摘要:

J . Chern. Soc., Furuduy Trans. I, 1989, 85(5), 1083-1089 Dielectric Relaxation in Concentrated Solutions of cis-Pol yisoprene Part 3.-Relationship between Friction Coefficient for Dielectric Normal-mode Process and Local Segmental Motions Keiichiro Adachi,* Yasuo Imanishi and Tadao Kotaka Department of Macromolecular Science, Faculty of Science, Osaka University, Toyonaka, Osaka 560, Japan The data of dielectric normal mode process (01,) and the segmental mode process ( a I I ) reported in parts 1 and 2 of this series were analysed to find a relationship between the friction coefficient [ for the normal mode process and the relaxation time z, for the segmental mode process. Assuming t, proportional to [, we attempted to superpose the t, us. molecular weight M , plots for solutions with different concentrations C.From the vertical and horizontal shift factors, we found that the monomeric friction coefficient [, is approximately proportional to P z , . To explain this result, we assumed that the effective size for the segmental motions expands with decreasing C. The molecular weight M,, for the unit of the segmental motion was estimated based on the computer simulation reported by Verdier and Stockmayer. The M , was found to be 77 for bulk &-PI and increased with decreasing C approximately in proportion to C-l. The friction coefficient for an arbitrarily defined segment is one of the key parameters for describing the large-scale molecular motions involved in polymer dynamics and rheological properties.' Mechanical and dielectric relaxation times for flexible polymers depend strongly on the concentration C and temperature 7".The change is generally ascribed to the change in <. As is well known, the relaxation time z for the large-scale motions is written as2 where f is the structural factor depending on the architecture of the chain molecule and is a function of the molecular weight M divided by the molecular weight between entanglement Me. The factor f depends also on C through the change in Me with C. The large-scale molecular motions may be a result of successive local segmental motions at various portions of the chain. Such local motions are termed the elementary process which governs the rate of the large-scale r n ~ t i o n s . ~ , ~ The relaxation time zo for such an elementary process may be proportional to 5.In part 1 of this ~ e r i e s , ~ we reported the dielectric normal mode process (aI) due to the fluctuation of the end-to-end distance in concentrated solutions of cis-polyisoprene (cis-PI). The C and M dependence of the relaxation time z, for the normal mode process have been determined. In part 2, we also studied the C dependence of the relaxation time z, for the dielectric segmental mode process (aII).6 In this paper, we examine the relationship between the < and z, in cis-PI solutions by analysing the data given in parts 1 and 2. To estimate the C dependence of [ for the aI process, we used the time-concentration shift factor, loga,.' It is also expected that the average size of the units for the elementary process can be estimated by using the relationship between the t, and z, obtained by computer simulation by Verdier and Stockmayer.* t = C(C7 W W I M J (1) 1083 37-21084 Friction Coeficien t of cis- Poly isoprene Solutions Results and Discussion Reduction to iso-M/M, State In part 1 of this series, we found that the dielectric normal-mode relaxation time z, for concentrated solutions of cis-polyisoprene (cis-PI) is proportional to M 2 in the range of molecular weight M lower than the characteristic molecular weight M,, while z, is proportional to M4.3 in the range of M > It was also found that the M , in solutions is inversely proportional to the concentration C and is equal to C-lK where the superscript 0 denotes the bulk state. Strictly speaking, C denotes the volume fraction of the polymer. For the analysis of the data, we assumed that C is equal to the weight fraction w since the density of solutions is almost independent of C.It is known empirically that M, is equal to 2Me,lv7 and hence Me is given by C-lM:. Replacing Me in eqn (1) by C - l q , we cast the structural factor f into a function of CM/M:. Therefore, we can reduce f for solutions of different C into an iso-M/Me state by using the reduced molecular weight M, given by M , = CM. (2) In the following analysis, we take the bulk state as the reference. It is noted that eqn (1) is valid for systems in which polymer chains are entangled. For non-entangled systems with M < M,,fin eqn (1) may be independent of Me. According to the Rouse t h e ~ r y , ~ the relaxation time z, for the normal mode process is given by ( 3 ) where Cb is the friction coefficient per bead; N , the number of beads; and b, the average distance between beads.Since N can be chosen arbitraril~,~ we assume that N = CN" where N" (= M/wb) is the number of beads in the bulk state. The molecular weight for the bead in the bulk is constant. The subscript b refers to the arbitrarily defined bead. Thus, we assume that the number of the bead ( N = kf/Mb) changes with concentration as discussed later. Based on these assumptions, the z, at different C were reduced to an iso-Me state by shifting the log z, us. log M , plot by log C along the log M , axis with the bulk state as the reference. The relaxation time z, for the segmental mode process was also shifted in the same manner.Z, = Cb N2b2/(3n2k, T ) Concentration Dependence of [ and z, The z, us. M, (= CM,) plots for 50 and 30 YO solutions were shifted by an amount loga, along the logz, axis so that the best superpositon between the plots for the solution and for the bulk taken as the reference is achieved. The master curve thus obtained is shown in fig. 1. Using the same shift factor, we also shifted z,. The time-concentration shift factor loga, is given in table 1 and shown in fig. 2 where the ratio of zi/z, is also shown for the sake of comparison. From these results, we conclude that both the z, us. MI. and z, us. M, plots for solutions with different C are superposable by the common shift factor loga,. Since the structural factor f has been reduced to the bulk state, we may regard loga, as totally attributable to the difference in [.In other words, a, is equal to p/C, where is the friction coefficient per bead in the bulk state. The results shown in fig. 2 and table 1 indicate that p/C is approximately equal to z,O/z,, and hence defined by eqn ( I H3) is proportional to 7,: where 5 is a constant depending only on the ME for the bulk sample. c = (3, (4)K. Adachi, Y. Imanishi and T. Kotaka 1085 Fig. 1. Master curve Table 1. Shift factor and the ratio of T : / T ~ 0.50 1.60 1.70 0.30 2.20 2.10 of cis-PI. at 273 K : Weight fraction of the polymer. Monomeric Friction Coefficient The monomeric friction coefficient (, is frequently used to compare the rheological relaxation times of different materials in different states.If we replace N in eqn (3) by the degree of polymerization x, the friction coefficient in eqn (3) is equal to c,. In the present case, the ratio of the monomeric friction coefficients of the bulk to the solution G/cm is given by t:/t,, in the range of M below M,. Since tn is proportional to M2, the shifting procedure by log C along the log M , axis corresponds to the vertical shift of 2 log C. Thus, the shift factor a: for the superposition of the t,, us. M , plots without reducing the molecular weight to MI. is given by a: = Czar ( 5 )1086 Friction Coeflcient of cis-Polyisoprene Solutions Fig. 2. Concentration dependence of the shift factor a, and the ratio z:/z, : where a, is the shift factor for the z, us. M,. plots given in the previous section.Therefore, we can describe the relation between C, and 7,: C m / C i = C2zs/z,". This equation is rewritten by eqn (7) as Cm = 5, C2zs (7) with the constant Cm being independent of temperature. For cis-PI, 5, is evaluated to be 1.33 x lo2 from the data reported in part 1. Time-Concentration Superposition We have found that for superposition of the double logarithmic plot of z, us. M , for the bulk cis-PI and that for the solutions was obtained by the following procedure: (1) the molecular weight is shifted by the factor of C and ( 2 ) the friction coefficient is shifted by z;/z,. The first procedure indicates that the chain length should be scaled by dynamical monomeric units M,, being equal to Mm/C where subscript 0 represents the unit of the elementary process and M , is the molecular weight of the chemical monomeric unit. Physically, this means that the size of the elementary process for the chain motion increases with concentration.Estimation of the Size of the Segment Involved in the Elementary Process In this section we estimate the size of the unit of the elementary process. For this purpose, we use the results of computer simulation reported by Verdier and Stockmayer,* who used a model shown in fig. 3. The test chain is composed of N beads. Each bead occupies a site in a simple cubic lattice. The bead jumps from one site to the opposite site (A to B) as illustrated in the figure when the bond angle is 90". The jump frequency is taken to be l/zo. If this model chain has the perpendicular component of the dipole moment, it should orient in the direction of the bisector of the bond angle.Therefore, the local jump of the1087 Fig. 3. Model of a chain molecule with the rectangular bond angle used by Verdier and Stockmayer. Table 2. Molecular weight M,, for elementary process 1 .oo 77 0.65 112 0.50 130 0.37 129 0.22 237 a Weight fraction of the polymer. model chain from A to B corresponds to the segmental motion in the real chain. This leads us to consider that z, in the model corresponds to z, observed dielectrically. Thus, the dynamical monomer unit M, proposed above corresponds to the bead of the Verdier-Stockmayer model. The relation connecting the longest relaxation time z, for the Rouse theory and zo may be given generally by where K and N are a constant and the number of beads, respectively.Verdier and Stockmayer8 determined K by computer simulation to be 0.85 to 1.1 which varied slightly depending on the number of beads used in the simulation. Here, we employ K = 1.1 for N = 32. The relaxation time for the normal mode process t, in the range of M < M , corresponds to 7,. Using the experimental values of z, and zs, we first evaluated N for cis-PI at various C and M . Then, we evaluated the molecular weight of the dynamical monomer unit as listed in table 2. For the bulk cis-PI, M , was determined to be 77. This value indicates that the number of the carbon atoms involved in the segmental motion is 4 to 5 in the bulk state. In the 50% solutions the size becomes ca. 9 carbon atoms. The size increases with decreasing concentration. Fig.4 shows the concentration dependence of M,. As shown in the figure, M , is approximately linear to C-' in the range 0.3 < C. However, at the lowest concentration of 2070, the plot deviates from this relation. We expect that in the semidilute region, the interaction between the segments and solvent molecules becomes dominant. Thus, eqn (6) does not hold in the concentration T, = K(N- 1 ) 2 ~ , (8)1088 Friction Coeficient of cis-Polyisoprene Solutions 7------ .1 I00 t I I 1 I I I I I 50 100 c (wt %) Fig. 4. Concentration dependence of the molecular weight M,, of the elementary process. range below a critical concentration Ccrit. Using the Rouse-Zimm theory, we have estimated Ccrit as follows. According to the free draining model, 4', is given by where qs is the viscosity of the solvent; [q], the intrinsic viscosity; NA, the Avogadro number.We assume that when [, given by eqn (6) becomes lower than 5, given by eqn (9), eqn (6) does not hold. Based on this assumption, Ccrit was evaluated to be 0.07 for cis-PIltoluene system. Here, we used [q] of cis-PI in benzene to calculate eqn (9). The deviation of the plot at C = 20 wt % from the trend that M , is proportional to C-' in fig. 4 is due to the onset of this crossover. Therefore, eqn (4) and (6) are valid in the concentration range above ca. 20 O/O. In part 2, we reported the dielectric relaxation time z(p) for rotation of the toluene molecules. It was pointed out that z, (aII process) is approximately proportional to zv). Thus, we expect that 4' in dilute solutions may be expressed, by an equation similar to eqn (8) using ~(p) instead of z,. Conclusions (1) The z, us.molecular weight curves for various concentrations are superposable with the vertical shift along the 5, axis z,O/z, and with the horizontal shift C. (2) The monomeric friction coefficient 4' for the normal mode process is proportional to C22,. (3) The size of the elementary process for bulk cis-PI is 77 and expands with decreasing concentration in proportion to C-' in the range C > 0.2. This work was partly supported by Grant-in-aid for Scientific Research by the Ministry of Education, Science and Culture (6055062). We also acknowledge the financial support from the Institute of Polymer Research, Osaka University. References 1 W. W. Graessley, Adz;. Polym. Sci., 1982, 47, 67. 2 G. C. Berry and T. G. Fox, Adz;. Polym. Sci., 1968, 5 , 261. 3 K. Iwata and M . Kurata, J . Chem. Phys., 1969, 50, 4008. 4 K. Iwata, J . Chem. Phys., 1974, 58, 4184.K. Adachi, Y. Irnanishi and T. Kotaka 1089 5 K. Adachi, Y. Imanishi and T. Kotaka, J. Chem. SOC., Faraday Trans. I , 1989, 85, 1065. 6 K. Adachi, Y. Imanishi and T. Kotaka, J. Chem. Soc., Faraday Trans. I , 1989, 85, 1075. 7 J. D. Ferry, Viscoelastic Properties of Polymers (Wiley, New York, 3rd edn, 1980), Chapter 17. 8 R. H. Verdier and W. H. Stockmayer, J . Chem. Phys., 1962, 36, 227. 9 P. E. Rouse, J . Chem. Phys. 1953, 21, 1272. Paper 8/02269K; Received 6th June. 1988

ISSN:0300-9599    

DOI:10.1039/F19898501083

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J . Cliem. Soc., Furuday Trans. I , 1989, 85(5), 1091-1097 Diaphragm Cell for High-temperature Diffusion Measurements Tracer Diffusion Coefficients for Water to 363 K Allan J. Easteal,? William E. Price* and Lawrence A. Woolf Atomic and Molecular Physics Laboratories, Research School of Physical Sciences, The Australian National Unioersity, Canberra, A.C. T . 2601 , Australia A modified diaphragm cell for diffusion measurements at high temperatures is described. It has been used to obtain tracer diffusion coefficients for 3HH0 and H,lsO in water from 298 to 363 K. Where comparisons are available, the results are in good agreement with literature data. The data have been converted to self-diffusion coefficients of water and combined with existing data to provide a series of equations covering the temperature range 242-498 K.The magnetically stirred diaphragm cell of the Stokes design' is a well established technique for measuring diffusion coefficients in liquid media,2 particularly at atmospheric pressure. It is a simple, elegant method which is very versatile in the range of systems which can be studied, and provides coefficients accurate to f 0.5 O/O or better. Although modifications of the design to allow continuous monitoring of the concentrations of electrolytes in the cell compartments during an experiment have been pre~ented,~ a versatile cell enabling measurements at high temperatures has not yet been developed. Some designs have been put forward,* but all have had various experimental difficulties associated with them or have been restricted in their use.The aim of the present work was to design and test a new diaphragm cell for use at high temperatures but which retained the essential versatility of the conventional Stokes type cell. One of the main motivations for this was to study the diffusion of caffeine in aqueous solution at elevated temperatures. This is of importance in the extraction of caffeine from natural products in the food and pharmaceutical industries5 The main requirement of a high-temperature cell, is to be able to obtain a sample from each cell compartment at the end of the experiment without the manipulation of the cell customary with standard design.6 This is necessary to avoid bulk flow occurring during the sampling process due to changes in temperature in the contents of the cell.Problems of evaporation also need to be minimised. In experiments with solutes in solution another requirement was to be able to fill the cell at high temperatures. This enables experiments using solutions where the concentration of the solute exceeds its solubility at room temperatures to be performed. This again has application in the study of caffeine where its solubility is low at 298 K but increases markedly with temperature. In order to test the new design, tracer diffusion coefficients of 3HH0 and H2180 in water were measured from 298 to 363 K. This enabled a comparison with the 3HH0 data of Mills7 (274-3 18 K) and the H2"0 results of Easteal and co-workers8 (278-323 K) to validate the experimental procedures. It then enabled the extension of the temperature range of the earlier data for the diffusion of the abovementioned species in water.? Department of Chemistry, University of Auckland, Auckland, New Zealand. 10911092 Diaphragm Cell for Difusion Measurements Experimental The modified cell is shown in fig. 1. The cell compartments were matched volumes,6 and were both 50 (kO.5) cm3. The major change in design was the bottom compartment opening and plug which facilitated the taking of the samples in situ in the bath without having to invert the cell as for a standard diaphragm cell experiment.6 The hollow tubes in the side-arm plug enabled liquid to be removed from the side cavity. It should be noted that the bottom compartment stirrer bar is enclosed within the cell during its construction and cannot be replaced.The stirrer must therefore be chosen in advance to cover the range of likely liquid densities for the diffusion experiments. The constant temperature bath, as depicted in fig. 2, was used in the experiments using water as the thermostat fluid. In order to minimise evaporation and improve temperature control a number of modifications of the standard design were employed. A closed unit was utilised with foam insulation in the bath walls and lid. All the controllers and stirrers were positioned so that they entered the bath from a side panel. This meant that the bath could be made with a very small liquid surface area. The top panel had a removable section to allow tracer material to be added to the cell and samples to be taken from it.A reservoir kept the liquid level constant and a coil from a circulating water bath (not shown in fig. 2) enabled fine tuning of the temperature control. The tracer diffusion experiments with water were carried out using the established solvent-filled diaphragm technique.6 The cell was filled while it was mounted on a rotating cradle inside a temperature controlled air bath. This also allowed the filling of the cell to be performed at (approximately) the temperature at which the experiment was to take place, thus minimising the formation of bubbles and enabling the use of solutions where the solute was sparingly soluble at room temperature. In practice, a filling temperature of up to 20 K below the experimental temperature, was sufficient. The cell was filled with distilled, deionised and degassed water, by sucking on the side-arm tube forcing liquid from the top compartment and attached reservoir volume through the diaphragm, ensuring that air was expelled from the sinter.To facilitate the removal of bubbles from the bottom compartment a small indentation by the capillary opening on the side arm was incorporated into the design. This permitted the cell to be oriented so that any offending bubbles became situated adjacent to the opening, thus enabling their removal. Once the cell had been filled and the plugs inserted, the hollow tubes in the side-arm plug enabled any liquid in the isolated side arm to be removed using a syringe. The side arm was then flushed with acetone and dried by flowing air through the system to remove any trace of liquid.On completion of the run, the sampling was carried out with the cell in situ in the temperature controlled bath. The top compartment sample was taken in the usual manner.6 All the remaining liquid in the upper compartment was then removed and a small amount of liquid from the lower one was drawn through the sinter, after loosening the side-arm plug. This liquid was then removed from the upper chamber. The bottom sample was then taken by drawing it through the capillary side arm. Temperatures were controlled to better than k0.02 K. The amounts of tracer used per run were 5 x moles of 3HH0 and 5 x lo-' moles for H,"O. Concentrations of 3H in the samples were measured using standard liquid scintillation counting techniques,6 and concentrations of l80 by isotope ratio mass spectrometry.8 Prior to the tracer diffusion experiments the cell was calibrated using a solution of 0.5 molar potassium chloride as previously described,6 to obtain an accurate cell constant.This was repeated at the conclusion of the investigation.A. J . Easteal, W. E. Price and L. A. Woolf A B 1093 Fig. 1. Design of high-temperature diaphragm cell : glass cell with glass (Sovirel) sinter and brass supporting base. Inset A: standard design of top plug.3 Inset B: new type of bottom plug made from PTFE and hollow stainless steel tubing. 6 6 0 8Q 0 <3po - l C Fig. 2. High-temperature bath for diffusion cell. A: Bath stirrer; B: Bevel gear stirrer with cell mounting for magnetic cell stirrers; C: Perspex viewing window; D: Bath controller.integrated heater and1094 Diaphragm Cell for Difusion Measurements Results and Discussion The experimental results for the tracer diffusion coefficients of 3HH0 and H,180 in water as a function of temperature are shown in table 1. Each listed value is a mean of in most cases three runs. The overall accuracy is estimated to be kO.68 YO for the 'HHO results and k0.2 O/O for H,180 or better at the lower temperatures. There is good agreement with the value at 298.15 K for 3HH0 with the accepted value of Mills.7 Each set of values were combined with previously obtained data for 'HHO-H,O (274-318 K) and H,180-H,0 (278-323 K) at lower temperatures,'.' and fitted by least squares to the function In ( 109D/m2 s-') = a, +a,[ 1000/( T/K)] +a,[ 1000/( T/K)],.The coefficients of eqn ( I ) for 'HHO and H2180 are given in table 2. The maximum deviation of the fitted curve from the experimental data is +_ 1 .O O/O for the :3HH0 values and k0.6 YO for the H,'*O data. For the fit for H,180 the present value at 323.15 K was used. The previously obtained value8 was some 3 YO lower, outside experimental uncertainty, and inadequate temperature control or insufficient stirring of the diaphragm cell compartments may be the cause of the discrepancy. Furthermore, the present value of 3.96 x lo-' m2 s-' is in better agreement with the other data of ref. (7). The successful combination of the high-temperature diffusion data with those of Mills7 and Easteal and co-workers8 shows that the new design of the diaphragm cell is a successful adaptation of the standard one.The two equations of the form of eqn ( I ) describe the diffusion behaviour of the two tracer species in water over essentially the whole liquid range at atmospheric pressure to k 1.0%. Table 3 gives a number of quantities from the equations at selected temperatures in the liquid water range. It may be seen that the percentage difference between D('HH0) and D(H,"O) is not constant with changing temperature, although the molecular masses of 'HHO and H , W are the same. This indicates differences in the diffusional behaviour of the two species with temperature due to effects of the position of the labelled atoms on intermolecular interactions. Activation energies E, of the diffusion processes were calculated from the derivatives of eqn (1).At lower temperatures E, for 3HH0 and H2180 are both 18-20 kJ mol-'. These are consistent with Rowlinson's value of the strength of a hydrogen bond in water,g with previously quoted values from this l a b ~ r a t o r y , ~ and also with values based on viscosity data." The decrease in E, with temperature shown in table 3 suggests a decrease in the effect of hydrogen bonding on the diffusion of water as temperature is raised, in keeping with the known breakdown of water structure with increases in temperature. A recent paper by Abraham and Abraham'" has used the Batschinski equation" to correlate the fluidity (l/viscosity) of water to its molar volume using a similar approach to that of Hildebrand.', They showed that for water the fluidity varied linearly with the molar volume over a limited temperature range (ca.313-373 K). As accurate tracer diffusion data for water are available for the complete normal liquid range, it is interesting to determine whether these data may be correlated in a similar manner. The molar volumes were calculated using an equation by Kell.13 This is shown in fig. 3 for 3HH0 in water; the plot is typical of all three isotopic species of water. As can be seen, a gentle curve may be fitted to the data with a good fit being obtained between 298-363 K. If the curve is extrapolated to zero diffusion a molar volume of 17.8 cm" mol-' is obtained, compared to a value of 17.6 cm" mol-' from the viscosity data. This agreement is within experimental error and confirms the usefulness of the free volume approach in summarising transport data in pure water.We combined the present tracer diffusion data with self-diffusion of water data in temperature regions not covered by the diaphragm cell work so as to cover the self- diffusion of liquid water along the coexistence curve from 242 to 498 K.A . J . Easteal, W. E. Price and L. A . Wooq 1095 Table 1. Tracer diffusion coefficients measured for diffusion of H2180 and 'HHO in ordinary water D/ 1 OP9 m2 s-' ~ ~ - - ~ temperature/K H,180 'HHO 298.15 - 2.236 323.15 3.960 3.844 338.15 5.138 5.069 353. I5 6.517 6.217 363.15 7.496 7.259 Table 2. Coefficients of eqn (1) for diffusion of H,"O and 3HH0 in ordinary water tracer a1 a2 a3 H,180 4.235 560 5 0.164 281 8 -0.352 540 8 ~ H H O 3.146 013 9 0.825 393 5 -0.454 338 7 Table 3.Diffusion coefficients for H2l8O and 3HH0 in ordinary water and the activation energies of diffusion from eqn (1) 273.15 283.15 293.15 298.15 303.15 313.15 323.15 333.15 343.15 353.15 363.15 373.15 1.082 1.484 1.964 2.234 2.523 3.155 3.857 4.620 5.437 6.300 7.200 8. I29 1.1 18 1.520 2.00 1 2.272 2.563 3.206 3.927 4.722 5.587 6.514 7.499 8.533 3.38 -20.8 -20.1 2.43 -19.8 -19.3 1.87 -18.9 -18.6 1.71 -18.5 -18.3 1.62 -18.1 -18.0 1.62 -17.3 - 17.4 1.83 -16.5 - 16.8 2.22 - 15.8 - 16.2 2.74 - 15.2 - 15.7 3.39 - 14.5 - 15.2 4.14 -13.9 -14.8 4.97 - 13.4 - 14.3 a Percentage difference between D(3HHO) and D(H,laO). The conversion of the present tracer data, embodied in eqn ( l ) , to self-diffusion values, is done by utilising the H,180 data rather than the tritiated values.The difference in the diffusion of H,160 and its l80 counterpart is entirely due to the isotopic mass difference. The temperature dependence of this effect should be small, less than the experimental uncertainty in the diffusion data. It thus seems reasonable to use the ratio D(H,l6O)/ D(H,"O) of 1.0105 at 298 K (where the values are most accurately known) to obtain good estimates for the self-diffusion of water over a temperature range of 274-363 K. The accepted value6 for D(H,160) is 2.299 x lop9 m2 s-'. Work covering the lower regime (242-298 K) has been reported by Gillen" and others using n.m.r. techniques. Their values above 273.15 K are on average 3.18 YO lower than1096 Diaphragm Cell for Diflusion Measurements I I I I I I I I 18.0 18.2 18.4 18.6 18.8 Fig.V, /an3 rnol-' 3. 3HH0 tracer diffusion coefficients in water between 274 and 363 K plotted against molar volume of water (V,). Table 4. Coefficients of eqn (2) temperature range/ K a0 a1 a2 a3 1 02dZ 278-498b 6.119 03 - 1.195 93 -0.052 29 -0.018 41 1.1 242-363' 30.831 1 -24.904 1 7.487 22 -0.813 27 2.8 2 4 2 ~ l 9 8 ~ 13.217 2 -9.086 02 2.808 83 -0.357 13 4.1 a Root mean square deviation of In (D) values. ref. (14). Present work and ref. (13). Present work and Present work and ref. (13) and (14). the accepted values of Mills.7 It thus seems reasonable to scale Gillen's data by this factor, The high-temperature range (275-498 K) has been covered, again by spin-echo techniques, by Krynicki and co-~orkers.'~ In the range of overlap, the latter group's results are an average of 1.7 O/O lower than the corrected present data and this factor has been used for the correlation.These adjustments are less than the quoted experimental uncertaintyf4* l5 for both sets of data ( 5 "/o). Table 4 shows the coefficients of the fitting of a third order polynomial over portions or all of the temperature regime together with estimates of the goodness of the fit. The fitted function was of the form given in eqn (2) ln(109D/m2 s-') = a,+al[1000/(T/K)]+a2[1000/(T/K)]2 +a,[1000/(T/K)]3. (2) It may be seen that the accuracy of the fit over the complete temperature range is rather disappointing. This is due largely to the scatter in Gillen's data ( & 6-7 Yo). However the separate equations covering the two overlapping ranges (242-363 K and 273-498 K) give better fitted values (k 2 '/o or better in most cases).The use of the accurate tracerA . J . Easteal, W. E. Price and L. A. Woow 1097 data to normalise the two sets of n.m.r. results should provide more accurate values of water diffusion coefficients. It is sensible therefore to suggest that these two latter equations be used for obtaining estimates of the self-diffusion of water between 242-273 K and 373498 K. It should be emphasised that in the temperature range 278-363 K the most accurate coefficients are obtained from eqn (1) for the H,180 data converted by the appropriate factor. Conclusions The new design of diaphragm cell has been shown to provide accurate diffusion coefficients for water at elevated temperatures.The nature of the design is such that it is suitable for the study of a wide range of systems. The present tracer diffusion data for water has enabled equations to be formulated for the temperature range 274-363 K for 3HH0 and 278-363 K for H,lsO. The correlation of the present results with earlier tracer and self-diffusion data has provided equations representing the combined data for the temperature ranges 242-498 K. The authors are grateful for the contributions of F. L. Wilson, for construction of the temperature-controlled bath and the cell plugs, and of C . J. Tomkins and H. Adler for glassblowing. W. E. P. expresses his thanks to the Commonwealth Government of Australia for the award of a Queen Elizabeth I1 Research Fellowship. References 1 R. H. Stokes, J . Am. Chem. Soc., 1950, 72, 763. 2 H. J. V. Tyrrell and K. R. Harris, DifSusion in Liquids (Butterworths, London, 1984). 3 A. F. Collings, D. C. Hall, R. Mills and L. A. Woolf, J . Physics E., 1971, 4, 425. 4 S. A. Sanni and H. P. Hutchison, J . Physics E., 1968, 1, 1101. 5 S. N. Katz, 9th. Colloq. Sci. Int. Cafe, 1980, 295. 6 R. Mills and L. A. Woolf, The Diaphragm Cell (Australian National University Press, Canberra, 7 R. Mills, J . Phys. Chem., 1973, 77, 685. 8 A. J. Easteal, A. V. J. Edge and L. A. Woolf, J . Phys. Chem., 1984, 88, 6060. 9 J. S. Rowlinson, Trans. Faraday Soc., 1949, 45, 974. 1968). 10 M. Abraham and M-C. Abraham, Electrochim. Acta., 1987, 32, 1475. I 1 A. J. Batschinski, Z . Phys. Chem., 1913, 84, 643. 12 J. H. Hildebrand, Science, 1971, 174, 490. 13 G. S. Kell, J . Phys. Chem. ReJ Data, 1977, 6, 1109. 14 K. T. Gillen, D. C. Douglass and M. J. R. Hoch, J . Chem. Phys., 1972, 57, 51 17. 15 K. Krynicki, C. D. Green and D. W. Sawyer, Faraday Discuss. Chem. SOL.., 1979, 66, 199. Paper 8/02326C; Received 10th June, 1988

ISSN:0300-9599    

DOI:10.1039/F19898501091

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J . Chern. Soc., Furuduy Trans. I , 1989, 85(5), 1099-1 I10 Fourier- transform Nuclear Magnetic Resonance Studies of the Effects of 2-Chloroethanol on the Association of N-Acetyl-L-amino Acid N',N'-Dimethylamides in Aqueous Solutions Kazuko Mizuno, Tomoko Takagi, Yohko Ikeda and Yohji Shindo" The Research Institute for Material Science and Engineering, Faculty of Etigineering, Fukui University, Bunkyo, Fukui 910, Japan The effects of 2-chloroethanol (CIEtOH) on the self-association of three N- acetyl-L-amino acid N',N'-dimethylamides (Val, Leu, Phe as amino acids) have been studied in water-deuterium oxide solutions by 'H and 13C F.t.n.m.r. These amides were found to self-associate through hydrogen bonding to a small extent. Hydrogen-bonding association of the amides takes place when [CIEtOH]/[amide] > 1.The hydroxyl proton peak of water-C1EtOH mixtures shifts to high field regardless of the presence of the amide owing to the disruption of water structure by CIEtOH. The decrease in the hydrogen-bonding capability of the solvent leads to the hydrogen- bonding association of the amide and ClEtOH induces the hydrogen- bonding association of the amides by decreasing the hydrogen- bonding capability of the solvent around the amides. An apparent dissociation of the amide complexes was found to occur at low concentrations of CIEtOH. This is attributed to the hydrophobic interaction between ClEtOH and the alkyl part of the amides to minimize the energy loss due to the disruption of water structure, and the interaction results in a further decrease in the hydrogen-bonding capability of the solvent around the amides and leads to the start of hydrogen-bonding association of the amides.2-Chloroethanol (ClEtOH) is well known as a strong helix-forming reagent for proteins. Timascheff and Inoue reported that ClEtOH binds preferentially to proteins and simultaneously induces a large change in protein conformation due to the interaction between non-polar groups of protein and C1EtOH2,3 in the same way as in the helix formation induced by a l k a n ~ l s . ~ . ~ In later studies on the preferential binding of ClEtOH to proteins and polypeptides,6 the same explanation has been presented. The solubility measurements of 2-substituted ethanols into hexadecane and octadecane showed that the hydrophobic property of ClEtOH is a little larger than that of ethanol (EtOH), but less than that of propan-1-01.~ We examined the denaturing abilities of several halogenoethanols and halogenopropanols, and compared them with those of the corresponding alkanok8 We discovered that the halogenoalcohols have almost the same denaturing ability irrespective of their hydrophobicity and that they induce large conformational changes in proteins at much lower concentrations than the corresponding alkanols.Furthermore, we studied the preferential binding of EtOH and propan-2-01 to lysozyme and found that the preferential binding does not occur up to 60 vol % of alcohol in these alkanol-water mixture^.^ These results suggest that we may not be able to attribute the preferential binding of ClEtOH to protein solely to the interaction between hydrophobic groups and that the mechanism of helix formation should differ between the halogenoalcohols and the alkanols. However, there are a few works studying the difference in the mechanism of helix formation induced by the halogenoalcohols and corresponding alkanols.'3 l1 10991100 F. T.N.M.R. Study of Self-association We noticed the proton-donating acidity of ClEtOH is much larger than that of EtOH because of the electron withdrawal by the chlorine atom.8 Thus, in the denaturation process of protein by ClEtOH, we can assume an intermediate state before helix formation, where ClEtOH forms a hydrogen bond with the carbonyl groups of the protein. Accordingly, we have studied the interaction between the alcohols and some amide compounds, as the model molecules of amino-acid residues within protein, in carbon tetrachloride solutions and found that ClEtOH and other halogenoalcohols have proton-donating acidity strong enough to dissociate very stable inter-amide hydrogen bonds.ti However, these findings are not necessarily applicable to aqueous systems, although these suggest the possibility of hydrogen-bonding interaction of ClEtOH with protein . In this work, we extend our study to the aqueous solutions of some amides, with the intention of presenting a mechanism of helix formation induced by ClEtOH. For this purpose, we investigated the effects of ClEtOH on the association of three N-acetyl-L- amino acid N',N'-dimethylamides in water-deuterium oxide (D,O) using F.t.n.m.r.We chose L-Val, L-Leu and L-Phe since these are known to have helix-forming capability.12 The effects of EtOH were also examined for comparison. The results are discussed from the following four standpoints : (1) whether inter-amide hydrogen-bonding association of the amides occurs, (2) whether hydrogen-bonding interaction occurs between ClEtOH and the amides, (3) whether a hydrophobic interaction occurs between ClEtOH and the amides and (4) whether ClEtOH causes some significant change in water structure and the change is in some way related to the association of the amides. The last point arose because we found it necessary to consider the interaction between ClEtOH and the solvent, water, to interpret our results. Experiment a1 The three N-acetyl-L-amino acid N',N'-dimethylamides were the same as those used in our previous work,' and are abbreviated as VMe,, LMe,, and PMe,.The specific rotations were as follows : LMe,; [a]:& - 26.6 (c = 1.055 in ethanol), VMe,; [a]::, - 2.27 (c = 0.967 in ethanol), PMe,; [a]& + 34.1 (c = 1.099 in ethanol). D,O (99.8 atom O/O D) was purchased from Aldrich. Water was prepared by distilling deionized water. ClEtOH and N-methylacetamide (Wako Chemicals, guaranteed-reagent grade) were doubly distilled under highly reduced pressure before use. EtOH (Wako, Spectrograde) was used as it was. The 'H and 13C F.t.n.rn.r. spectra were measured with a JEOL GX-270 spectrometer operating at 270 MHz for 'H and 67.5 MHz for 13C. The ratio of D,O to H,O was 1 : 4 by volume. The chemical shifts are referred relatively to external sodium trimethyl- silylpropanesulphonate dissolved in D,O confined to a coaxial tube with the sample tube.Spin-lattice relaxation times (q) were obtained by the usual (1 80°-t-900-T), pulse sequence. The difference in the nuclear Overhauser effect (NOE) of the two carbonyls arising from irradiation of NH or OH protons was measured under off-noise-decoupling by repeating alternative accumulations with and without the irradiation. Results and Discussion Self-association of the Amides through Hydrogen Bonds We measured the 'H and 13C n.m.r. spectra of each amide with increasing concentration to investigate whether the amides associate through inter-amide hydrogen bonds. Fig. 1 shows the 'H chemical shifts of NH protons us.concentration. The NH peaks shift to higher field slightly with increasing the concentration. The CH,CO and NCH, protons also shift very slightly to higher field. In fig. 2, the chemical shifts of the two carbonyls0.3 0.6 [amide] /mol dm-3 0 Fig. 1. The 'H chemical shift of NH of each amide 27s. the concentration at 24k0.2 "C: 0, PMe,; ., VMe,; A, LMe,. 176.5 176.0- - FL ,a CQ 1 75.5. 175.0 0 0.2 0.4 0.6 0.8 1.0 [amide]/mol dm-3 Fig. 2. The plots of the 13C chemical shifts of the two carbonyls of each amide us. the concentration at LMe,, where the open symbols are for CH,CO and the filled symbols are for >NCO. 2 4 ~ 2 0 ~ : ( 0 ~ 0 ) m e , , (0, m) m e , and (A, A) Y L ,s1102 F. T. N.M. R. Study of Self-association 4 . 7 4 1 O 7 14 21 [ClEtOH]/[amide] Fig.3. Effects of increasing ClEtOH content on 'H chemical shifts of NH (a) and OH (6) of each amide solutions at 24k0.2 "C: 0, [PMe,] = 0.37 mol dm-3; H, WMe,] = 0.30 rnol dm-3; A, [LMe,] = 0.80 mol dm-3. are plotted us. concentration. The assignment of the two carbonyls can be easily made by observing the fine structure of the CH,CO carbonyl due to the long-range C-H couplings. The extent of the high-field shift of the NH and the carbonyls is very small, but it is ten times larger than the spectral resolution under the measuring conditions. Thus, we conclude that the shifts are large enough to be analysed. We consider that the N-H and C=O bonds of the amides are polarized to a great extent in very dilute aqueous solutions because of the large hydration of the monomer.The formation of an inter-amide hydrogen bond >NH -.. O=C<should result in a decrease in the polarization of both N-H and C=O bonds. This must be more prominent for the C=O bond, because NH is thought to be a weaker proton donor than the water hydroxyl. The upfield shift observed for both NH and carbonyl groups can be attributed to inter-amide hydrogen bonding. Thus, we conclude that hydrogen-bonding association of the amides occurs in aqueous solutions, although the extent of the association is small.K. Mizuno et al. I 0 5 10 1103 [CEtOHI /[amidel Fig. 4. Effects of increasing ClEtOH content on 13C chemical shifts of the two carbonyls of each amide at 24k0.2 "C: (0, 0 ) [PMe,] = 0.37 mol dm-3, (0, .) WMe,] = 0.30 mol dm-Y, and (A, A) [LMe,] = 0.80 mol dm-3, where the open symbols are for CH,CO and the filled symbols are forSNCO.Effects of ClEtOH on the Self-association of the Amides We then measured the chemical shifts of the NH and the two carbonyls of each amide by increasing the content of ClEtOH under a constant amide concentration. The percentage by volume of ClEtOH was ca. 40 O/O at the highest concentration. The results are shown in fig. 3 and 4. Both the NH and the carbonyls shift to high field with increasing ClEtOH, which can be interpreted as the result of inter-amide hydrogen bonding on the basis of the reasoning mentioned above. This means that ClEtOH promotes inter-amide hydrogen bonding in aqueous solutions, contrary to the dissociation of the amide complexes observed in carbon tetrachloride solutions in our previous work.8 In order to confirm this conclusion, we measured the 13C NOE difference of the carbonyls resulting from the selective irradiation of the NH proton with increasing ClEtOH concentration under a constant amide concentration.The values obtained (abbreviated as 13C-NOE(NH)) can be directly correlated to the extent of inter-amide hydrogen bonding, while the values observed by irradiating the OH proton (abbreviated as l3C-NOE{OH)) can be correlated to the extent of hydration of the carbonyls. The 13C- NOE{OH) values are larger than those of 13C-NOE(NH} and so more suitable for quantitative analysis. Fig. 5 shows the changes in the *3C-NOE{OH) of PMe,. A very1 I04 F. T.N.M.R. Study of Self-association 0.5 0.4 s 0.3 2 8 0 2 0.2 m d 0.1 0 Y 3 6 [ROHl/[amide] 0 Fig.5. Effects of increasing the content of ClEtOH or EtOH on l3C-NOE(OH) values of the two carbonyls of PMe, at [PMe,] = 0.65 mol dm-3 and 15k0.1 "C; ClEtOH (0, CH,CO; .,>NCO), and EtOH (0, CH3CO; o,>NCO). similar result was obtained for VMe, and LMe,. By adding a small amount of ClEtOH, the 13C-NOE{OH} increases slightly in =NCO and remains almost unchanged in CH,CO, indicating that the amide complexes dissociate to a small extent and the carbonyls are hydrated or coordinated to ClEtOH hydroxyl groups. By adding more ClEtOH, the 13C-NOE(OH} values began to decrease, and simultaneously the 13C- NOE{NH) values began to increase, although a quantitative treatment was not possible. These results support the promotion of the inter-amide hydrogen bonding by ClEtOH.It was necessary to confirm the initial dissociation of the amide complexes concluded from the 13C-NOE{OH} measurements. Thus we carried out q measurements of the methyl protons of the amides with increasing ClEtOH concentration under a constant amide concentration. As of the methyl protons is taken to be a measure of the freedom of rotation of the methyl groups, it reflects the extent of intermolecular association of the amides, i.e. becomes larger or smaller as the extent of the association decreases or increases. The results are shown in fig. 6. We note that q increases initially and has a maximum at a concentration ratio of ca. 1 irrespective of the amide concentration, thus supporting the conclusion derived from the 13C-NOE measurements. Effects of ClEtOH on Water Structure So far we have investigated whether hydrogen- bonding interaction occurs between ClEtOH and the amides in aqueous solutions.However, what we observed in this work is the hydrogen-bonding interaction between amide molecules promoted by ClEtOH. To discover the decisive factor determining the extent of inter-amide hydrogen-bonding association in various solvents, we re-examined the experimental results obtained in our previous studies. We examined the self-association of VMe, and PMe, in carbon tetrachioride solutions using n.m.r. and i.r.13 The NH protons shifted 50 times more than in this work and the degree of association calculated from the i.r. data was 0.21 inK. Mizuno et al. 1105 4 0 5 10 15 [ClEtOHl/[PMe2 1 2 1 A 0 5 10 15 I- t k 0 1 2 Fig.6. Effects of increasing ClEtOH content on of the methyl protons of each amide at 24k 0.1 "C at (a) [PMe,] = 0.37 mol drn-,, (6) [VMe,] = 0.30 mol dmP3 and (c) [LMe,] = 0.80 rnol dm-,: 0, CH,CO; +, NCH,; A, phenyl protons of PMe,, and CCH, protons of VMe, and LMe,. VMe, and 0.14 in PMe, at 0.3 mol dm-3 and 21 "C. Although we cannot determine the degree of association in D,O-water solutions in this work, it is reasonable to assume much smaller values than those in carbon tetrachloride solutions. Another study was performed on the association of amide compounds in alcohokarbon tetrachloride mixtures.8 Dissociation of the inter-amide hydrogen bond occurs with increasing ClEtOH content in the mixed solvent, whereas the degree of inter-amide hydrogen bonding remained unchanged with increasing EtOH concentration.From all these results, we find that the hydrogen-bonding capability of solvents is the factor determining the extent of the hydrogen-bonding association of amides. The1106 F. T.N.M.R. Study of Self-association volume fraction of ClEtOH 0 0.5 0 1.0 mole fraction of ClEtOH Fig. 7. The 'H chemical shifts of the OH in the water-ClEtOH system with different molar ratios at 205-0.1 "C. degree of association increases with decreasing hydrogen-bonding capability of the solvent and vice versa. This is because a solvent with a strong hydrogen-bonding capability solvates amide groups strongly through hydrogen bonding and disturbs the inter-amide hydrogen bonding, whereas inter-amide hydrogen bonding predominates in solvents which have weak hydrogen-bonding capability.We measured the chemical shift of the OH proton in water-ClEtOH mixtures at a series of different mole fractions of ClEtOH. We observed a single hydroxyl peak only, the chemical shift of which is a weighted average of the values for the alcohol and water hydroxyls. The data obtained are plotted in fig. 7. The hydroxyl of pure ClEtOH is at slightly lower field than the pure-water hydroxyl. The addition of ClEtOH to water does not shift the coalesced hydroxyl peak to lower field than the pure-water hydroxyl but does shift it to higher field over the wide range of the mole ratio of ClEtOH. Since the high-field shift occurs over the entire concentration range of ClEtOH, we consider that both the water and ClEtOH are responsible for the high-field shift of the coalesced peak.Thus, we know that the hydrogen-bonding capabilities of water and ClEtOH are decreased by mixing, therefore ClEtOH acts as water structure-breaker. l4 Gierycz et al. measured the enthalpy of mixing of the water-ClEtOH system and observed that mixing is an endothermic process over the wide range of C1EtOH.l5 As the weakening of hydrogen bonds is an endothermic process, the results of our n.m.r. data are consistent with those based on the thermodynamic data. We measured the chemical shift of the hydroxyl peak in water-ClEtOH solutions of the amides with increasing ClEtOH concentration. The water and alcohol hydroxyls coalesce into a peak over the alcohol concentrations measured (fig.3). The hydroxyl peak shifts to high field with increasing ClEtOH concentration, indicating that ClEtOH weakens the hydrogen-bonding capability of water as well as water in amide solutions. It has been reported that alkanols stabilize water structure at low c~ncentrationl~ and that the low-field shift of the water hydroxyl is evidence of the enhancement of water Coccia et al. attributed the low-field shift to hydrogen bonding betweenK . Mizuno et al. 1107 8.3 8.2 8.1 n E ,a Lo 4.95 4.90 4.85 Y 0 2 4 6 8 [EtOH]/[ amide] Fig. 8. Effects of increasing EtOH content on 'H chemical shifts of NH (a) and OH (6) of each amide solution at 22k0.2 "C. 0, [PMe,] = 0.65 mol dm-3; ., [VMe,] = 0.45 mol dm-3; A, [LMe,] = 0.70 mol dm-3.water and EtOH and to the enhanced hydrogen bonding between water molecules owing to the hydrophobic hydration of ethyl groups.18 We also examined the effects of EtOH on both the association of the amides and the water structure of the amide solutions for the comparison with those of ClEtOH. We measured the chemical shifts of NH and OH protons with increasing EtOH (fig. 8). The NH peak hardly changes over the entire concentration range examined, so we know that EtOH hardly affects the association of the amides. The water and EtOH hydroxyls coalesce over the alcohol concentrations measured and shift to lower field with increasing EtOH concentration. Accordingly, we know that EtOH acts as a water structure-maker in the amide solutions. We also measured the 13C-NOE(OH) of the two carbonyls of PMe, with increasing EtOH concentration (fig. 5).The 13C-NOE(OH) values of the two carbonyls increase slightly at low EtOH concentrations, indicating the dissociation of the amide complexes to a small extent. Further increase in EtOH concentration does not induce any changes in the 13-NOE(OH} values, in agreement with the constant chemical shift of NH shown in fig. 8. We interpret all these results as follows. The increase in ClEtOH concentration in aqueous solutions of each amide leads to a decrease in the hydrogen-bonding capability of the water-ClEtOH mixtures because of the disruption of the water structure and the1108 F.T.N.M.R. Study of Self-association 4 VJ 23 2 0 1 2 3 4 [ ClEtOH] / [ N MA] Fig. 9. Effects of increasing ClEtOH content on the association of NMA at WMA] = 3 mol drn-,.(a) Changes in of the CH,CO (0) and NCH, (A) protons, (b) changes in the lH chemical shift of NH and (c) changes in the 'H chemical shift of the water hydroxyl. increase in the hydrophobic property arising from the alkyl part of ClEtOH. This results in promotion of the hydrogen-bonding association of the amides, as demonstrated experimentally above. In the water-EtOH mixtures, on the other hand, we consider two opposite factors affecting the hydrogen-bonding capability of the mixture, viz. enhancement of the water structure and increase in the hydrophobic property of the system. At low EtOH concentrations, water-EtOH mixtures have increased hydrogen- bonding capability because of the large contribution from the enhancement of the water structure, leading to the initial dissociation of the amide complex as shown in fig.5. However, further addition of EtOH increases the hydrophobic property of the mixture and offsets the hydrogen-bonding capability of the mixture. Consequently, the extent of amide association remains almost unchanged (fig. 5). A Mechanism for Disruption of Water Structure by ClEtOH We must consider the reason why ClEtOH behaves as a water structure-breaker, whereas EtOH acts as a structure-maker. So far we have focussed only on the proton-donating acidity of alcohols. However, from the fact that EtOH acts as a structure-maker, theK. Mizuno et al. 1109 enhancement of water structure seems to be correlated to the proton-accepting basicity of the hydroxyl oxygen of alcohols.Kamlet and Taft defined the hydrogen-bond accepting basicity of many organic solvents based on the data obtained by using solvatochromic compari~on.'~ They reported that EtOH has a much greater basicity than water because of the electron-repelling capability of its ethyl group. Thus, the hydroxyl oxygen of EtOH should induce polarization of the neighbouring water molecules and result in the enhancement of the water structure. On the other hand, the basicity of ClEtOH is only slightly larger than that of water.lg Therefore, we can assume that the hydroxyl oxygen of ClEtOH hardly affects the polarization of the neighbouring water molecules. Thus, we should take the presence of the chlorine atom into account. It is well known that C1-, Br- and I- ions act as a water structure-breakers and their strength as such is in the order C1- < Br- < I-.'4 The surface charge density of these ions is in the order C1- > Br- > I-.As there is a good correlation between the surface charge density and their capability as structure-breakers, we expect an ion with a lower charge density to be a stronger structure-breaker. We consider that the chlorine atom of ClEtOH carries a weak negative charge because of its electron-withdrawing property, but its surface charge density is smaller than that of Cl-. Thus, the chlorine atom of ClEtOH should be a stronger structure-breaker than C1-. Accordingly, we consider that the chlorine atom is responsible for the disruption of the water structure by ClEtOH.Hydrophobic Interaction between ClEtOH and the Amides As shown by the 13C-NOE{OH} and measurements in fig. 5 and 6, dissociation of the amide complexes occurs at a ratio of [ClEtOH]/[amide] = 1. Since we cannot explain this in terms of the change in the hydrogen-bonding capability of water-ClEtOH mixtures described above, we should find another cause for this. The three amides used have rather large alkyl groups. On the other hand, ClEtOH is more hydrophobic than EtOH, but less hydrophobic than propan- 1-01.' ClEtOH may then be able to interact hydrophobically with the amides, while it behaves as a water structure-breaker. To discover whether the initial dissociation of the amide complexes can be correlated with the hydrophobic interaction, we examined whether a similar initial dissociation occurs in the complexes of N-methylacetamide (NMA), which seems to have too small an alkyl moiety to interact hydrophobically with CIEtOH.We set the concentration of NMA to 3 mol dm-3 and confirmed that inter-amide hydrogen- bonding association occurs to some extent. This was done by measuring the chemical shift of the NH proton and of the methyl protons and comparing them with the data of very dilute solutions of NMA. Then we measured the chemical shifts of the NH and OH protons and q with increasing ClEtOH (fig. 9). The high-field shift of both NH and OH observed indicates that the inter-amide hydrogen bonding is promoted by ClEtOH for the same reason as mentioned above. However, decreases monotonically, showing that the initial dissociation of the amide complexes does not occur in NMA.From these results, we can correlate the initial dissociation of the amide complexes with the hydrophobic interaction between ClEtOH and the amides. We consider that the interference of the interaction between the amide molecules from ClEtOH should lead to the apparent dissociation of the amide complexes. The fact that the initial dissociation occurs at the same ratio (ca. 1) irrespective of the amide concentration also supports the idea that the dissociation is correlated to molecular interaction between the amides and ClEtOH. Note that the interaction between the amide and ClEtOH leads to a decrease in the ClEtOH content in the bulk water-ClEtOH mixture and to the resultant energetic stabilization of the bulk mixture (fig.7). On the other hand, the increase in ClEtOH content in the neighbourhood of the amide brings about a decrease in the hydrogen-1110 F. T.N.M. R. Study of Self-association bonding capability of the ClEtOH-water mixtures around the amide, and consequently leads to the promotion of the inter-amide hydrogen-bonding association. Thus, we observe apparent dissociation of the amide complexes only at low concentrations of ClEtOH. Conclusion Our results can be summarized as follows: (1) ClEtOH promotes inter-amide hydrogen bonding in aqueous solutions, (2) ClEtOH behaves as a water structure-breaker and decreases the hydrogen-bonding capability of water, (3) the promotion of inter-amide hydrogen bonding can be attributed to a decrease in the hydrogen-bonding capability of the water-ClEtOH mixture around the amide molecules, (4) at very low ClEtOH concentrations, ClEtOH may interact hydrophobically with the large alkyl part of the amides, resulting in the apparent dissociation of the inter-amide hydrogen bond.The relationship between the conformational change in proteins and the structure of water-organic solvent systems has been investigated using alkanols, polyols, amides and urea.14*20-22 However, we have not as yet been able to draw clear conclusions concerning the relative importance of the different contributions, such as hydrogen bonding, ionic interactions and the hydrophobic interaction. 23 With the intention of presenting a mechanism of helix formation in proteins induced by ClEtOH, we have carried out our study focussing on the effects of ClEtOH on the inter-amide hydrogen bonding between the amide molecules.Compared with proteins, the three amides used in this study are very small molecularly. Our results, however, suggest that the change in the hydrogen-bonding capability of a solvent is an important factor in determining the conformational changes in proteins, and is closely related to the formation and dissociation of the inter-amide hydrogen bonds within proteins. References 1 S. Lapanje, Physicochemical Aspects of Protein Denaturation (Wiley-Interscience, New York, 1978). 2 S. N. Timascheff and H. Inoue, J. Am. Chem. SOC., 1968,90, 1890. 3 H. Inoue and S. N. Timascheff, Biopolymers, 1972, 11, 737. 4 E. E. Schrier, R. T. Ingwall and H. A. Scheraga, J. Phys. Chem., 1965, 69, 298. 5 T. T. Herzkovits, B. Gadegbeku and H. Jaillet, J. Biof. Chem., 1970, 245, 2588. 6 M. Morcellet and C. Loucheux, Biopolymers, 1980, 19, 2177. 7 C. L. De Ligny, N. J. Koole, H. D. Nelson and G. H. E. Nieuwdorp, J. Chromatogr., 1975, 114, 63. 8 K. Mizuno, H. Kaido, K. Kimura, K. Miyamoto, N. Yoneda, T. Kawabata, T. Tsurusaki, N. Hashizume and Y . Shindo, J. Chem. SOC., Faraday Trans. I , 1984, 80, 879. 9 Y. Shindo and K. Kimura, J. Chem. SOC., Faraday Trans. I , 1984, 80, 2199. 10 K. Hamaguchi and A. Kurono, J. Biochem. (Tokyo), 1963, 54, 497. 11 S. Ebina, M. Suzuki, I. Naitoh, K. Yamauchi and Y. Nagai, Int. J. Macromof., 1982, 4, 406. 12 P. N. Lewis, N. Go, M. Go, D. Kotelchuck and H. A. Scheraga, Proc. Natl Acad. Sci. USA, 1970,65, 13 K. Mizuno, S. Nishio and Y. Shindo, Biopolymers, 1979, 18, 693. 14 K. D. Collins and M. W. Washabaugh, Q. Rev. Biophys., 1985, 4, 323. 15 P. Gierycz, M. Denda and K. Nakanishi, Thermochim. Acta, 1985, 88, 241. 16 M. D. Zeilder, in Water, A Comprehensive Treatise, ed. F. Franks (Plenum Press, New York, 1973), 17 J. M. Harvey, S. E. Jackson and M. C. R. Symons, Chem. Phys. Lett., 1977, 47, 440. 18 A. Coccia, P. L. Indovina and V. Viti, Chem. Phys., 1975, 7, 30. 19 M. J. Kamlet and R. W. Taft, J. Am. Chem. Soc., 1976, 98, 377. 20 F. Franks and D. Eagland, CRC Crit. Rev. Biochem., 1975, 3, 175. 21 F. Franks, NATO ASI Ser., Ser. E, 1985, 90, 1. 22 M. P. Tomb, NATO ASI Ser., Ser. E, 1985, 90, 25. 23 J. L. Finney, in Interactions of Water in Ionic and Nonionic Hydrates, ed. H. Kleeberg (Springer Verlag, 810. vol. 2, p. 529. Berlin, 1987), p. 147. Paper 8/02342E; Receitled 13th June, 1988

ISSN:0300-9599    

DOI:10.1039/F19898501099

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J . Chern. SOC., Faraday Trans. I, 1989, 85(5), 1 1 11-1 116 Infrared Study of NH,/CO Reactions on Fe/A1,0, and Promoted Fe/SiO, Catalysts Colin Johnston, Norman Jorgensen and Colin H. Rochester* Chemistry Department, The University, Dundee DDl 4HN Infrared spectra are reported of Fe/AI,O,, Fe/SiO, and K,O, AI,O, and doubly promoted Fe/SiO, after treatment at 298-723 K in a gaseous mixture of carbon monoxide and ammonia. Product species included isocyanate groups on iron, surface cyano complexes on iron and isocyanate groups either on the oxide supports or involved in interaction with aluminium ions in promoted iron. The formation of FeNCO was enhanced by both promoters separately and was doubly enhanced when both promoters were present simultaneously. The formation of AlNCO(ads) for Fe/SiO, promoted by Al,O, provides evidence for the existence of aluminium ions in the promoted iron surface.Migration of NCO from sites on iron to the oxide support was more facile for alumina support than for silica. The non-dissociative and dissociative adsorption of carbon monoxide on silica- supported iron is enhanced by the presence of potash or alumina promoters, although the effects are not additive for doubly promoted catalyst.'*2 Infrared spectroscopy has shown that ammonia adsorbs non-dissociatively on unpromoted Fe/SiO,, and is also dissociatively adsorbed to form N adatoms which can react with carbon monoxide to form FeNCO surface species., Spillover of NCO to the silica support occurred at elevated temperatures and surface decomposition reactions led to the appearance of infrared bands ascribed to surface cyanide.The present study of ammonia/carbon monoxide reactions over Fe/Si02 catalysts containing potash and alumina was undertaken to try and gain information about the effects of promoters on the decomposition reaction of ammonia to N adatoms on the iron surface. Experimental Promoted iron catalysts were prepared from homogeneous pastes of aerosil silica (176 m2 g-') with aqueous iron(rr1) nitrate containing appropriate amounts of potassium nitrate and aluminium(rr1) nitrate. The pastes were dried at 348 K for 2 days, pressed into 25 mm diameter self-supporting discs (ca. 80 mg), mounted in an infrared cell and reduced in hydrogen as before2 with a final reduction temperature of 628 K. Fully reduced catalysts contained 5 wt YO iron.Three promoted catalysts were prepared containing 0.5% K,O, 2.5% A120, and 0.5% K20+2.5% A120,. These latter percentages are defined by K 2 0 (YO) = 100 (wt K20/wt Fe,O,) and A1,0, (Yo) = 100 (wt A120,/wt Fe,O,) which compares the K 2 0 and A120, promoter contents with the hypothetical Fe,O, content if all the iron was present as magnetite. Alumina (surface area ca. 200 m2 g-') supported iron containing 5 wt Yo iron was prepared in the same way as the Fe/Si02. Ammonia (> 99.98%) was treated with solid KOH and then Na before removal of permanent gases by freeze-thaw cycles. Carbon monoxide ( > 99.997 Yo) was passed through a trap at 77 K before use. 11111112 NH3/C0 on Iron data station. Spectra were recorded with a Perkin-Elmer 68 1 spectrometer linked to a 3600 infrared Results The addition of CO followed by ammonia or of ammonia followed by CO to K,O, A1,0, or doubly promoted Fe/SiO, gave similar results to those4 for unpromoted Fe/SiO, at ambient temperature. Maxima at 3380, 3290 and 1610 cm-' ascribed3 to vibrations of non-dissociatively adsorbed ammonia were more intense after exposure of catalysts to CO and ammonia than after exposure to ammonia alone.The intensity promotion effect was greatest (ca. 325 YO increase) for unpromoted Fe/SiO, and least (ca. 165 %) for Fe/SiO, containing K,O. If ammonia was added first followed by addition of CO with ammonia still present then no infrared bands due to adsorbed CO were observed. The reverse experiments in which ammonia was added to promoted catalysts already exposed to CO resulted in the disappearance of the sharp maximum at 2020-2040 cm-' (depending on coverage)'.2 due to linearly adsorbed CO. As before,' the intensities of the maxima due to linearly adsorbed CO were, in terms of promoter content, in the sequence K,O > K,O+A1,0, > Al,03 > none.The addition of ammonia enhanced the intensities of infrared bands in the range 1850-2000 cm-' due to bridge-bonded CO' for unpromoted4 and A1,0, and K,O + A1,03 promoted Fe/SiO,. In contrast, Fe/SiO, containing K,O alone did not show this effect, the residual spectrum in the 1850- 2000 cm-' region after ammonia addition agreeing closely with the band envelope' for CO adsorption alone. Fig. 1 compares the results of heating unpromoted and promoted Fe/SiO, in CO/ammonia mixtures.All the results reported were characteristic of the total systems and were not obtained if iron or either of the gaseous components were absent. The results for unpromoted Fe/SiO, (fig. 1 A) were consistent with previous, more detailed data which have been reported el~ewhere.~ The formation of FeNCO(ads), characterised by the strong band at 2220cm-', was enhanced by increasing the temperature from ambient to 473 K, but at 573 K decomposition of FeNCO(ads) to surface cyano complexes of iron (bands at 2055 and 2120 cm-') occurred. Spillover of isocyanate groups to the silica surface gave an infrared band at 2320 cm-l due to SiNCO(ads) which, after treatment at 623 K, was accompanied by a band at 2375 cm-l also attributed to SiNCO(ads) species.Several features of the infrared results for promoted catalysts were similar to the corresponding behaviour for Fe/SiO, alone. Treatment of all four catalysts at 473 K [fig. 1 (f)] maximised the existence of FeNCO(ads) on the iron surface as this temperature led to maximum intensities of the band at 2220cm-'. The subsequent decomposition of FeNCO(ads) at higher temperatures was accompanied by the concomitant appearance of the bands at 2055 and 2120cm-' due to cyano complexes. Bands at 2320 and 2375 cm-' ascribed to SiNCO(ads) were present in spectra of all four catalysts after treat- ment at the highest temperatures studied. The most significant difference between the spectra for the four catalysts involved the appearance of a band at 2270 cm-' for discs containing A1,0, promoter (fig.1 C and D). This is emphasised by comparison of spectra after heating the four catalysts in CO/ammonia at 473 K (fig. 2A) or 573 K (fig. 2B). The band at 2270 cm--' grew in intensity approximately in parallel with the growth in intensity of the stronger maximum at 2220 cm-l due to FeNCO(ads). At treatment temperatures above 523 K the new band diminished in intensity and became weak after treatment at 673 K [fig. 1 C(j)]. The reaction of alumina-supported iron with CO/amrnonia mixtures was investigated in order to confirm that the presence of A1,03 promoter in Fe/SiO, discs was responsible for the appearance of the maximum at 2270 cm-'. Carbon monoxide adsorbed on 5% Fe/Al,O,.gave a spectrum similar in general appearance to corresponding spectra for Fe/Si0,,'q2 with a sharp maximum atC.Johnston, N . Jorgensen and C. H. Rochester 1113 .r( 8 i z m b 2400 2000 2400 2000 2400 2000 ' I I D u 100 2000 v w avenumber/cm- ' Fig. 1. Spectra of 5 % Fe/SiO, exposed to ammonia (10 kN m-,) and CO (10 kN mV2) mixtures (a) before addition of gases, (b) addition of gases at 298 K, (c)-(k) after heat treatments for 15 min (consecutively for each series A, B, C, D and the specific temperatures for which spectra are shown for each series) at temperatures of (c) 323, ( d ) 373, (e) 423, ( f ) 473, (g) 523, (h) 573, (i) 623, (1) 673 and ( k ) 723 K. A, Unpromoted Fe/SiO,; B, K,O promoter; C, A1,0, promoter; D, K,O and A1,0, promoters. 2030 cm-' due to linearly adsorbed CO and weaker broader maxima at 1955 and 1900cm-' due to bridged CO on iron.Removal of CO by evacuation at ambient temperature caused a ca. 8 % reduction in the intensity of the linear CO band, which shifted to 2020 cm-', and small increases in the intensities of the other two bands, which shifted to 1945 and 1875 cm-l. Spectra of Fe/Al,O, in CO/ammonia mixtures contained no detectable bands due to adsorbed species until the temperature of the sample was raised to 323 K and above. By 423 K in particular, a band at 2220 cm-' characteristic of FeNCO(ads) had appeared together with two further bands at 2270 cm-' [fig. 2C(d)] and 1465 cm-' (not shown), which may be ascribed to vibrations of AlNCO(ads) species on the alumina upp port.^,^ The last two bands did not appear for alumina alone in the absence of iron, suggesting that migration of NCO groups from the iron surface to the alumina support' had occurred in Fe/A1,0, catalyst.The maxima at 2220 and 2270 cm-' continued to grow in intensity with increasing treatment temperature up to 523 K [fig. 2C(f)]. However, at 573 K FeNCO(ads) was apparently decomposed to give cyano complexes of iron responsible for infrared bands at 2035 and 2120 cm-' [fig. 2C(g)]. The band at 2220 cm-' became a shoulder on the more intense maximum at 2270 cm-'. The spectra of Fe/Al,O, also contained bands at 3380, 3290 and 1610 cm-', ascribed to NH,(ads) on the iron surface., Notable absentees from the spectra for Fe/A1,0, were the bands at 2320 and 2375 cm-' attributed to SiNCO(ads) on the silica support in Fe/SiO, catalysts.* Although the presence of promoters in Fe/SiO, did not apparently alter the temperature (ca.473 K) at which the generation of NCO species adsorbed on iron was optimised, the maximum attainable amounts of FeNCO(ads) were influenced by the 38 F A R I1114 NHJCO on Iron 2400 2000 wavenumber/an- I 2400 2003 Fig. 2. Spectra of (a) unpromoted, (b) K,O-promoted, (c) Al,O,-promoted, ( d ) K,O- and A1,0,- promoted catalysts after heat treatments in ammonia and CO at 473 K (A) and 573 K (B). Spectra C are of 5 % Fe/Al,O, exposed to ammonia (10 kN m-,) and CO (10 kN m-,) mixtures (a) before addition of gases, (b) in the presence of gases at 298 K, (c)-(g) after consecutive heat treatments (15 min) in the gaseous mixture at temperatures of ( c ) 323, ( d ) 423, (e) 473, (f) 523 and (g) 573 K.promoters. This is not obvious from the spectra in fig. 1 (and hence also fig. 2) because each set of spectra A, B, C and D were recorded in the percentage transmission mode under different conditions of attenuation. However, absorbance values at 2220 cm-' for catalysts after heat treatment at 473 K [fig. 1 cf)] were calculated using the infrared data station relative to the baseline spectra [fig. l(a)] each (A, B, C or D) under identical conditions of attenuation. The absorbance values were 0.26 for promoter-free Fe/SiO,, 0.50 for K,O/Fe/SiO,, 0.48 for Al,O,/Fe/SiO, and 0.85 for Fe/SiO, containing both K,O and A120,. Both promoters therefore approximately doubled the amount of FeNCO(ads) formed, double promotion giving a bigger enhancement than the combined separate effects of K,O and Al,O, alone.Fe/SiO, containing K 2 0 promoter gave a very weak shoulder or band at 2270 cm-l (fig. lB), which was not present in spectra of unpromoted Fe/SiO, after exposure to CO/ammonia at temperatures 2 473 K. Discussion The results of heat treatment of Fe/SiO, and Fe/A1,0, catalysts in CO/ammonia mixtures strongly support the contention5-' that the formation of isocyanate groups on transition-metal surfaces in oxide-supported catalysts is generally followed, particularly at elevated temperatures, by migration of isocyanate to the supporting oxide surface. The infrared band at 2270 cm-', ascribed to isocyanate groups on alumina, compares with a corresponding band at 2290 cm-l in spectra of Al(NCO), and bands in the rangeC.Johnston, N . Jorgensen and C. H. Rochester 1115 2259-2272 cm-' attributed to AlNCO surface species in a variety of alumina-supported transition metak7 The appearance of the band at 2270cm-' due to AlNCO(ads) for Fe/Al,O, heated in CO/ammonia at 323 K [fig. 2C(c)] contrasts with the lack of appearance of the band at 2320 cm-' due to SiNCO(ads) in Fe/SiO, catalyst until the treatment temperature was raised to 573 K [fig. 1 A(h)]. Migration of NCO from iron to alumina was much more facile than migration to silica. This effect is in accordance with the observation that migration of NCO from chromia to silica in Cr,O,,/SiO, catalyst was much slower than migration to alumina in Cr0,/A1,0,.7 Rasko and Solymosi7 concluded that the migration process is greatly influenced by the surface character of the supporting oxide.For Fe/SiO, catalysts containing alumina promoter the infrared bands at 2270 cm-' due to AlNCO(ads) and at 2320cm-' due to SiNCO(ads) appeared after treatment temperatures of 298 and 573 K, respectively, again confirming the more facile formation of the AlNCO species. The location of the alumina promoter in iron catalysts has aroused considerable attention. Possibilities include the occlusion of FeAI,04,8 Fe2A1049 or A120310 into the a-iron lattice or the existence of aluminium on the iron surface." Ertl et a/.'' used scanning Auger electron spectroscopy to show that an appreciable A1 concentration exists in the surface region of reduced iron particles. The appearance of the infrared band at 2270 cm-' for Fe/SiO, catalyst containing AI,O, promoter can be similarly ascribed to the existence on the iron surface of aluminium ions which are available as sites for the adsorption of NCO to form AlNCO(ads).There is much evidence that the iron surface in K,O-promoted catalysts is partly covered by potassium" and therefore the weak band at 2270 cm-I in spectra of Fe/SiO, containing K,O (fig. I B) is tentatively ascribed to surface KNCO species. The addition of K,O promoter to Fe/SiO, enhances the dissociative adsorption of dinitrogen to give N adatoms which exist at sites otherwise available for the linear or bridged adsorption of CO and which may react with CO to form FeNCO(ads) species.' The inhibiting effect of N adatoms on CO adsorption was most pronounced for linearly adsorbed CO.Similarly ammonia undergoes dissociative adsorption to give N adatoms at surface sites which are also active for the linear adsorption of CO in the presence of CO a10ne.~ The promoting effect of K,O on the subsequent reaction with CO to form FeNCO(ads) correlates with the enhancement in the linear adsorption of CO promoted by K,0.1*2 A similar, but less satisfactory, correlation between NCO formation and bridge-bonded adsorption of CO exists for CO(ads), giving an infrared band at 1970 cm-' but not for bridge-bonded CO giving a band at 1880 cm-'.'V2 Alumina promoter was not as efficient as K,O in enhancing the formation of infrared- detectable complexes of CO on iron.'T2 However, FeNCO(ads) formation from CO/ammonia over Fe/SiO, was promoted to nearly the same extents at 473 K by 0.5 O/O K,O and 2.5% Al,O,.Volumetrically determined total uptakes of CO by iron were nearly as great for alumina-promoted Fe/SiO, as for K,O-promoted catalyst., The overall results imply that there is no single identifiable type of site for CO adsorption which also constitutes the only type of site for the dissociative adsorption of ammonia and subsequent formation of isocyanate in the presence of CO. This conclusion is strengthened by the data for doubly promoted catalyst. The effects of double promotion on CO adsorption were clearly not additive (K,O alone generally having a greater effect that K,O + A1,0,),1,2 whereas FeNCO(ads) formation was enhanced to a greater extent by K,O + AI,O, than the sum of the separate effects of K,O and AI,O, alone. We thank the S.E.R.C. for two CASE studentships, ICI Agricultural Division for collaboration and financial support and Drs J. R. Jennings and S. A. Topham for helpful discussions. 38-21116 NHJCO on Iron References 1 N. Jorgensen and C. H. Rochester, Appl. Catal., 1986, 25, 69. 2 C. Johnston, N. Jorgensen and C. H. Rochester, J. Chem. SOC., Faraday Trans. I , 1988, 84, 309. 3 C. Johnston, N. Jorgensen and C. H. Rochester, J. Chem. SOC., Faraday Trans. 1, 1988, 84, 2001. 4 C. Johnston, N. Jorgensen and C. H. Rochester, J. Chem. SOC., Faraday Trans. 1, 1988, 84, 3605. 5 D. Gutschick and H. Miessner, React. Kinet. Catal. Lett., 1983, 22, 221. 6 F. Solymosi and J. Sarkany, J . Appl. Surf. Sci., 1979, 3, 68. 7 J. Rasko and F. Solymosi, J. Chem. SOC., Faraday Trans. I , 1980, 76, 2383. 3 N. Pernicone, G. Fagherazzi, F. Galante, F. Garbassi, F. Lazzerin and A. Mattera, Proc. 5th Int. 9 H. Ludwiczek, A. Preisinger, A. Fischer, R. Hosemann, A. Schonfield and W. Vogel, J. Catal., 1978,51, Congr. Catal., (North Holland, Amsterdam, 1973), vol. 2, p. 1241. 326. 10 H. Topsse, J. A. Dumesic and M. Boudart, J . Catal., 1973, 28, 477. I I G. Ertl, D. Prigge, R. Schloegl and M. Weiss, J . Catal., 1983, 79, 359. Paper 8/02472C; Received 21st June, 1988

ISSN:0300-9599    

DOI:10.1039/F19898501111

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J . Chern. Soc., Faraduy Trans. I , 1989, 85(5), 1 1 17-1 128 Infrared Study of the Adsorption of Acetone, Acrolein, Ethanoic Acid and Propene-NO Mixtures on Rh/Al,O, Catalysts James A. Anderson and Colin H. Rochester* Chemistry Department, The University. Dundee DDI 4HN Infrared spectra are reported of adsorbed products from exposure of Rh/Al,O, at ca. 300 and 473 K to propene-NO mixtures, ethanoic acid, acetone and acrolein. Ethanoic acid was dissociatively adsorbed, giving ethanoate ions on alumina and a rhodium carbonyl species. The latter species was also formed from acrolein which was adsorbed on alumina in four ways: hydrogen bonding with hydroxyl groups, ligating to A13+ sites and oxidation to acrylate or 2-formylethanoate anions. Acetone was associatively adsorbed via hydrogen bonding or coordinative interactions and was oxidised to ethanoate or methanoate anions on alumina.Carbonyl complexes were formed on rhodium. The oxidation of propene by NO over rhodium in Rh/Al,O, catalysts specifically gave acrylate anions as the dominant adsorbed reaction product. Reactions between hydrocarbons and oxides of nitrogen (NO,) over heterogeneous catalysts are of interest in relation to NO, conversion to nitrogen and hydrocarbon conversion to water and carbon dioxide in exhaust gases from internal combustion engines.' The use of rhodium as a catalyst for the reduction of NO, by carbon monoxide or hydrogen is well documented, but few studies have been reported of the catalysed reduction of NO, by hydrocarbons and the concomitant hydrocarbon oxidation reactions.' This paper reports an infrared study of the reaction of nitric oxide and propene over a Rh/Al,O, catalyst and of the adsorption of acetone, acrolein and ethanoic acid.The oxidation of propene by oxygen over supported rhodium catalysts leads not only to carbon dioxide and water but also to the partial oxidation products ethanol, ethanoic acid, acrolein and acetone. 2* Methanoate, ethanoate and acrylate anions adsorbed on oxide surfaces are reported to be formed by the oxidation of ~ r o p e n e . ~ - ~ The important role of surface ally1 species as intermediates in the oxidation reactions has been identified.3-5 Experimental Catalyst precursor was prepared by evaporating to dryness slurries of aqueous rhodium nitrate (5.44 w/w %, Johnson Matthey) and Alon fumed alumina (Cabot Corporation) with surface area 102 m2 g-'.The rhodium content was 1.0 wt YO after complete reduction to Rho. Self-supporting discs (80 mg, 25 mm diameter) of alumina impregnated with rhodium nitrate were pressed at 50 MN m-2 (ca. 2 ton on die) and mounted in an infrared cell with fluorite windows, an external furnace and glassblown to a conventional high vacuum (ca. Catalyst pretreatment involved heating in oxygen (27 kN m-,, 673 K, 30 min) followed by reduction (773 K, 1 h) in a flowing (200 cm3 min-') hydrogen-argon mixture (101 kN m-2) containing 3.5 v/v % hydrogen. The cell was evacuated (15 min) with the disc at 773 K before cooling of the catalyst to ambient temperature in vacuum. N m-,) apparatus. 11171118 Infrared Study of Rh/Al,O, 2500 ZOO0 1800 1600 1400 1200 w avenumber /an- ' Fig.1. Infrared spectra of Rh/Al,O,: (a) background spectrum, (0) exposed to propene, (c) partial removal of propene and addition of NO, (dHf> exposed to a propene (4.3 kN m-2)-N0 (3.3 kN m-,) mixture (d) at ca. 300 K, (e) after 40 min at 473 K, (f) 16 h at 473 K and evacuation at 473 K. Nitric oxide was purified by repeated fractional distillation7 and the other adsorbates were high-purity commercial samples. Spectra were recorded using a Perkin-Elmer 683 infrared spectrometer and a 3600 infrared data station with catalyst discs at the ambient temperature (ca. 300 K) in the spectrometer beam. Results Spectra of Rh/A1,0, in the presence of propene vapour contained bands which were similar to those for propene molecules in the gas phase [fig.l(b)]. Partial removal of propene using a cold trap (77 K) followed by admission of nitric oxide [fig. 1 (c)] and removal of the cold trap to allow both gases to mix at ambient temperature over the catalyst [fig. 1 (d)] led to the appearance of additional bands at 1915, 1823 and 1740(sh) cm-' due to adsorbed NO,' an enhanced maximum at 1640 cm-' assigned to the C=C stretching vibrations of an adsorbed reaction product, and a new band at 1695 cm-l. The latter became more intense and shifted to 1690 cm-' when Rh/A1,0, was heated in propene-nitric oxide mixtures for moderate lengths of time [fig. 1 (e)]. Decreases in intensity of the bands at 1823 and 1740(sh) cm-l assigned to gem-dinitrosyl species8 were accompanied by an increase in intensity of the band at 191 5 cm-' due to RhNO+.NewJ . A . Anderson and C. H . Rochester 1119 2500 ZOO0 1800 1600 1400 1200 2 9 0 2000 1800 1600 1400 121 wavenumber/cm-' Fig. 2. Infrared spectra of Rh/AI,O,: (a) background, (b) after heating (473 K, 17 h) in propene 5.9 kN m-'), (c) propene condensed out, (d)+h) after ( d ) addition of NO (0.9 kN m-2) at 300 K, (e) heating in propene-NO (473 K, 2; h), (.I) evacuation (473 K , 1 h), ( g ) evacuation (673 K, 2 h), ( h ) addition of oxygen (16 kN m-?, 300 K). bands also appeared at 1575 and ca. 1480 cm-' and became the dominant features of the spectrum at 1575 and 1464 cm-' after prolonged heat treatment at 473 K [fig. I ( f ) ] . A small band at 2250 cm-' and a shoulder at 2185 cm-' may be ascribed to surface isocyanate adsorbed on the alumina support and on rhodium, respectively.' Bands at 3794,3727 and 3677 cm-' due to hydroxyl groups on the alumina surface were obscured by a broad more intense band envelope centred at 3582 cm-' after reaction between NO and propene over Rh/A1,0,.A sharp band at 1648 cm-' persisted in spectra even after evacuation at 473 K. All the observed infrared bands after prolonged heating of Rh/Al,O, in NO and propene could be attributed to adsorbed species since the spectrum was unaffected by evacuation of the sample at temperatures up to 473 K [fig. 1 ( f ) ] . The reactions between NO and propene were only catalysed by Rh/AI,O, and not by alumina alone. Spectra of Rh/A1,0, after heat treatment in propene alone at 473 K exhibited infrared bands similar to those for propene vapour and an additional maximum at 1575 cm-' due to an adsorption product [fig.2(h)]. Removal of propene vapour by cold-trapping (77 K) also showed that adsorbed species gave bands at 1653, 1470, 1450 and I38 I cm-I [fig. 2(c)]. Subsequent addition of NO gave weak bands at 1823 and 1740 cm-' due to a gem-dinitrosyl complex of rhodium [fig. 2 (41. Re-admission of propene followed by heat treatment of Rh/AI,O, in the gaseous mixture at 473 K gave two intense maxima at 1575 and 1464 cm-l, a shoulder at 1660 cm-' and weaker bands at 1275, I3 1 I , I38 I , 1395, 1915 (due to RhNO+), 2250 (AINCO) and 2185(sh) cm-' [fig. 2(u)]. Evacuation at 473 K had no effect on the dominant maxima at 1575 and 1464 cm-' or on the other weaker bands except the shoulder at 1660 cm-' which disappeared [fig.2(,f)]. The shoulder was probably largely due to bands in the spectrum of propene vapour [fig. 2(b)]. Heat treatment at 473 K enhanced the formation of isocyanate groups giving a band at 2250 cm-', but increasing the temperature to 673 K removed isocyanate groups and led to the appearance of a new band at 2064 cm-' attributed to linearly adsorbed1120 Infrared Study of Rh/Al,O, I I I I I I I I 1 \ 2500 2000 1800 1600 1400 1200 wavenumtxr /cm- Fig. 3. Spectra of Rh/AI,O,: (a) background, (b) after adsorption of ethanoic acid at cia. 300 K, (c), ( d ) subsequent evacuation at ( c ) 473 K ( 1 h), ( d ) 673 K (30 min). CO on rhodium [fig. 2(g)].l0," The strong bands at 1578 and 1464 cm-' were greatly weakened in intensity and all the weaker bands disappeared.The effect of oxygen on the adsorbed CO was tested. The band at 2064cm-l was replaced by a maximum at 2108 cm-' possibly due to CO ligated to a rhodium atom also ligated to an oxygen atom.", 13 [fig. 2(h)]. A weak band due to surface isocyanate reappeared at 2250 cm-'. The adsorption of ethanoic acid on Rh/A1,0, at ambient temperature generated intense infrared bands at 1584 and 1475 cm-', with a shoulder at 1425 cm-l and a weak maximum at 1336 cm-' [fig. 3(b)]. Closely similar spectra were recorded after the adsorption of ethanoic acid on alumina alone which had been subjected to identical pretreatment procedures to those used for the preparation of reduced rhodium on alumina.The infrared bands may be ascribed to adsorbed ethanoate anions formed by the dissociative adsorption of ethanoic acid on alumina. 14-18 Subsequent evacuation with the sample at 473 K had little effect on the infrared bands due to adsorbed ethanoate but in the case of Rh/Al,O, led to the apearance of a band at 2019 cm-l [fig. 3(c)] which may be assigned to linearly adsorbed CO molecules at rhodium sites. The band did not appear in the corresponding spectrum for alumina in the absence of rhodium. Evacuation at 673 K removed a high proportion of the adsorbed ethanoate from the alumina surface and also caused the disappearance of the infrared band at 2019 cm-' [fig. 3(d)]. The maxima due to residual ethanoate were at 1564 and 1481 cm-'. Adsorption of ethanoic acid on alumina or Rh/AI,O, led to the replacement of the bands at 3794,3727 and 3677 cm-l due to hydroxyl groups on alumina by a broader and more intense band envelope with a maximum at 3510 cm-' and a shoulder at ca.3000 cm-'. Very weak bands also appeared due to CH stretching vibrations of adsorbedJ . A. Anderson and C. H . Rochester 1121 Lo00 I 3500 3000 I 2300 2000 1w wavenumtxr/cm- Fig. 4. Spectra of Rh/Al,O,: (a) background, (b) after adsorption of acrolein at ca. 300 K, (c) after 20 min contact time, (d), (e) subsequent evacuation at ca. 300 K and heating in a static vacuum at (d) 373 K (30 min), (e) 473 K (90 min). species. The spectral changes involving hydroxyl groups induced by adsorption of ethanoic acid were completely reversed by desorption in vacuum at 673 K.Spectra of Rh/Al,O, exposed to acrolein at ambient temperature exhibited bands at 3098(vw), 2931, 2887(sh), 2720, 1728(sh), 1678, 1648, 1567, 1456, 1378 and 1275 cm-' [fig. 4 (b)] which may be assignated primarily to vibrations of associatively adsorbed acrolein molec~les'~ or of surface acrylate*O formed by oxidation of acrolein. Surface hydroxyl groups on the alumina support were perturbed or displaced and a broad band centred at ca. 3571 cm-' appeared in the spectrum. Two bands at 2092(vw) and 2019 cm-' are attributed to rhodium carbonyl species formed by the decomposition of acrolein at rhodium sites. A time-dependent effect was that a band at 1728 cm-' grew in intensity apparently at the expense of the band at 1678 cm-' [fig. 4(c)]. No change in the spectrum was observed when the cell was evacuated at ca.300 K, but heating the disc in a vacuum at 373 K led to the prominence of a maximum at 1698 cm-' and enhanced growth of the bands at 1567 and 1456 cm-' due to acrylate [fig. 4(d)]. The formation of acrylate was further enhanced by heat treatment at 473 K [fig. 4(e)], which removed most other adsorbed species from the sample. An additional residual band remained at 1728 cm-' and a new band appeared at 131 1 cm-'. The band at 2019 cm-' diminished in intensity and broadened towards lower wavenumbers with a maximum at 1993 cm-'. Spectrum 4(e) was recorded after heating the Rh/Al,O, disc in a static vacuum and cooling to ca. 300 K in the static vacuum. Desorbed species might therefore have been re-adsorbed during the cooling process.A Rh/A1,0, disc, which after the adsorption of acrolein was subjected to heat treatments at 373 K and 473 K with continuous pumping before being cooled to ca. 300 K in the dynamic vacuum for spectroscopic observation, gave a spectrum similar to fig. 4(e) except for the absence of a band at 1993 cm-'. Spectra of acrolein adsorbed on alumina which had been subjected to the standard1122 Infrared Study of Rh/Al,O, 4000 3300 2600 1800 1600 1400 1200 wavenumber/crn-' Fig. 5. Spectra of alumina: (a) background, (b) after addition of acrolein at ca. 300 K, (c) 20 min contact time, (d)-(g) subsequent evacuation at (d) ca. 300 K (6 min), (e) 373 K (30 min), cf) 473 K (30 min), (g) 673 K (30 min). (for Rh/A1,0,) reduction procedure in hydrogen were similar [fig. 5(b), (c)] to the corresponding spectra for acrolein on Rh/Al,O, except for the absence of the bands at 2092 and 2019 cm-' due to rhodium carbonyl species. Subsequent heat treatments also led to similar results [fig.5(e) (f)] to those for Rh/A1,0,, showing that the dominant spectroscopic effects for Rh/Al,O,, apart from those in the spectral range 18OG 2300 cm-', could be ascribed to surface species on the alumina support. After evacuation at 673 K the two most intense bands remained in spectra, although some broadening occurred, in particular for the maximum at 1456 cm-' which split into two maxima at 1478 and 1450 cm-' [fig. 5(g)] possibly due to two distinguishable forms of surface carboxylate complex. Spectra of Rh/Al,O, in contact with acetone vapour (0.67 kN m-,) contained a broad band envelope, with a maximum at 3582 cm-', due to OH species on the alumina surface [fig.6(6)]. Bands at 2965, 2931 and 2876 due to CH stretching vibrations compare with corresponding bands at 2972, 2937 and 2867 cm-' in spectra of acetone vapour.,l The bands remained in the spectra after removal of acetone vapour [fig. 6(d)] and were therefore due to adsorbed species. In contrast a band at 1739 cm-' due to unperturbed carbonyl groups disappeared when the cell was evacuated. A maximum at 1703 cm-' in spectra of Rh/Al,O, in the presence of acetone is ascribed to the v,-.=~ vibrations of weakly adsorbed acetone molecules involved in hydrogen-bonding interactions with surface hydroxyl groups on a l ~ m i n a . ' ~ .~ ~ Weakly bonded acetone was desorbed by evacuation at ca. 300 K. More strongly adsorbed species giving a band at 1628 cm-' were resistant to desorption. This band compares with a band at 163&1640 cm-' previously ascribed to the v,=, vibrations of acetone molecules ligated to Lewis-acidic A13+ sites in the alumina surface.22 Danyushevskii et aL2, also ascribed a band at 1580 cm-' to ligated acetone. A similar band at 1584 cm-' existed in the present spectra [fig. 6(d)] but must at least in part have resulted from decomposition of acetone to give surface ethanoate anions. Formation of ethanoate was time-dependent [fig. 6 (6)-(d)]J. A . Anderson and C. H. Rochester ~ 4000 3300 26c I I I 2000 1800 1600 1400 1 I123 waven umber/m- Fig. 6. Spectra of Rh/AI,O,: (a) background, (6) after addition of acetone at c'u.300 K, ( c ) 20 min contact time, ( d ) - ( f ) subsequent evacuation at ( d ) c'u. 300 K (6 rnin), (e) 473 K (30 rnin), (f') 673 K (30 min). and enhanced by evacuation at 473 K, after which spectra contained prominent maxima at IS84 and 1467 cm-l characteristic of adsorbed ethanoate [fig. 6 ( e ) ] . Coordinatively ligated acetone was unaffected by evacuation at 373 K , which led to a spectrum resembling that in fig. 6 ( d ) , but after evacuation at 473 K had either been desorbed or had reacted to give ethanoate anions [fig. 6(e)]. Partial desorption of ethanoate occurred at 673 K [fig. 6 ( f ) ] which also resulted in loss of the broad band envelope due to OH groups and restoration of the bands typical of hydroxyl groups on an alumina surface in the absence of adsorbates.In one sequence of experiments acetone was adsorbed on Rh/AI,O,, and the system was evacuated with the disc at cu. 300 K to give a spectrum similar to that in fig. 6 ( d ) . Subsequent heat treatment at 373 K in a closed system (as opposed to a dynamic vacuum with continual pumping) had no effect on the spectrum. However, heating at 473 K in a closed system gave a spectrum with one significant difference from that in fig. 6(e). Acetone which was desorbed during heating was re-adsorbed as the disc was cooled to ca. 300 K for spectroscopic examination and a maximum appeared in the spectrum at ca. 1700 cm-'. Studies of acetone adsorption on alumina alone were carried out with exactly the same procedures as those adopted for Rh/AI,O,.All the spectroscopic effects reported above for Rh/Al,O, were also observed for support alone, showing that the detectable adsorption and reaction behaviour was occurring on the alumina rather than the rhodium surface. However, two additional spectroscopic effects were observed for Rh/AI,O, which did not occur for alumina alone. First, two weak bands at 1970 and 1810cm-' [fig. 6 ( d ) ] may be attributed to adsorbed species resulting from the decomposition of acetone on rhodium and retention of product species on either rhodium or the support. The band at 1810 cm-l was at its strongest when Rh/Al,O, treated with acetone was evacuated at cu. 300 K, heated in a static vacuum at 473 K and cooled in the static vacuum to ca.300 K. The band at 1970 cm-' was also observed after1124 Infrared Study of Rh/A1,0, this sequence of treatments but heating in a dynamic vacuum at 473 K removed species responsible for both the infrared bands from the surface [fig. 6(e)]. Secondly, residual hydrocarbon species after evacuation at 673 K gave an infrared band at 3054 cm-' [fig. 6(f)], suggesting the presence of unsaturated hydrocarbon on rhodium and a band at 2920 cm-' became more intense than the residual band at 2965 cm-' due to a vibration of methyl groups. Discussion In accordance with the present results previous infrared of ethanoic acid on alumina established that dissociative adsorption occurred, giving surface ethanoate species. With some caution23 the difference Avco between the positions of the infrared bands due to the vassm (CO,) and vsym (CO,) vibrations of ethanoate complexes may be used to gain information about the chelating configuration of the ethanoate anions.24 The values of Avco for ethanoic acid on alumina were 109 cm-' (maxima at 1584 and 1475 cm-') for high surface coverages and 83 cm-' (1564, 1481 cm-') after partial desorption by heat treatment at 673 K.These values apparently rule out the possibility of a unidentate configuration and would be more consistent with a bidentate chelating configuration of each ethanoate anion to a single surface A13+ site than with each ethanoate anion being bridged between pairs of adjacent exposed A13' s i t e ~ . ~ ~ , , ~ An inelastic electron tunnelling spectroscopic study of ethanoic acid on alumina gave Avco = 125 cm-' for surface ethanoate, which was interpreted in terms of a bridging rather than a chelating c~nfiguration.,~ One possibility in the present study involving alumina which had been reduced in hydrogen before admission of ethanoic acid was that at high surface coverages bands due to both bridging and chelating ethanoate anions at different sites contributed to the overall vCo2(as) and vco,(s) band envelopes ; whereas after subsequent heat treatment only chelating species remained, giving a lower value for Avco. The appearance of a band at 2019 cm-' due to a carbonyl complex of rhodium after heat treatment of adsorbed ethanoic acid on Rh/Al,O, resembles the appearance of bands at 2030-2050 cm-' after heat treatment of supported rhodium catalysts in the presence of methanoic acid.26 Similar bands at 2020-2035 cm-' have also been observed after the reaction of hydrogen and carbon dioxide over Rh/Al,O, and have been attributed to a rhodium carbonyl hydride surface species in which the carbonyl group responsible for the infrared band is ligated to an exposed rhodium atom simultaneously bonded to a hydrogen a d a t ~ m .~ ' ~ ~ ~ There is, however, controversy about this interpretation of the ~ p e c t r a ~ ~ , ~ ~ since carbon dioxide alone on Rh/Al,O, has been reported to adsorb dissociatively, giving surface carbonyl species responsible for an infrared band observed at 2020,' or 2015 cm-'.8 The rhodium carbonyl hydride explanation would only be tenable in relation to CO, adsorption alone if the catalyst had retained sufficient hydrogen on either rhodium or alumina after reduction in hydrogen to be able to generate the complex from the retained hydrogen and CO formed by dissociation of CO,.Solymosi et af.26*32933 proposed that the adsorption of methanoic acid on Rh/A1,0, involved the formation of adsorbed methanoate on the support, but that methanoate was unstable on rhodium. The decomposition of methanoate, primarily to CO, and H,, was thought to occur at rhodium sites and possibly, at least in part, involved migration of methanoate anions from the support to rhodium where reaction occurred. The present results would be consistent with similar behaviour for the decomposition of ethanoic acid over Rh/Al,O,. The resulting infrared band at 2019 cm-' derives from the formation of CO,(ads) on rhodium,26 which may subsequently dissociate to give CO(ads). If CO(ads) and H(ads) on a single Rh atom are required to produce the species giving the 2019 cm-' band in accordance with the proposals of Solymosi et af.26-28330 thenJ .A . Anderson and C. H. Rochester 1125 H(ads) must also be an intermediate26*28 in the decomposition of ethanoic acid on supported rhodium. The present bands at 1703, 1628 and 1584 cm-' in spectra of acetone on alumina (fig. 6) resemble similar bands observed by Kiselev and U ~ a r o v ' ~ at 1692 cm-' (attributed to acetone involved in hydrogen-bonding interactions with surface hydroxyl groups), 1625 and 1600 cm-l (due to acetone on two types of exposed aprotic site). Acetone decomposition at 523 K was previously recordedI4 through the appearance of infrared bands due to ethanoate species at 1575 and 1465 cm-' which compare with present maxima at 1584 and 1467 cm-' [fig. 6(e)].Acetone decomposition to ethanoate anions on an oxide surface might be expected also to lead to adsorbed methanoate anions.34 Infrared bands have been reported at 1585, 1395 and 1373 cm-' for methanoate ions formed by the adsorption of methanoic acid on Rh/A1,0,.26 Bands at 1386 and 1372 cm-' in the present spectra of acetone on Rh/A1,0, may be ascribed to the formation of methanoate. A band at 1585 cm-' due to methanoate could not be separately identified because of overlapping or coincident bands due to ethanoate anions or coordinately adsorbed acetone. The adsorption of acetone on oxide surfaces is sometimes accompanied by enolisation and dimerisation to mesityl For example, bands at 1545 and 1430 cm-' for acetone on rutile have been ascribed to enolic species and bands at 1655 and 1595 cm-' were attributed to mesityl Whether similar bands contributed to the presently observed spectra of acetone on Rh/Al,O, (fig. 6) or alumina could not be recognised in view of the existence of several other maxima in the same spectral region.However, comparison of the spectra of acetone and ethanoic acid on Rh/Al,O, and A1,0, suggested that there was an additional band at ca. 1530 cm-' which contributed to the spectra of adsorbed acetone but not of adsorbed ethanoic acid. The band was a separate maximum after adsorption of acetone on alumina and evacuation at 473 K and may be indicative of the formation of enol or a related adsorbed product.The presence of rhodium favoured enhanced decomposition of acetone to unsaturated hydrocarbon species, giving the band at 3054 cm-' [fig. 6 0 1 , a rhodium carbonyl species giving a band at 1970 cm-' [fig. 6(d)] probably indicative of a bridged onf figuration,''*^^ and an adsorbed species giving a very weak band at 18 10 cm-' [fig. 6(d)]. The latter is rather low to be due to a rhodium carbonyl complex, although bands due to CO on rhodium have been recorded down to 1820 cm-' and in one case at 1780 cm-'." One possibility would be that an organic carbonyl complex of rhodium was formed by partial decomposition of acetone to an adsorbed product retaining both the CO group and an hydrocarbon component.An alternative would be that decomposition at sites on rhodium was followed by spillover on to the alumina support giving organic- bridged carbonate species on alumina. Bands at around 1800 cm-' in spectra of carbon dioxide on alumina have been attributed to bridged ' organic-like' carbonate 39 An infrared study of acrolein adsorption on tin oxide established that non-dissociative coordination to Sn4+ surface sites was accompanied by chemisorptive interactions giving surface a~rylate.~' Acrolein was not coordinatively bound to the surface of magnesium oxide, but chemisorption led to adsorbed acrylate and 2-formylethanoate anions, the latter being oxidised to malonate ions at high temperat~re.~' Infrared bands at 3098 (CH stretch) and 1648 cm-' (C=C stretch) in spectra of acrolein on alumina (fig.5) or Rh/Al,O, (fig. 4) indicate the presence of adsorbed species containing CH2=CH- groups.41 The bands at 1698 and 1678 cm-' resemble similar bands at 1710 and 1675-1680 cm-' for acrolein on tin which were assigned to vco vibrations of coordinatively bound acrolein molecules at Sn4+ surface sites. The present band at 1678 cm-' is similarly ascribed to acrolein molecules ligated to A13+ sites. However, a maximum at 1693 cm-' in spectra of acrolein adsorbed on Rh/SiO, catalyst4, closely resembles the present maximum at 1698 cm-l. The two maxima are assigned to the vco1126 Infrared Study of Rh/A1,0, vibrations of acrolein molecules involved in hydrogen- bonding interactions with surface hydroxyl groups on the silica or alumina supports, respectively. Maxima at 1567, 1456, 1378 and 1275 cm-' [fig.4(b)] closely agree with bands at 1540-1570 [ v ~ ~ ~ ( ~ ~ J 1450 [ V ~ ~ , ( ~ J , 1368 [CH bend] and 1285 cm-' (vC-J in spectra of sodium acrylate,lg confirming that chemisorption of acrolein on alumina led to adsorbed acrylate anions. The maxima at 1567 and 1456 cm-' must also have contained contributions from bands due to the v , - . ~ , ( ~ ~ ) and vco2(s) vibrations of 2-formylethanoate anions on the alumina surface. Strong evidence for this conclusion is provided by the maxima at 2931 and 1728 cm-' [fig. 4(b)] which correspond to bands at 2928 (vcH) and 1725 cm-l (vC=J ascribed by Niwa et al.41 to vibrations of CH,CHO groups in 2-formylethanoate ions formed by the adsorption of acrolein on magnesia.The growth of the bands at 293 1 and 1728 cm-' with time for alumina [fig. 5(c)] or Rh/Al,O, [fig. 4(c)] at ca. 300 K suggests that the generation of 2-formylethanoate by reaction of acrolein with oxide ions in the alumina surface is an activated process. The 2-formylethanoate ions apparently existed at A13+ sites which after very short contact times acted as Lewis-acidic centres, forming coordinative interactions with associatively adsorbed acrolein molecules. The latter gave the infrared band at 1678 cm-l which was depleted in intensity as conversion to 2- formylethanoate occurred. The band at 2019cm-l in spectra of Rh/Al,O, exposed to acrolein at ca. 300 K (fig. 4) was identical to the band at 2019 cm-' for Rh/Al,O, which had been treated with ethanoic acid at ca.300 K and then subjected to evacuation at 473 K [fig. 3 (c)]. The same band appears in spectra of acrolein adsorbed on Rh/SiO, at ca. 300 K,42 suggesting that the decomposition of acrolein to form a rhodium carbonyl complex occurred on the rhodium surface and not on the alumina followed by spillover of decomposition products on to rhodium. The appearance of a weak infrared band at 2092 cm-' provided evidence for the formation of a second carbonyl complex of rhodium, possibly either a gem-dicarbonyl or a linear carbonyl at rhodium atoms in the + 1 oxidation state.10'27,37 The formation of carbonyl complexes from acrolein on rhodium might also be expected to produce a further adsorbed product derived from the alkene fragment of the acrolein molecule.Solymosi and Lancz3, have shown that adsorbed ethylene and CO can coexist on the rhodium surface in Rh/A1,0, catalysts. The gem-dicarbonyl band at 2100 cm-' for CO on rhodium in the absence of ethylene remained at 2100 cm-', albeit at lower intensity, when ethylene had been preadsorbed before the addition of CO. In contrast the band due to linearly bonded CO shifted from 2068 cm-' for CO/Rh alone to 2026 cm-' in the presence of ethylene. The second weak band expected at ca. 2035 cm-l for a Rh(CO), complex" would have been obscured in the present spectra by the more intense band at 2019 cm-' due to linearly adsorbed CO which was perturbed by co- existence with hydrogen adatoms2'* 28 or adsorbed hydrocarbon species,,, also resulting from acrolein decomposition.Infrared bands due to rhodium carbonyl surface complexes have been previously observed for Rh/Al,O, catalysts heated with methanol vapour at 160 K or 230 K.43 Comparison of fig. 1 (f) and 4(e) shows that the main product of oxidation of propene by nitric oxide over Rh/A1,0, was acrylate anions adsorbed on the alumina surface. The formation of acrylate occurred at 473 K, but not at ambient temperature. Weak infrared bands characteristic of acrylate were even observed [fig. 2 (c)] after Rh/A1,0, was heated at 473 K in propene alone, showing that the initial catalyst surface was capable of effecting a low level of oxidation in the absence of nitric oxide. Acrylate formation was considerably enhanced by the addition of nitric oxide oxidant at 473 K [fig.2(e)]. Although the adsorbed acrylate existed on the alumina surface, the catalytic reactions leading to acrylate must have occurred on rhodium sites since no reactions took place at 473 K for alumina alone. The formation of acrylate in the present systems is consistent with a report that acrolein was the dominant partial oxidation product fromJ. A. Anderson and C. H. Rochester 1127 the reaction of propene and oxygen over Rh/Al,O, catalysts., Propene has also been observed to adsorb on tin(rv) oxide-palladium oxide in part as a ~ r y l a t e . ~ Ethanoic acid was also a product of propene-oxygen reaction over Rh/AI,0,3 and ethanoate or methanoate species are often formed on oxide surfaces after contact with propene and oxygen.5.6.44 However, spectra containing prominent bands due to v ~ ~ , ( ~ ~ ) and vco,(s).vibrations of carboxylate ions after propene-NO reaction over Rh/Al,O, did not exhibit bands due to methanoate ions nor a band at 1336 cm-' [fig. 3(6)] due to a methyl-group deformation vibration of ethanoate anions.18 Nitric oxide oxidation of propene was apparently selective in the generation of acrylate and not ethanoate or methanoate. The maximum at 1690 cm-' in spectra [fig. I (e)] after heating propene-NO mixtures over Rh/AI,O, must be due to the v,,, vibration of an adsorbed carbonyl compound, probably acrolein. Acetone is unlikely as it would lead to the formation of ethanoate and methanoate ions on further oxidation.34 Acrolein is the logical precursor of acrylate (fig.,).lo A notable absentee from the spectra was the band at 1678 cm-' [fig. 4(h)] due to coordinatively ligated acrolein at A13+ sites on which 2-formylethanoate anions subsequently appeared. No 2-formylethanoate was generated by the propene-NO reaction, in accordance with the specificity of reaction to give adsorbed acrylate. There was no infrared evidence for a n-ally1 species which might have been an intermediate in the conversion of propene to acrolein3 or might have existed on the alumina The bands at 1915 (RhNO+), 1823 and 1740 cm-' [Rh(NO),] due to NO adsorbed on rhodium after pretreatment or heat treatment with propene resemble similar bands for NO alone on Rh/A120,.8 Pre-adsorption of propene prevented the formation of RhNO' [fig. I(c) and 2(d)] and hindered the formation of Rh(NO), particularly after heat treatment in propene at 473 K.However, for Rh/AI,O, in propene-NO mixtures the Rh(NO), species disappeared and the band due to RhNO+ was observed. Evidence for the concomitant dissociative adsorption of NO and oxidation of propene is provided by the infrared bands at 2185 cm-' (due to RhNCO)9 and 2250 cm-' (AINCO formed by spillover from Rh to Al,O,).' Dissociation of NO gives an N adatom, which reacts with a CO species derived from propene oxidation to give NCO and an oxygen atom which either may act as the reactive species in propene oxidation or remains as an 0 adatom possibly bonded to the same rhodium atom as an NO molecule contributing to the infrared band at 191 5 cm-'.8 The Rh(NO)(O) complex replaces the gem-dinitrosyl Rh(NO), on the surface, accounting for the disappearance of the bands at 1823 and 1740 cm-'.We thank Johnson Matthey Research Centre for helpful discussions and for a sample of aqueous rhodium nitrate. References 1 B. Harrison, M. Wyatt and K. G. Gough, Curalysis (Specialist Periodical Report, Royal Society of Chemistry, London, 1982), vol. 5. p. 127. 2 N. W. Cant and W. K. Hall, J. Catal., 1970, 16, 220. 3 N. W. Cant and W. K. Hall, J. Catal., 1971, 22, 310. 4 P. G. Harrison and B. Maunders, J . Chem. Soc., Faruday Trans. I , 1985. 81, 1329. 5 M. Iwamoto and J. H. Lunsford, J . Phys. Chem., 1980. 84, 3079. 6 Y. Kubokawa, M. Miyata, T. Ono and S . Kawasaki, J . Chem. Soc., Chem. Commun., 1974, 655. 7 R. E. Nightingale, A. R. Downie, D. L. Rotenberg, B.Crawford and R. A. Ogg, J . Phjx Chem., 1954, 58, 1047. 8 E. A. Hyde, R. Rudham and C. H. Rochester, J . Chem. Soc., Furuduy Truns. 1. 1984. 80, 531. 9 F. Solymosi and J. Sarkany, Appl. Surf. Sci., 1979, 3, 68. 10 E. A. Hyde, R. Rudham and C . H. Rochester, J . Chem. Soc.. Furudav Truns. I . 1983. 79, 2405. 1 I N. Sheppard and T. T. Nguyen, Ado. Infrared Raman Spectrosc., 1978. 5, 67.1128 Infrared Study of Rh/Al,O, 12 A. C. Yang and C. W. Garland, J. Phys. Chem., 1957, 61, 1504. 13 M. Primet, J. Chem. Soc., Faraday Trans. 1, 1978, 74, 2570. 14 A. V. Kiselev and A. V. Uvarov, Surf. Sci., 1967, 6, 399. 15 F. Bozon-Verduraz and G. Pannetier, Bull. SOC. Chim. Fr., 1970, 3856. 16 S. Hayashi, T. Takenaka and R. Gotoh, Bull. Inst. Chem., Res. Kyoto Unil;., 1969, 47, 378.17 M. Falk and T. A. Ford, Can. J. Chem., 1966, 44, 1699. 18 E. Spinner, J. Chem. SOC.. 1964, 4217. 19 R. K. Harris, Spectrochim. Acta, Part A , 1964, 20, 1129. 20 W. R. Feairheller and J. E. Katon, Spectrochim. Acta, Part A , 1967, 23, 2225. 21 G. Dellapiane and J. Overend, Spectrochim. Acta, Part A, 1966, 22, 593. 22 V. Y. Danyushevskii, L. I. Lafer, V. I. Yakerson and A. M. Rubinshtein, Xzvest. Akad. Nauk SSSR, 23 D. A. Edwards and R. N. Hayward, Can. J. Chem., 1968, 46, 3443. 24 K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounds (Wiley- 25 H. E. Evans and W. H. Weinberg, J. Chem. Phys., 1979, 71, 4789. 26 F. Solymosi and A. Erdohelyi, J. Catal., 1985, 91, 327. 27 F. Solymosi and M. Pasztor, J. Catal., 1987, 104, 312. 28 F. Solymosi, A. Erdohelyi and M. Kocsis, J. Cafal., 1980, 65, 428. 29 T. Iizuka and Y. Tanaka, J. Catal., 1981, 70, 449. 30 F. Solymosi and A. Erdohelyi, J . Catal., 1981, 70, 451. 31 Y. Tanaka, T. Iizuka and K. Tanabe, J. Chem. Soc., Faraday Trans. I , 1982, 78, 2215. 32 F. Solymosi and M. Lancz, J. Chem. Soc., Faraday Trans. I , 1986, 82, 883. 33 F. Solymosi and J. Kiss, J . Catal., 1983, 81, 95. 34 H. Miyata, K. Hata, T. Nakajima and Y. Kubokawa, Bull. Chem. SOC. Jpn, 1980, 53, 2401. 35 H. Winde, 2. Chem., 1970, 10, 64. 36 D. M. Griffiths and C. H. Rochester, J. Chem. SOC., Faraday Trans. 1, 1978, 74, 403. 37 C. A. Rice, S. D. Worley, C. W. Curtis, J. A. Guin and A. R. Tarrer, J . Chem. Phys., 1981, 74, 6487. 38 G. Busca and V. Lorenzelli, Muter. Chem., 1982, 7, 89. 39 N. D. Parkyns, J. Phys. Chem., 1971, 75, 526. 40 P. G. Harrison and B. Maunders, J . Chem. SOC., Faraday Trans. I , 1985, 81, 1345. 41 M. Niwa, Y. Tanaka and Y. Murakami, J. Colloid Interface Sci., 1982, 89, 571. 42 J. A. Anderson and C. H. Rochester, J. Chem. SOC., Faraday Trans. 1, 1989, 85, 1129. 43 J. T. Yates and R. R. Cavanagh, J. Catal., 1982, 74, 97. 44 R. Delobel, M. Le Bras, M. Traisnel and J-M. Leroy, in Vibrations at Surfaces, ed. R. Caudano, J-M. Gilles and A. A. Lucas (Plenum Press, New York, 1982), p. 315. 45 T. A. Gordymova and A. A. Davydov, Kinet. Catal., 1979, 20, 604. Ser. Khim., 1972, 2089. Interscience, New York, 1978), p. 232. Paper 81024745; Received 21st June, 1988

ISSN:0300-9599    

DOI:10.1039/F19898501117

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J. Chern. Suc., Faraduy Trans. I , 1989, 85(5), 1129-1138 Infrared Study of the Adsorption of Ethanoic Acid, Acrylic Acid, Acetone, Acrolein and Propene-NO Mixtures on Rh/SiO, Catalysts James A. Anderson and Colin H. Rochester* Chemistry Department, The University, Dundee DDI 4HN, Scotland Infrared spectra of the adsorbed products of reaction of propene and NO over Rh/SiO, are compared with spectra of adsorbed acetone, acrolein, acrylic acid and ethanoic acid. Acetone and acrolein formed hydrogen bonds with surface silanol groups and underwent decomposition on rhodium to give rhodium carbonyl species. The carboxylic acids also gave rhodium carbonyl species, but for ethanoic acid this only occurred at high temperature. Both acids were adsorbed on silica either through chemi- sorption giving a surface ester or as a weakly held dimer.Heat treatment of Rh/SiO, in propene/NO mixtures gave acrylic acid as the dominant partial oxidation product. However, addition of propene before NO to the reaction system led to the generation of a surface poison which inhibited the oxidation of propene. Acrylate anions on the surface of the alumina support were the major adsorbed product after exposure of Rh/A1,03 catalysts to propene/NO mixtures and heat treatment at 473 K, although infrared spectra also showed maxima due to surface rhodium carbonyl species.' Spectra of adsorbed ethanoic acid, acetone and acrolein were dominated by bands due to species on alumina, making products on rhodium, apart from rhodium carbonyl complexes, impossible to identify.' Ethanoic acid,' a ~ e t o n e , ~ acrolein and other possible products of oxidation of propene over supported rhodium catalysts are less likely to be adsorbed on silica than alumina.* In the present study of Rh/SiO, catalysts it was hoped to obtain additional information about adsorbed intermediates or products on rhodium in the absence from spectra of intense infrared bands due to adsorbed species on the oxide support.Experimental Catalyst precursor was prepared by evaporating to dryness slurries of aqueous rhodium nitrate (5.44 w/w YO Johnson Matthey) and type TK900 silica (Degussa) with surface area 148.6+ 1.1 m2 g-'. The rhodium content was 1.0 wt YO after reduction to Rho. Impregnated discs (25 mm diameter, 80 mg weight) of rhodium nitrate on silica were compressed at 30 MN rn-, (1.5 ton on die) and mounted in an infrared cell attached to a conventional high-vacuum system.After heat treatment in oxygen (27 kN m+, 673 K, 30 min) catalysts were reduced (773 K, 1 h) in a flowing (200 cm3 min-') hydrogen- argon mixture (101 kN m-') containing 3.5 v/v% hydrogen. The cell was evacuated with the sample at 773 K before cooling the catalyst to ambient temperature in vacuum. Other experimental details were as bef0re.l All spectra were recorded with the samples at ca. 300 K and are presented here after background subtraction in order to eliminate bands in the bulk spectrum of the silica support. I1291130 I.R. Study of Rh/SiO, Catalysts Results Exposure of Rh/SiO, to low pressures of ethanoic acid gave spectra containing a dominant band at 1751 cm-' [fig.1 (a)]. There was no change in the intensity of a band at 3744 cm-' due to isolated silanol groups on the silica support. This manifests itself as no band at this position in fig. 1 (a) because of the background subtraction used. At higher ethanoic acid pressure the band at 3744 cm-' was replaced by a broader band envelope at 3209 cm-' showing that silanol groups were forming hydrogen bonds with adsorbed ethanoic acid molecules. The v(C = 0) vibration of adsorbed molecules gave a strong band at 1726 cm-l [fig. 1 (b)]. A second strong band at 1417 cm-l suggested that acid dimers were present in the adsorbed state., Subsequent evacuation with the sample at ambient temperature removed acid dimer and ethanoic acid molecules involved in hydrogen-bonding interactions from the surface.The maxima at 1726 and 1417 cm-l disappeared, leaving maxima at 1751, 1433 and 1381 cm-' [fig. 1 (c)], which compare with bands at 1760, 1433 and 1381 cm-' in spectra of ethanoic acid adsorbed on silica alone., These maxima are ascribed to surface ester-like ethanoate species formed by dissociative chemisorption of ethanoic acid., The removal of hydrogen bonded ethanoic acid on evacuation was further shown by the disappearance of the band envelope centred at 3209 cm-l and the reappearance of the maximum at 3744 cm-' almost to its original intensity before admission of ethanoic acid. A further band was revealed at 3538 cm-' [fig. l(c)] and existed at lower intensity in spectra of Rh/SiO, in the presence of low pressures of ethanoic acid [fig.1 (a)]. The changes in intensity of this band apparently paralleled intensity changes of the bands at 1751 and 1381 cm-', therefore the band is ascribed to water or silanol groups generated by the dissociative adsorption of ethanoic acid. Heat treatment of Rh/SiO,, which had been exposed to ethanoic acid and evacuated before raising the temperature to 473 K in a closed vacuum, led to decomposition of surface ethanoate [fig. 1 (d)] and the appearance of a band at 2047 cm-' due to linearly adsorbed CO on rhodium. The band at 3538 cm-' became a very weak band at 3560 cm-'. Subsequent dynamic evacuation at 473 K removed adsorbed CO from the rhodium surface [fig. 1 (e)]. The intensity of the maximum at 3744 cm-' due to silanol groups was identical to its intensity before ethanoic acid was initially admitted to the cell, showing that the number of isolated silanol groups was unaffected by the chemical reactions taking place. After evacuation at 623 K, residual weak bands due to surface ethanoate remained in the spectra [fig.1 0 1 together with a weak maximum at 1703 cm-'. Throughout the sequence of experiments involving the adsorption of ethanoic acid there were no clearly detectable bands due to vCH vibrations. Bands at 2653 and 2576cm-1 present in some spectra correspond to overtone or combination bands in spectra of ethanoic acid., The adsorption of acetone on Rh/SiO, gave spectroscopic data characteristic of the formation of hydrogen bonds between surface silanol groups and adsorbed acetone m01ecules.~~~ The band at 3744 cm-l was replaced by a broader, more intense maximum at 3399 cm-l, and a band due to perturbed carbonyl groups appeared at 1714 cm-' [fig.2(a)]. Additional bands at 3015,2970,2937, 1425 and 1376 cm-' correspond closely with previously reported bands for acetone on silica.5 A side-band at 1740 cm-' was due to the v(C0) vibration of acetone vapour.6 All these bands were also present in spectra of acetone adsorbed on silica alone, which had been subjected to an identical reduction pretreatment procedure as used for Rh/SiO, catalyst. Furthermore a weak band at 2140 cm-' [fig. 2(a)] was also present in spectra of acetone.6 However, weak maxima at 1981 and 1492 cm-' were recorded only for acetone on Rh/SiO, and not for silica in the absence of rhodium.The former was similar to a band at 1970 cm-' for Rh/Al,O,, which was attributed to a bridged carbonyl complex of rhodium formed by the dissociative reaction of acetone.' Evacuation at ambient temperature largely removedJ . A. Anderson and C. H. Rochester 1131 I I I I I I I LOO0 3500 3000 2 9 0 2500 2000 1800 1600 1400 1200 wavenumtm/crn- Fig. 1. Spectra of Rh/SiO, (a) exposed to a low vapour pressure of ethanoic acid at cu. 300 K, (b) with a higher vapour pressure of ethanoic acid, ( ( 8 ) after evacuation (6 min, cu. 300 K), ( d ) subsequent heat treatment (1 h, 473 K) in a static vacuum, followed by evacuation at (e) 473 K (30 min), and ( f ) 623 K (30 min). acetone involved in hydrogen-bonding interactions, leaving only a very weak band at 1714 cm-l [fig.2(6)]. The sharp band at 1492 cm-I remained in the spectrum, but the band at 1981 cm-' was replaced by a band at 2019 cm-' due to a linear surface carbonyl complex of rhodium. Very weak residual bands were also revealed at 1575 and 1400 cm-'. Evacuation at 3 13 K had little further effect on the spectrum [fig. 2 (c)]. Evacuation at 523 K completely removed all adsorbed species from the Rh/SiO, surface. Spectra of silica exposed to acrolein showed that hydrogen bonds were formed between isolated surface silanol groups and the carbonyl groups of adsorbed acrolein molecules. Perturbed silanol and carbonyl groups gave maxima at 3460 and 1693 cm-I, respectively, with a shoulder at 173 1 cm-' due to the v(C0) vibration of acrolein vapour. Three maxima at 2826,2776 and 271 5 (vw) cm-' were combination bands' for adsorbed acrolein.Nearly all the acrolein was desorbed by evacuation at ca. 300 K, although very weak residual bands at 2960 and 2880 cm-' suggested that some chemical reaction involving acrolein had occurred. All the spectral features for acrolein on silica were also observed for acrolein on Rh/SiO, [fig. 3 (u)-(c)]. However, two additional features could be attributed to adsorption involving rhodium. First, a band at 2019 cm-' may be ascribed to a linear carbonyl complex of rhodium which was also formed when Rh/ A1,0, was exposed to acrolein.' Secondly, a band at 1492 cm-', which would have been1132 I.R. Study of Rh/SiO, Catalysts t - - 7 4000 3300 2700 I I I I I I I I zoo0 1800 1600 1400 wavenumber/cm- Fig.2. Spectra of Rh/SiO, (a) exposed to acetone vapour (4.0 kN m-2, ca. 300 K) and subsequently evacuated at (6) ca. 300 K (6 min) and (c) 313 K (30 min). ZOO0 1800 1600 wavenumber/cm-' Fig. 3. Spectra of Rh/Si02 exposed to acrolein vapour at ca. 300 K for (a) ca. 2 min (:b) 20 min and subsequently evacuated at (c) ca. 300 K (6 or 25 min) and ( d ) 313 K (30 min). obscured by intense bands due to surface acrylate on alumina for Rh/Al,O,, corresponded to the same band in spectra of Rh/SiO, exposed to acetone (fig. 2). However, unlike the result after acetone adsorption, the band at 1492 cm-' disappeared after evacuation at 313 K [fig. 3(d)]. The maximum at 2019 cm-l was replaced by a weaker band at 2030 cm-l, also due to a linear carbonyl surface complex of rhodium, and very weak residual bands due to organo-carbonyl species were present at 1780 and 1739 cm-'.Spectra resulting from the interactions between NO and propene over Rh/SiO, are shown in fig. 4. The adsorption of NO gave surface rhodium nitrosyl species withJ . A. Anderson and C. H. Rochester I I I I I I 1 I ,oo 2000 1wx) 1600 1400 9 10 3500 3000 wavenumber/cm- I133 Fig. 4. Spectra of Rh/SiO, (a) exposed to NO (5.3 kN m-2), (b) partial removal of NO and addition of propene (8.0 kN m-*), (cHg) exposed to an NO (5.1 kN m-2)+propene (7.6 kN m-2) mixture (c) at ca. 300 K, (d) after 30 min at 373 K, (e) 30 min at 473 K, cf) 4.5 h at 473 K, (g) 19 h at 473 K and (h) evacuation at ca. 300 K. characteristic' infrared bands at 1917, 1843 and 1753 cm-' and a weaker band at 1680 cm-I due to a negatively charged nitrosyl specie^^*'^ [fig.4(a)]. The addition of propene depleted the intensities of bands due to adsorbed NO and gave a series of bands at 2800-3 100, 1668, 1642, 148 1 and 1445 cm-' due to vibrations of propene gas [fig. 4 (b), (c)]. The only new feature after exposure of Rh/SiO, to NO+ propene at ca. 300 K was a weak band at ca. 1720 cm-l. Raising the temperature to 373 K had negligible further effect [fig. 4(d)]. However, at 473 K, slow reactions took place giving several adsorbed and gaseous products [fig. 4(e)-(g)]. A broad maximum at 3500 cm-l may be attributed either to water, which would have also contributed to the growth in absorption intensity at ca. 1620 cm-l or to silanol groups forming hydrogen bonds with reaction products.A maximum at 2330cm-' and a doublet at 2242 and 2208cm-l were due to carbon dioxide and nitrous oxide, respectively. Subtraction of spectrum 4(b) from spectrum 4 0 showed that there had been no detectable changes in the 2800-3200 cm-' region of the spectrum, therefore only a very small proportion of the propene added to the system had undergone reaction. The bands at 1668, 1642, 1481 and 1445 cm-' due to propene gas must also have remained unchanged by the sequence of heat treatments. The most prominent new maximum was at 1709 cm-', which suggests the formation of an organic carbonyl compound as product. There was also broad growth in absorption intensity in the range 1550-1600 cm-l. Apparent bands at 1987 and 1851 cm-' [fig.4(g)] arose from imperfect cancellation of intense combination bands in spectra of the bulk silica support. Prolonged heat treatment of discs slightly altered the baseline characteristics of the1134 I.R. Study of Rh/SiO, Catalysts I I I I I I I 10 3500 3000 2500 2000 1800 1600 1400 wavenumber/cm-' Fig. 5. Spectra of Rh/SiO, exposed to a propene (5.4 kN m-2, added first)+NO (5.1 kN m-*) mixture (a) at ca. 300 K, (b) after 25 min at 473 K, ( c ) 60 min at 473 K, ( d ) 19 h at 473 K, (e) 30 min at 623 K, cf) subsequent evacuation at ca. 300 K. samples and made cancellation inexact. The dominant residual bands after evacuation at ca. 300 K were at 3554 and 1575 cm-' [fig. 4(h)]. Hyde et al.8 showed that the adsorbed products of reaction of CO + NO mixtures over supported rhodium differed depending on whether CO or NO was first added to the catalyst sample.Similarly here the order of addition of NO and propene influenced the reactions which subsequently occurred at 473 K. The results in fig. 4 were for the initial addition of NO followed by propene. Fig. 5 shows results which are typical for experiments in which the order of addition was reversed. The 1350-1700 cm-l spectral region showed a broad featureless band envelope which not only contained bands due to propene gas but also a maximum at ca. 1550-1 575 cm-l which persisted throughout the experiments and was also present when NO had been added to the system first (fig. 4). Apart from a weak band at 1917 cm-' [fig. 5(a)] there was little evidence for the non- dissociative adsorption of NO on rhodium, showing that metal sites had become occupied by preadsorbed propene or its adsorption products.Subsequent heat treatment at 473 K [fig. 5(b)--(d)] failed to generate carbon dioxide or nitrous oxide in the gas phase or give the adsorbed carbonyl compound responsible for an infrared band at 1709 cm-'. Spectrum 5 (4, for example, contrasts with spectrum 4(g) which gave clear evidence for these three reaction products. A novel feature in fig. 5(b)-(d) was a maximum at 2178 cm-', which was not present in spectra of the gas phase in contact with the disc. This band is ascribed to a surface complex of rhodium, possibly a cyanide,l'. l2 which acts as a poison of sites which might otherwise be active for the oxidation of propene.Heat treatment at 623 K destroyed the species, giving the band at 2178 cm-l, and led to the appearance of infrared bands due to carbon dioxide and nitrous oxide [fig. 5(e)]. Evacuation removed gaseous reactants and products and left a spectrum with broad maxima at ca. 3500 and 1575 cm-l and very weak residual bands at 2214 and 2158 cm-' [fig. 5 0 1 . Bands at 1680-1690 cm-' due to a rhodium nitrosyl species have been reported inJ . A . Anderson and C. H. Rochester 1135 I I I I I I I e .r( .i ii b I I I 1 I 1 1 I I 4000 3500 3000 2500 2ooo 1800 1600 1400 w avenumtxr/cm- Fig. 6. Spectra of Rh/SiO, ( a ) after contact with propene + oxygen (473 K, I h), (h)--(d) fresh disc exposed to increasing vapour pressures of acrylic acid at ca. 300 K, (e)-(g) after subsequent evacuation at (e) cn.300 K (6 min), (I) 353 K (30 min) and (g) 473 K (30 min). spectra of Rh/SiO, after high-temperature treatment in mixtures of NO with hydrogen, CO or hydrogen-CO mixture^.^. lo* l3 Confirmation that the band at I709 cm-' in spectra of Rh/SiO, exposed to NO-propene mixtures [fig. 4(g)] was an organo-carbonyl species rather than a surface nitrosyl complex of rhodium was gained by brief study of propene oxidation by oxygen over Rh/SiO,. A disc was exposed to a propene-oxygen mixture at 473 K. Spectra of the gas phase after reaction confirmed that carbon dioxide (2347 cm-') was the only detectable desorbed product. Bands due to adsorbed products appeared at 3443 cm-' (due to molecular water or surface silanol groups), 1709, 1550 and 1431 cm-' [fig.6(a)]. The maximum at 1709 cm-' was almost entirely removed from the spectrum when the sample was evacuated at ca. 300 K, although the maximum due to surface hydroxyl groups was only slightly depleted. The band at 3443 cm-' is therefore primarily attributed to the formation of hydrogen-bonded silanol groups on the silica surface. Molecular water formed in the oxidation process4 was dissociatively adsorbed at strained siloxane group sites, giving adjacent interacting silanol groups. The maximum at 1709 cm-' closely resembled the corresponding band resulting from the oxidation of propene by NO [fig. 4 0 , (g)] and therefore the two bands are attributed to the same weakly adsorbed reaction product. Comparing the spectra from the propene- oxygen reaction with spectra of adsorbed acrolein and acetone showed that acrolein could not have been responsible for the band at 1709 cm-', but that adsorbed acetone could at least have contributed to the intensity of a band at this position. However, the overall spectral fit for acetone was not good, showing that although acetone might have been present it was not the dominant adsorbed partial oxidation product.In particular a strong band,6 appearing at 1376 cm-' for acetone on Rh/SiO, [fig. 2(a)], was present only at the most as a weak shoulder after the propene-oxygen reaction [fig. 6(a)] and was absent from spectra after the reaction of propene with NO [fig. 4u),(g)]. The observation' that surface acrylate was the dominant partial oxidation product from propene and NO over Rh/AI,O, suggested that acrylic acid might have been formed in1136 I.R.Study of Rh/SiO, Catalysts the present systems. Spectra of Rh/SiO, exposed to acrylic acid vapour were therefore recorded. The adsorption of acrylic acid on Rh/Si02 paralleled the results for ethanoic acid, giving a surface acrylate ester species by dissociative adsorption on silica and weakly adsorbed dimer molecules. A pair of dominant infrared bands existed in spectra of both species, at 1728 and 1414 cm-' for the ester and cu. 1710 and 1428 cm-' for the dimer. The bands are at 1705 and 1432 cm-' for liquid acrylic acid.14 As for ethanoic acid (fig. 1) the ester was the main product adsorbed during the initial stages of adsorption [fig. 6(b)] or after desorption of the dimer by evacuation [fig.6(ek(g)]. Dimer was present at pressures of acrylic acid approaching its saturated vapour pressure at ca. 300 K. The dimer bands were not distinguishable as separate maxima from the ester bands in spectra of Rh/SiO, exposed to acrylic acid but were detected through their band broadening and shift effects [fig. 6(c), (41. Computer-calculated difference spectra, e.g. spectrum 6 (c) minus spectrum 6 0 , eliminated the ester bands and unambiguously revealed the two maxima due to adsorbed dimer. The close similarity between the dimer bands and the main maxima at 1709 and 1431 cm-' appearing after the propene-oxygen reaction over Rh/SiO, [fig. 6(a)] strongly suggests that acrylic acid was the dominant partial oxidation product and was responsible for the infrared band at 1709 cm-' after oxidation of propene by either oxygen or NO.The absence of appreciable bands due to the acrylate ester in spectra of the oxidation products on Rh/SiO, is attributed to the chemisorption of product water at surface sites on silica which would otherwise be available for the dissociative adsorption of acrylic acid. An additional feature of the adsorption of acrylic acid on Rh/SiO, was the appearance of very weak bands at 2019 and 1982 cm-' due to rhodium carbonyl species [fig. 6(e)l- Discussion The spectroscopic results for the adsorption of ethanoic acid on the silica component of Rh/SiO, catalysts were consistent with previous data for ethanoic acid on silica alone., However, the suggestion2 that surface ethanoate was formed by a condensation reaction between silanol groups and ethanoic acid, with the elimination of water, is not compatible with the present results since the formation of ethanoate [fig. l(u),(c)] was not accompanied by a decrease in intensity of the infrared band due to isolated silanol groups.Dissociative adsorption of ethanoic acid more probably occurred at strained siloxane group sites in accordance with the reaction. 0 The concomitant generation of adjacent hydroxyl and ethanoate groups which can interact laterally via hydrogen-bond formation accounts for the infrared band at 3538 cm-' due to perturbed surface silanol groups. The formation of rhodium carbonyl species from ethanoic acid over Rh/SiO, or Rh/Al,O,' occurred less readily than the formation of rhodium carbonyl from either acetone or acrolein.Both acetone and acrolein, but not ethanoic acid, gave carbonyl species at ambient temperature. The band at 1981 cm-' for acetone on Rh/SiO, compares with a corresponding band at 1970 cm-' for Rh/A120,, whereas acrolein gave a maximum at 2019 cm-' for both Rh/Si02 and Rh/Al,O,. The band at 1970 cm-' for acetone on Rh/Al,O, was ascribed to a bridged carbonyl species,l although a band as low as 1990 crn-' in HREELS spectra of CO on Rh(II1) crystal faces has been attributedJ . A . Anderson and C. H. Rochester 1137 to CO bonded to a single Rh atom.'% One possibility herel is that the positions of the infrared bands due to carbonyl species were influenced by the coexistence of CO with hydrogen adatornsl6* l i or adsorbed hydrocarbon species" at the same or adjacent rhodium sites.Acetone may give bridging carbonyl species and acrolein may generate linear carbonyl. However, a plausible alternative is that acetone, which contains two more hydrogen atoms than acrolein, leads to rhodium carbonyl with a higher surrounding surface concentration of other (hydrido or hydrocarbon) decomposition products and thereforel6-l8 the carbonyl stretching vibration gives an infrared band at a lower (in cm-') position in the spectrum. Partial removal of other decomposition products by evacuation shifted the band at 1981 cm-' for acetone on Rh/SiO, to 2019 cm-' in agreement with the band position following acrolein adsorption. The implication that acetone and acrolein give similar adsorption products on rhodium is further borne out by the weak band at 1492 cm-' in spectra of Rh/SiO, after exposure to either acetone or acrolein at ambient temperature.The attribution of this band is uncertain, but it is probably due to a hydrocarbon species derived from decarbonylation of the carbonyl compounds. The formation of nitrous oxide from propene-NO mixtures over Rh/SiO, is not surprising because NO alone generates nitrous oxide over supported rhodium at elevatedlg but not ambient' temperatures. The appearance of infrared bands due to carbon dioxide and water was also expected since Cant and Hall' showed that these were the dominant products of oxidation of propene by oxygen over Rh/SiO,. However, their kinetic study also identified a series of partial oxidation products which were present in similar relative proportions after catalysis by either Rh/Si0,4 or Rh/A1,0,.20 The present infrared study has also identified both complete and partial oxidation products derived from propene by reaction with NO.The dominant partial oxidation product observed here for Rh/SiO, was adsorbed acrylic acid which compares with the formation of acrylate ions adsorbed on the alumina support for Rh/Al,O,.' Cant and Hall4, 2o studied the propene-oxygen reaction in flow systems over Rh/SiO, and Rh/AI,O, and found that acetone was the main product after short reaction times. However, after ca. 1 h steady-state conditions were achieved and acrolein became the main partial oxidation product. Traces of acrylic acid product were also detected. Heat treatment of Rh/SiO, or Rh/Al,O, in static propene/NO mixture in the present systems allowed any acrolein formed to be further oxidised to acrylic acid which became the dominant product.Propene may adsorb on rhodium to form a symmetrical n-ally1 complex which is believed to be an intermediate in the dominant mechanism for the formation of acrolein from propene-oxygen mixtures. 2o Alternatively, a surface propylidyne species is formed at ambient temperatures but undergoes fragmentation through carbon-carbon bond scission and loss of hydrogen above ca. 420 K.15 Coverage of rhodium by propene is shown by the subsequent inhibition of NO absorption and by the displacement of non- dissociatively preadsorbed NO by propene. The infrared band at 2178 cm-l [fig. 5 (h)-(d)] is ascribed to an adsorbed species on rhodium resulting from reaction between NO and scission products of propylidyne decomposition at 473 K.The band is close to previously observed maxima at 2185 and 2170 cm-l, which where attributed to rhodium isocyanate species generated by reaction between CO and NO over r h o d i ~ m . ~ , ' However, the formation of metal isocyanate in silica-supported metals is invariably accompanied by spillover of isocyanate to the silica support, with the resulting appearance of a very strong maximum at ca. 2300 cm-l due to SiNCO.' The absence of this maximum from the present spectra preclude's the assignment of the band at 2178 cm-' to RhNCO. The band is also in the same spectral region as bands due to dinitrogen ligated to or adsorbed on rhodium However, dinitrogen is unlikely to constitute a plausible catalyst poison at 473 since it is too weakly adsorbedz4 and there is no logical reason why adsorbed dinitrogen should be formed from NO in the presence1138 I.R.Study of Rh/SiO, Catalysts but not the absence of propene. Bands close to 2178 cm-’ exist in spectra of cyano complexes of rhodium,”. l2 therefore the present result is tentatively ascribed to surface cyanide formed by the reaction between carbon atoms generated by the breakdown of propylidyne species15 and nitrogen atoms generated by the dissociative adsorption of NO.” Hydrogen cyanide is a reported product of reaction of NO with CO/hydrogen mixtures over rhodium dispersed in alumina. 25 We thank Johnson Matthey Research Centre for helpful discussions and a sample of aqueous rhodium nitrate.References I J. A. Anderson and C. H. Rochester, J. Chem. Sac., Faraday Trans. I , 1989, 85, 1 1 17. 2 R. P. Young, Can. J. Chem., 1969, 47, 2237. 3 A. V. Kiselev and A. V. Uvarov, Surf. Sci., 1967, 6, 399. 4 N. W. Cant and W. K. Hall, J. Catal., 1970, 16, 220. 5 T. Bernstein, D. Michel, H. Pfeifer and P. Fink, J. Colloid Interface Sci., 1981, 84, 310. 6 G. Dellepiane and J. Overend, Spectrochim. Acta, 1966, 22, 593. 7 R. K. Harris, Spectrochim. Acta, 1964, 20, 1129. 8 E. A. Hyde, C. H. Rochester and R. Rudham, J. Chem. Soc., Faraday Trans. I , 1984, 80, 531. 9 W. C. Hecker and A. T. Bell, J. Catal., 1983, 84, 200. 10 W. C. Hecker and A. T. Bell, J. Catal., 1985, 92, 247. 1 1 W. P. Griffith and G. Wilkinson, J. Znorg. Nucl. Chem., 1959, 7, 2757. 12 W. P. Griffith and G. Wilkinson, J. Znorg. Nucl. Chem., 1958, 7, 297. 13 W. C. Hecker and A. T. Bell, J. Catal., 1984, 88, 289. 14 W. R. Feairheller and J. E. Katon, Spectrochim. Acta, Part A, 1967, 231, 2225. 15 R. J. Koestner, M. A. Van Hove and G. A. Somorjai, Chemtech., 1983, 13, 376. 16 F. Solymosi and M. Pasztor, J. Catal., 1987, 104, 312. 17 F. Solymosi, A. Erdohelyi and M. Kocsis, J. Catal., 1980, 65, 428. 18 F. Solymosi and M. Lancz, J. Chem. Soc., Faraday Trans. I , 1986, 82, 883. 19 A. A. Chin and A. T. Bell, J. Phys. Chem., 1983, 87, 3700. 20 N. W. Cant and W. K. Hall, J. Catal., 1971, 22, 310. 21 G. A. Ozin and A. Vander Voet, Can. J. Chem., 1973, 51, 3332. 22 Y. G. Borod’ko and V. S. Lyutov, Kinet. Catal., 1971, 12, 202. 23 V. S. Lyuotov, V. A. Redosimov and Y. G. Borod’ko, Russ. J. Phys. Chem., 1972, 46, 973. 24 H. P. Wang and J. T. Yates, J. Phys. Chem., 1984, 88, 852. 25 R. J. H. Voorhoeve, C. K. N. Patel, L. E. Trimble and R. J. Kerl, J. Catal., 1978, 54, 102. Paper 8/02860E; Received 15th July, 1988

ISSN:0300-9599    

DOI:10.1039/F19898501129

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J. Chern. Soc., Faraday Trans. I , 1989, 85(5), 1139-1 147 X-Ray Emission Spectra and Electronic Structure of the Sulphate and Methyl Sulphonate Anions, Dimethyl Sulphone and the Trimethylsulphoxonium Cation [(CH3)nS04-n]n-2, (n = 0, 1 , 2 and 3) Renate Foercht and David S. Urch* Chemistry Department, Queen Mary College, Mile End Road, London El 4NS The S K/? X-ray emission spectra have been measured for the series [(CH3),rS01-,r]’L-2r ( n = 0, 1, 2 and 3). Peak shifts and changes in relative peak intensities are rationalised in terms of changes in electronic structure. S 1s and other sulphur orbitals become less tightly bound as methyl substitution increases, which causes a decrease in S 3p participation in ligand 2s orbitals and a decrease in the relative intensity of KF.Evidence is found for increased ‘ hyperconjugation ’ between C-H and S-0, S-C r~ bonds as the number of methyl groups increases. When an atom is bombarded with electrons of more than a few hundred electron volts or irradiated with X-rays a core electron is ejected leaving a vacancy in an inner shell. This excited ion can relax either by the emission of an X-ray or by the ejection of an Auger electron.’ In the former case the vacancy is transferred to an outer orbital and the energy of the X-ray is the difference in excitation energies of the initial (core-hole) and final (outer orbital hole) states. Proviged that the wavelength of the X-radiation is greater than atomic dimensions (say 2 A, equivalent to a photon energy of ca. 6000 eV) then its emission is, to a good approximation, governed2 by the electric dipole selection rule, A1 = 1 ( I = atomic orbital angular momentum quantum number).Thus if the initial vacancy were in a sulphur 1s level, only transitions from the 2p and 3p levels would be allowed, giving rise to the S K G C ~ , ~ and S KP,,, emission lines, respectively. However, sulphur 3p orbitals are in the valence band and contribute to molecular orbitals whenever sulphur forms chemical bonds. This perturbation will lead to 3p character being distributed in varying degrees over many orbitals with different energies. The dipole selection rule applies to the atomic components of molecular orbitals just as it does to the individual atomic orbitals, so that the fine structure observed in valence X-ray emission spectra (i.e. X-ray spectra where the final state is in the valence band) is directly related to the electronic structure of molecular orbital^.^ A simple, convenient but approximate model to describe chemical effects in valence X-ray spectra can be based on the one-electron frozen-orbital ’ appr~ximation.~ Molecular orbitals are represented as linear combinations of atomic orbitals (LCAO appro xi mat i on), 71 wi = C uri Q r r [where uri is the coefficient describing the contribution of atomic orbital Qr (there being n such orbitals in the system) to the ith molecular orbital wJ.The amount of $,. present t (nee Horn) ; present address : Surface Science Laboratory, Natural Sciences Centre, University of Western Ontario, London, Ontario N6A 5B7, Canada. I I391140 Electronic Structure of [(CH,),SO,-,]"-' in lyi is given by (ari)', or alternatively this quantity represents the probability of an electron in lyi being at $,..If the LCAO-MO representation is used for ly,, the orbital in which the vacancy resides after valence X-ray emission, and if it is assumed that one electron orbital function can be used to represent the initial and final states then it can be shown that the relative intensities of component peaks in a valence X-ray spectrum, involving orbital d,. are directly proportional to (arf)', where different values offindicate the different molecular orbitals in the valence band.5 Recent calculations by Larkins et al.' have shown that when such approximations are made quantitative results are compromised owing to the neglect of relaxation effects and, for bonds between small first-row atoms, the neglect of 'cross-over ' orbital overlap effects.The latter limitations are less important when two larger non-first-row atoms are involved. The simple one- electron LCAO-MO model can then be used successfully to develop a qualitative framework of molecular orbitals which can be used to rationalize the general features and the fine structure of X-ray emission and X-ray photoelectron spectra.' Such a model will be used in this paper to interpret changes in the sulphur KP spectra associated with the successive replacement of oxygen in the sulphate anion by the methyl group to form, the methyl sulphonate anion, dimethyl sulphone and the trimethylsulphoxonium cation. Experimental The compounds sodium sulphate, sodium methyl sulphonate, dimethyl sulphone and trimethylsulphoxonium iodide, were used as supplied without further purification.Samples were prepared by pressing each compound into a disc with terephthalic acid. The X-ray emission spectra were measured using a Philips PW 1410 single-crystal X-ray spectrometer. Excitation was achieved by bombardment of the sample with X-rays from a chromium anode X-ray tube, run typically at 50 kV and 40 mA. The characteristic X-rays were dispersed by a single Si(lI1) flat crystal (2d = 627.1 pm) and detected in a proportional counter fitted with 1 pm Mylar window. Even though fine collimation was used (blade separation) ( I 50 pm), it was necessary to insert an extra collimator,s with the blades perpendicular to those of the main collimator, to control ' horizontal divergence ' and so eliminate low-energy tailing of the emission peaks.Under such conditions it was found that the spectra obtained were of comparable quality to those previously obtained using double The output from the proportional counter was amplified using Harwell 2000 series electronics and recorded by a CBM PET microcomputer for which programs were developed to enable multiple scanning and consequently improve the spectral statistics. The 28 angle scanned during the experiments was typically from 105.5 to 108.5' in 0.02" 28 steps, which corresponds to an energy range of 2435 to 2485 eV. Data analysis involved background removal and normalization of the experimental data to enable direct comparison of spectral profiles.Detailed analysis was achieved by Lorentzian fitting and the calculation of Lorentzian peak areas for each of the spectra. Programs were developed for this purpose for use on the CBM PET microcomputer. or curved crystal spectrometers.12 Results The S KP spectra of the compounds listed in table I are shown in fig. 1 (a)-(d). These spectra are the averages of five scans for fig. 1 (a), (h) and (c), but of 50 scans for fig. 1 ( d ) , trimethylsulphoxonium iodide. This compound showed a very low intensity signal and considerable volatility in the spectrometer. The main S KP peak from the tetrahedral sulphate anion derives from S 3p character in triply degenerate t, molecular orbitals. As these orbitals all have the same energy a sharp peak should result which can be used to indicate the basic peak shape and minimum peak width to be anticipated under the experimental conditions used in thisR.Foerch and D. S. Urch 1141 Table 1. Components of the S KP peaks Na,SO, peak type area (%) Na[CH,SO,] peak type area (YO) 2468.0 C 2 2469.1 C 4 2465.0 B 69 2465.7 B 14 2464.2 B 58 2457.2 A 3 245 1.3 A 29 2450.3 A 21 (CH,),SO, peak type area (%) [(CH,),SOII peak type area (YO) 2469.1 C 5 2467.9 C 7 2467.1 C 5 2465.8 C 18 2464.8 B 33 2464.1 B 36 2463.4 B 33 2462.6 B 22 2456.7 A 13 2456.4 A 15 2450.0 A 1 1 2448.8 A 2 work. The peak was found to conform well to a Lorentzian curve with a full-width at half maximum of 2.8eV, a width which is made up of the natural line width of the S 1s hole (0.59 eV),13 the effect of the ‘fine’ collimator (0.14’ 28, equivalent to 2.1 eV), and broadening due to crystal imperfections and sulphur-oxygen bond vibrations (clearly these latter two factors make only a small contribution). Table 1 shows the relative intensities of component peaks as calculated by the areas of Lorentzians fitted to the experimental curves and given as percentages of the total peak area.This procedure includes background removal and normalization of spectra, to allow the direct comparison of component area values for the range of samples. Whilst the proposed deconvolution cannot be unique, the overall agreement between the experimental data and the sum of the proposed Lorentzian components corresponds to a standard deviation of + 2 %. Discussion The sulphur KP spectra in fig.1 are, where comparison is possible, in good agreement with previously reported work : (a) sodium ~ulphate,~. 12, l4-l5 (6) sodium methyl sulphonate, (c) dimethyl sulphonelo, 14-17 and ( d ) trimethylsulphoxonium iodide. Each spectrum shows X-ray emission in three regions: (A) low-energy satellite peak(s) in the range 2448-2460 eV, (B) main peak, of increasing width but centred at ca. 246k 2466 eV and (C) peak(s) or shoulders at energies above 2466 eV, whose intensity, relative to the main peak increases with the number of methyl groups. The division into three parts follows from the disposition of S 3p character that is anticipated by a simple molecular orbital model for the electronic structure of the sulphate anion. The tetrahedral symmetry requires that molecular orbitals with S 3p contributions transform as the t, irreducible representation of the & point group.In the valence band these orbitals ( 1 t, and 2t, are allocated to core orbitals 0 1s and S 2p) are: 3t2 mostly ligand 2s, some contribution from S 3p. 4t2 ligand 2p, S 3p, main o bond m.0.s. 5t2 ligand 2p ‘lone pairs’, with some contribution from S 3d (‘71.-bonding’). Only the principal contributions to the valence band m.0.s are indicated above. In particular the division into o(4t2) and n(5t2) is an approximation based on overlap considerations and not symmetry. ‘0-n mixing’ will lead to S 3p entering 5t, and S 3d being found in 4t,.1142 Electronic Structure of [(CH,)nS0,-n]n-2 110 99 88 77 66 33 22 11 0 2445 2447 2449 2451 2453 2455 2457 2459 2461 2463 2465 2467 2469 2471 2473 2475 energy /eV 11 0 99 88 77 6 6 33 22 1 1 0 2445 2447 2449 2451 2453 2455 2457 2459 2461 2463 2465 2467 2469 2471 2473 2475 energy lev Fig.1 (a), (b). For caption see facing page. This model provides a direct rationalization of the observed S K spectrum for sulphate with the peaks at (A) 2451, (B) 2465 and (C) 2472 being due to S 3p character in 3t2, 4t2 and 5t,, re~pectively.~. '** l9 That this pattern is retained on methyl substitution points to the basic similarity in electronic structures of all the compounds considered in this paper. The various sections of the spectra will now be considered in detail.R. Foerc'h and D. S. Urch 1143 99 88 77 66 33 22 11 0 2445 2447 2449 2451 2453 2455 2457 2459 2461 2463 2465 2467 2469 2471 2473 2475 energy /eV 110 99 88 77 66 33 22 11 0 2445 2447 2449 2451 2453 2455 2457 2459 2461 2463 2465 2467 2469 2471 2473 2475 energylev Fig.1. S KP spectra for (a) sodium sulphate, (h) sodium methyl sulphonate, ( ( 2 ) dimethyl sulphone and ( d ) trimethylsulphoxonium iodide. In all cases the experimental data are shown as small squares: the vertical scale is in arbitrary units (see text). The smooth lines show proposed Lorentzian components (2.8 eV f.w.h.m.) and their sum, for each compound. Details of peak positions and areas are summarised in table 1.1144 Electronic Structure of [(CH,),SO,-,]"-' A. 2448-2460 eV The peaks in this region arise from the presence of S 3p character in orbitals that are primarily ligand 2s in c h a r a ~ t e r .~ ' ~ ~ The peak at 2451.3 eV in sulphate moves to lower energy and becomes less intense (table 1) as the number of oxygen ligands decreases. (This peak will be denoted as S KPi to differentiate it from the other satellite peak associated with C 2s-S KP:..) This decrease in relative intensity is to be expected on simple numerical grounds, but note that it is more precipitate than mere numbers would indicate. The shift to lower energies of this S KPi peak cannot reasonably be ascribed to changes in the bonding between 0 2s and S 3p, since the 3t2 orbitals are predominantly 0 2s and their ionisation energy will be determined by oxygen and not sulphur. Successive replacement of oxygen by methyl will, however, reduce the S Is ionisation energy.methyl being less electronegative than oxygen. Thus transitions to S 1s from orbitals such as 3t, whose energy can be expected to be indifferent to oxygen replacement should suffer a reduction in energy. In the absence of direct XPS data on such core-level shifts it should be remembered that C 1s becomes less tightly bound by ca. 2.5eV in going from carbonyl (>c=o), to methyl (CH,-), in acetone;" just the same shift that is observed from S KP;. The shift to lower ionisation energy by S 1s will be followed by the other sulphur atomic orbitals. Thus S 3p orbitals will become less tightly bound as methyl substitution increases. As the efficiency with which atomic orbitals interact to form molecular orbitals is inversely related to the difference in their ionisation energies such a shift in the S 3p energy will lead to a reduction in amount of S 3p character to be found in the 'oxygen 2s' orbitals.This is in accord with the observed trend in S KPi intensity noted above. A second low-energy satellite peak, which is absent from sulphate, is observed with increasing relative intensity at ca. 2456-2457 eV. By analogy with S KP: it seems reasonable to rationalise this peak as due to S 3p character in orbitals from methyl that are dominated by C 2s, i.e. S KPL. This would explain the increase in relative intensity of this peak which roughly follows the number of methyl groups, and its shift to lower energies with increasing methyl substitution, which is due to the decrease in the S Is ionisation energy. B. Main Peak If only the immediate ligand atom environment of the sulphur atom is considered then the effective symmetries of sulphate (q), methyl sulphate (C3\) and dimethyl sulphone (C2,) are such that at least one S 3p orbital in all three systems will be exclusively engaged in S-0 a-bond formation.Although there will be small changes in overlap from one system to another, it is reasonable to assume that the nature of this orbital will remain constant, the same as for sulphate, and that it should still be possible to identify it as giving rise to a specific peak, S KP,, in the X-ray spectra of the methyl-substituted systems. This peak can be located by its position relative to S K/3' as the S KP,-S KP: energy difference (13.7 eV) can be expected to remain ~ o n s t a n t . ~ The S KP, peak will therefore decrease in energy, following S KP:, as the number of methyl groups increases and the S KP components at 2464.2 eV in CH,SO, and at 2463.4 eV in (CH,),SO, can be identified as due to transitions from S-0 a bonds.The percentage of S 3p character observed in these bonds diminishes as the number of oxygen ligands in sulphate decreases : sulphate, ( : t2) three bonds, 69 % ; methyl sulphonate (C,\ : e) two bonds, 58% and in dimethyl sulphoxide (C2\:b1), just one bond, 33%. The methyl group has local C,, symmetry and so the C-H bonding will be of two types. The e orbitals will be exclusively C 2p, H Is, whilst the a, orbitals will have mixedR. Foerch and D. S. Urclz 1145 C 2s, 2p as well as H 1s character. Even so it would appear that the two a, orbitals have quite distinct energies and that the most tightly bound has considerable C 2s character.Evidence for this assertion comes from the observation S KP:, as a distinct peak, analogous to S KPA. The less tightly bound a, orbital will also be orientated along the C-S direction and will interact with S 3s and 3p orbitals to form o bonds. Transitions from orbitals of the latter type will give rise to peaks in S KP. The local symmetry about sulphur is such that in the trimethylsulphonium cation (C3”, e) there will be two and in dimethyl sulphone (C2v, b,) one orbital exclusively associated with bonds of this type. In the absence of an S(Mea),+ cation whose S KP peak could be used to locate transitions from such orbitals, it is necessary to look to the trimethylsulphoxonium cation and to identify the principal S KP component at 2464.1 eV with S-C o orbitals.As it is reasonable to expect that the electronegativity of methyl should be less than that of oxygen and that the ionisation energy of the methyl ‘ C 2p, H 1s’ a,o should be less than that of 0 2p, it follows that the S-C o orbitals should be less tightly bound than the corresponding S-0 orbitals. It also follows that 0 orbitals with both carbon and oxygen character as well as sulphur should have ionisation energies between S-0 o and S-C 0. In all the compounds considered in this paper, except sulphate, there are a, orbitals of this type. It is clear from the narrowness and lack of structure of the KP peak that all the orbitals involved in o bonding, S-0, S-C and 0-S-C, all have quite similar energies, within 2 eV of each other.Thus given the experimental minimum peak width of 2.8 eV it is not possible to expect that the individual o molecular orbitals will give rise to distinct features in the KP spectrum. For CH,SO; and OS(CH3)i the asymmetric peak profile indicates the proposed deconvolution, but for (CH,),SO, the broadened, yet otherwise symmetric S KP peak could equally well (but without justification) have been presented as the sum of three rather than two components, i.e. the best that can be said is that the observed S KP spectrum for (CH,),SO, is not inconsistent with the presence of S KP character in three closely spaced molecular orbitals. The deconvolution of spectral envelopes cannot be expected to indicate unequivocally the energies of component peaks, neither can it give the relative energies of those components with any great accuracy.Even so the general ordering of intensities proposed in table 1 is in accord with the simple m.0. model. C. High-energy Peaks Emission intensity in this region of the spectrum grows remarkably as oxygen is replaced by methyl. The very weak feature observed in the S KP spectrum of sulphate can be rationalised as due to weak interaction between 4t,o orbitals and 5t, 0 2p ‘lone pair’ orbitals, 0-n mixing. The substitution of just one oxygen by methyl is sufficient to double the intensity of this little peak. The methyl group (C3”) is isoelectronic with an oxygen ligand, the e lone pairs being equivalent to e orbitals derived from C-H bonds.Thus when oxygen is replaced by methyl then 0-n mixing, which leads to the weak peak at 2469 eV, can, in principle, continue. As the carbon nuclear charge is less than that of oxygen, carbon 2p orbitals can be expected to be rather more diffuse than those of oxygen. This should enhance the possibility of S-C or S-0 0 bond interaction with C-H ‘n’ bonds, and in particular interactions between S-C 0 and C-H bonds. Such ‘ hyperconjugative ’ interaction, which would necessarily increase with the number of methyl groups, thus provides an explanation for the observed increase in relative intensity of peaks in the high-energy part of the S KP spectrum. As the amount of S 3p character ‘transferred ’ to CH bonds in this way increases, the contribution to S-C and S-0 o bonds will decrease, causing a reduction in the relative intensity of the main S KP peak (B).Ultraviolet photoelectron spectroscopy on a related molecule, dimethyl sulphoxide,22 39 F A R I1146 Electronic St r uct w e of [ (C H 3)n SO, - n] n- * indicates that molecular orbitals involved in C-H and S-C bonding have comparable ionisation energies and that these energies are ca. 3-5 eV greater than the ionisation energies of electrons in oxygen lone-pair orbitals. The ‘ hyperconjugative ’ a-7~ interactions discussed above will lead to S 3p character being found in C-H and 0 (lone pair) orbitals and thus to peaks being found in the S KP spectrum at energies corresponding to the ionisation energies of these orbitals. This is indeed the case (fig.1). Thus two peaks should be found in this region of the S KP spectrum; a higher energy peak reflecting (S-0, S-C)a-0 (lone pair) interactions and a lower energy peak reflecting (S-0, S-C)a-(C-H) interactions. The final orbital energy separation for (S-C)u-(C-H) is of the order of 2 eV, whilst for the former (S-0)a-0 (lone pair) it is nearer 5 eV. In dimethyl sulphone the two peaks have equal intensities, as might be expected, whilst for trimethyl sulphoxonium the lower energy peak, 2465.8 eV, is more than twice as intense as the peak at 2467.9 eV; and there is an overall increase in relative intensity in this region associated with the increased methyl content of the system. Conclusions The observed shift in the S KP peaks, to lower energies with increased methyl substitution, can be rationalised as due to a decrease in sulphur atomic orbital ionisation energies. This has the concomitant effect of decreasing the interaction between S 3p and ligand-based 2s orbitals so that K/?’ peaks become, relative to the main peak, less intense.Conversely methyl substitution is associated with an increase in X-ray emission at energies greater than the KP peak. An effect most simply interpreted as due to an increase in ‘hyperconjugation’ between C-H a and S-0, S-C 0 bonds. These results show that, although it neglects polarization and relaxation effects, the simple one- electron molecular model can be used successfully to interpret X-ray emission spectra and permits qualitative conclusions to be drawn concerning electronic structure.The authors thank the Royal Society, the S.E.R.C. and the Central Research Fund of London University for numerous grants for the purchase of equipment. They are also most grateful to British Coal for their support. References 1 ( a ) E. H. S. Burkop and W. N. Asaad, Adv. At. Mof. Phys., 1972, 8, 163; ( b ) E. Eyring, J. Walter and G. E. Kimball Quantum Chemistry, (Wiley, New York, 1944), p. I l k 1 16. 2 B. K. Agarwal, X-Ray Spectroscopy (Springer-Verlag, Berlin, 1979), p. 78 et seq. 3 D. S. Urch, Q . Rev., 1971, 25, 342. 4 R. Manne. J . Chem. Phys., 1970, 52, 5733. 5 D. S. Urch, J . Phys. C, 1970, 3, 1275. 6 F. P. Larkins and T. W. Rowlands, J . Phys. B, 1986, 19, 591. 7 D. S. Urch, in Chemical Bonding and Spectroscopy in Mineral Chemistry, ed. F. J. Berry and D. J. Vaughan, (Chapman-Hall, London, 1989, chap. 2. 8 D. Haycock and D. S. Urch, J . Phys. E, 1982, 15, 40. 9 K. Taniguchi, Rev. Sci. Instrum., 1983, 54, 559. 10 S. Yasuda and H. Kakoyama, Spectrochim. Acta, Part A , 1979, 35, 485. 1 1 R. Manne, M. Karras and E. Suoninen, Chem. Phys. Lett., 1972, 15, 34. 12 K. I. Narbutt, Izv. Akad. Nauk, SSSR Ser. Fiz., 1974, 38, 548. 13 M. 0. Krause and J. H. Oliver, J . Phys. Chem. ReJ Data, 1979, 8, 329. 14 R. Horn and D. S. Urch, Spectrochim. Acta, Part B, 1987, 42, 1177. 15 A. P. Sadovskii, G. N. Dolenko, L. N. Mazalov, V. D. Yumatov, E. S. Gluskin, Yu. I. Niknnorov and E. A. Gal’tsova, Isv. Akad. Nauk SSSR Ser. Fiz., 1974, 38, 606. 16 Y. Takahashi and K. Yabo, Bull. Chem. SOC. Jpn, 1969, 42, 3064. 17 A. Meisel, R. Szargan, G. Leonhardt and H. J. Koehler, J . Phys. (Paris), 1971, 32 (C4), 301. 18 D. S. Urch, Arab. J . Sci. Eng., 1988, 13, 211. 19 N. Kosuch, G. Wiech and A. Faessler, J . Electron Spectrosc. Relat. Phenom.. 1980, 20, I 1R. Foerch and D. S. Urch 1147 20 E. 1. Esmail, C. J. Nicholls and D. S. Urch, Anrrlj:vf (London), 1973, 98, 725. 21 K. Siegbahn, C. Nordling, G. Johansson, J. Hedman, P. F. Heden. K . Hamrin, U . Gelius. T. Bergmark, L. 0. Werme, R. Manne and Y. Baer, ESCA Applicd t o Frcv M o k m k s (North-Holland, Amsterdam, 1969). 22 K. Kimura, S. Katsumata, Y. Achiba, T. Yamazaki and S. Iwata. Hutidhook 01' H c I Pliofoclwtron Spectra of Fundamental Organic molecule.^ (Japan Sci. Soc. Press, Tokyo, 198 I ) . Puper 8/02657B : R c c i . i r 4 4th Juky. 1988 39-2

ISSN:0300-9599    

DOI:10.1039/F19898501139

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J. Chem. Sor., Faradajl Trans. I, 1989, 85(5), 1149-1158 Infrared Spectroscopic Studies on Carbon Dioxide Adsorption in Alkali-metal and Alkaline-earth-metal Ion-exchanged A-type Zeolites Part 1 .-General Features of CO, Interaction with A-type Zeolites Horst Forster" Institute of Physical Chemistry, University of Hamburg, Bundesstr. 45, 0-2000 Hamburg 13, Federal Republic of Germany Monika Schumann Physikalisch- Technische Bundesanstalt, Bundesallee 100, 0-3300 Braunsch weig, Federal Republic of Germany The influence of different alkali-metal and alkaline-earth-metal ions on the formation of physisorbed and chemisorbed carbon dioxide in A-type zeo- lites has been studied by F.t.i.r. spectroscopy. Various species of both kinds can be distinguished, the mutual correlations of which are determined by the exchange cation.In the case of physisorbed CO, the vz band and the 2v,/v, Fermi diads, which become i.r. active upon adsorption, can be observed, and possible sites and geometries of the sorption complex are discussed. Compared to X- and Y-type zeolites the v3 frequency shift shows a poorer correlation with the charge:radius ratio of the cation, so rendering the detection of sorption sites difficult. The general loss of rotational fine structure points to a strong hindrance to rotation. Satellite bands of the v3 fundamental band are tentatively assigned to external motions. From a multitude of papers it is well known that the sorption of carbon dioxide in zeolites yields physisorbed CO, species as well as ~arb0nate.l-l~ In the case of alkali- metal and alkaline-earth-metal ion-exchanged X and Y zeolites adsorption of CO, has been thoroughly investigated by infrared spectroscopy,l-1° but in the case of A-type zeolites there are only a few such studies, essentially dealing with NaA and NaCaA Moreover, only the behaviour of the v, vibration of physisorbed CO, has been examined, while the vl, v2 and 2v, bands were reported by only a few authors.7~8~11*12,14 To elucidate the oxidation of CO on A-type zeolites,15 CO, sorption in its lithium, sodium, magnesium and calcium forms has been studied by i.r.spectroscopy, and will be dealt with in three papers. In this first paper the general features of CO, interaction with A-type zeolites will be outlined and the formation of physisorbed CO, and carbonate compared on the alkali-metal and the alkaline-earth- metal forms.Experimental The starting material was Nal,A zeolite from Union Carbide, with the desired forms obtained by ion exchange. The lithium ion-exchanged zeolite was prepared by stirring 5 g of Nal,A zeolite in 20 cm3 of a saturated LiCl solution three times for 12 h at room temperature. In the case of the calcium form, during ion-exchange some carbonate was already developed, which might perturb subsequent investigation. It was found that the 11491150 Physisorbed Carbon Dioxide Table 1. Influence of temperature maintained during ion exchange of A-type zeolite with aqueous CaCI, solutions on the amount of carbonate formed absorbance of carbonate bands/cm-' */K 1492 1444 1435 1384 295 with C 0 2 exclusion 0.030 0.204 0.204 0.065 266 0.005 0.099 0.099 0.019 273 - 0.105 0.105 - 295 - 0.192 0.192 - 343 - 0.498 0.498 - Table 2.Unit-cell composition, designation and lattice constant of the zeolites used composition designation lattice cons tan t/ nm Na12Si12A112048 NaA 1.230 If: 0.002 Lil 1. 4Na0. GSi l Z A 1 1Z04S LiA 1.204 f 0.002 Ca.5.4Na1.2Si12A112048 CaA 1.225 f 0.002 Mg3.4Na5.2Si12A11Z048 MgA 1.226 f 0.003 temperature maintained in the course of the exchange procedure influenced carbonate formation. For a more thorough investigation, the parent zeolite (5 g) was stirred three times for 12 h in a solution of 10 g CaC1, in 50 cm3 water at four different temperatures. The amount of carbonate formed was estimated from the intensities of the bands at 1444/1435 cm-', characteristic of Ca2+-containing zeolite^.^ From table 1 it can be deduced that carbonate formation increases with temperature and that the application of a protective gas has a minor influence only.To keep the carbonate content low, ion- exchange was performed at 266 K. The magnesium-containing zeolite was prepared by stirring 5 g of Na,,A for 12 h in MgCI, solutions with increasing salt concentration (0.75, 3 and 10 g MgCI, in 20 cm3 water), in order to arrive at a higher degree of ion exchange without hydroxide precipitation. The compositions of the samples, determined by neutron activation analysis and atomic absorption spectroscopy, are listed in table 2. Carbon dioxide (99.995 %, from Linde) was used without further purification. The experiments were performed in different infrared cells, connected to a sorption pump/ion pump vacuum system.The amount of adsorbate admitted could be quantified using a dosing volume and monitoring the pressure change by a capacitance pressure meter (MKS Baratron). The glass cell (fig. l), consisting of a tube of Pyrex glass, was attached to the vacuum line via a Conflat flange. The lower part of the cell is terminated by a glass frame of 4 mm thickness, onto which two KBr windows are glued by Silvac (Curtis Associates, San Diego). The substrate wafer is inserted into a gold frame clamped to a sample holder made from glass, which contains a glass-covered iron rod, enabling movement of the sample up and down using an external magnet. The wafer can be lifted to a position above the windows for thermal treatment, via a hinged tubular oven.For running the i.r. spectra the sample is lowered into the i.r.-transmitting part of the cell. In both positions the sample holder can be fixed by glass bolts, enclosing iron rods such that the bolts can be shifted by means of a ring magnet.H. Forster and M . Schumann 1151 (0) Fig. 1. The glass cell for recording infrared spectra of adsorbed species. (a) Outer cell body with Silvac-glued KBr windows, glass-covered iron rods and Conflat flange, (6) Sample holder from glass supplied with an iron rod, (c) Gold frame. The all-metal cell, composed of two parts, is made of stainless steel (fig. 2). The outer shell consists of two concentric cylinders, welded together with an interspace evacuated for insulation.At the lower end it is fitted with two flanges, to which the KBr windows are Silvac-mounted. The internal part can be sealed by copper gaskets to the outer one and contains the sample holder surrounded by a heating coil and bored for the flow- through of cold nitrogen gas, enabling measurements to be performed between 100 and 700 K. The temperature is controlled by an iron-constantan thermocouple attached to the sample holder. The zeolite powder is pressed into self-supporting wafers of 5-8 mg cm-2 and activated at 673 K and Pa. The i.r. spectra were recorded between room temperature and 673 K on Digilab FTS 14E and 20E F.t.i.r. spectrometers and ratioed against the zeolite background. The amount of adsorbed carbon dioxide was estimated by dosing a distinct amount of gas and metering the pressure drop after adsorption.From where p c and p D are pressures in the cell and in the dosing volume, respectively, Vc and VD are cell and dosing volumes, respectively, and p is the pressure after equilibration, the amount adsorbed, nads, could be derived and hence the sorption isotherm obtained. The error was within 1 '/o up to a pressure of 133 Pa.1152 Physisorbed Carbon Dioxide -f 1 ange line f o r cool nl t rogen gas 1 anges to vacuun line (b) Fig. 2. The metal cell for recording infrared spectra of adsorbed species. (a) Internal cell part equipped with the sample holder, wrapped by a heating coil surrounded by a cylindrical container for cold nitrogen gas. (6) Outer shell with evacuated interspace between cell walls, glued infrared windows in the lower part and Conflat flanges for sealing the internal cell body and connecting the cell to the vacuum line.Results and Discussion Competition between Formation of Physisorbed CO, and Carbonate As already mentioned, carbon dioxide inside the zeolite framework forms loosely bound CO, and carbonate. The former can be discerned by bands in the vicinity of those of gaseous CO,. Carbonate species give rise to absorptions in the range 1200-1800 cm-'. Our first aim was to find out to what extent the CO, admitted is distributed between physisorbed carbon dioxide and carbonate, depending on the exchangeable cation. The formation of either species was followed by selecting one characteristic i.r. band, the intensity of which was monitored us.the amount of CO, introduced. For physisorbed CO, we chose the vg band near 2350cm-'. In case of carbonate several species were formed. The number and positions of their i.r. bands, depending on the cations, will . . L- 1 ---- : L - A :- A,.4,.:1 :- n,...t -c +L:- ,,...:,, u,,, +LA A ---.-- "-A- .-+A ,,,t --,. 4,. ._.-- followed by the change of the absorbance of the most intense band in each case. Normally, two isotherms were measured subsequently. Between the two runs discussed the zeolite samples were evacuated at room temperature only in order to remove the physisorbed gas. In contrast to the results of Delaval and Cohen de Lars', we found that chemisorbed and physisorbed CO, influence each other, depending on the cation and the temperature of exposure to the substrate.H .Forster and M . Schumann 1153 A A A A A I , I I I 0 0.2 0,4 0.6 0.8 1.0 molecules per unit cell Fig. 3. Initial stages of the carbonate formation with CO, loading in A-type zeolite detected by the development of the 1600 cm-' band at room temperature. Run 1 (A) was extended to higher pressure (not shown in the figure). After evacuation (at room temperature) run 2 (@) indicates the limit for carbonate production. Alkali-metal Ion-exchanged Zeolites In case of the alkali-metal ion-exchanged NaA and LiA zeolites both species are observed at room temperature. Initially only carbonate was formed. After admission of ca. 0.2 molecules CO, per unit cell in NaA zeolite and 0.05 in the case of LiA zeolite the bands of the physisorbed carbon dioxide also emerged, proving that in the case of LiA zeolite carbonate formation was largely suppressed.The intensity of the bands of the two species increased with further loading until maximum adsorption was reached. After evacuation in the second loading run only physisorbed CO, was formed, while the intensities of the carbonate bands remained constant. This implies that only a limited amount of carbonate can originate inside the cavities of alkali-metal ion-exchanged A- type zeolites. In case of NaA zeolite the maximum amount of carbonate formed was estimated by extrapolation of the initial curve to approximately 0.3 carbonate molecules per cavity (fig. 3). Alkaline-earth-metal Ion-exchanged Zeolites On the alkaline-earth-metal ion-exchanged forms CaA and MgA only physisorbed CO, was found at room temperature.The absorbance of the v, band develops linearly with the amount adsorbed and the isotherms of the first and the second run are nearly congruent. Carbonate was formed when higher temperatures were used ; its formation started at ca. 323 K and increased steeply up to 623 K, as has been shown by Gallei and S t ~ m p f . ~ There was no indication of a limit to carbonate formation over this temperature range. In the case of MgA zeolite a small amount of carbonate was already formed at room temperature, probably due to the Na+ content, as this zeolite was only partly Mg2+ ion-exchanged. Here the isotherms of the consecutive runs lay close together, showing a conspicuous similarity to LiA zeolite, which may reflect the diagonal relationship of these two elements.1154 Physisorbed Carbon Dioxide 4nOO w avenumtm/cm-' 10 Fig.4. M.i.r. spectrum of carbon dioxide sorbed at room temperature in CaA zeolite. Physisorbed Carbon Dioxide In accordance with the D,, point group, carbon dioxide has four fundamentals: the symmetric stretching vibration v,, the doubly degenerate bending vibration v, and the asymmetric stretching vibration v3. Owing to symmetry, v, and v3 are only infrared active, whereas v1 is only Raman active. v, and the Raman-active 2v, vibration are in Fermi resonance with each other, therefore their eigenfunctions are mixed and an unambiguous assignment is no longer possible. In contrast to former interpretations,16 in the gas phase the high-frequency Fermi diad is attributed to 2v, and the low-frequency band to ~ , ." 3 ' ~ This assignment is reversed for CO, under high pressure and in liquid, solid or aqueous phases,lg and might be also applied to the adsorbed state. Additionally the combination vibrations v,+v, and v3+2v, can be observed in the i.r. spectrum of gaseous CO, (fig. 4). From the analysis of the i.r. spectra of adsorbed CO, inside different forms of A-type zeolite information can be inferred on the geometry of the sorption complex, the sorption site of CO,, the mobility in the adsorbed state and external and internal bonding of the adsorbate. Furthermore, we can obtain indications of the influence of different exchanged ions on the degree of Fermi resonance. These results will be discussed in Part 3 of this series.The frequencies of the bands of physisorbed CO, are listed in table 3, from which the following results can be drawn: the v, and 2v, vibrations become infrared-active in all samples under investigation ; their intensity ratios as well as their frequency shift depend on the exchanged cation; the v3 band is accompanied by satellite bands; in the case of the v3 and v1 fundamental the bands are shifted to higher wavenumbers and in the case of the v, and 2v, vibrations to lower wavenumbers, compared to gaseous CO,. In the LiA, CaA and MgA zeolites only one single adsorbed species could be clearly identified (species A). On NaA zeolite two species (species A and B) could be observed. On LiA and CaA zeolites a second but weaker v3 band is clearly observable, for which the corresponding v,, v, and 2v, absorptions are missing.Assuming that this band refers to1155 H . Forster and M. Schumann Table 3. Band positions and frequency shifts with respect to the gas phase (v- vgJ of physisorbed C02 (in cm-') vibration gas NaA LiA CaA MgA v, + 2v2 v3 + v1 v3 satellite v3 A v3 B v3 c v, satellite 3716 3609 - 2349 - - 1388 (R) 1285 (R) - 667 - 3726 3610 2468 2434 236 1 235 1 2293 2286 1246 1240 1383 648 6521648 - + l o 3732 +16 3720 3603 + 119 2399 +50 2453 +12 2364 +15 2368 - + 1 - +85 - +2 - - - - - - 2347 -2 2362 -56 2297 -52 2301 -63 - - - -142 1259 -129 1257 -148 - - - +98 1383 +98 1383 -19 650 -17 648 -19 - - - + 4 -6 + 104 + 19 + 13 - 48 - - - - 131 - + 98 - 19 3728 +12 3614 + 5 2442 +93 2364 +15 - - - - 2297 -52 - - 1269 -119 - - 1381 +96 646 -21 a second species it will be tentatively labelled as species C; although the MgA zeolite was only 57 O h ion-exchanged, separate bands attributable to two different CO, sorption complexes, either interacting with Mg2+ or Na+ (as has been found with other adsorbates like CO or butenes20*21) were not observed.Geometry of the Sorption Complex The CO, symmetry is lowered upon adsorption. Depending on the kind of orientation, different symmetries of the sorption complex are to be expected. Assuming (i) a linear (i.e. not bent) CO, molecule and (ii) a location near the cation, the symmetry should be C,, in a linear, C,, in a T-shaped arrangement with a cation or C, on a site between the cation and the zeolite lattice. C,, symmetry should be discernible from the vl fundamental becoming i.r.-active.Further symmetry reduction should be distinguishable from the splitting of the v, band into Al+B, species in the case of C,, or into 2A' species in the case of C, symmetry. In the adsorbed state the v1 and 2v, bands become i.r.-active on all samples under investigation, which may be interpreted as CO, linearly adsorbed in front of the sorption site. In case of NaA zeolite two physisorbed CO s ecies can be distinguished, as has been already found by Delaval and Cohen de La:a.' However, in our investigation in addition the two v3 bands are paralleled by two v, and 2v, bands. In case of the v, fundamental, fortuitously absorbing in a 'window ' of the very intense absorption of the zeolite skeleton, at a low coverage of ca.one CO, molecule per cavity, a broad band arose with a maximum at 648 cm-l. With increased loading a shoulder developed on the high-energy side. From the difference spectrum two bands at 648 and 652 cm-l can be distinguished (fig. S), for which two interpretations are possible: (1) both bands are single bands belonging to different species and (2) the 648 cm-' band is the unsplit band of one C,, species. Owing to lower symmetry, the second species has a split v, band, one absorbing at 652 cm-l, while the second cannot be resolved from the 648 cm-' absorption of the former associated with a C,, species. From a broadening of the low- frequency side at high coverage, at present the second interpretation may be preferred. Concerning the geometry of the sorption complex, a linear arrangement of the CO, molecule with the cation in C,, symmetry is expected in accordance with the negative sign of its quadrupole moment, if the interaction with the electric field gradient1156 Physisorbed Carbon Dioxide 0.Q 8 B e 2 % 0 A 0.2 0 B 600 700 600 700 w avenumber/cm- wavenumber/cni’ Fig.5. (A) v 2 fundamental band at low (a) and high CO, coverages (b) in zeolite NaA. (B) Difference spectrum (a)-(b). contributes to physisorption. Thus although experimental proof is sparse, different geometries can also be taken into account, e.g. an interaction via both oxygen atoms with two cations or with one cation and the wall of the zeolite cavity. This linear arrangement seems to be true with all species including species A in the case of the NaA zeolite, while in the case of species B, because of the observed splitting of the v, fundamental, a lower symmetry may be assumed, i.e.a T-shaped complex or adsorption between the cation and the framework exists. Sorption Sites of Physisorbed CO, From the number of i.r.-active vibrations the manner in which the CO, molecule is attached to its sorption site can be derived. Questions then arise as to the type of adsorption site. Owing to the electronic structure, CO, is a hybrid of electron donor-acceptor properties and therefore an interaction with the adsorbent is possible via the oxygen as well as via the carbon atom. Carbonate, for example, will be formed, if the carbon atom is attached to framework oxygen, as will be discussed in Part 2 of this series.For physisorbed CO,, preferred sorption sites in front of the cations may be assumed. In this case the frequency shift should correlate qualitatively with the charge : radius ratio, characteristic for the polarizing power of the cation, as has been found with CO and n-butenes in different A-type zeo1ites.20*21 For the related faujasites, a linear dependence of the CO, v, frequency on the electric field strength has been observed.,’ The degree of the displacement is different for X and Y zeolites, and was explained assuming not only the cations but also the zeolite lattice in total was responsible for the frequency shift. Additionally the approach of the cations by CO, could be probed for transition-metal ion-exchanged A-type zeolites from the alteration of the coordination sphere indicated in the corresponding u.v.-visible spectraz3 upon interaction with CO,.In the present case the i.r. bands of physisorbed CO, are shifted up and down scale according to the vibrational mode (see fig. 6), but are in less good correlation with the charge:radius ratio when compared to the analogous faujasite data. In the case of polyatomics the correlation of the frequency shift with the strength of interaction is inH . F6rster and M . Schumann 2380 2370 - 2360 - 2350 d I I I Mq Na L i Ca 1157 1290 1280 (4 1270 1260 1250- 1240; 1230 I I I I Na L i Ca Mg 5 (b) 6 1390-, . . . 1380- Mq 1370 1 I I I NJ L i Ca I (4 660 - 650 - 640 I I I Mn , Nd L i Ca I ' Fig. 6. Attempt to correlate the v3 (a), 2v, (b), v1 (c) and v, ( d ) band shift of CO, with charge:radius ratio (expressed as q / r ) of the cation.0 10 20 30 40 50 general more complicated, compared to the behaviour of simpler diatomics or quasi- diatomics and will be understood only by quantum-chemical calculation supplemented by vibrational analysis.24 The results of these calculations on a system modelled by a CO, molecule adsorbed in front of a Ca2+ ion surrounded by a six-ring cluster of 37 zeolite framework atoms were as follows. An electronic charge transfer takes place from the molecule via the Ca2+ ion to the zeolite framework. The equilibrium distance between the Ca2+ ion and the resultant positively charged CO, molecule is 225 pm. The CO, molecular axis lies parallel to the threefold axis of the oxygen six ring.The most significant feature is a bond-bond charge transfer within the adsorbed molecule from the bond adjacent to the cation to the terminal bond. The last case could be independently proved by normal-coordinate analysis. The force constant of the bond between the carbon atom and the terminal oxygen farthest from the cation increases compared to that in the free molecule, while the corresponding force constant of the C-0 bond adjacent to the cation decreases. There is an excellent correspondence between the observed and calculated frequencies. Therefore the frequency shifts upon adsorption seem mainly due to the internal charge redistribution in the adsorbed molecule. Satellite Bands of the v3 Fundamental On the zeolite samples investigated all bands of adsorbed carbon dioxide are sharp bands, having lost their rotational fine structure, which points to a strong hindrance to rotation.The v3 band is always accompanied by two small satellites at nearly equal distances with respect to the main band, as has been previously o b ~ e r v e d . ~ ~ 11* l2 While the low-frequency (LF) satellites are narrow and sharp, the high-frequency (HF) satellites are broad. Further, the change of their intensity with the amount adsorbed parallels that of the main band. (In case of NaA zeolite at high coverages two additional shoulders could be observed, pointing to a second species.) From this it must be concluded that the same species is responsible for the appearance of the triplet.1158 Physisorbed Carbon Dioxide Compared to the main band, the LF satellite is displaced by an amount equal to the isotope shift of gaseous 13C0, and therefore can be assigned to the v, band of physisorbed 13C02, present in carbon dioxide in natural abundance (table 3).Owing to their spectral pattern, the satellites have been discussed elsewhere as originating from motion external to the molecule, being combinations of the v3 fundamental and the frustrated translation with respect to the zeolite surface.,, l1 During detailed investigation of CO physisorption in A-type zeolites, the appearance of a similar pair of satellites has been observed, which most likely could be assigned (with the support of interaction potential calculations) to librational transitions," although it remains questionable whether translation-libration coupling is also involved. In comparison with these findings the broad HF satellite may be similarly attributed, while the LF counterpart may be masked by the v3 band of 13C0,.The support given by the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie is gratefully acknowledged. References 1 M. Hair, Infrared Spectroscopy in Surface Chemistry (Marcel Dekker, New York, 1967). 2 L. Bertsch and H. W. Habgood, J . Phys. Chem., 1963, 67, 1621. 3 J. W. Ward and H. W. Habgood, J. Phys. Chem., 1966, 70, 1178. 4 C. L. Angell and M. V. Howell, Can. J . Chem., 1969, 47, 3831. 5 E. Gallei and G. Stumpf, J. Colloid Interface Sci., 1976, 55, 415. 6 C. Mirodatos, P. Pichat and D. Barthomeuf, J. Phys. Chem., 1976, 80, 1335. 7 P. A. Jacobs, F. H. van Cauweleart, E. F. Vansant and J. B. Uytterhoeven, J. Chem. Soc., Faraday 8 P. A. Jacobs, F. H. van Cauweleart, E. F. Vansant and J. B. Uytterhoeven, J . Chem. So"., Faraday 9 0. A. Sinitsyna, I. F. Moskovskaya, A. A. Kubasov and K. V. Topchieva, Vest. Mosk. Univ., Ser. 2, Trans. 1, 1973, 69, 1056. Trans. 1, 1973, 69, 1973. Khim., 1977, 18, 652. 10 L. Kh. Tas-ool, A. A. Kubasov and K. V. Topchieva, Kinet. Catal., 1978, 19, 200. 11 H. Forster, M. Schuldt and R. Seelemann, Z. Phys. Chem. N.F., 1975, 97, 329. 12 Y. Delaval and E. Cohen de Lara, J . Chem. Soc., Faraday Trans. 1, 1981, 77, 869; 879. 13 T. Masuda, K. Tsutsumi and H. Takahashi, J . Colloid Interface Sci., 1980, 77, 232. 14 Y. Delaval, R. Seloudoux and E. Cohen de Lara, J . Chem. Soc., Faraday Trans. 1, 1986, 82, 365. 15 H. Forster, W. Frede, G. Peters, M. Schumann and U. Witten, J . Chem. SOC., Chem. Commun., 1981, 16 G. Herzberg, Infrared and Raman Spectra of Polyatomic Molecules (Van Nostrand, New York, 1945), 17 G. Amat and M. Pimbert, J. Mol. Spectrosc., 1965, 16, 278. 18 H. E. Howard-Lock and B. P. Stoicheff, J . Mol. Spectrosc., 1971, 37, 321. 19 J. F. Bertran, Spectrochim. Acta, Part A , 1983, 39, 119. 20 H. Bose, H. Forster, W. Frede and M. Schumann, in Proc. 6th Int. Zeolite Conf., Reno, 1983, ed. D. Olsen and A. Bisio (Butterworths, Guildford, 1984), p. 201. 21 H. Forster, and R. Seelemann, J. Chem. Soc., Faraday Trans. 1, 1979, 75, 2744. 22 C. L. Angell, J. Phys. Chem., 1966, 70, 2420. 23 H. Forster and U. Witten, Zeolites, 1987, 7 , 517. 24 M. Grodzicki, 0. Zakharieva-Pencheva and H. Forster, J . Mol. Struct., 1988, 175, 195. 1064. p. 178. Paper 8/02697A; Received 5th July, 1988

ISSN:0300-9599    

DOI:10.1039/F19898501149

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J . Chem. Soc., Faraday Trans. 1, 1989, 85(5), 1159-1 168 Surface-enhanced Raman Spectroscopy of NAD+ and related Compounds Janina C. Austin and Ronald E. Hester" Chemistry Department, University of York, Heslington, York YO1 5DD Surface-enhanced Raman spectra (SERS) are reported for nicotinamide adenine dinucleotide (NAD', oxidised form) adsorbed on colloidal silver at concentrations between lop3 and mol dm-3, and using various excitation wavelengths. Adenosine diphosphate (ADP) gave an SER spectrum similar to that of NAD+, except for the absence of the nicotinamide SER band observed for NAD' at 1030cm-'. NADH (reduced form of NAD') did not give an SER spectrum at the alkaline pHs necessary to prevent decomposition. No SER spectrum was obtained for NAD' when bound to the enzyme GAPDH; the NAD' is too deeply buried in the protein for effective surface enhancement.The applicability of the surface selection rules arising from the electromagnetic enhancement mechanism to determine NAD' orientation at the silver surface is discussed in the light of the probable contribution to SERS intensity from a charge-transfer mechanism. While the selection rules do not, in this case, appear useful for determining NAD' orientation at silver, wavenumber shifts do indicate surface interaction with N, of adenine. The general surface geometry of NAD' adsorbed on silver at low (< mol dmP3) concentrations is deduced as being extended, with phosphate groups binding directly to silver, and ribose, nicotinamide and adenine moieties all in close proximity to the surface.The coenzyme nicotinamide adenine dinucleotide (NAD', oxidised form) is essential to a large class of biological redox systems. Many techniques have been employed in the study of the structure and bonding of this coenzyme, both when complexed with metal ions and when bound to enzymes.' Raman and resonance Raman (RR) spectroscopies have recently yielded new information on the conformation and bonding of NAD+ and NADH (reduced coenzyme) when bound to the dehydrogenase enzymes alcohol dehydrogenase2. and glyceraldehyde-3-phosphate dehydrogenase3 (GAPDH). The vibrational bands in the normal (off-resonance) Raman (NR) spectra of the free coenzymes have been analysed in terms of the contributions of the adenine, nicotinamide, ribose and phosphate m~ieties.~ The vibrational spectra of adenine and various adenine derivatives have also been interpreted using normal-coordinate approaches.5 7 SERS has proved to be an excellent tool for obtaining the vibrational spectra of molecular species at very low concentration. The haemproteins cytochrome c, myoglobin,' haemoglobins and cytochrome P-450'. lo have all been successfully studied as adsorbates on silver surfaces by SERS in combination with resonance enhancement, i.e. SERRS, enabling detailed vibrational analyses of the prosthetic groups to be made with haem concentrations of 10-6-10-9 mol dmp3. The flavoproteins have not shown such a large degree of surface enhancement," since the flavin appears to be unable to interact directly with the silver surface due to shielding by the protein. Silver colloids appear to be the most convenient, effective, and least denaturing medium [depending on the exact preparation, see ref.(S)] for SERS, although silver and gold electrodes lend themselves particularly well to studies of electrochemical reduction or oxidation. Without the extra resonance enhancement afforded by a visible 11591160 SERS of NAD' I I 700 1000 1300 1600 3000 3400 wavenumber/cm- Fig. 1. (a) NR spectrum of NAD' (50 mmol dm-3, pH 6.0), (6) SER spectrum of NAD' ( I mmol dm-3), (c) SER spectrum of ADP ( 1 mmol dmP3). Acetone (20% v/v) peaks are marked (+). chromophore, SERS can be used to provide information on vibrational modes of species at sub-millimolar concentrations. SERS of amino a c i d ~ , ' ~ , ' ~ DNA14 and DNA bases13 have been reported, and orientations of these species at the metal surface have been suggested.From a SERS study of NAD+ at gold and silver electrodes15 the orientations of this molecule with respect to the gold or silver surface also have been suggested. The information on molecular orientation has been deduced on the basis of surface-enhancement selection rules; these determine which vibrational modes of a molecule will be surface-active in a particular molecular orientation. The theoretical and experimental bases for these rules have recently been reviewed,16 and it has been shown that these are, in general, considerably more complex than the corresponding rules for infrared reflectance spectra. In this paper we demonstrate that the commonly used simplification of these rules may give incorrect information on surface orientation.We point out the difficulty in determining surface orientation from SERS data where the relative intensity contributions from the molecular resonance or charge-transfer (CT) mechanism and the electromagnetic enhancement mechanism are unknown. The possibility of obtaining SERS of NAD+ when bound to the enzyme GAPDH has been investigated. The burial of NAD+ deep within this dehydrogenase enzyme, as seen in its X-ray ~tructure,'~ has been shown to preclude SERS enhancement.J. C. Austin and R . E. Hester 1161 500 550 600 650 excitation w avelengthhm Fig. 2. Raman Excitation Profiles of SER bands of NAD': (a) 730 cm-' band, (b) 1330 cm-I band, (c) 1030 cm-' band. Experimental GAPDH, NADH, NAD+ and adenosine diphosphate (ADP) were purchased from Sigma and used without further purification. The NAD+ content of GAPDH was determined spectroscopically from its ratio of absorbance at 280 and 260 nm, A28,,/A260.18 Silver colloids were prepared by reduction of a boiling solution of silver nitrate (90 mg in 500 cm3 water) by trisodium citrate (10 cm3 of a 1 % w/v solution), as in the method described by Lee and Meisel." Deionised or doubly distilled water was used for all colloid preparations.The silver sols were murky grey in appearance, with strong absorption maxima at 405410 nm, and were stable for many months. The sols darkened in colour on addition of the SERS adsorbate and potassium nitrate, These sols gave broad absorption maxima at 55&650 nm in addition to the maximum at 405410 nm, the former band being attributable to aggregation." The pH of the sol was adjusted as necessary by addition of potassium hydroxide solution.For measurement of spectra at different NAD' concentrations, and at different excitation wavelengths, acetone was added to 20 YO v/v as an internal intensity standard. All NAD' Raman band intensities were normalised to the acetone peak at ca. 800 cm-'. Further corrections were made for differential spectrometer response at wavenumber values far from 800 cm-'. which involves measurement of the intensities of the v, and v,, bands of pyridine. For these measurements samples were made up in the normal manner in the presence of a low concentration (ca. lo-* mol dm-3) of pyridine.Raman spectra were obtained using laser excitation from either an Ar+ (Spectra Physics model 2025) or a Kr+ (Spectra Physics model 170) laser. Raman-scattered light was dispersed and detected using either a Jobin-Yvon Ramanor HG2 double monochromator with cooled photomultiplier tube and photon-counting electronics, Estimates of surface potential were made according to the method of Wetzel et1162 10- SERS of NAD+ a 0 0 0 n Y .r. c .CI t. m .B 1 10- 0 1 1 2 3 4 5 6 NAD c~ncentration/lO-~ mol dm-3 Fig. 3. Concentration dependence of (a) 1330 and (b) 1030 cm-' SER bands L,; NAD+. controlled by a Nicolet 1074 computer, or a Spex 1403 double monochromator with cooled photomultiplier tube and photon-counting electronics controlled by a Spex SCAMP computer.In all cases, 90" illumination was used, with the sample in a rotating cell. Laser powers of up to 100 mW at the sample were used, with a spectrometer bandpass of ca. 6 cm-'. Results The NR spectrum of NAD+ (50 mmol dm-3) is compared with the spectra of NAD+ and ADP (both at 1 mmol dm-3) on colloidal silver in fig. 1. No Raman signal could be obtained for NAD+ alone at 1 mmol dm-3 under comparable conditions; thus the spectra in fig. 1 (b) and (c) are assumed to be entirely the SER spectra. The NAD+ SER spectrum shows particular enhancement of bands at 730, 1030 and 1320-1340 cm-l with weaker bands at 620, 790, 820, 925, 955, 11 15, 1244, 1399, 1463, 1509, 1570 and 2940 cm-l. Broad features at 620 and 925 cm-' were not observed in all spectra; these bands were attributed to the sol itself.The 132&1340 cm-' band in the SER spectrum is broad and consists of two unresolved components, one at ca. 1325 cm-' and one at 1335 cm-'.J . C. Austin and R. E. Hester 1163 I 700 900 1100 1300 1500 w avenumber/cm-' Fig. 4. SER spectrum of NAD' (2 x mol dm-3). Acetone (20% v/v) peaks are marked (+). The 1378 cm-' band in the NR spectrum of NAD+ is not prominent in the SER spectrum. The pHs of NAD+- and ADP-containing sols were in the range 4.5-6. At this pH, the surface potential of the silver was estimated (see Experimental section) to be in the range -0.1 to -0.3 V US. SCE. At pH 9.5 the surface potential dropped to Fig. 2 shows the dependence of the SERS band intensities on excitation wavelength for the 730, 1030 and 1330 cm-' bands of NAD+ (Raman-excitation profiles).No significant changes in the relative intensities of any other NAD+ bands were observed over this wavelength range; i.e. all bands showed a similar wavelength dependence to that of the three bands shown in fig. 2. Fig. 3 shows the dependence of the SER band intensity on NAD+ concentration for the 1030 and 1330 cm-l SER bands of NAD+. The 1330 cm-l band shows a general decrease in intensity with decreasing concentration, and the intercept is non-zero. A similar decrease in intensity with concentration was observed for the 730 cm-' band, the corresponding plot (not shown) having the same intercept value. Fig. 4 shows an SER spectrum of NAD+ at very low (2 x mol dm-3) concentration. NADH did not give comparable SER signals at millimolar concentrations on silver.At neutral pH, weak SER bands were observed at ca. 730 and 1330 cm-'. At higher pHs (pH > 7.5) little or no SER signal was observed. NAD+ gave no SER spectrum when bound to GAPDH at concentrations of 10-4-10-5 mol dmP3, although good spectra could be obtained at these NAD+ concentrations in the absence of the enzyme. At higher enzyme concentrations, the sols precipitated. -0.55 V US. SCE. Discussion The SER spectrum of NAD+ shows strongest enhancement of adenine in-plane ring vibrations at 730 and 1330 cm-'. The aliphatic C-H stretching vibrations (2940 cm-')1164 SERS of NAD+ of the ribose moiety are moderately enhanced, whereas the aromatic C-H stretching vibrations (expected at ca. 3080 cm-') are not. The 1030 cm-' band in the SER spectrum of NAD+ [fig.1 (b)] is far weaker than the corresponding band in the NR spectrum, in which it has been assigned to a vibration of the oxidised nicotinamide moiety.' Since the SER spectrum of ADP shows no similar band at 1030 cm-', we may assume that the band at 1030 cm-' in the NAD+ SER spectrum is due to nicotinamide and is not a coincidentally close adenine band. The wavenumber changes observed in the NAD+ spectrum on adsorption onto the silver surface [cf. fig. l(a) and (b)] are qualitatively similar to the changes observed4 in the NR spectrum of NAD+ when the pH is lowered below 4. The N, of adenine becomes protonated below pH 4 (pK, 3.8,,); this is assumed to be the cause of the spectral changes. However, as previously stated, the pH of the NAD+-containing sol is above pH 4.5; thus the changes are unlikely to be due to adenine N, protonation.The fact that the spectrum in fig. 1 (h) is similar to the protonated NAD' spectrum may indicate strong interaction between the silver surface and the adenine ring, with a particular involvement of N,. The SER spectrum of NAD+ on colloidal silver is substantially different from that reported for NAD+ adsorbed on a silver electrode at potentials between 0 and - 1.0 V us. SCE;15 this showed strong bands only at 735 and 1335 cm-l. No enhancement of the nicotinamide band at 1030 cm-' was reported and other features were weak. Since our value for the silver surface potential (- 0.1 to - 0.3 V us. SCE) is well within the range used in the electrode study, a simple potential-induced re-orientation cannot be used to explain the differences between colloid and electrode spectra.Indeed, in the same electrode study, a 1030 cm-' nicotinamide band was observed when gold electrodes were used. Clearly, the nicotinamide ring approaches more closely, and possibly with a different orientation, to the colloidal silver surface than to the silver electrode surface. The results from the electrode study were interpreted as showing a close contact between adenine and the silver surface, with the adenine ring in a perpendicular orientation, binding to the surface via its NH, and N, nitrogens. This contradicts a surface reflectivity study which was interpreted in terms of a flat orientation of the adenine ring.23 The basis for the assignment of a perpendicular orientation of adenine on the silver electrode was a simplified view of the surface electromagnetic enhancement selection rules; namely, that only vibrations in a plane containing the surface normal, z , are enhanced.Since both strong bands observed in the electrode study were in-plane ring vibration^,^^^ it was assumed that the plane of the ring must be normal to the surface. However, we shall show that the surface selection rules cannot generally be interpreted in such a straightforward manner. The surface selection rules arising from the electromagnetic enhancement theoryl6. 24-26 state that with excitation at, or to the red of the dipole resonance maximum of the silver particle (or surface roughness feature in the case of electrodes), I vibrations with components of polarisability ax, and a,, will be enhanced by a factor of up to IcI2, while bands from the polarisability component a,, will be enhanced by a factor of up to compared with those vibrations with only azy, axx or a,, components, where E is the effective dielectric constant of the metal relative to the surrounding medium (the difference in values of E at incident and scattered wavelengths being neglected).In the case of adenine (C, symmetry, molecule in xy plane), the totally symmetric (in-plane) modes may contain a varying degree of a,, character. Considering a vibration involving the whole Ic-electron system, i.e. a vibration of the C-N skeleton, there might be expected to be a small degree of a,, character out of the ring plane as the n-electron distribution expands and contracts over the ring.An exocyclic vibration, e.g. C-H or C-NH,, involving only a-type electrons, would contain far less uzz character. Accordingly, adsorption of adenine flat onto a surface would result in the C-H in-plane modes in particular not being enhanced, whereas C-N skeleton modes may be enhanced to varying extents. Such a treatment has been developed for benzene andJ. C. Austin and R . E. Hester 1165 extended to adenine SERS by Suh and Moskovits.13 Out-of-plane adenine modes would, of course, be more enhanced, but still might be weak in the SER spectrum, as they are very weak in the NR spectrum. Conversely, adsorption perpendicular (or partially so) to the surface would result in the in-plane modes, especially C-H modes, becoming much more enhanced (as the bulk of their polarisability derivative would become a,,).Out-of-plane modes would show no enhancement. The data presented in the electrode study unfortunately do not show the high- wavenumber region where, if the adenine moiety of NAD+ is indeed perpendicular to the surface, C-H stretching modes at ca. 3080 cm-’ would be expected to be strongly enhanced. The observation of enhancement of only in-plane ring modes does fit with the electromagnetic surface-enhancement selection rules, but it is not the only possible explanation. Indeed a totally different explanation of a similar spectrum, that of adenine itself on a silver colloid, has been propo~ed.’~ Using the same selection rules, adenine was deduced to be adsorbed parallel to the surface.The main evidence for this flat orientation was the observation that the C-H mode at ca. 3080 cm-’ was not enhanced. However, out-of-plane modes also were not enhanced in this SER spectrum, contrary to expectations. The 739 and 1339 cm-’ bands were strongly enhanced and, although they both arise from similar vibrations involving the C-N skeleton in the NR spectrum, the authors reassigned the 739 cm-l band to a coincidentally close strongly enhanced band. The new assignment of this band to a coupled NH, deformation and ring vibration seems rather unnecessary, the amount of a,, character in these ring vibrations being unquantified. It might be expected that, similar to the example of benzene,27 these modes might contain sufficient az, character to be considerably enhanced when the molecule is adsorbed flat on to the surface.Complicating the picture is the wavelength dependence of the electromagnetic enhancement selection rules. At wavelengths to the blue of the dipole resonance maximum, modes containing axx, a,? and axy polarisability components may become more enhanced, and indeed the situation can be reached where the rules are completely reversed. The lack of major changes in relative intensities of the NAD+ SER bands with excitation wavelength implies that this ‘ reversal ’ situation is not reached for silver sols using excitation wavelengths greater than 457.9 nm, or that SER intensity from other enhancement mechanisms is dominating the excitation profile. Neither of the two studies mentionedl3,l5 have made use of the surface resonance Raman (RR) enhancement theory to explain their results.Briefly, an A-term RR mechanism is thought to be the most likely contributor to any surface RR enhancement. l6 Surface enhancement by an A-term RR mechanism would selectively enhance totally symmetric vibrations in the ‘chromophoric’ part of the molecule; in particular, those vibrations along whose normal coordinates the molecule would relax if the (virtual) electronic transition involved in the RR process were real. There appears to be no direct evidence for surface-orientation selectivity in enhancement by this mechanism, in contrast to enhancement by the electromagnetic enhancement (EM) mechanism. The electronic transition most likely to be active in the surface RR process is a metal + ring n* charge transfer.This enhancement mechanism is consequently often called the charge-transfer (CT) mechanism to avoid confusion with the usual RR phenomenon; this title will be used here. It should not be confused with strong ‘first- layer’ EM enhancement arising from chemisorption, which was proposed as a possible explanation for the particular enhancement of the adsorbate molecules in the first monolayer, but has recently been dismissed due to lack of supporting experimental evidence.2s According to the CT enhancement mechanism then, vibrations of the ring skeleton could be enhanced, whereas exocyclic vibrations (e.g. C-H) could not. It is likely that the CT contributions to the SER spectra of NAD+ presented here, and to the SER spectra of NAD+ and adenine previously reported,15* l3 are large. Recent estimates of the size of CT enhancement in SERS have varied.An SERS study of pyridine1166 SERS of NAD+ adsorbed on a silver-rhodium substrate attributed an absolute enhancement factor of 15-65 to the CT me~hanism.~' Another silver island film study estimated a CT enhancement factor of lo3, compared with a factor of 10-102 from an EM mechanism.30 The observation of good SER spectra at very low NAD+ concentrations (fig. 3 and 4) is in itself evidence for a significant contribution from a CT enhancement mechanism. Such strong enhancement is unlikely to arise from the EM mechanism alone. In addition, the observation of enhancement of only ring modes (not exocyclic C-H modes) of the adenine moiety is in harmony with a CT mechanism.A large CT enhancement contribution would also adequately explain both previous studies of the adenine moiety at silver, similar enhancements of adenine modes being observed. Particular evidence for the presence of a CT mechanism arises from studies of SER band intensity variations with electrode potential. At strongly negative potentials the metal- donor level is raised closer to the ligand-acceptor level. Thus SER CT bands are able to gain intensity with lower-energy red excitation as the electrode potential is made more negative. The electrode study of NAD+, which used red excitation, does in fact appear to show an increase in band intensities as the electrode potential is lowered from 0 to -0.6 V US.SCE. After this point, the spectra start to lose intensity, probably due to counteracting desorption effects. It is clear that unless the CT mechanism can somehow be dismissed as not contributing to the SER intensity, the orientation of adenine on silver cannot be determined with certainty. Our low-concentration ( mol dmP3) data point to some contribution from a CT mechanism, as does the strong selective enhancement of just two adenine totally symmetric modes. Thus the surface selection rules provided by the EM mechanism cannot be used in isolation to deduce the orientation of the adenine ring with respect to the silver surface. The evidence cited previously for specific binding of the adenine moiety to the silver surface via N, may imply a tilted geometry with respect to the surface. A flat geometry would be expected also to affect the vibrational modes involving N,, N, and the exocyclic amino group, for which the spectra present no strong features.The data shown in fig. 3 shed some light on the nature of the relative proximity to the surface of the adenine and nicotinamide rings. The adenine band intensities show a clear concentration dependence [fig. 3(a)], but the nicotinamide band [fig. 3(b)] does not. The only adequate explanation for this observation is that the NAD+ molecule undergoes reorientation as the concentration is lowered. At high concentrations the nicotinamide ring is further away from the surface than the adenine, and so its Raman spectrum is relatively weakly enhanced. As the concentration is lowered, the nicotinamide moiety is able to approach closer to the surface and its SER band becomes more enhanced. This counteracts the loss of signal due to the lowering of concentration, resulting in the apparent independence of the 1030 cm-' band intensity on concentration [fig.3 (b)]. Also of interest from fig. 3 is the non-zero intercept value (see Results section). This indicates that only molecules giving relatively weakly enhanced Raman bands are being lost over the concentration range 10P3-10-5 mol dmP3 and that these are the less strongly bound NAD+ molecules. This is consistent with previous observations of just a few very strongly enhancing sites on the silver surface amongst many moderately enhancing ~ i t e s . ~ ' The observation of no SER enhancement for NADH at alkaline pH is difficult to explain at first sight in terms of adsorption via adenine and nicotinamide. Above pH 4 the adenine moiety carries no charge in either NAD+ or NADH.The nicotinamide moiety is positively charged in NAD+, but uncharged in NADH. The surface of the colloidal particles has a distinctly positive character at pH 4-5 (PZC for silver is ca. -0.9 V US. SCE), but loses its surface charge at alkaline pH (see Results section), presumably due to the formation of coordinated hydroxide. Superficially, NADH might be expected to adsorb as readily to the silver surface as NAD+, giving a similar SER spectrum, excepting the 1030cm-' nicotinamide band. In order to account for theJ . C. Austin and R. E. Hester 1167 observed behaviour, we propose that the primary binding of NAD' and NADH to silver is via the negatively charged phosphate groups, with only weaker binding via adenine.At alkaline pH, the reduction of the positive charge on silver inhibits adsorption of NADH, adenine binding alone being too weak for effective adsorption to take place. Coordination via the phosphate groups also can explain the strong enhancement of the ribose C-H vibrations. With adsorption uia phosphate, the adenine ribose of NAD+ is necessarily located very close to the surface. In the tetrameric enzyme GAPDH, NAD+ is deeply buried. Although the X-ray structure" shows it to be located at the edge of a subunit, each subunit shields the others from solvent (and the silver surface) and thus entrains the coenzyme within the protein envelope.Accordingly, any SER signal would have to arise from either a large electromagnetic enhancement or from protein denaturation (either gross structure change or mere dissociation into monomer units). The lack of SER signals thus provides further proof that silver colloids prepared by citrate reduction do not denature proteins. It is possible that the lack of SER enhancement is due to a lack of adsorption onto silver. In view of the numerous charged groups at the exterior of the enzyme, however, adsorption seems probable. The lack of enhancement demonstrates rather that the separation of the NAD+ from the silver surface is too great for even electromagnetic enhancement to be effective. Conclusion NAD+ adsorbs to silver primarily via its phosphate groups, with weaker adsorption via the adenine ring.As the solution concentration is reduced, the nicotinamide moiety approaches closer to the silver surface, as shown by its greater relative SER enhancement. Thus, at low concentrations, the NAD' molecule is quite extended along the silver surface, with nicotinamide, phosphate, ribose and adenine groups all in close proximity to the surface. In contrast to the strong enhancement seen for NAD+ on silver, little enhancement is seen for NADH. This is due to adsorption being prevented by coordination of hydroxide ions to the silver at the alkaline pH needed for stability of NADH in solution, resulting in a more negative surface potential at the higher pH values. NAD+ did not give an SER spectrum when bound to GAPDH.The separation between silver and NAD+ caused by the protein completely prevents enhancement by any CT mechanism. The separation is sufficiently large for enhancement by the electromagnetic enhancement mechanism to be negligible. The future for study of NAD+ by SERS when bound to similar large dehydrogenase enzymes appears bleak. We thank Dr J. A. Creighton for helpful discussions, and Mr Reuben Girling for experimental assistance. We thank the S.E.R.C. for financial support. Note added in proof: Since this paper was accepted for publication, closely related work has been published by 0. Siiman et al., Inorg. Chern., 1988, 27, 3940. References 1 Pyridine Nucleotide-dependent Dehydrogenases, ed. H . Sund (Walter de Gruyter, Berlin, 1977). 2 D. Chen, K. T. Yue, C.Martin, K. W. Rhee, D. Sloan and R. H. Callendar, Biochemistry, 1987, 26, 4776. 3 J . C. Austin, C . W. Wharton and R. E. Hester, Biochemistry, 1989, in press. 4 K. T. Yue, C. Martin, D. Chen, P. Nelson, D. Sloan and R. H . Callendar, Biochemistry, 1986, 25, 5 M. Majoube, J . Raman Spectrosc., 1985, 16, 98. 494 1.1168 SERS of NAD+ 6 M. Tsuboi, Y. Nishimura, A. Y. Hirakawa and W. L. Peticolas, in Biological Applications of Raman 7 T. M. Cotton, S. G. Schultz and R. P. Van Duyne, J. Am. Chem. SOC., 1980, 102, 7960. 8 J. de Groot and R. E. Hester, J . Phys. Chem., 1987, 91, 1693. 9 K. Kelly, B. N. Rospendowski, W. E. Smith and C. R. Wolf, FEBS Lett., 1987, 222, 120. Spectroscopy, ed. T. G. Spiro (J. Wiley, New York, 1987), vol. 2, p. 109. 10 P. Hildebrandt, R. Greinert, A. Stier, M. Stockburger and H. Taniguchi, FEBS Lett., 1988, 227, 76. 11 R. E. Holt and T. M. Cotton, J. Am. Chem. Soc., 1987, 109, 1841. 12 E. R. Nabiev, V. A. Savchenko and E. S. Efremov, J. Raman Spectrosc., 1983, 14, 375. 13 J. S. Suh and M. Moskovits, J. Am. Chem. SOC., 1986, 108, 4711. 14 V. Brabec and K. Niki, Stud. Biophys., 1986, 114, 11 1. 15 I. Taniguchi, K. Umekita and K. Yasukouchi, J. Electronal. Chem., 1986, 202, 315. 16 J. A. Creighton, in Advances in Spectroscopy: Spectroscopy of Surfaces, ed. R. J. H. Clark and R. E. 17 T. Skarzynski, P. C. E. Moody and A. J. Wonacott, J. Mol. Biol., 1987, 193, 171. 18 J. B. Fox and W. B. Dandliker, J. Biol. Chem., 1956, 221, 1005. 19 P. C. Lee and D. Meisel, J. Phys. Chem., 1982, 86, 3391. 20 J. A. Creighton, C. G. Blatchford and M. G. Albrecht, J. Chem. SOC., Faraday Trans. 2, 1977,75,790. 21 H. Wetzel, H. Gerischer and B. Pettinger, Chem. Phys. Lett., 1982, 85, 187. 22 W. Saenger, Principles of Nucleic Acid Structure (Springer-Verlag, Berlin, 1984), p. 109. 23 (a) K. Takamura, A. Mori and F. Kusu, Bioelectrochem. Bioenerg., 1981, 9, 229; (h) K. Takamura, A, 24 J. A. Creighton, Surf. Sci., 1985, 158, 21 1. 25 M. Moskovits, J. Chem. Phys., 1982, 77, 4408. 26 M. Moskovits and J. S. Suh, J. Phys. Chem., 1984, 88, 5526. 27 M. Moskovits and D. P. DiLella, J. Chem. Phys., 1980, 73, 6068. 28 C. Pettenkofer, I. Mrozek, T. Bornemann and A. Otto, Sure Sci., 1987, 188, 519. 29 X. Jiang and A. Campion, Chem. Phys. Lett., 1987, 140, 95. 30 H. Yamada, H. Nagata, K. Toba and Y. Nakac, Surf. Sci., 1987, 182, 269. 31 P. Hildebrandt and M. Stockburger, J . Phys. Chem., 1984, 88, 5941. Hester (J. Wiley, Chichester, 1988), chap. 2. Mori and F. Kusu, Bioelectrochem. Bioenerg., 1982, 9, 499. Paper 8/02794C; Receitled 1 1 th July, 1988

ISSN:0300-9599    

DOI:10.1039/F19898501159

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J . Chem. Soc., Furuduy Trans. I, 1989, 85(5), 1169-1 179 Mechanistic Study of P-Hydroxy Elimination from [Tetra sulphophthalocyanine CO~~~-CR,R,CR,R,OH] in Aqueous Solutions A Pulse Radiolysis Study Yacov Sorek,"? Haim Cohen*$ and Dan Meyerstein*tj Ben-Gurion University of the Negev, Beer-Sheva, Israel The reactions of cobalt(r1) tetrasulphophthalocyanine, [Co"(t~pc)]~- with 'CH,, 'CH,CH,OH, 'CH(CH,)CH,OH, 'CH(CH,)CH(CH,)OH and 'CH,C(CH,),OH free radicals have been studied. The results indicate that the first product formed in these reactions has the form [C~'~-(tspc-R)]~-, where R is the aliphatic free radical, the exact nature of the bonding of 'R to tspc is not clear. The complex [Co"-(t~pc-R)]~- rearranges into [(tspc)C~"'-R]~- in a process obeying a first-order rate law.[(tspc)Co'" -CHJ- decomposes into [Co"'(t~pc)]~- + CH,, whereas the complexes [(~~~c)CO"'-CR,R,CR,R,OH]~- decompose into [Co'"(tsp~)]~- + R,R, C=CR,R, +OH-. We have recently reported' that the mechanism of reaction of 'CH,C(CH,),OH radicals with cobalt(r1) tetrasulphophthalocyanine, [Co"(tspc)I4-, in neutral aqueous solutions involves the following reactions :T[ (1) (2) [ CO I '( t SPC)] 4- + C H C( C H J , 0 H -, [ (t SPC) CO I ' ' - CH C( CH ,) ,OH] 4- [ ( t SPC)CO I I I- C H , C( CH ,) 0 HI ' - -+ [ ( t SPC) CO I ' I - C H=C( CH ,) ,] 4- + H ,O H 2o [(~S~C)CO"~--CH=C(CH,),]~- + [Co"l(t~p~)]~- + CH,=C(CH,), + OH-. (3) It was also suggested that the products [Co"'(tsp~)]~- and CH,=C(CH,), remain complexed to each other and that the latter complex dissociates in a consecutive reaction.' In alkaline solutions, pH > 9.5, it was suggested' that reaction (2) is followed [(~S~C)CO"'-CH=C(CH~)~]~- -+ [C~'(tspc)]'- + HOCH=C(CH,), + H,O+.(4) Again the products seemed to remain complexed and to dissociate in a consecutive process. This mechanism was based on the observation that reactions (3) and (4) cause a change in the conductivity of the solution in accord with the consumption or formation by H ,o + Nuclear Research Center Negev. R. Bloch Coal Research Center. 9 Chemistry Department. In this study, as in the previous one,' all rates of reaction were calculated with the assumption that all the [Co"(tspc)14- is in its monomeric form; however, under the experimental conditions most of it exists in the solutions in the dimeric form. Thus k,, is clearly greater than indicated.In none of the systems studied were different properties for any of the intermediates formed in the reaction of the aliphatic free radicals with [C~"(tspc)]~- and [Co"(tspc)]!- detected. Thus, as in the previous study,' the equations are written for the monomeric form, although one cannot rule out the possibility that analogous intermediates for the dimeric form are present, but have similar physical and chemical properties. I1691170 Mechanistic Studies of p-Hydroxy Elimination of one equivalent of H,O+, respectively, and on an analogous mechanism suggested earlier for the decomposition of [(H,0),Cr111-CH,C(CH3)20H]2+, where the tran- sient observed was identified as [(H,0)5Crr11-CH=C(CH3)2]2+.2 We have recently ~ h o w n , ~ by an isotope-labelling study, that the mechanism of decomposition of [(H,0)5Crr11-CH,C(CH,),0H]2- occurs via the reaction : -OH- \ / [(H,0),Cr111-CH,C(CH,),0H]2+ + [(H20)5Cr11r- I I ],+ + [Cr(H,0)6]3+ + CH,=C(CH,), (5) and not via the mechanism suggested previously.If the p-elimination step in the decomposition of [(~s~c)CO~~'CH~C(CH,),OH]~+ follows an analogous mechanism then it is accompanied by a conductivity change, and therefore the proposed mechanism has to be modified. Owing to the interest in metal phthalocyanines as catalysts and photocatalysts (including plausible phototherapeutic uses4' 5, in systems that probably involve free radicals, and the importance of P-hydroxy elimination reactions we decided to reinvestigate the reaction of 'CH,C(CH,),OH with [Co"(tspc)I4-.We extended the study to the reactions of 'CH,CH,OH, 'CH(CH,)CH,OH and *CH(CH,)CH(CH,)OH in order to study the effect of methyl substituents on the reaction mechanism and rates of reaction. Finally, we studied also the reaction of 'CH, radicals with [Co"(tspc)I4- as a ' blank ' system. Experimental The tetrasodium salt of Cor1-4,4',4"4"'-tspc was prepared according to the procedure reported by Weber and Bush.' The u.v.-visible absorption spectrum was checked and was identical with that reported in the literature,' calculated: C, 37.82; Co, 5.80%; found : C, 37.98 ; Co, 5.86 %. Pulse-radiolytic experiments using electron linear accelerators were carried out both at the Hebrew University of Jerusalem ( 5 MeV, 200 mA, 0.1-1.5 ps per pulse, dose per pulse 1-50 Gy) and at Argonne National Laboratory (12-1 8 MeV, 4 4 0 ns per pulse, dose per pulse 1-50 Gy).The details of the experimental technique including solution preparation and data analysis were identical to those in the previous study.' High integral dose irradiations were performed with a 'OCo y source with a dose rate of 2.5 x lo3 Gy h-l. Preparation of Radicals The radiolysis of water and dilute aqueous solutions can be summed up by the equation' 7, e- H,O + eiq, 'OH, 'H, H,, H,O,, H30+, OH-. (6) These products are homogeneously distributed in the solution within < 100 ns of the absorption of radiation. Their yields are G(eJ = G(0H) = 2.65, G(H) = 0.60, G(H,) = 0.45, G(H,O,) = 0.75, G(H,O+) z 3.65 and G(OH-) z 1.0.' (The G values are in units of molecules of product formed per 100 eV absorbed in the solution.) In concentrated solutions the yield of the free radicals is somewhat higher and that of H,O, is somewhat smaller.' The reactions of eiq, 'OH and 'H with the solutes used in this study and their specific rates are summarised in table 1.From the data in the table it is evident that when N,O-saturated solutions, 2.2 x lop2 mol drn-,, at pH > 3.0 containing 0.01-1 .O mol dm-, of an organic solute and (1-10) x mol dm-, Co"(ts~p)~-, are irradiated the dominant free radicals produced are those formed in the reactions of 'OH with the aliphatic solutes. Ca. 10 % of the free radicals are formed via the reaction of hydrogen atoms with the aliphatic solutes. However, in most cases the same free radicals are formed (table 1).Y. Sorek, H .Colien and D . Meyerstein Table 1. Specific rates of reaction of the primary free radicals with several solutes 1171 ____ reactants eiq + H,O+ e, + N,O 0- + H,O eag + CR,R,OH(R=H,CH,) eag + R,R,C=CR,R,(R=H,CH,) 'OH + C(CH,),OH 'H + C(CH,),OH 'OH + C,H,OH 'H + C,H,OH 'OH + C,H, 'H + C,H, 'OH + C,H, 'H +C,H, 'OH + CH,CH=CH(CH,) 'OH + (CH,),SO (CH,),S(O)OH eag + Co"(tsp~)~- 'OH + C~"(tspc)~- ___.~___ products 'H N, + 0- 'OH+OH- - 'CH,C(CH,),OH + H,O 'CH,C(CH,),OH + H, 'CH(CH,)OH + H,O (CU. 85 Yo) 'CH,CH,OH + H,O (CU. 13 Yo) 'CH(CH,)OH + H, 'CH,CH,OH ' CH (CH ,)CH ,OH a ' CH (C H ,), 'CH(CH,)CH(CH,)OH 'CH, + CH,SOOH Co"(tsp~,OH)~- 'C2H5 (CH3)2S(o)oH Co'(tspc)S- k/dm3 mo1-' s-' ref. 2.3 x 1Olo 21 8.7 x 109 21 < I x 105 21 < 2.5 x lo6 21 4.2 x loR 22 8 x 104 23 1.7 x 109 22 2.5 x 107 23 4.8 x 109 22 3 x 109 23 6.5 x 109 22 4 x 109 23 7 x 109 24 1.5 x lo7 (s-l) 24 3 x 109 12 3 x 109 12 a E.s.r.data show that in this reaction only 'CH(CH,)CH,OH and no 'CH,CH(CH,)OH is formed.25 Solutions containing alkenes were prepared by saturation of the required solution with the alkene and then by mixing, using the syringe technique with N,O-saturated water. Usually a mixture of 90 YO of the alkene-containing solution and 10 YO N,O-containing solution was prepared. However, for all systems other ratios were also prepared and the results were shown to be independent of the ratio. Analysis The yields of methane as well as those of alkenes were determined by gas chromatography using a thermal-conductivity detector.Blank experiments were always performed in the absence of [Co"(tsp~)]~-. Qualitative identification of the alkenes by F.t.i.r. spectroscopy was used. A Nicolet MXS i.r. spectrophotometer with a 14 cm gas- sampling cell was used. Mass-spectral analysis, using a Balzers quadrupole mass spectrometer, was carried out in order to check isotope enrichment where necessary. Heavier organic products were determined by g.c.-m.s. on a 30 m OB-wax column. Results and Discussion Reaction of [C~~~(tspc)I~- with Methyl Free Radicals N,O-saturated solutions containing (1-5) x lop5 mol dm-, [Co"(tspc)]", 0.1 mol dm-3 (CH,),SO in the pH range 3.0-12.0 were irradiated by pulses producing ca. 3 pmol dmP3 free radicals. Under these conditions 'CH, radicals are the dominant free radicals formed by the pulse (table 1).Two processes were observed spectrophotometrically under these conditions; a typical result at 450 nm is shown in fig. 1. The first process observed obeys a first-order rate law, with a specific rate of k = (2.5 f 0.5) x lo3 s-'. This rate and the absorption change caused by it are independent of the complex concentration, pH, pulse intensity and the wavelength of observation. The1172 Mechanistic Studies of /I- Hydroxy Elimination 4 1 1 4 400 800 tIcLT 0.8 1.6 tls Fig. 1. Light intensity changes observed at 450 nm in an N,O-saturated solution containing 2 x lop5 mol dm-3 [Co"(t~pc)]~-, 0.1 mol dmp3 (CH,),SO at pH 6. I after a pulse producing 3.0 x mol dm-3 free radicals. observation that the rate of this process is independent of the complex concentra- tion, that it is considerably slower than that of the reactions of [Co"(tsp~)]~- with 'CH,C(CH,),OH,' 'CH2CH20H," 'CH(CH:5)CH20H,9u 'CH(CH:,)CH(CH:,)OH,!'" 'CH,C(CH,),CO,HSb and .CH2C(0)CH3,9b and that the observed rate is somewhat slower than expected for the dimerisation of methyl radicals in the absence of the complex, 2k(2CH3 -+ C,H,) = 3.2 x lo9 dm3 mo1-' s-'," prove that the process observed is not the reaction of 'CH, with [Co"(tsp~)]~~.Indeed a careful analysis of fig. 1 shows a very fast small change in the light transmittance occurring prior to the process discussed. The absorption changes in this fast process are too small at 380 < A/nm < 540 to be studied. Therefore the first process observed corresponds to the second reaction occurring in the system. The spectrum of the transient formed in this reaction is plotted in fig.2. The second process observed, which therefore corresponds to the third reaction occurring in the system, also obeys a first-order rate law with k = 3.0k 1.0 s-l. The specific rate of this reaction is independent of [Co"(tspc)I4-, pH, pulse intensity and wavelength of observation. The spectrum of the product of this reaction (fig. 2) is identical to that observed when [Co"(tsp~)]~- is oxidized by Br,.'" G.c. analysis showed that methane is the major organic product [G(CH,) = 5.0+0.5] and only traces of ethane are observed. The mechanism of reaction of [Co"(tspc)14- with Bri- has recently been reinvestigated'" and shown to be: [C~"(tspc)]~- + Bri- -+ [C~"(tspc+)]~- + 2Br- followed by [Co"(tspC+)]3- -+ [Co'"(tspc)]3-.The observation that the spectrum of the final product in the present study is identical to that of [Co'"(tsp~)]~- and that methane is the major organic product suggests that the second transient observed is [(tspc)Co"'-CH,]", which decomposes via heterolysis to the final products. We therefore suggest that scheme 1 represents the detailed reaction mechanism of methyl radicals with [Co"(tsp~)]~-.Y. Sorek, H . Cohen and D . Meyerstein 1173 400 500 Unm Fig. 2. Difference spectra between the spectrum of the second intermediate and that of the final product relative to that of the solution before the pulse. N,O-saturated solution contain- ing 2 x 10-s mol dm-3 [Co"(tspc)l4-, 0.1 mol dm-3 (CH,),SO at pH 6.1.Pulse producing 3.0 x mol dm-3 free radicals. a, 2 ms after the pulse; A, 2 s after the pulse. [CO"(~SPC)]~- + 'CH, + [Co"(t~pc-CH,)]'- [CO"(~SPC-CH,)]~- + [(~sPc)CO"'-CH,]"; k , , = 2.5 x lo3 S-' [(~SPC)CO'"-CH,]~- -+ [CO"'(~SPC)]~-+CH, + OH-; k,, = 3.0 S-' (9) (10) ( 1 1) Scheme 1. The absorption spectrum of [C~"(tspc-CH,)]~- is similar, but not identical, in the narrow spectral window which is experimentally accessible to that of [Co"(tspc)14- and therefore reaction (9) could not be studied. The nature of the binding of 'CH, to (tspc) in [C~"(tspc-CH,)]~- is not clear. The binding has to be different from that of OH1' and C112 formed in the reaction of [C~"(tspc)]~- with 'OH and Cl,, respectively, as in the latter cases [Co"'(tspc)I3- is not the final product.We tentatively suggest that the methyl is bound to one of the nitrogen atoms and then migrates to the central cobalt ion [reaction (lo)] by analogy with reactions observed for porphyrin c~mplexes.'~ Note that reaction (1 1) is considerably faster that that of a variety of analogous reactions for other LCo"'-CH, complexes, e.g. L = c ~ r r i n s , ~ ~ bisdimethygly~xime,'~1174 Mechanistic Studies of p- Hydroxy Elimination 2,3,9,lO-tetramethyl- 1,4,8,11 -tetra-azacyclotetradecane- 1,3,8,lO-tetraene," 5,7,7,12,14, 14-hexamethyl- 1,4,8,11-tetra-azacyclotetradecane'7 and 5,7,7,12,14,14-hexamethyl- 1,4, 8 , l l -tetra-azacyclotetradeca-4,11 -diene." The mechanism of reaction of methyl radicals with [Co"(tsp~)]~- can alternatively be described by scheme 2.where 1 = tspc Scheme 2. According to this mechanism the first short-lived transient formed is a hepta-coordinated cobalt(m) complex. Kinetically one cannot differentiate between schemes 1 and 2. However, it is difficult to argue why the transient formed in reaction (12) has a very similar spectrum to that of [Co"(tsp~)]~-, whereas the transient formed in reaction (1 3) has a considerably different spectrum ; furthermore it is difficult to envisage why reaction (13) should be so slow. We therefore are tempted to prefer scheme 1 to scheme 2. As for the reaction of 'CH,C(CH,),OH radicals' it is suggested that the dimeric form of [Co"(tsp~)]~- reacts in an analogous mechanism to the monomeric one. Reaction of [Co"(tspc)]*- with 'CH,C(CH,),OH Free Radicals The kinetics of this reaction were reported earlier in detail.' We have now added two experiments in order to check whether the mechanism suggested originally [reactions (1) - (3)] is correct.(a) If the decomposition of [(~s~c)CO"'-CH,C(CH,),OH]~- occurs via reactions (2) and (3), as suggested earlier,' CHD=C(CH,), is expected as the product when the reaction is carried out in D20. We therefore irradiated N,O-saturated deuterated aqueous solutions (D > 97 %) containing 4 x lo-, mol dm-3 [Co"(tspc)]*- and 1 .O mol dm-3 C(CH,),OH in the pH range 3.5-9.5 and analysed the methylpropene product mass-spectrometrically. The results of this experiment point out that the 2-methyl- propene is not labelled by deuterium above the natural abundance.Thus, by analogy with the decomposition of [(H20),Cr111-CH2C(CH3)20H]2+ we have to conclude that the D-elimination step involves the elimination of OH-, and not of H20, and is thereforeY . Sorek, H . Cohen and D . Meyerstein 1175 accompanied by a conductivity decrease below pH 6.0, i.e. the a-elimination step is the third process observed and not the second, as was proposed earlier. It is therefore suggested that scheme 3 describes the reaction mechanism of [Co"(tspc)]'- with 'CH,C(CH,),OH free radicals. [CO"(~SPC)]~- + 'CH,C(CH,),OH -, [CO"(~S~C-CH,C(CH,),OH)]~- ; k = 4.5 x lo9 dm3 mol-' s-' (14) [CO"(~S~C-CH,C(CH,),OH)]~- + [(~s~c)CO"~-CH,C(CH,),OH]~- ; k = 2000 S-' (1 5 ) (16) \ / [ ( t SPC)CO I' '-C H ,C( C H 3) ,OH] 4- + [ ( t s pc) C 0 'I - 11 ],-+OH-; k = 15 S-' 1 1 1 ' ' [(tspc)Co 11 1,- + [CO"'(~S~C)]~- + CH,=C(CH,),; k = 4 S-'.(1 7) Scheme 3. Reactions (14) and (15) in scheme 3 are analogous to reactions (9) and (10) for the methyl radical. Instead of reactions (14) and (1 5), analogous processes to reactions (12) and (1 3) in scheme 2 may be proposed. Reactions (1 6) and ( I 7) are analogous to reaction ( 5 ) for the decomposition of [(H,0)5Cr"'-CH,C(CH3)20H]2+. Thus scheme 3 is in accord with all the experimental observations and with processes observed in analogous systems. Naturally the spectra of the transients reported in the previous study2 have to be attributed to the transients indicated in scheme 3. (b) In alkaline solutions [Co'(tspc)I5-, and not [Co"'(tsp~)]~-, is the final product.' The organic product in alkaline solutions has not been determined previously.We have therefore irradiated N,O-saturated solutions containing 4 x lop3 mol dmP3 [Co"(tspc)]'-, 1 .O mol dm-, C(CH,),OH at pH 1 1.5. G.c.-ms. analysis indicated that 2-methyl- propane-1,2-diol is the major organic final product. Its yield is similar to that of (-CH,C(CH,),OH), formed under similar conditions, but in the absence of [Co" (tspc)I4-. We cannot rule out the possibility that 1,2-epoxy-2-rnethylpropane is formed and hydrolyses prior to the analysis into the diol. (Epoxyethane is the product in the decomposition of CuI"- CH,CH,0H2+,, into Cutaq + CH2CH29la. The first two reactions in alkaline solutions are identical to those observed at lower pH1 and are therefore identified as reactions (14) and (15).The third process observed is accompanied with a decrease in conductivity equivalent to the consumption of one equivalent of OH-. Therefore reactions ( 18) and ( 1 9) are plausible explanations for this process. [(~S~C)CO'~'-CH,C(CH,),OH]~- + H,O + [CO'(~SPC)]~- + HOCH,C(CH,),OH + H,O+ ; k = 19 S-' (18) [(~S~C)CO"'-CH,C(CH,),OH]~-+ [CO'(~S~C)]~- + OCH,C(CH,), + H,O+; k = 19 s-'. - (19) However, as the third process is followed by a fourth process during which the spectrum of [Co'(tspc)I5- appears,' one has to assume that either the organic product, HOCH, C(CH,),OH or QCH,C(CH,), remains somehow complexed to [C~'(tspc)]~- and has a major effect on its spectrum or that the third process in alkaline solutions is CH2 [(~s~c)CO"'-CH,C(CH,),OH]~-+ [(tspc)Co"' ' 'C(CH3)J5-+H3O+; k = 19 s-! (20) '0'1176 This process is then followed by Mechanistic Studies of /?- Hydroxy Elimination CH[( tspc)Co'" ' ' C(CH,),]'-+ [Co1(tspc)l5- + HOCH,C(CH,),OH \*/ or OCH,C(CH,),; k = 4 s-' (21) V The spectral changes observed, cf fig.5 and 3 in ref. (2), seem to fit better a reaction mechanism involving reactions (14), (15), (20) and (21). The cyclic transient suggested in this scheme is analogous to that proposed in the epoxidations by cytochrome P-450 and its model compounds. Reaction of [(Co"-(tspc)14- with 'CH,CH,OH, 'CH(CH,)CH,OH and ' CH( CH,)CH( CH,)OH Free Radicals The reactions of the free radicals 'CH,CH,OH, 'CH(CH,)CH,OH and 'CH(CH,)CH(CH,)OH, produced by the addition of hydroxyl radicals to the corresponding alkenes, with [Co"(tspc)]'- at pH 6.1 were studied.For each of these radicals three consecutive processes are observed. The first process obeys a first-order rate law, the rate being proportional to [Co"(tspc)I4-. In the casz of 'CH,CH,OH this process causes only small changes in the light absorbance, only slightly larger than in the case of 'CH, radicals, and therefore only a rough estimate of the specific rate of reaction is obtained. The spectrum of the second intermediate and the final product in this system are plotted in fig. 3. The two following processes obey first-order rate laws, the rates of reaction are independent of [C~"(tspc)]~-. The final product in all these systems is [Co"[(tsp~)]~-. The specific rates of reaction obtained in these systems are summarised in table 2.For the reaction of 'CH,CH,OH three further experiments were carried out. (a) N,O-saturated solutions containing 4 x I OP3 mol dm-" [Co"(tspc)]"-, I .O mol dmP3 C,H,OH at pH 6.1 were irradiated. Under these conditions ca. 13 O/O of the 'OH radicals yield 'CH,CH,OH free radicals (table I). The final products were analysed by g.c. and G(C,H4) = 0.55&0.10 was obtained. This result proves that [(~s~c)CO"~CH,CH,OH]~- decomposes via /?-elimination, as expected. The experiment was repeated in D,O, and no labelling of the ethene was observed. (b) The reaction of 'CH,CH,OH with [Co"(tsp~)]~- was studied as a function of pH in the range pH 3.0-1 1.5. The specific rates of reaction were found to be pH-independent. The yield of [Co"'(tsp~)]~- is also pH-independent above pH 5.5, G([C~'I~(tspc)]~- = 5.5 +_ 0.5, i.e.all the 'CH,CH,OH radicals oxidize [Co"(tspc)I4-. The yield of [Col"(tspc)]'- dropped with pH to ca. 2.2 and ca. 0.6 at pH 5.0 and 3.0, respectively. This drop in the yield is attributed to the reaction [ C o I I ( t s pc)] 3- + [ ( t s pc) Co I I - C H , C H , 0 H ] * -, 2 [ C 0 I ' ( t s pc)] '- + ox i d a ti on prod uc t s . (22) An analogous process was observed in the same pH region in the 'CH,C(CH:,),OH system.' We checked this point by adding 1 x mol dm-, [Co"'(tsp~)]~-, which does not react with 'CH,CH,OH free radicals under the experimental conditions, to solutions saturated with a mixture of C,H, and N,O and containing 2 x lop5 mol dm-" [(Co"(tspc)]*- in the pH range 4.5-6.0.In these solutions the radiolytic yield of [C~*''(tspc)]~- was indeed smaller than in the absence of [Co"'(tspc)]".Y. Sorek, H. Cohen and D. Meyerstein 1177 500 Alnm 4 00 Fig. 3. Difference spectra between the spectrum of the second intermediate and that of the final product relative to that of the solution before the pulse. 9 : 1 = C,H4 : N,O-saturated solution containing 2 x mol dmP3 [Co"(tsp~)]~-, at pH 6.2. Pulse producing 3.2 x mol dm-3 free radicals. 0 , 4 ms after the pulse; A, 2 s after the pulse. Table 2. Specific rates of reaction for the reactions of [Co"(t~pc)]~- with free radicals with a P-hydroxyl group, at pH 6.1 CH,C(CH,),OH (4.5 f 0.5) x lo9 (2.0 k 0.2) x 103 15f3 4k 1 CH,CHCH(OH)CH, (8 & 2) x lo8 (2.4 f 0.4) x lo3 6+2 - 'CH(CH,)CH,OH (1.5k0.3) x lo9 (l.OkO.3)~ lo3 8+2 - 'CH,CH,OH 2 5 x lo9 (1.2ko.3)x 103 5 f 2 - 'CH, > 5 x 109 2.5 x 1 0 3 - - (c) The conductivity change in the reaction of 'CH,CH,OH with [Co"(tspc)I4- was studied at pH 4.1.Only one reaction was observed, the specific rate of which corresponds to the last reaction observed spectrophotometrically. The reaction observed consumes H,O+ with a yield of G(H,O') = 1.3 & 0.2 This finding is in full accord with the yield of [Co"'(t~pc)]~- at this pH and proves that the third process observed 40 F A R I1178 Mechanistic Studies of @-Hydroxy Elimination spectrophotometrically is the @-elimination reaction. All these experiments clearly point out that the three processes observed spectrophotometrically in the reactions of 'CH,CH,OH, 'CH(CH,)CH,OH and 'CH(CH,)CH(CH,)OH with [C~'~(tspc)]~~- corre- spond to the reactions (where Ri = H,CH,) (23) (24) [(~s~c)CO"'-CHR,CR,R~OH]~- + [CO"'(~S~C)]~- + CHR,=CR,R, + OH- (25) in complete analogy to the 'CH,C(CH,),OH system.Only in the latter system is [Co1(tspc)I5- formed in alkaline solutions and an intermediate complex between [Co"'(tspc)J3+ and the alkenes is observed. The latter result is in accord with observa- tions for the decomposition of (H,0)5Cr'1'-CHRlCR,R,0H2f.'9 [CO"(~S~C)]~- + 'CHR,CR,R,OH + [CO"(~S~C-CHR,CR,R,OH)]~- [ C o ( t SPC- C H R , C R , R , 0 H )] - + [ ( t s PC) C o ' I I- C H R C R , R ,O H ] 4- Conclusion It is of interest to note the large difference in the spectra of the first transient observed for the 'CH, and 'CH,CH,OH systems on one hand and for 'CH(CH,)CH,OH, 'CH,C(CH,),OH and *CH(CH,)CH(CH,)OH on the other.The similarity between the spectra in the 'CH, and 'CH,CH,OH systems suggests that the spectral changes are due to steric interactions between the alkyl residue and the phthalocyanine ring and not to inductive effects. As the exact nature of the first intermediate is unknown further speculations at this stage seem unjustified. A comparison of the specific rates of reaction in table 2 suggests that the rate of reaction (23) is slightly lowered owing to steric hindrance by methyl substituents on the a carbon. The specific rates of reactions (24) and (25) seem to be little affected by methyl substituents. This observation is somewhat surprising in regard to reaction (25) as the rates of the analogous @-hydroxyl elimination reactions from (H,0),Cr'11-CRlR,CR,R40H2+ l9 and CU~~-CR~R,CR,R~OH~~ origin of this discrepancy in behaviour is not clear at present.are considerably affected by methyl substituents. The We are indebted to Dr D. Meisel for carrying out the conductometric experiment, to Mrs E. Norton for performing the g.c.-m.s. analysis, and to Mr A. Hemmy for technical assistance. This study was supported by grants from the Israel-United States Binational Science Foundation (B.S.F.), Jerusalem, and from the Planning and Granting Committee of the Council of Higher Education and the Israel Atomic Energy Commission. D.M. expresses his appreciation to Mrs Irene Evens for her continuous interest and support. References 1 Y.Sorek, H. Cohen, W. A. Mulac, K. H. Schmidt and D. Meyerstein, Inorg. Chem. 1983, 22, 3040. 2 H. Cohen and D. Meyerstein, Inorg. Chem., 1974, 13, 2434. 3 H. Cohen, D. Meyerstein, A. J. Shusterman and M. Weiss, J . Am. Chem. Soc., 1984, 106, 1876. 4 I. Rosenthal, C. M. Krishna, P. Reisz and E. Ben-Hur, Radiat. Res., 1986, 107, 136. 5 E. Ben-Hur, I . Rosenthal, S. G. Bowrand, D. Phillips, in Photomedicine, ed. E. Ben-Hur and I. 6 J. H. Weber and D. H. Busch, Znorg. Chem., 1965, 4, 469. 7 L. C. Gruen and R. J. Blagrove, Aust. J . Chem., 1973, 26, 319. 8 M. S. Matheson and L. M. Dorfman, Pulse Radiolysis (M.I.T. Press: Cambridge, MA, 1969). 9 (a) see below, table 2; (b) Y. Sorek, Ph.D. Thesis (Ben Gurion University of the Negev, Beer-Sheva, Rosenthal (CRC Press, Boca Raton, in press). 1986); (c) A. B. Ross and P. Neta, Natl Bur. Stand. Ref. Data Systems, NSRDS-NBS, 1982, 70.Y. Sorek, H. Cohen and D. Meyerstein 1179 10 D. K. Geiger, G. Ferraudi, K. Madden, J. Granifo and D. P. Rillema, J. Phys. Chem., 1985, 89, 11 G. Ferraudi and L. K. Patterson, J. Chem. Soc., Dalton Trans., 1980, 476. 12 G. Ferraudi, S. Oishi and S. Muralidharan, J. Phys. Chem., 1983, 88, 5261. 13 D. Mansuy, J. P. Battioni, D. Dupre and E. Sartori, J . Am. Chem. SOC., 1982, 104, 6159 and references 14 J. F. Endicott and G. Ferraudi, J. Am. Chem. Soc., 1977, 99, 243. 15 J. Haplern, Acc. Chem. Res., 1982, 15, 238. 16 K. Farmery and D. H. Busch, Znorg. Chem., 1972, 11, 2901. 17 T. S. Roche and J. F. Endicott, Znorg. Chem., 1974, 13, 1575. 18 G. V. Buxton, J. C. Green, R. Higgins and S. Kanji, J. Chem. Soc., Chem. Commun., 1976, 158. 19 R. Ish-Shalom, H. Cohen and D. Meyerstein, to be published. 20 H. Cohen and D. Meyerstein, J. Chem. SOC., Faraday Trans. I , 1988, 84, 4157. 21 M. Anbar, M. Bambenek and A. B. Ross, Nut1 Bur. Stand. Ref. Data System, NSRDS-NBS, 1973, 43. 22 Farhataziz and A. B. Ross, Natl Bur. Stand. ReJ Data System, NSRDS-NBS, 1977, 59. 23 M. Anbar, Farhataziz and A, B. Ross, Nut1 Bur. Stand. ReJ Data System, NSDRS-NBS, 1975, 51. 24 D. Veitwisch, E. Janata and K. D. Asmus, J. Chem. SOC., Perkin Trans. 2, 1980, 146. 25 W. E. Griffiths, G . F. Longster, J. Myatt and P. F. Todd, J. Chem. SOC. B, 1967, 530. 3890. cited therein. Paper 8/02920B ; Received 19th July, 1988 40-2

ISSN:0300-9599    

DOI:10.1039/F19898501169

出版商:RSC    

年代:1989

数据来源: RSC