown that the slopes of these linear isotherms may be estimated from the individual single-component isotherms by use of the multicomponent Polanyi-based model of Greenbank and Manes. More recently, Gu and Manes5 have shown that the simple calculation of multicomponent adsorption from individual isotherms is not limited by the total loading, relative loadings or the number of components. Although the Polanyi-based approach of Crreenbank and Manes is adequate in explaining the linear isotherms for multicom- ponent adsorption from solution, it is not necessarily the only theory to approach this problem.In this paper the results are explained using a Langmuir-type mixed- adsorption equation. 185186 MULTICOMPONENT ADSORPTION ONTO SILICA GEL LANGMUIR-TYPE EQUATION FOR MIXED ADSORPTION The Langmuir equation for a single adsorbate is where the adsorption, ns, and the limiting adsorption, (nS),, are expressed in mmol g-l and c is the equilibrium concentration of adsorbate in solution. The adsorption coefficient, 6 , can be written as6 (2) where b, is a constant and E is the molar adsorpton energy of the adsorbate. At low concentrations, or more correctly at small adsorptions, bc can be neglected in comparison with 1 in the denominator and the equation reduces to Henry’s law: b = b,, exp (EIRT) ns = ( n S ) , hc = Hc.(3) However, note that the Henry coefficient His constant only if the quantity b, which is proportional to exp(E/RT) according to eqn (2), is constant. In general, owing to the heterogeneity of the adsorbent surface, the energy of adsorption E normally falls with increasing adsorption. Therefore eqn (1) will not be obeyed if the adsorbent surface is heterogeneous, particularly in the low-concentration range, where E falls sharply with increasing adsorption. In other words, Henry’s law is usually not obeyed even in the low-concentration range; this is true for the adsorption of many organic solutes from water onto activated carbon7-10 and from organic solvents onto silica gel.ll7 l 2 However, when dealing with multicomponent adsorption, the Langmuir-type mixed adsorption equation may lead to the expectation that multiple trace adsorbates should exhibit linear and independent adsorption i~0therms.l~ The Langmuir-type equation for multicomponent adsorption from solution is where n? is the amount of solute i adsorbed at its equilibrium concentration ci by unit weight of adsorbent (expressed in mmol g-l) and (n:), and bi are the same constants occurring in eqn (1) for a single solute.It follows from eqn (4) that one adsorbate always decreases the adsorption of the other. Note that enhancement of the adsorption by another adsorbate cannot be explained on the basis of the Langmuir model. We now consider the cases of multiple trace adsorbates 2, 3, ... i in the presence of a dominate adsorbate 1.If in the denominator (1 + X i + , bi c i ) is negligible compared with b,c,, and if one keeps c, constant, then eqn (4) reduces to where Hi is the Henry coefficient In this case each trace adsorbate is adsorbed as though the other trace adsorbates were not present; the adsorption isotherms of each trace adsorbate obey Henry’s law,ZHENGUO ZHAO AND TIREN GU 187 provided that bi and b, are independent of the amount of solute adsorbed. From eqn (2) one can write bi and b, as b, = bi, exp (Ei/RT) and b, = bl, exp (EJRT). Substituting eqn (7) into eqn (6) we obtain (7) Although, as mentioned previously, E, and El usually decrease with increasing adsorption, E, - El is likely to remain constant, at least approximately. Moreover, the adsorption of trace adsorbate, nz, is extremely small in the presence of a dominant adsorbate; this makes E,-E, a constant in the concentration range of the trace adsorbate.Thus, for a given c1 one can expect that multiple trace adsorbates should exhibit linear and independent adsorption isotherms in the presence of a dominant adsorbate, provided only that the displacing power (i.e. b,c,)* of the dominant adsorbate is much higher than that of the trace adsorbates, or more correctly b, c, 1 + Ci + , bi ci. Moreover, if eqn (1) holds for single adsorbate and c1 is constant, one can use eqn (6) to calculate Hi using only the values of parameters of the Langmuir equation for single adsorbates. EXPERIMENTAL The silica gel was essentially the same as that used in earlier studies.l1.12, 14. l5 It had a specific surface area of417 m2 g-l (by B.E.T. nitrogen adsorption) and an average pore radius of 4.5 nm. The silica gel was activated in an oven at 200 "C for 4 h before use. Cyclohexane and carbon tetrachloride were of AnalaR quality and were further dried over 5A molecular sieve for over one week, then freshly distilled before use. The organic adsorbates were essentially the same as in an earlier study.12 Solvent purity and organic-component purity were verified by gas chromatography. The adsorption experiments were conventional shaker-bath experiments at 25 "C (& 0.5 "C), using 25 cm3 flasks, with 0.2 g silica gel samples, 10.0 cm3 solution and a 6 h shaking time (previous experiments showed that equilibrium was achieved within 4 h).Following settling, the supernatant liquid was analysed by gas chromatography (Shimadzu GC-1 B instrument) using direct injection of 10 pm3 samples and a succinate polyester column (2.25 m x 6 mm) at uniform temperatures of 50-1 50 "C, depending on the system. The systems were chosen for ease of analysis. The desorption experiments were carried out as follows. After adsorption equilibrium, a measured volume (e.g. 8 cm3) of supernatant solution was removed from the flask by pipette, analysed by gas chromatography and a known volume (e.g. 10 cm3) of pure solvent was added to the flask, which was then shaken for a further 6 h and the supernatant solution was again analysed by gas chromatography. Since the concentrations of adsorbate in the adsorbed phase at such dilute solutions as were used here were very much higher than those in the bulk solution, the loading (nS) was calculated according to VAc.m n; = 2 where Aci is the change in concentration of solute i following adsorption, V is the total volume of solution and m is the mass adsorbing, i.e. no correction for the amount of adsorbate in solution has been applied. For multicomponent adsorption the data were plotted as the loading (in mmol g-l silica gel) of each trace component against its concentration (in mol dmP3), without reference to the concentration of any other minor components. However, runs with single, double and triple minor components are distinguishable in the figures. * Note that in the Polanyi-based treatment the displacing power is measured by the adsorption potential density, which correlates strongly with the refractive index.'188 MULTICOMPONENT ADSORPTION ONTO SILICA GEL RESULTS AND DISCUSSION The single-component adsorption isotherms are plotted in fig.1 and 2 according to the Langmuir equation, eqn (1). Fig. 1 and 2 show that, except for very dilute solutions, eqn (1) holds. The values of the parameters of eqn (1) for various adsorbates and two solvents are shown in table 1. Thus we can use eqn (6) to calculate the Henry coefficient of individual minor components for multicomponent adsorption in the presence of a dominant adsorbate. * 0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 ( 0.010 0.008 0.00 6 z -. 0.004 0.002 I I I I 1 I I I I I I I I I 6 0.0 6 0.05 0.04 0.03 0.02 0.0 1 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 c/mol dm-3 Fig.1. Single-component adsorption isotherms of ethyl alcohol (EA), n-butyl alcohol (BA), n-hexyl alcohol (HA) and n-octyl alcohol (OA) from cyclohexane on silica gel at 25 "C according to the Langmuir equation. The open symbols represent adsorption results and the filled symbols represent desorption results. The multicomponent systems studied comprise n-butyl alcohol (BA), n-hexyl alcohol (HA) and n-octyl alcohol (OA) in cyclohexane solutions (0.1 and 0.3 mol dm-3) of ethyl alcohol (EA), and n-amyl acetate (AA), cyclohexanone (CH) and n-heptyl alcohol (HPA) in carbon tetrachloride solutions (0.1 and 0.3 mol dmP3) of methyl ethyl ketone (MEK). The results are shown in fig. 3 and 4. In these two figures the points are experimental values and the straight lines are least-squares plots fitted to the single-component experimental points and to the origin.The points for the multiple minor components are multiply plotted, so that each individual loading may be compared with the corresponding loading for a single (minor) component at the same concentration; the points are coded so that all individual loadings and concentrations are available for each run. From fig. 3 and 4 one may conclude that when a more concentrated solute of comparable displacing power is present and in equilibrium with silica gel in an organic solvent, one should obtain adsorption isotherms that are both linear and independentZHENGUO ZHAO AND TIREN GU 189 0.06 0.05 0 .a $ 0.03 0.02 0.01 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.0 0 8 1 I I I I I I I I 10 0 .I6 0.12 0.08 0.04 0 0.004 0.008 0.012 0.016 0.0 20 clmol dm-3 Fig.2. Single-component adsorption isotherms of methyl ethyl ketone (MEK), n-heptyl alcohol (HPA), cyclohexanone (CH) and n-amyl alcohol (AA) from carbon tetrachloride on silica gel at 25 "C according to the Langmuir equation. The open circles represent adsorption results and the filled circles represent desorption results. Table 1. Values of the Langmuir parameters for single-component adsorption isotherms (adsorbent, silica gel; temperature, 25 "C) solvent adsorbate ns/mmol g-' b/dm3 mol-l cyclohexane ethyl alcohol n-butyl alcohol n-hexyl alcohol n-octyl alcohol n-amyl acetate cyclohexanone n-heptyl alcohol carbon tetrachloride methyl ethyl ketone 1.54 0.99 0.95 0.97 0.64 0.4 1 0.62 1.04 433 1260 1140 644 172 168 20 3 151 of each other, with slopes that depend on the concentration of the dominant component.These results are similar to those for multicomponent adsorption from water and from organic solvents onto activated From the slopes of the linear isotherms in fig. 3 and 4 experimental Henry coefficients for the individual minor components in the presence of a dominant190 MULTICOMPONENT ADSORPTION ONTO SILICA GEL 0.16 0.14 - - 0.12 - * I On 0.10 - - E ;;;- 0.08 - E: 0 0.002 0.004 0.006 0.008 0.1 c/mol dm-3 10 Fig. 3. Adsorption of n-butyl alcohol (BA), n-hexyl alcohol (HA) and n-octyl alcohol (OA) from solutions containing 0.1 and 0.3 mol dm-" ethyl alcohol (EA). The open circles represent runs with a single minor component.The half-filled circles represent runs with two minor components, each symbol being plotted twice. The filled circles represent runs with three minor components and are each plotted on three curves for 0.3 mol dmP3 EA. 0.14 0.12 - I - On 0.10 0 E 9 0.08 e 0.06 0.04 0.02 , 0 0.002 0.004 0.006 0.008 0.010 clmol d ~ n - ~ Fig. 4. Adsorption of n-heptyl alcohol (HPA), cyclohexanone (CH) and n-amyl acetate (AA) from solutions containing 0.1 and 0.3 mol dm-" methyl ethyl ketone (MEK). The open circles represent runs with a single minor component. The half-filled circles represent runs with two minor components, each symbol being plotted twice.ZHENGUO ZHAO AND TIREN GU 191 Table 2. Experimental and predicted Henry coefficients (adsorbent, silica gel ; temperature, 25 "C) - Ho,,/cm3 g-' Ho.3/cm3 g-' Ho.1 / Ho.3 _ _ _ _ ~ predicted predicted component exptl by eqn (6) exptl by eqn (6) exptl predicted ( I ) from cyclohexane - (a) in 0.1 mol dm-3 ethyl (b) in 0.3 rnol dm-3 ethyl alcohol alcohol n-butyl alcohol 27.8 k 1.3 29 6.27 k 0.33 9.6 4.43 3.0 n-hexyl alcohol 24.4+ 1.5 25 5.09 fO.50 8.3 4.79 3.0 n-octyl alcohol 20.7 -I 1.9 14 4.47 & 0.29 4.8 4.63 3.0 av. 4.62 f- 0.18 (2) from carbon tetrachloride ~~ ~ (a) in 0.1 mol dm-3 methyl ethyl ketone (b) in 0.3 mol dm-3 methyl ethyl ketone n-amyl acetate 2.39 f 0.30 4.0 1.25 & 0.26 1.3 1.91 3.0 cyclohexanone 10.7 f 1 .O 7.3 5.71 f0.29 2.4 1.87 3.0 n-heptyl alcohol 25.9 f 3.9 9.1 12.3 & 1 . 1 3.0 2.11 3.0 av. 1.96k0.13 adsorbate can be evaluated. The results, together with the Henry coefficients predicted from eqn (6), are shown in table 2.In view of the fact that the dominant components increase the equilibrium concentrations by ca. 20-1 00-fold over the single-component concentrations at the same loading, the agreement between the predicted and experimental values of the Henry coefficients is good, except for n-heptyl alcohol (in carbon tetrachloride solutions of methyl ethyl ketone), where eqn (6) underestimates the Henry coefficient by a factor of ca. 3 4 . According to eqn (6) the Henry coefficient should be inversely proportional to the concentration of dominant component (cl). Note that the slope ratio (i.e. the ratio of Henry coefficients) of the linear isotherms of any minor component in 0.1 and 0.3 mol dmP3 solutions of the major component are nearly constant.It follows that the displacing power (ie. b,c,) of the major component is the predominant factor influencing the slope of the minor components. However, the slope ratio (i.e. Ho.l/Ho.3 in table 2) is not equal to the predicted value (3.0), but either higher or lower. Similar results have been obtained by Gu and Manes,4 who studied the adsorption of 4,4'-dichlorobiphenyl from toluene in the presence of naphthalene as dominant component onto activated carbon. As the concentration of naphthalene was raised from 40 to 100 g dm-3, the Henry coefficient (average value) of 4,4-dichlorobiphenyl decreased from 0.0069 to 0.0043, with a ratio of 1.6 rather than the predicted value of 2.5. The most obvious reason for this is that we neglect the non-ideality of the solution and use the concentration c instead of the activity a.This is probably a good approximation for minor components, but may well be far from satisfactory for the major component. Finally, fig. 1 and 2 show that the desorption points lie essentially on the curves of the adsorption isotherms. This means that most of the adsorption is reversible. Note192 MULTICOMPONENT ADSORPTION ONTO SILICA GEL that, since the desorption experiments are difficult to carry out accurately at very low concentrations, there is still a possibility that some irreversible adsorption (e.g. of alcohol) is present. However, whether or not there is such irreversible adsorption, the adsorption isotherms of minor components from multicomponent systems in the presence of a dominant adsorbate are linear. Tiren Gu thanks Prof. Milton Manes (Kent State University, U.S.A.) for encour- agement and editorial assistance. ' M. Greenbank and M. Manes, J . Phys. Chem., 1981, 85, 3050; 1982, 86, 4216. T. Gu and M. Manes, J . Phys. Chem., 1982, 86, 4221. T. Gu and M. Manes, J . Colloid Interface Sci., 1983, 91, 591. T. Gu and M. Manes, Environ. Sci. Technol., 1984, 18, 55. T. Gu and M. Manes, J . Phys. Chem., 1983, 87, 3334. A. W. Adamson, Physical Chemistry of Surfaces (Wiley, New York, 4th edn, 1982), chap. 11 Z. Jean and T. Gu, Acta Chim. Sin., 1966, 32, 140. Z. Cheng and T. Gu, Acta Chim. Sin., 1966, 32, 153. C. Chang and T. Gu, Ke Xue Tong Bao (Science), 1973, 18, 223. lo H. Wang, T. Chow and T. Gu, Ke Xue Tong Bao (Science), 1973, 18, 266. I I C. Chang and T. Gu, J. Colloid Interface Sci., 1981, 82, 254; Acta Chim. Sin., 1981, 39, 287. l3 T. Gu, Environmental Chemistry (in Chinese), 1984, 3(2), 1. l4 P. Li and T. Gu, Sci. Sin., 1979, 22, 1384. Z. Zhao and T. Gu, Acta Chim. Sin., 1981, 39, 503; 1983, 41, 1091. M. Dai, Y. Gao, Z . Zhao and T. Gu, Chem. J. Chinese Unioersities, 1981, 2, 495. (PAPER 4/703)

ISSN:0300-9599    

DOI:10.1039/F19858100185

出版商:RSC    

年代:1985

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1985, 81, 193-205 Nuclear Magnetic Resonance and Molar-volume Studies of the Complex Formed between Aluminium(II1) and the Sulphate Anion BY J. W. AKITT* AND JON A. FARNSWORTH School of Chemistry, The University of Leeds, Leeds LS2 9JT AND PIERRE LETELLIER Physicochimie des Solutions, ENSCP, Universite Pierre et Marie Curie, 11 rue Pierre et Marie Curie, 75321 Paris 05, France Received 2nd May, 1984 Aluminium-27 n.m.r. spectroscopy monitors the concentration of a first-sphere sulphate complex which is formed in aqueous aluminium sulphate solutions. The proportion of complexed aluminium increases markedly with temperature and a heat of reaction for complexation of 33.1 kJ mo1-1 is obtained. Line broadening occurs as the temperature is increased and this gives a value of 49.8 kJ mol-l for the activation energy of exchange of the sulphate complex.The lifetime of the aqua cation is 0.47 s at 298 K, ca. 4 that in AlCl, solution. At constant temperature the proportion of first-sphere complex formed is insensitive to concentration > 0.01 mol dmP3 and its formation constant is much greater than is calculated from the n.m.r. integrals of the two resonances observed. This paradox can be resolved if it is assumed that the first-sphere complex signal simply monitors the amount of aluminium with sulphate in the second sphere, it being known that the second-sphere complex is very strongly formed. Molar-volume measurements confirm that the more concentrated solutions are highly associated. Spectral changes obtained on varying the proportion of sulphate to aluminium indicate that up to two sulphate anions may reside in the second sphere.Line broadening of the resonances which is observed at the highest dilutions is associated with auto-hydrolysis. That a complex forms between [A1(H20)6]3+ and [SO4I2- in aqueous solution has been known for many years,l-lo although it was not clear whether the species detected might be a second-sphereg or first-sphere c ~ m p l e x . ~ - ~ The 27Al n.m.r. spectrum of aluminium sulphate solutions contains two resonances, one due to the hexa-aquo cation and a minor one, 3.3 ppm to higher field, due to a complexll9 l2 which must be a first-sphere complex because of its long lifetime. Its formation constant, calculated from the n.m.r.integrals, is much lower than found in the earlier work,13 a finding which would be consistent with the hexa-aquo aluminium having sulphate complexed also in the second sphere. Only one complex resonance was seen and this was taken to indicate that we were observing the first member of the series [A1(H,0)6-,(S0,),](3-2~)+ with x = 1. In principle it should be possible to determine the number of complexed water molecules by proton n.m.r. spectroscopy but we will show here that this is prevented by the reduction in concentration of complex which occurs as the temperature is reduced.? One surprising feature of the 27Al resonance of the complex is that its line is only marginally broader than that of the regularly t There is evidence for monodentate [SO,]'- in CdSO, solutions in R.Caminti and G . Johansson, Acta Chem. Scand., Ser. A , 1981, 35, 373. 193194 AL'JMINIUM-SULPHATE COMPLEXES octahedral [A1(H,0)J3+, whereas the loss in symmetry would be expected to lead to significant broadening. One of us has shown recently that this need not be the case,14 and that a narrow line is compatible with a monodentate, monosulphate complex. The earlier work,13 which was carried out in the continuous-wave mode (c.w.), indicated that the proportion of first-sphere complex was not very sensitive either to salt concentration or to added sodium sulphate, although the range of concentration over which this could be meaningfully tested is limited with C.W. spectroscopy by the low signal-to-noise ratio of the complex resonance. The advent of Fourier-transform techniques, however, gives a marked improvement and has allowed us to investigate a much wider concentration range.The n.m.r. technique is even then limited to relatively high concentrations for accurate data, and we have chosen to investigate the lower concentrations using molar-volume determinations also. The molar volumes of several aluminium salts have been measured previously,15+ l6 although the sulphate has been avoided because of its known tendency to form ion pairs. A 4; value for A1,(S04), can be calculated from the data and this, when compared with the value actually found and interpreted in the light of the n.m.r. results, may allow some conclusions to be made as to the influence of ion pairing on molar-volume measurements. EXPERIMENTAL ALUMINIUM-27 SPECTRA These were obtained principally at 23.45 MHz on a Bruker HFX 90 instrument in the Fourier-transform mode.Samples were held in 8 mm tubes placed in 10 mm tubes with D,O in the annulus to act as field-frequency lock. A 90" pulse (20 ps) and 500 Hz sweep-width were used, and data were collected in 1 k of memory. This gave an acquisition time of 1.024 s, which is long compared with the longest relaxation time observed of ca. 0.09 s (narrowest linewidth 3.5 Hz). These conditions provide well defined spectra with no differential saturation of the two lines and are ideal for quantitive measurements. In the case of the most dilute solutions (< 0.01 mol dm+) it was found that self-hydrolysis led to a broadening of the resonances to the extent that the two lines could no longer be satisfactorily resolved to allow quantitative measurements.These solutions were therefore examined on a Bruker HX400 at 104.2 MHz. Variable-temperature measurements were made using the Bruker accessory (HFX 90) ; because the proportion of complex was very sensitive to the temperature, the temperature of the sample was checked during all room-temperature measurements and then corrected to 23 "C. Temperature variations were the major source of experimental errors. The quantitative estimation, however, also had its problems. The signal-to-noise ratio of the smaller complex resonance was always noticeably poor, although the number of scans taken was determined principally by the need to measure this line accurately. A small exponential multiplication of the free-induction decay was given to reduce the noise although the amount of reduction was limited by the need to maintain resolution of the two lines.Phase correction also caused problems since the smaller resonance (,l,th of the larger) appears in the skirts of the larger at 23.45 MHz, and estimates of area based on the formula area = width at half height x height contain subjective errors in positioning the sloping baseline of the smaller peak. Curve analysis was also used,17, IN but small departures from a Lorentzian shape of the strong resonance mean that this line cannot be fitted correctly in the skirt, where it influences the fit of the smaller one. The two peaks had to be fitted in separate operations, and again subjective error is inevitable.MOLAR-VOLUME MEASUREMENTS These were made using a Picker densimeter vibratometer type 03D, supplied by Sodev,lg and 0.002 K using a thermostat manufactured by Setaram. the temperature was maintained at 298 The density of water was taken as equal to 0.997047 g cm-3.J. W. AKITT, J. A. FARNSWORTH AND P. LETELLIER Table 1. Effect of varying concentration of Al,(SO,), A1 as linewidth/Hz [A~z(SO,),I complex if high /moldrn-, pH (% ) A13+ AlSOl field 1 .oo 1.98 7.5 0.80 2.40 8.2 0.60 2.46 7.8 7.6 0.46 0.40 8.4 8.3 0.36 0.30 2.58 7.4 7.8 0.27 0.15 2.72 7.8 0.075 2.88 7.6 0.038 3.02 7.9 0.019 3.16 7.6 0.010 7.9 0.009 4 3.30 8.0 0.005 0 6.5 5.5 0.002 5 - 0.001 25 - 5.8 0.000 94 3.74 CQ. 2.2 - - - - - - 22.5 24.0 12.5 17.0 9.0 14.5 10.4 13.0 10.4 16.0 8.8 13.0 10.0 14.0 6.8 10.4 8.5 14.5 8.0 14.5 9.0 17.0 8.0 17.0 10.4 20.0 18.0 36.0 20.8 30.0 28.5 37.5 56.2 104 not resolved 195 RESULTS EFFECT OF ALUMINIUM SULPHATE CONCENTRATION The proportion of complex present varies very little with concentration from 1 .O down to 0.01 mol dm-3 in A1,(S04),, where the use of the high-field spectrometer becomes necessary owing to the onset of auto-hydrolysis and line broadening by exchange with the increasing concentration of hydrolysis products.A reduction in proportion of complex is just detectable at the lowest concentrations. The linewidths observed and chemical shifts are in table 1. Temperature effects (0.3 percentage unit K-l) limit the accuracy of the determinations. EFFECT OF ADDING HCl TO AN ALUMINIUM SULPHATE SOLUTION It has already been shown that HC1 suppresses the formation of the sulphate c0mp1ex.l~ Some further data are presented in table 2 together with the calculated concentration of free sulphate anion.The proportion of complex formed is roughly proportional to [SO:-]. Note also the marked reduction in linewidths which occurs with the initial acid addition owing to the parallel suppression of hydrolysis and/or proton exchange. EFFECT OF TEMPERATURE The spectra illustrated in fig. 1 demonstrate the changes which occur with increasing temperature. At 278 K the resonances are relatively broad and the proportion of complex is small, being difficult to measure accurately. Increasing the temperature causes the resonance first to narrow, as the increasing rate of motion in solution decouples the aluminium quadrupole from the surrounding lattice, but then to broaden as the rate of the exchange process [Al(H20)J3+ * [Al(H2O),(SO4)]+ becomes comparable with the inverse frequency separation of the two lines.196 ALUMINIUM-SULPHATE COMPLEXES Table 2.Effect of adding HCl to a 0.15 mol dmP3 Al,(SO,), solution ~~ A1 as linewid th/ Hz [HClI complex [SO:-],,,, /moldmP3 pH (% ) /mol dmP3 ~ 1 3 + Also,+ 0.0 2.72 8.5 0.44 8.5 14.5 0.1 1.60 8.3 0.32 4.0 8.5 0.2 1.28 7.3 0.25 3.5 7.5 0.4 1.02 4.9 0.18 5.0 7.5 0.8 0.88 4.1 0.15 4.0 8.0 1.6 0.40 2.0 0.06 5.0 7.5 3.2 0.16 1.7 0.04 7.0 12.0 Fig. 1. 27Al spectra of 0.25 mol dm-3 Al,(SO,), solution at temperatures of (a) 277, (b) 297, (c) 317 and ( d ) 337 K. The proportion of complexed A1 also increases linearly from near 0% at 273 K as the temperature is increased. Similar changes occur with a solution of reduced [SOq-]/[Al] ratio, although the proportion of complexed A1 is smaller.Spectral coalescence occurs at ca. 350 K for this solution and ca. 340 K for the Al,(SO,), solution. The chemical shift of the single resonance observed above coalescence is temperature dependent because the proportion of complexed A1 continues to increase. The data are given in table 3. EFFECT OF CHANGES OF SULPHATE CONCENTRATION AT CONSTANT [All The ratio [SO;-]/[Al3+] was varied by preparing a series of solutions of constant aluminium concentration, these being either AlC1, + Al,(SO,), mixtures or Al,(SO,), with added sodium suiphate. The proportion of complex varies strongly with sulphate content where this is small, but reaches an ill defined limit when the ionic ratio is ca.2 and where further addition of sulphate produces no change (fig. 2 and table 4).J . W. AKITT, J . A. FARNSWORTH AND P. LETELLIER 197 Table 3. Effect of temperature on % complex in solutions 0.5 mol dmP3 in A1 Al,(SO,), solution Al,(SO,), . ,C13., solution A1 as linewidth/Hz A1 as linewid th/ Hz complex complex T / K (% 1 ~ 1 3 + AlSOZ (% 1 ~ 1 3 + AlSO: 3 26 317 316 307 306 297 296 287 286 277 276 - - 15.4 14.7 11.3 11.0 9. I 8.1 5.1 5.0 3.5 2.8 - -~ 12.8 13.2 9.7 10.6 9.3 8.8 9.7 8.8 11.0 11.0 - __ 28.6 28.0 20.3 19.8 16.7 14.1 12.8 12.3 12.8 12.3 - 9.0 7.4 5.8 - - __ - 4.1 2.2 1.3 - - 10.7 9.4 8.5 - - - - 9.4 10.4 11.3 - - 12.3 16.5 19.8 - - - - 16.5 13.2 12.3 - - 6 - 4 - 2 - Fig.2. :{ A1 complexed by sulphate in the first sphere as a function of the ratio of sulphate to aluminium in solutions containing mixtures of Al,(SO,), with either AIC1, or Na,SO,. The results encompass total A1 concentrations of between 0.5 and 0.075 mol dmP3.198 ALUMINIUM-SULPHATE COMPLEXES Table 4. % Complex as a function of [SO:-]/[Al3+] in Al,(SO,), +Na,SO, mixtures [A13+] [so:-] [so:-] /mol dm-3 /mol dmP3 /[A13+] A1 as linewidth/Hz complex PH (% 1 A13+ AlSO; 0.5 0.187 0.5 0.245 0.3 0.15 0.15 0.075 0.075 0.038 0.5 0.3 0.5 0.375 0.5 0.413 0.5 0.45 0.5 0.488 0.3 0.3 0.15 0.15 0.075 0.075 0.5 0.525 0.5 0.563 0.5 0.60 0.5 0.675 variousb 0.5 0.80 0.5 0.90 0.5 1 .oo 0.5 1.10 0.5 1.25 0.3 1 .oo 0.15 1-00 0.375 0.49 0.5 0.5 0.5 0.6 0.75 0.825 0.90 0.975 1 .00 1 .oo 1 .00 1.05 1.125 I .20 1.35 1 S O 1.60 1.81 2.00 2.20 2.50 3.33 6.67 2.0 3.1 2.7 2.8 3.1 4.1 4.3 5.4 5.2 5.5 5.4 6.4 6.0 5.3 6.1 7.0 7.I 7.P 8.3 8.5 9.6' 9.0' 9.Ic 9.6 9.3 8.9 6.7 - - - - 9.1 9.5 9.8 9.3 - - - 10.2 10.7 10.7 10.7 8.4 10.2 8.9 8.9 8.9 - - - 10.7 14.2 - - - - 13.3 16.0 15.1 14.7 - - - 14.2 16.0 16.9 16.0 13.3 16.0 14.2 13.3 13.3 - - - a Nominally at 296 K. Average of all high-concentration values in table 1 . At 293 K, corrected. LINEWIDTH AND CHEMICAL-SHIFT CHANGES Linewidth measurements of the two resonances for a variety of solution conditions are given in tables 1-4. In almost all samples the complex linewidth is greater than that of A13+(aq), although the ratio varies from ca. 2 to only just greater than unity in the most concentrated Al,(SO,), solution.The linewidths then are not intrinsic properties of the species but reflect chemical interactions in the solutions. The fact that a marked decrease in linewidth occurs on addition of acid indicates that hydrolysed forms of the ions probably pay an important part in this via exchange of aluminium environments, and this conclusion is reinforced by the changes that occur with concentration (table 1). Both resonances are broad at high concentration (where the pH is lowest), so that viscosity is likely to be the determining factor and, surprisingly, gives nearly equal linewidths. The lines narrow as the solution is diluted, although A13+ narrows faster than AlSO: and their widths become different. At a concentration of 0.01 mol dm-, the lines again start to become broader and indeed broaden to the extent that they cannot be resolved.The pH range over which this happens (3.3-3.74) is consistent with the line broadening being caused by the presence of hydrolysed species. These trends are evident despite the low accuracy of theJ . W. AKITT, J. A. FARNSWORTH AND P. LETELLIER 199 Table 5. Molar-volume measurements [AI,(SO,), * 18H,O] (concentration): 4v $v- 108.52mi /mol kg-' /(mol kg-l); /cm3 mo1-' 0.332 2 0.186 5 0.152 6 0.109 6 0.085 9 0.074 0 0.046 1 0.035 7 0.028 2 0.022 0 0.017 1 0.013 7 0.010 6 0.009 66 0.007 5 0.007 36 0.005 5' 0.002 55' 0.001 16' 0.576 0.432 0.390 0.33 1 0.293 0.272 0.2 15 0.189 0.168 0.148 0.131 0.1 17 0.103 0.0983 0.0866 0.0858 0.074b 0.050b 0.034h - 3.22 - 9.85 - 11.77 - 16.00" - 16.41 - 17.32 - 20.96 -21.72 - 22.85 - 24.45 - 26.27 - 27.37 - 28.56 - 29.18 - 30.54 - 30.30 - 32.25' - 37.43b - 40.96' - 64.8 - 56.7 - 54.1 -51.9" - 48.2 -46.8 - 44.3 -42.2 -41.1 -40.5 -40.5 -40.1 - 39.7 - 39.8 - 39.9 - 39.6 - a This result appears to carry an unusually high error and has been ignored in calculating The three results obtained at the lowest concentrations have intrinsically high error the slopes.(0.2, 0.4 and 2.0 cm3 mol-l, respectively) and have not been used in assessing the data. linewidth determinations, which are correct only to ca. 1 Hz and seem to contain other systematic errors. Clearly, full and T2 determinations for this system would be even more informative. The chemical shift between the two resonances falls generally in the range 3.3 f 0.1 ppm.There are, however, a few results outside these limits which correlate roughly with the SO:- concentration. Thus the most concentrated A12(S0,), solutions (> 0.6 mol dm-,) have shifts of 3.4-3.5 ppm and solutions 0.5 mol dm-, in A1 with [S0,2-]/[A13+] c 0.9 have shifts of 3.14-3.26 ppm. The inference is that the sulphate influences the relative shift in some way, although the effect is weak and may not be real. STOPPED-FLOW 2 7 ~ 1 N.M.R. SPECTROSCOPY Because it seemed from early 170 data that the rate of water exchange on AP+ was rather slow (see below) it seemed that it would be possible to measure the rate of formation of the complex in a stopped-flow apparatus in which AlCl, and MgSO, solutions were mixed and successive 27Al spectra taken to monitor the appearance of the complex resonance.The apparatus used has been described previously.20 The complex was fully formed within the first second of the experiment and seemed to be formed much faster than the rate of water exchange would suggest. MOLAR-VOLUME MEASUREMENTS The raw data are given in table 5 and is plotted as a function of mi in fig. 3. Above 0.04 mol kg-l the slope of this plot is ca. 47 but increases as the concentration200 ALUMINIUM-SULPHATE COMPLEXES -1 0 - $ -20 E E m --. s' -30 -40 0.1 0.2 0.3 0.4 0.5 m+/ rn 01: kg -b Fig. 3. Apparent molar volumes, q5,, of Al,(SO,), as a function of m:. The dashed line is drawn at the Debye-Huckel limiting slope for a 3 : 2 electrolyte. -4 0 -. - I 2 5 m -72 E '? 0 -50 s' m - I - 10 I 1 I I 0 0.1 0.2 rn/mol kg-' Fig. 4.4"- 108.5rna as a function of Al,(SO,), molality, rn.The intercept at rn = 0 corresponds to a 4: value of ca. 39 cm3 mol-l. The slope b, at low concentrations is -70. The point at m = 0.1 1 is of doubtful validity. is decreased below 0.04 mol kg-l, which is the concentration range where the n.m.r. results indicate that the complex becomes progressively dissociated. The limiting slope of this plot which may be attained for the most dilute solutions is given by the Debye-Huckel limiting law, using the relation16+ 21 plim = 0.66(v+ + v-) Z+Z-[V+(Z+)~ + V - ( Z - ) ~ ] ;J. W . AKITT, J. A. FARNSWORTH AND P. LETELLIER 20 1 for an electrolyte [A;+,] [Bt:]. The limiting slope predicted for Al,(SO,), is then 108.45.It is evident from fig. 3 that the experimental points are hardly sufficient to define this value; indeed, strictly they never reach it, so that we have to seek other means to extrapolate to a value of 4;. We have adopted the method proposed by Redlich and Rosenfeld2,7 23 and plot the quantity - 108.45rn4 as a function of solution molality. This is shown in fig. 4 and for concentrations < 0.04 mol kg-l gives a straight line of slope b, = - 70 cm3 kg molP2 and an intercept at infinite dilution of 4; = - 39 cm3 mol-1 which should be compared with the value of -42.5 cm3 m o t 1 derived from the determinations made on other ~a1ts.l~. l6 DISCUSSION Aluminium sulphate solutions have been investigated recently using relaxation techniques which have identified three rate processes, two fast ones with zI = 3 x 10-lo s and zII = 1.6 x s2, and a slower one with rates of the order of 1 s ., ~ These results are applied to the following reaction scheme: I. HOAlWL ,+ HOAlL + H+ 4 3 k43 where W represents a water molecule separating the ions and L2- is the sulphate anion. The results obtained were k12/k,, = 0.2, k,, = 2.5 s-l, El, = 89.9 kJ mol-1 and E,, = 75.2 kJ mol-l, at 298 K. Some values were also given for the hydrolysis reactions, k,, K24 = 0.025 mol dm-3 s-l. It is clear that steps I and I1 are not accessible to n.m.r. study but that the line broadening and change in proportion of complex which occur with temperature give information about equilibrium 1 t-) 2 and line- width changes at high dilution or upon addition of acid are influenced by the equilibria 1-3 and 2-4.The exchange of water molecules on A1(H20)i+ has also been investigated using 1 7 0 or lH n.m.r. spectroscopy. The former gives a lifetime of 4.5 ms at ca. 355 K which extrapolates to 7.5 s at 298 K, an extrapolation over three orders of magnitude of 7.26 The proton work gives z = 5555 s at 243.5 K.27 Both measurements were made in the absence of sulphate. Extrapolation between these two sets of results gives a lifetime of 1.6 s at 298 K, which is probably better than the value obtained from 1 7 0 alone. It compares reasonably well with the value of 2.0 s calculated from the relaxation work ( k I 2 = 0.5 s-l) for loss of H 2 0 to form the complex cation. It should also be recalled that proton exchange on the aquo cation is very f a ~ t ~ ~ 9 28 and is catalysed by hydrolysis.Two separate exchange processes exist with rates of ca. lo5 and lo9 s-l. Thus reactions 1 - 3 and 4- 2 should be very fast processes. The aluminium-27 data give a small formation constant for the complex of 0.06 dm3 mol-l, independent of concentration > 0.01 mol dm-3 and over four orders of magnitude lower than the accepted value of 2000 dm m01-i.25729 Such a weakly formed complex should dissociate easily upon dilution, but table 1 indicates that this does not occur and that we are observing a weak complex behaving as a strong one. We can resolve this paradox by assuming that the n.m.r.-observable complex is formed weakly from a non-observable one which is formed strongly (in the reaction202 ALUMINIUM-SULPHATE COMPLEXES scheme above, the second-sphere complex).Such a model, with formation constants of 1000 and 2000dm3mol-l for the second-sphere complex and 0.085 for the first-sphere complex, can be shown satisfactorily to fit the data. The reduction in complex formation which occurs upon HCl addition is proof that the complexing anion is SO:- rather than HSO;. Some information on the way the second-sphere complex is formed can be obtained from the way the first-sphere complex concentration varies as the ratio [SO:-]/[A13+] is varied, since the formation constants are such that the signal of the complex in effect monitors the concentration of second-sphere complex too. It is clear that there is no change in the slope of the percentage complex/ion ratio plot of fig.2 at an ion ratio of unity, and that second-sphere complex formation continues up to the point where the ratio is two. The inference is that the solutions contain both LWAlWL- and LWAl+. The 2: 1 complex will be quite stable because of the large size of the ions involved, which means that the electric field due to one sulphate is almost negligibly small at another sulphate anion placed on the other side of the aquo cation. The overall picture is then of solutions in which all the cation is complexed by anion, and all the anions appear to reside near the cations. Linewidth changes occur on dilution at constant temperature or on warming a given solution, and these two sets of data give access to different processes. The variable-temperature results are particularly informative.The concentration of first-sphere complex is near zero at 273 K and rises more or less linearly with temperature. The equilibrium constant for the reaction LWAl+ $- LAIS K is easily calculated from the results and a plot of In K against 1/T gives the heat of reaction as 33.1 1 .O kJ mol-l, both solutions that were measured giving closely similar results. A number of interesting conclusions are possible. First, the first- sphere complex is not present at low temperatures, so its solvation number cannot be measured by low-temperature proton n.m.r. spectroscopy. Secondly, at sufficiently high temperatures, > 470 K and therefore at high pressure, all the cations are likely to have sulphate coordinated in the first sphere.This suggests very clearly a mechanism whereby alunite, (H,O)Al,(SO,),(OH),, in which aluminium octahedra are bridged by sulphate anions, is formed by heating aluminium sulphate solution at 520 K under p r e ~ s u r e . ~ ~ ~ 31 Thirdly, we can conclude that the lifetime of A1 in the aquo cation falls more rapidly with an increase in temperature than does that of the sulphate complex. The coalescence temperature is higher in the solution weaker in sulphate because the complex concentration is less and so the average lifetime of the species is higher. In addition to the proportion of first-sphere complex which is formed, we can also measure the changes in linewidth which occur with temperature below coalescence and so in principle measure the exchange rates of the- individual species using the r e l a t i ~ n s h i p ~ ~ k = n(excess linewidth/Hz).There are, however, problems in applying this relationship to quadrupolar nuclei. The linewidths of such nuclei depend upon temperature even in the absence of exchange, owing to changes in viscosity of the solutions and to changes in the correlation times of motion of the species carrying the nucleus being observed. Some way of allowing for these changes has thus to be found before meaningful exchange parameters can be obtained. The linewidths are plotted in fig. 5 as a function of temperature. That of the aquo cation falls initially with temperature in the way normal for a quadrupolar nucleus, and we have therefore extrapolated this part of the data linearly to act asJ.W. AKITT, J. A. FARNSWORTH AND P. LETELLIER 203 1 I I I I I 70 290 310 330 TI K Fig. 5. Linewidths of the *'A1 resonances of both the sulphate complex, (a), and the aqua ion, (b) and (c), in 0.25 mol dmP3 solutions of Al,(SO,),, (a) and (b), and A12(S04)l.,C13.6, ( a ) and (c). The lowest, straight, line was used as a datum from which to calculate the excess line broadening due to slow exchange between the two species of cation. the baseline from which the excess linewidths may be calculated. A plot of Ink against 1 / T then gives, in principle, the activation energy of the exchange on the two species. It is clear from the data that the broadening of the sulphate complex resonance is independent of the proportion of complex present since the two mixtures give essentially the same results.We can therefore assume that, even though the proportion of complex changes with temperature, this change does not contribute to its lifetime. We can therefore accept that the activation energy E,, = 49.8 2 kJ mol-1 obtained for the sulphate complex is valid. On the other hand, the lifetime of the aqua complex is strongly dependent on the sulphate content of the solution, and it would be unwise to derive any activation energy. Combination of E,, with the heat of reaction gives E,, = 82.9 kJ mol-l, which compares reasonably well with the relaxation value. The value of E,, derived here is, however, much lower, in agreement with the much lower stability of the complex revealed by the n.m.r. data. At 298 K we find K,, = 0.0989, or less than half the relaxation value.25 The linewidths at 298 K give the rate constant for decomposition of the sulphate complex as k,, = 21.5 5.0 s-l, an order of magnitude faster than the relaxation data.The n.m.r. lifetime of the aqua-ion to sulphate exchange can be calculated from this: k,, = K,, k,, = 2.13 f 0.4 s-l, so giving a lifetime of 0.47 s. This is about one-third the lifetime for water-molecule exchange obtained from the lH and li0 n.m.r. data by interpolation. This explains why the stopped-flow experiment was unable to follow the formation of the complex. The entry of sulphate into the first sphere therefore does not depend on water dissociation but is presumably assisted by the association in the second sphere. It is surprising that it is then lost so rapidly, but this may be accounted for by the steric repulsion between the non-coordinated sulphate oxygens and the remaining first-sphere water molecules.204 ALUMINIUM-SULPHATE COMPLEXES The changes in linewidth which occur with increasing dilution are related to auto-hydrolysis due to the increase in pH.Unfortunately, these solutions are probably not the best in which to attempt to study this process because of the complications introduced by the exchange between the two complex species present, and a full study is in progress with dilute AlCl, The molar-volume measurements provide clear evidence supporting the conclusions based on spectroscopy. The plot in fig. 3 is curved but may be considered to be made up of two straight-line portions with the slope changing in the concentration range 0.04-0.01 mol kg-l, the region where the results indicate that dissociation on dilution commences.The value of 4; obtained from the Redlich and Rosenfeld plot of fig. 4 agrees adequately with the values calculated from the earlier data.l5* l6 A line is plotted through this point on fig. 3 at the Debye-Huckel limiting slope for a 3: 2 electrolyte and passes through the data for the lowest concentrations. The deviation parameter b, obtained from fig. 4 is - 70 cm3 mo1P kg and is unusually large.l69 *l This seems to be associated with the predominance of ion pairing, and the large value presumably reflects a progressive change in the amount of such association which occurs throughout this region of concentration. We also of course have hydrolysis occurring in this region and this should result in a reduction in the total electrostriction as the average charge on A1 is reduced and a corresponding movement of the data to more positive volumes.This does not happen, even when the high-error data obtained at the lowest concentrations is taken into account, and we must conclude that the proportion of hydrolysed cations remains low. The value of 4; for A1 is more negative in fact than some recent values for similar cations where hydrolysis was specifically avoided by acid addition.34 The onset of total ion pairing which occurs as the concentration is raised from 0.01 through 0.04 mol kg-' causes a marked curvature in fig. 3 and a distinct discontinuity in fig. 4. This is consistent with a change in the make-up of the electrolyte, which is now in effect [AlWSO,],f [SO,]", ignoring the small proportion of first-sphere complex.The slope of the plot in the higher concentration region of fig. 3 should be 9.706 but is in fact 46.8. This high value is probably an indication that the second-sphere complex is not quite so definite a species as is implicit in the above formulation but retains strong electrostriction effects due to its dipolar field, which varies significantly as the Al3+-SO;- distance changes with concentration. Note that very similar volume results have been obtained for BeSO, and for MgS0,,35* 36 since both are believed to be salts which form ion pairs, and their volumetric data exhibit high curvature at low concentrations. We thank the S.E.R.C.for time on the very-high-field n.m.r. spectrometer situated at Sheffield University and 0. W. Howarth for obtaining the stopped-flow spectra at Warwick University. J. W. A. thanks Prof. R. Schaal for hospitality at E.N.S.C.P. A. W. Thomas and A. T. Whitehead, J . Phys. Chem., 193 1, 35. 27. M. Eigen and K. Tamm, Z . Elektrochem., 1962, 66, 93; 107. E. Matijevic, J. Leja and R. Nemethy, J . Colloid Interfuce Sci., 1966, 22, 68. L. J. Stryker and E. Matijevic, J . Phys. Chem., 1969, 73. 1484. J. Miceli and J. Stuehr, J. Am. Chem. SOC., 1968, 90, 6967. B. Behr and H. Wendt, 2. Elektrochem., 1962, 66, 223. A. K. Nanda and S. Aditya. Z . Phys. Chem., 1962, 35, 139. T. Nishide and R. Tsuchaya. Bull. Chem. SOC. Jpn, 1965, 38, 1398. a M. le Peintre, C.R. Acud. Sci., 1964, 223, 1004.lo J. E. Salmon and J. G. L. Wail, J. Chem. Soc.. 1958, 1128. l 1 J. W. Akitt, N. N. Greenwood and G. D. Lester, J . Chem. Soc. A , 1969, 803. l 2 B. F. Epperlein and 0. Lutz, 2. Nuturforsch.,Tril A , 1968, 23, 141 3. l 3 J. W. Akitt. N. N. Greenwood and B. L. Khandelwal, J . Chem. Soc., Dalton Truns., 1972, 1413.J . W. AKITT, J. A. FARNSWORTH AND P. LETELLIER 205 l 4 J. W. Akitt and W. S. McDonald, J. M a p . Reson., 1984, 58, 401. l5 R. Zana and E. Yeager, J. Phys. Chem., 1967, 71, 521. l6 F. J. Millero, Water and Aqueous Solutions, ed. R. A. Horne (Wiley, Chichester, 1972), chap. 13. l 7 J. W. Akitt, Appl. Spectrosc., 1975, 29, 493. l R J. W. Akitt, Comput. J., 1977, 20, 84. l9 P. Picker, E. Tremblay and C . Jolicoeur, 1. Solution Chem., 1974, 3, 377. 2o J. W. Akitt, A. Farthing and 0. W. Howarth, J. Chem. Soc., Dalton Trans., 1981, 1609. 21 F. J. Millero, Chem. Rev., 1971, 71, 147. 22 0. Redlich and P. Rosenfeld, Z. Elektrochem., 1931, 37, 705. 23 0. Redlich and P. Rosenfeld, Z. Phys. Chem., Abt. A , 1931, 155, 65. 24 A. Bonsen, W. Knoche, W. Berger, K. Giesel and S. Petrucci, Ber. Bunsenges. Phys. Chem., 1978,82, 678. 25 C. Kalidas, W. Knoche and D. Papadopoulos, Ber. Bunsenges. Phys. Chem., 1971, 75, 106. 26 D. Fiat and R. E. Connick, J. Am. Chem. Soc., 1968, 90, 608. 27 A. Takahashi, J. Phys. Soc. Jpn, 1970, 28, 207. 28 Dodd-Wing Fong and E. Grunwald, J. Am. Chem. SOC., 1969,91, 2413. 29 L. G. Sillen and A. E. Martell, Stability Constants (The Chemical Society, London, 1964). 30 Athanesco, C.R. Acad. Sci., 1886, 103, 271. 31 Y . Wing, M. La1 and A. T. Howe, Muter. Res. Bull., 1980, 15, 1649. 32 M. L. Martin, J-J. Delpuech and G. J. Martin, Practical NMR Spectroscopy (Heyden, London, 1980). 33 J. W. Akitt and J. M. Elders, unpublished work. 34 T. W. Swaddle and M. K. S. Mak, Can. J. Chem., 1983, 61, 473. 35 F. Kuschel and J. Seidel, Z . Phys. Chem. (Leipzig), 1981, 262, 967. 36 K-H. Ducker, Ber. Bunsenges. Phys. Chem., 1970, 74, 416. (PAPER 4/709)

ISSN:0300-9599    

DOI:10.1039/F19858100193

出版商:RSC    

年代:1985

数据来源: RSC