摘要:

J. Chem. Soc., Furuday Trans. 1, 1988, 84(1), 321-330 Adsorption and Decomposition of Ammonia on an Fe(1 x 1) Overlayer on an Ru(OO1) Surface with or without Co-adsorbed Oxygen Chikashi Egawa, Kyoichi Sawabe and Yasuhiro Iwasawa* Department of Chemistry, Faculty of Science, University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113, Japan The adsorption and decomposition of ammonia on Fe( 1 x 1) overlayers on an Ru(OO1) surface with or without co-adsorbed oxygen have been investigated by means of Auger electron spectroscopy, low-energy electron diffraction and thermal desorption, and also by kinetic studies. The adsorption of ammonia at 420 K on an epitaxial Fe overlayer deposited on an Ru(001) surface gave a newly-ordered ( ~ ' 7 x 2/7) R19" structure over a wide range of Fe coverages from 0.14 to 2 monolayers.The desorption of N, from the ordered adsorbate layers showed a sharp peak at 870 K which was replaced by a peak at 970 K above 0.6 monolayer Fe coverage, showing similarity to the decomposition of surface nitride, Fe,N, observed for Fe single-crystal planes. The decomposition of ammonia on Fe/Ru(OOl) surface proceeded uiu two consecutive reaction steps that were dynamically balanced. The activation energy of the formation of atomic nitrogen varied from 33 to 8 kJ mol-' with an increase in Fe coverage from 0 to 1 monolayer, while the rate of desorption of N, was reduced. Accordingly, the steady-state rate of decomposition of ammonia exhibited an optimum Fe coverage to reach the maximum rate. In contrast, a 4 4 x 2 ) ordered overlayer appeared upon the adsorption of oxygen on an annealed Fe/ Ru(001) surface. The steady-state rate of decomposition of ammonia was enhanced 3.5 times by co-adsorbed oxygen, because nitrogen was effectively activated by surrounding oxygen atoms in c(4 x 2) mixed overlayers.The metals iron and ruthenium are both good catalysts for ammonia synthesis and decomposition. Studies of the adsorption of nitrogen on single-crystal planes of Fe have revealed that the dissociative adsorption of nitrogen takes place preferentially on C, sites present on the ( I 11) surface,' where the activation energy for dissociative adsorption is nearly zero. However, the bonding energies between the metal and nitrogen are similar on the three Fe single-crystal planes, taking the difference in activation energy for dissociation into consideration.This suggests the formation of similar surface nitrides on these different planes by the reconstruction of the topmost atomic layers, where various LEED patterns have been observed. The Fe(ll1) surface has also been demonstrated to be most active for ammonia synthesis,2 because the dissociative adsorption of N, is a rate-limiting step in the reaction. The interaction of ammonia with Fe3 and Ru* surfaces using single crystals and the catalytic decomposition of ammonia on transition metals5-' have also been investigated. The decomposition of ammonia on an Ru(OO1) surface1' takes place at ca. 400 K and proceeds through a reaction mechanism of dynamic balance ; two consecutive reactions, the formation of atomic nitrogen from ammonia and the subsequent desorption of surface nitrogen, are balanced to give the overall reaction rate and the corresponding surface nitrogen coverage at steady state, where no inhibition by hydrogen of the reaction rate or the surface nitrogen coverage was observed.Similar behaviour is also observed on Fe foi1,l' where the 32 1 11-2322 Adsorption and Decomposition of Ammonia on Fe reaction starts at 580 K and the activation energies vary from 200 to 0 kJ mol-1 as the reaction temperature is increased from 640 to 830 K. In contrast to these monometallic surfaces, we have recently employed an Fe/Ru(001) bimetallic system and observed the epitaxial growth of Fe overlayers on an Ru(OO1) surface.l2? l3 By depositing Fe on an Ru(OO1) surface at 420 K (an epitaxial surface), Fe atoms grew in two-dimensional islands, and a commensurate Fe p(1 x 1) structure similar to the hexagonal structure of an f.c.c.Fe( 11 1) surface was formed in the first monolayer. On a commensurate Fe(1 x 1) overlayer at 1 monolayer coverage the bonding energy of CO was found to be strengthened owing to the specific expanded ye overlayer structure, where the nearest-neighbour distance of the Fe overlayer (2.706 A) is very close to that for the (111) surface of f.c.c. Fe,N, although the second Fe monolayer changed to an overlayer structure similar to that of a b.c.c. Fe(ll0) surface, exhibiting extra diffraction spots due to a (6d3 x 6d3) R30" overlayer structure. Annealing the epitaxial surface with Fe coverages below 1 monolayer to 1030K (annealed surface) caused a dispersion of Fe atoms from the islands to the bare Ru surface and led to the formation of a variety of mixed sites composed of Fe and Ru atoms, although upon heating to 900 K the modification of electronic structure occurred with little change in the surface structure of the Fe p( 1 x 1) overlayer. On the contrary, annealing the epitaxial surface with Fe coverages above 1 monolayer to 1030 K induced the diffusion of excess Fe atoms into the Ru substrate and left 1 monolayer of Fe coverage on the surface.In the present paper the bonding states of nitrogen, the local chemical interaction of co-adsorbed oxygen and nitrogen and the effect of oxygen on the ammonia decomposition rate on these epitaxial and annealed surfaces with Fe coverages below 1 monolayer are reported in relation to the exploration of fundamental active factors for heterogeneous catalysis as well as properties of bimetallic surfaces.l3 Experimental Experimental procedures were similar to those described previously. lo, l2 Auger electron spectroscopy (AES) and low-energy electron diffraction (LEED) measurements were carried out using a four-grid retarding-field analyser. Thermal desorption of N, following the adsorption of ammonia and N, partial pressures during the decomposition of ammonia were monitored using a mass filter. Results and Discussion Adsorption of NH, on the Fe/Ru(001) Surface Adsorption of ammonia [lo L (1 L = 1.33 x Pa s)] on Fe-evaporated Ru(001) surfaces (epitaxial surface) at 420 K produced extra diffraction spots in addition to the (1 x 1) spots due to the commensurate Fe p(1 x 1) lattice, as shown in fig. 1 (a).The LEED pattern corresponds to a (2/7 x d 7 ) R19' surface structure, where filled circles represent diffraction spots composed of three domain orientations due to the superstructure and small open circles are derived from multiple scattering between the adsorbate and the substrate lattice. The pattern is different from the diffused p(2 x 2) structure observed with a clean Ru(OO1) surface4*'0 and has not been observed for b.c.c. Fe single-crystal plane^.^ In the adsorption of N, on Fe single-crystal planes, a 4 2 x 2) structure appeared on the Fe( 100) surface, while more complex LEED patterns were observed on Fe( 11 1) and Fe( 110) p1anes.l The adsorption of N, on an Fe( 1 11) surface led to the formation of a series of ordered structures of the form (3 x 3), (d19 x d19) R23.4', (d21 x 2/21)R10.9', (32/3 x 32/3)R30° and (2 x 2), and on Fe(ll0) similar complex LEED patterns such as (2x 3) and (; !) structures were reported.These structures could not be correlated with distinct ranges of surface concentrations, butC . Egawa, K. Sawabe and Y. Iwasawa 323 0 (a 1 (b ) (c) Fig. 1. (a) LEED pattern of the (2/7 x 4 7 ) R19" structure obtained from NH, adsorption on epitaxial Fe overlayers on Ru(001). (b) LEED pattern of the ( 4 3 7 x 2/37) R25" structure observed with NH, adsorption on a 0.5 monolayer Fe/Ru(OOl) surface pre-annealed to 900 K or on an Ru(OO1) surface with 0.5 monolayer of Fe and co-adsorbed oxygen.(c) Surface-structure model of the ( 4 7 x 47)R19" structure. 0, Ru atom; 0, 0, Fe atoms; Q, N atom. were considered to be related to the reconstruction of the topmost atomic layers due to the formation of surface nitrides such as Fe,N. In contrast to Fe single-crystal planes, the ordered structure due to NH, adsorption appeared on the Ru(001) surface with Fe coverages as relatively low as 0.14 monolayer, and was continuously present for the whole range of Fe coverages up to 2 monolayers, which is consistent with the epitaxial growth of an Fe overlayer in two-dimensional commensurate p(1 x 1) islands and also with the results of CO adsorption on epitaxial Fe 0ver1ayers.l~ This is supported by the result that the peak area of N, desorption in the thermal desorption spectra (fig.2) increased almost linearly with surface Fe coverage. An exposure of 10 L is sufficient to saturate the desorption states on the epitaxial surfaces. The surface nitrogen coverage at 1 monolayer of Fe was found to be 0.15, where the calibration was performed using the value of CO desorption from a ( 4 3 x 4 3 ) structure (Oco = 0.33). It is in reasonable agreement with the coverage of 0.14 evaluated from a ( 4 7 x d7) R19" overlayer structure. The ordered structure is due to surface nitrogen atoms, since the disappearance of diffraction spots coincides with the onset of N, desorption, as shown in fig. 2; below Fe coverages of 0.5 monolayer the additional pattern disappeared at 800 K, while above 0.5 monolayer coverage it remained at 900 K.The LEED spots due to the ( d 7 x d7)R19" structure were always as sharp as those for the substrate lattice. This indicates that the Fe overlayer structure is perfectly matched to the substrate Ru lattice structure during the growth of two- dimensional Fe islands [see fig. l(c)]. On the other hand, in accordance with the development of the second Fe overlayer, similar to b.c.c. Fe( 110) structure above 1 monolayer, the extra spots due to the ( 4 7 x 47)R19" nitrogen overlayer became diffused, but were sharpened by annealing to 800 K. Fig. 2 shows a series of N, thermal desorption spectra obtained from ammonia adsorption on an epitaxial surface at 420 K as a function of Fe coverage. For Fe coverages below 0.5 monolayer the desorption peak at 870 K grew with an increase in Fe coverage.On increasing the coverage to ~ 0 . 5 monolayer, a new desorption peak at 970 K developed in addition to the peak at 870 K, and the former became predominant above 0.7 monolayer. The drastic change in the thermal desorption spectrum at an Fe coverage of ca. 0.5 monolayer is entirely different from the continuous peak shift towards higher temperatures observed with the thermal desorption spectra of C0,13 which may be caused by a change in the electronic structure of the Fe overlayer resulting from the size of the islands.324 Adsorption and Decomposition of Ammonia on Fe - m Do .- z 600 700 800 900 1000 T/K Fig. 2. N, thermal desorption spectra from NH, adsorption on epitaxial Fe overlayers on an Ru(001) surface. 10 L exposure; = 8.3 K s-l; Fe coverage: (a) 0.15, (b) 0.5, (c) 0.6, ( d ) 0.75 and (e) 1.5 monolayer. The desorption peaks in each region are characterized by their narrow width (70 K), and exhibit peak maxima at essentially the same temperature regardless of coverage.Therefore, the desorption of N, is likely to be rate-limited by the decomposition of surface nitrides such as ‘Fe4N’, proposed in the adsorption of N, on Fe( 11 1) and (1 10) surfaces,l where the desorption of N, by the decomposition of the surface nitride appeared in the temperature range 850-950 K, depending on the single-crystal plane, while bulk nitrides contributed only to a continuous increase in background at temperatures above 1000 K. Although there are differences in the complex LEED patterns on the (1 lo), (100) and (1 1 1) planes, the metal-nitrogen bonding energies on the three planes become similar through a combination of the activation energies for adsorption with those for desorption.In contrast, a sudden shift in the N, thermal desorption spectra from the Fe overlayer on the Ru(OO1) surface shown in fig. 2 was seen, despite the observation of the same LEED structure for nitrogen atoms over a wide range of Fe coverages on the Ru(OO1) surface. Although the drastic change in the thermal desorption spectra is not easily interpreted at present, the possibility is precluded that the higher-temperature peak may be attributed to the desorption of bulk nitride, because it showed a sharp peak width which is completely different from that on Fe single-crystal surfaces.Moreover, it was observed even below an Fe coverage of 1 monolayer while the lower-temperature peak was still absent, which cannot be interpreted by othe bulk nitride alone. Note that the nearest-neighbour metal-metal distance (2.68 A) at the (1 11) suzface of the f.c.c. Fe4N bulk nitride is close to that on the h.c.p. Ru(OO1) plane (2.706 A). Accordingly, the ( d 7 x d 7 ) R19” surface structure of the nitrogen overlayer can be stabilized without the reconstruction of an Fe p(1 x 1) overlayer, as displayed in fig. 1 (c), where each nitrogen atom occupies threefold hollow h.c.p. sites, binding to the three nearest Fe atoms (filled circles) of the Fe(1 x 1) commensurate overlayer. On this adsorption structure the interaction with Ru atoms in the underlying substrate layer can also occur.In contrast, the (d7 x d 7 ) ordered diffraction pattern was not observed on the surface pre-annealed to 1030 K, where (1 x 1) spots were visible, as in the case of the (2 x 2)C. Egawa, K. Sawabe and Y. Iwasawa 325 ordered structure due to D-CO states (dissociated CO).13 This is interpreted by the dispersion of Fe atoms from islands over the surface, leading to the formation of various mixed sites composed of Fe and Ru atoms. The situation is consistent with a change in the N, thermal desorption spectra from the annealed surface, where a broadening of peak shape and a shift towards lower temperatures were observed. On the other hand, following the adsorption of ammonia on the surface pre-annealed to 900 K, an intermediate ordered structure was observed as shown in fig.1 (b). Since the diffraction spots due to the adsorbate structure are restricted on the ring containing the first fundamental spots, an unambiguous assignment could not be made; however, it is probably derived from a ( d 3 7 x 2/37)R25" structure. Since a study of CO adsorption has revealed that such a treatment (annealing to 900 K before adsorption) mainly causes a modification of the electronic structure without any change of the surface structure of the Fe over la ye^,'^ the treatment may induce a change in the attractive and/or repulsive interactions among adsorbates and hence lead to the formation of a differently ordered structure. Decomposition of Ammonia on the Fe/Ru(001) Surface From studies of the steady-state decomposition of ammonia over Ru(OO1)l' and Fe foil'' at pressures around 1 x Pa and temperatures between 400 and 900 K under flow-reactor conditions, the reaction mechanism may be written as follows : ( i ) NH3(g) + N(ads) + 3H(ads) Since both the reaction rate and surface nitrogen coverage at steady state are independent of hydrogen partial pressure, the backward rate in step (i) is not fast and equilibrium is not established. In this mechanism, with an increase in surface nitrogen coverage, the rate of the first step decreases while the rate of step (ii) increases, and thus a steady-state reaction rate and nitrogen coverage are attained under conditions where both reaction rates are equal.The phenomena observed in fig. 2, which are related to the metal-nitrogen bonding energies, were also observed on the annealed surface.Here, step (i) for the formation of nitrogen atoms from ammonia molecules (ca. 20 L) was examined on a surface pre-annealed to 1030 K. The rate of formation of nitrogen atoms was obtained by integrating the area of the desorption peak of nitrogen. The experiments were carried out on surfaces with Fe coverages between 0 and 1 monolayer (less than half the saturation coverage of nitrogen) in order to obtain a rate of ammonia dissociation as near to zero nitrogen coverage as possible. Typical results are shown in fig. 3 for the surface annealed with 0.3 monolayer of Fe. As the adsorption temperature was raised from 300 to 420 K, the integrated desorption area, i.e. the concentration of nitrogen atoms, increased, with the peak maximum shifting towards lower temperatures.This may be interpreted as a first-order desorption with activation energy varying with coverage, which is caused by the formation of various mixed sites of Fe and Ru atoms. Since the formation of surface nitrogen is thought to be an activated process, the rate of formation of surface nitrogen was plotted against the reciprocal temperature in order to evaluate the activation energy. The results are shown as a function of Fe coverage in fig. 4. As the iron coverage increased from 0 to 1 monolayer, the activation energies decreased from 33 to 8 kJ mol-I. The value obtained for the clean Ru(OO1) surface was a little smaller than that (48 kJ mol-') reported in the literat~re.~ Assuming an adsorption equilibrium for ammonia on the surfaces (which is reasonable on the basis326 Adsorption and Decomposition of Ammonia on Fe 700 800 900 Fig.3. N, thermal desorption spectra from NH, adsorption on an annealed Ru(001) surface with 0.3 monolayer Fe coverage. 20 L exposure; adsorption temperature: (a) 300, (b) 320, (c) 335, ( d ) 360, (e) 370, cf) 395 and (g) 420 K. TIK 40 7 - 30 E 0 c, \ Y x 20 5 K .* * m > .- * 2 10 0 \ O\O I . . . . l 0 0.5 1 Fig. 4. Activation energies for the formation of surface nitrogen from ammonia as a function of Fe coverage. Fe coverage/monolayer that the desorption temperature of ammonia is below 250 K), the decrease in the activation energy is partially explained by an increase in the adsorption energy of molecular ammonia on the Fe overlayer, because the desorption temperature (250 K) of ammonia on Fe surfaces3 is higher than that (183 K) on an Ru(OO1) ~urface.~ This indicates that the adsorption energy of ammonia increased by 25 kJ mol-l. In addition, there is a stabilization of the surface nitrogen which reduces the activation energy of the dissociative adsorption of ammonia for Fe overlayers on the Ru(001) surface, as shown in fig.2. As revealed above, both elementary surface reaction rates were affected by an increase in Fe coverage. The rate of decomposition of ammonia at steady state was measured under ca. 5 x Pa of ammonia at three different temperatures (620, 735 and 850 K)C. Egawa, K. Sawabe and Y. Iwasawa 327 0.5 1 Fe coverage/monolayer I " ' " ' " " 0.5 1 Fe coverage/monolayer Fig.5. Decomposition rates of ammonia on Fe/Ru(001) surfaces. Reaction temperature : 0, 0, 620; A, A, 735; 0, ., 850 K. (a) Epitaxial overlayer, (b) overlayer pre-annealed to 1030 K. 735 800 900 1000 T/K Fig. 6. N, desorption curves from the steady state of reaction at 735 K as a function of Fe coverage. (a) 0.11, (b) 0.33, (c) 0.5, ( d ) 0.66 and (e) 0.83 monolayer. as a function of Fe coverage, as shown in fig. 5 for both sets of epitaxial ( a ) and annealed (b) surfaces. Although some scatter was present in the data, the behaviour was similar for both surfaces. The decomposition rates increased with Fe coverage and then decreased. The coverage of Fe giving the maximum decomposition rate shifted to higher values with an increase in the reaction temperature.On the Ru(OO1) surface'' the maximum rate was observed at ca. 600 K, so that the decomposition rate decreased with an increase of the reaction temperature employed in the present study. The presence of an Fe overlayer at low coverages enhanced the overall reaction rates by accelerating the rate of formation of surface nitrogen from ammonia, whereas the rate of desorption of328 Adsorption and Decomposition of Ammonia on Fe (a) ( b ) Fig. 7. (a) LEED pattern of the 44 x 2) structure obtained from 0, adsorption on a 1030 K pre- annealed Fe/Ru(OOl) surface. (b) Surface-structure model of the 44 x 2) structure on 0.75 monolayer Fe overlayer. 0, Fe atom; 0, 0 atom; 0, N atom. N, was reduced owing to the increased stability of the surface nitrogen; the decomposition was suppressed at higher Fe coverages.Since the activation energy for the desorption of N, is much larger than that for the formation of surface nitrogen from ammonia, and moreover since the overall reaction rate is limited by the desorption step, the maximum reaction rates shifted towards higher Fe coverages as the reaction temperature increased. Since the formation of surface nitrogen from ammonia and the desorption of nitrogen are dynamically balanced at steady state, the steady-state surface nitrogen coverage, which is related to the degree of stabilization, is expected to increase with Fe coverage, as is demonstrated in fig. 6. Moreover, the effect of varying the partial pressure of hydrogen on the reaction rate was not observed for PH,/PNH, ratios up to 2, and surface hydrogen coverage at steady state was negligible (below the detection limit of coverage, similar to the results of ammonia decomposition on Ru(OO1)" and Fell surfaces.Decomposition of Ammonia on an Fe/Ru(001) Surface with Co-adsorbed Oxygen The adsorption of oxygen molecules (3.6 L) at 420 K on a surface pre-annealed to 1030 K with Fe coverages > 0.5 monolayer gave sharp, ordered extra diffraction spots as shown in fig. 7(a). It corresponds to a c(4 x 2) ordered structure on the h.c.p. (001) surface as depicted in fig. 7(b), where atomic oxygen is located on threefold hollow sites similar to a p(2 x 2) oxygen structure on a clean Ru(OO1) surface.14 The surface oxygen coverages for both surface structures are estimated to be 0.25.The subsequent adsorption of ammonia on the oxygen-preadsorbed surface at 420 K did not change the extra spots arising from the c(4 x 2) structure. On the other hand, following the adsorption of O,, the p(2 x 2) LEED pattern, which was sharp on a clean Ru(OO1) surface, became streaky and diffused on the epitaxial surface as the Fe coverage was increased. Subsequent exposure to ammonia produced the ( d 7 x 2/7)R19" LEED pattern, and an intermediate LEED pattern as shown in fig. 1 (b), probably with a (d37 x 437) R25" structure, appeared for the surface with 0.5 monolayer Fe coverage. Taking the island growth of the Fe overlayer into consideration, these results indicate that oxygen atoms preferentially adsorbed on the Ru patches and surface nitrides on bare Fe domains uncovered by oxygen coexist with disordered overlayers of atomic oxygen.Consistently, the sticking coefficient of 0, on an Fe( 1 1 0)15 decreases to 0.1 above 0.1 monolayer coverage, whereas a relatively constant value (0.6-0.8) is reported on Ru(001).16 The decomposition of ammonia on a surface with pre-adsorbed oxygen has also been investigated. The effects of co-adsorbed oxygen on the desorption of N, at both epitaxialC. Egawa, K. Sawabe and Y. Iwasawa 329 600 700 800 900 1000 TI K Fig. 8. N, thermal desorption spectra from NH, adsorption on modified Fe/Ru(OOl) surface with Fe coverage of 0.75 monolayer. 10 langmuir NH, exposure, /? = 8.3 K s-l; (a) surface annealed to 1030 K : (b) surface pre-adsorbed with 0,, followed by H, reduction; (c) surface annealed to 1030 K and pre-adsorbed with 0, followed by H, reduction.(6) and annealed (c) surfaces are shown in fig. 8. For comparison with steady-state decomposition of ammonia, the adsorption of oxygen (3.6 L) on the epitaxial and annealed surfaces was followed by H, treatment (2.7 x lop5 Pa at 900 K for 3 min) before NH, adsorption to remove excess oxygen and to make the surface equivalent to steady-state reaction conditions. From AES measurements the surface oxygen coverage on both (b) and (c) surfaces was estimated to be 0.2. The temperature for the maximum desorption of N, was successively lowered in the order ( a ) (pre-annealed and without pre-adsorbed oxygen), ( b ) and (c), with a concomitant reduction in surface nitrogen coverage. This indicates that the bonding energy of the nitrogen atom was reduced by 30 kJ mol-' by the co-existence of oxygen atoms.The trend of lowering of desorption temperatures was closely related to the change in the rate of decomposition of ammonia at steady state on these surfaces. The relative decomposition rates at 735-780 K were observed to be (a):(b):(c) = 1 :2.5:3.5. This is consistent with the result that the desorption of nitrogen is a rate-limiting step in the decomposition of ammonia.'O As shown in fig. 8, the enhancement of nitrogen desorption by the co-adsorbed oxygen was most effective on surface (c), where the ordered c(4 x 2) structure was observed for the mixed overlayer of nitrogen and oxygen. The ordered structure disappeared in accordance with the desorption of N,.In contrast, p(1 x 1) spots due to the substrate lattice were only observed with high background intensity on surface (b), which indicates a disordered structure for the oxygen-nitrogen mixed overlayer. Accordingly, it is likely that the nitrogen atom occupying the central position of a c(4 x 2) unit structure is effectively activated by the surrounding co-adsorbed oxygen atoms, as 2hown in fig. 7 (b), where the distance between nitrogen and oxygen atoms is 4.69 or 5.41 A. This separation may be a reasonable one to provide the interaction between neighbouring adatoms. Thus in the c(4 x 2) structure the co-adsorbed oxygen atoms may electronically modify the Fe overlayer and also activate the surface nitrogen, resulting in an enhancement of the rate of decomposition of ammonia.330 Adsorption and Decomposition of Ammonia on Fe Conclusions The results obtained in this study are summarized as follows.(1) Adsorption of ammonia at 420 K on epitaxial Fe overlayers on an Ru(OO1) surface produced an ordered ( 4 7 x 2/7)R19' structure over the range of Fe coverage from 0.14 to 2 monolayer owing to surface nitrides similar to Fe,N proposed in Fe single-crystal surfaces. (2) The surface nitride was desorbed.at 870 K below an Fe coverage of 0.5 monolayer, while this takes place above 970 K on the surface with coverage above 0.6 monolayer. (3) The activation energy for the formation of surface nitride from ammonia decreased from 33 to 8 kJ mol-1 with an increase of Fe coverage from 0 to 1 monolayer. (4) The rate of formation of surface nitrogen from ammonia increased with increasing Fe coverage, while that of N, desorption decreased.As a result, the maximum rate was achieved at an optimum Fe coverage. (5) Adsorption of 0, on an annealed Ru(OO1) surface with an Fe coverage >0.5 monolayer led to a c(4 x 2) ordered structure. Subsequent ammonia adsorption produced a c(4 x 2) mixed overlayer (nitrogen and oxygen) structure. (6) The rate of ammonia decomposition was enhanced 3.5 times by co-adsorbed oxygen in the c(4 x 2) mixed structure. References 1 F. Bozso, G. Ertl, M. Grunze and M. Weiss, J. Catal., 1977, 49, 18; F. Bozso, G. Ertl and M. Weiss, 2 N. D. Spencer, R. C. Schoonmaker and G. A. Somorjai, J. Catal., 1982, 74, 129. 3 M. Grunze, F. BOZSO, G. Ertl and M. Weiss, Appl. Surf. Sci., 1978, 1, 241; M. Weiss, G. Ertl and 4 L. R. Danielson, M. J. Dresser, E. E. Donaldson and J. T. Dickinson, Surf. Sci., 1978, 71, 599. 5 J. J. Vajo, W. Tsai and W. H. Weinberg, J. Phys. Chem., 1985,89,3243; W. Tsai, J. J. Vajo and W. H. 6 W. L. Guthrie, J. D. Sokol and G. A. Somorjai, Surf. Sci., 1981, 109, 390. 7 A. P. C. Reed and R. M. Lambert, J. Phys. Chem., 1984, 88, 1954. 8 R. W. McCabe, J. Catal., 1983, 79, 445. 9 A. Vavere and R. S . Hansen, J. Catal., 1981, 69, 158. J. Catal., 1977, 50, 519. F. Nitchke, Appl. Surf. Sci., 1979, 2, 614. Weinberg, J. Phys. Chem., 1985,89, 4926. 10 C. Egawa, T. Nishida, S. Naito and K. Tamaru, J. Chem. Soc., Faraday Trans. I , 1984, 80, 1567; 11 G. Ertl and M. Haber, J. Catal., 1980, 61, 537. 12 C. Egawa, T. Aruga and Y . Iwasawa, Surf. Sci., in press. 13 C. Egawa and Y. Iwasawa, S u f . Sci., in press; C . Egawa and Y. Iwasawa, Chem. Lett., 1987, 959. 14 C-M. Chan and W. H. Weinberg, J. Chem. Phys., 1979,71,2788; T. S . Rahman, A. B. Anton, N. R. 15 T. Miyano, Y. Sakisaka, T. Komeda and M. Onchi, Surf. Sci., 1986, 169, 197. 16 T. E. Madey, H. A. Engelhardt and D. Menzel, Surf. Sci., 1975, 48, 304; G. Praline, B. E. Koel, 1595. Avery and W. H. Weinberg, Phys. Rev. Lett., 1983, 51, 1979. H-I. Lee and J. M. White, Appl. Surf. Sci., 1980, 5, 296. Paper 71484; Received 17th March, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400321

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1988, 84(1), 331-342 The Dehydration of Na,S,O, 5H,O Single Crystals as studied by Thermal Analysis and Optical Microscopy Giulio G. T. Guarini" and Sandro Piccini Department of Chemistry, University of Florence, Via G . Capponi, 9, 50121 Firenze, Italy The dehydration of cleaved single crystals of sodium thiosulphate pentahydrate in vacuum or in flowing dry N, has been studied employing optical microscopy and thermal methods. Evidence of the formation of a dihydrate as intermediate has been obtained in both dehydration and rehydration experiments. Comparative studies were made of reactions on single crystals cleaved parallel to three different crystallographic planes, but the dependence of the dehydration on face was ascertained only for the initial portions of vacuum or high-N, flow experiments.Microscopic evidence and the initial splitting of the constant-temperature dehydration thermal curves indicated that, at the onset of dehydration, the formation of a fully dehydrated layer on the crystal surface takes place. The subsequent evolution of the transformation is determined by the properties of this layer. These in turn appear to depend on the perfection of the cleaved surface as well as on the efficiency of the dehydrating agents. Dehydration mechanisms for strong and mild dehydrating conditions are suggested, based on the presence of the dehydrated surface layer and on the formation of the intermediate dihydrate. Kinetic analysis of the constant-temperature thermal dehydration responses gave a good fit to the data of the pentahydrate-to-dihydrate transformation by a contracting circle equation, while for the dihydrate-to-anhydrous reaction an n = 2 Avrami-Erofeev law was obeyed.The sensitivity of the transformation to the defect content of the crystals and to the changes in N, flow and temperature prevented the evaluation of reliable Arrhenius parameters, particularly for the first dehydration stage. However, this provides further support for the surface- layer mechanism proposed here, which is reputed to apply more generally to other dehydration reactions in place of the preceding nucleation-and- growth model. Studies of the dehydration of salt hydrates have been of great imp~rtancel-~ in the development of interpretative models for reactions of the type solid = solid+gas.After crystals containing water of crystallization have been submitted to dehydrating conditions, microscopy reveals that nuclei of dehydrated material have been formed on as-grown and/or cleaved surfaces ; thus nucleation and growth schemes have been advanced to account for the experimental observations. In such schemes it is usually assumed that the product forms at the interface between reactant and product (i.e. at the nucleus boundaries), so that the onset and progress of the transformation can be traced back to the origin of such an interface and to its progressive advance inside the unreacted material. Recent r e p o r t ~ , ~ - ~ originating through a reinvestigation of the classical systems studied by Garner,' such as the dehydration of alums, have, however, shown that for these reactions the role hitherto attributed to the nuclei now requires critical 33 1332 Dehydration of Na,S,O, * 5H,O Single Crystals reconsideration.Hence another sequence of events capable of accounting for the same kind of phenomena had to be sought. Experiments concerned with the rehydration of dehydrated surfaces studied by thermal methods were particularly informative, since they revealed a splitting5? of the rehydration exotherm into two distinct portions attributed to reactions in the surface layer and in the bulk. These findings, coupled with microscopic evidence concerning the surface changes termed ' orange-peel formation ' and ' enhanced nucleation ',4-67 indicated that such dehydration/rehydration reactions are controlled through the participation of an interphase layer having properties differing from those of the bulk phase of the same crystal.To find direct evidence of this layer and to characterize its role in the reactivity of solids, the present article reports a study of the dehydration of cleaved crystals of sodium thiosulphate pentahydrate. This reactant was also identified as being of interest through offering the possibility of observing the behaviour of three different cleavage surfaces (and therefore of three different two-dimensional structures) reacting under the same experimental conditions. Experimental Crystals of pro analysi sodium thiosulphate pentahydrate, as supplied by either Merck or Erba, were used directly. These well crystallized, regular-habit crystals were sufficiently large (ca.2 x 3 x 5 mm) to enable cleavage surfaces of sizes suitable for examination to be prepared easily. Cleavage was performed by impact using suitably oriented sharpened blades. Fracture was particularly facile for the (01 0) planes ; cleavage in the (001) and { 11 l} planes was comparatively much more difficult, and the surfaces obtained included a large number of defects such as steps, etc. The cleavage planes were identified by optical goniometry on a number of specimens. The surfaces thus obtained remained unaltered during exposure to the laboratory atmosphere for short times (ca. 10-15 min at room temperature), and therefore the cleavage was always performed immediately before submitting the cleaved crystals to the conditions of dehydration.In preliminary observations by optical microscopy of surface texture changes upon dehydration of (010) faces, a number of dehydrating agents were tested, ranging from absolute ethanol and propanol to silica gel, concentrated sulphuric acid and vacuum treatment. The behaviour of the observed surfaces changed remarkably with the effectiveness of the dehydrating agent. Hence, throughout the present report, both optical microscopy and thermal analysis dehydrations were performed using either nitrogen dried over silica gel, flowing at various rates (in the range 5-30 cm3 min-l) over the samples, or a (ca. 1 Pa) vacuum. Rehydrations of partially or wholly dehydrated crystals were invariably performed by exposing them to flows of nitrogen containing known amounts of water vapour added by bubbling the gas through thermostatted aqueous solutions of 0, 40, 50 and 60 wt% sulphuric acid.Details of individual experiments are given in the text. A Reichert Zetopan optical microscope, equipped with interference contrast and an Olympus OM2 camera, was used, mainly in the reflection mode, for all the optical microscopy experiments performed at ambient temperature (293 f 1 K) ; the observation cell employed containing the crystal has already been de~cribed.~ Dehydration and rehydration thermal curves were recorded by means of a Mettler TA 2000 thermal analyser, the dehydrating and rehydrating conditions being the same as in the optical microscopy experiments. The use of crystal slabs, obtained by cleavage of crystals of approximately the same size and weighing ca.12 mg, ensured good thermal contact with the flat surface of the aluminium sample pans. Corrections to account for the change in sensitivity of the thermocouples with temperature and for the time constant (z) were introduced, when necessary, during the computer analysis of the data. Linearly increasing temperature dehydration curves gave no information relevant to theJ. Chem. SOC., Faraday Trans. 1, vol. 84, part 1 Plate 1 ( C) ( d ) Plate 1. Sequence of micrographs illustrating the behaviour typical of (001) and { 11 1) cleavage faces during the initial stages of vacuum dehydration. See text (magn x 75). G. G. T. Guarini and S. Piccini (Facing p . 332)J . Chem. SOC., Faraday Trans. I , uol.84, part 1 Plate 2 (9) ( h ) Plate 2. Sequence of micrographs illustrating the behaviour of {OlO} cleavage faces in the initial stages of vacuum dehydration. Formation and growth of acicular reliefs is evident. See text (magn x 80). G. G. T. Guarini and S. PicciniJ . Chem. Soc., Faraday Trans. 1, vol. 84, part 1 Plate 3 ( b 1 Plate 3. (a) Correspondence of acicular reliefs on matched cleavage halves, indicating the effectiveness of crystal defects in the present transformation (magn x 65); (b) evidence of alignments of elongated reliefs also thought to be indicative of the presence and action of crystal defects (magn x 130). G. G. T. Guarini and S . PicciniJ . Chem. SOC., Faraday Trans. 1, vol. 84, part 1 Plate 4 (c 1 (d 1 Plate 4. Sequence of micrographs recording the formation and evolution of the reliefs on a (010) cleavage face under the action of a high nitrogen flow.See text (magn x 75). G. G. T. Guarini and S. PicciniJ. Chem. SOC., Faraday Trans. 1, vol. 84, part 1 Plate 5 ( d ) (el Plate 5. Sequence of micrographs recorded during the dehydration of a (010) cleavage face by a low nitrogen flow. See text (magn x 75). G. G. T. Guarini and S. PicciniJ. Chem. SOC., Faraday Trans. 1, uol. 84, part I Plates 6 and 7 Plate 6. (010) cleavage surface retexturing observed during the dehydration by low nitrogen flow and simulating the shape of a grown relief (magn x250). ( b ) Plate 7. Micrographs illustrating the resumed growth of dehydration nuclei after the extension to the whole surface of the retexturing shown in plates 5 and 6 (magn x 100).G. G. T. Guarini and S. PicciniJ . Chem. Soc., Faraday Trans. 1, vol. 84, part 1 Plate 8 ( C 1 (d 1 Plate 8. Sequence of micrographs illustrating a partial vacuum dehydration of a (010) cleavage face [(a) and (b)] followed by rehydration by means of nitrogen bubbled in 40 % sulphuric acid [(c) and (41. The reorganization of the surface upon rehydration should be compared with that reported in plates 2 ( g ) and (h) (magn x 80). G. G. T. Guarini and S. Piccini333 G. G. T. Guarini and S. Piccini present report except for the determination of the expected6 enthalpic change of ca. 280 kJ mol-l, corresponding to complete dehydration, which was deduced from some experiments at the lowest heating rate of 0.1 K min-’.For a determination of kinetic parameters, the thermal curves corresponding to isothermal dehydration and rehydration experiments were recorded at a number of temperatures in the range 290-313 K comprising, for the sake of comparison, the temperature of the corresponding runs by optical microscopy. Results Optical Microscopy Experiments To present systematically the large amount of data it is convenient to classify under separate headings the behaviour observed under different dehydrating conditions. Vacuum Dehydrations The behaviour of the (001) and { 11 l } surfaces was found to be almost identical, and a typical record for a { 11 l} cleavage surface is reported in plate 1. Soon after the onset of evacuation the surface undergoes reorganization, starting near steps or on the bare surface [A and B, respectively, in plate 1 (b)].This change of texture spreads rapidly over the surface while crystallization (nucleation) of dehydrated product begins at localized spots [e.g. C in plate 1 (b) and (c)]. These areas then spread (growth) until they cover the whole surface with a microcrystalline material that strongly diffuses light, thereby preventing further observation of the changes accompanying reaction. The initial behaviour of the (010) surface in vacuum is remarkably different from that described above, but later becomes very similar, as shown by the sequence of photographs in plate 2. As soon as the prevailing pressure of gas is reduced at the start of reaction, a number of iso-oriented rectangular acicular reliefs having their long side parallel to the direction of the c axis appears [plate 2(b)].These grow rapidly (longitudinal growth faster than transverse), eventually giving an undulated aspect to the surface [plates 2(c) and (d)]. Only at this stage is the reorganization observed on the (00 1) and { 1 1 1) surfaces initiated [the left-hand side of plate 1 (d) and the central portion of plates 2 ( f ) , ( g ) and ( h ) should be compared], and this advances across the undulating surface [plates 2 ( 4 , (e) and (f)], again followed by the formation of ‘nuclei’ of dehydrated material [plates 2 ( g ) and (h)]. While discussion concerning the possible nature of both surface reorganizations will be postponed, it is appropriate to mention that the correspondence of acicular reliefs on matched cleavage halves [plate 3(a)] and the strong alignments observed [plates 3(b) and 4(b)] suggest control by lattice defects.Dehydration in a Flow of Dry Nitrogen The textural changes observed depended sensitively on the precise reaction conditions relating to flow rate and temperature. The results reported below are presented in order of decreasing flow rates. At high flow rates of dry nitrogen ( 3 30 cm3 min-l) the behaviour of the (001) and { 1 1 1) surfaces was similar to that described for vacuum dehydrations. The formation of acicular reliefs on the (010) faces proceeds as before, but the subsequent development differs from that observed under vacuum. Not only is the observed growth of the reliefs much slower (as expected), but almost all of them undergo a superficial retexturing that appears as indentation.This does not prevent their growth but results, when these acicular reliefs eventually meet, in a very grained surface [plates 4(a)-(d)]. In some cases334 Dehydration of Na,S,03 * 5H,O Single Crystals [A in plate 4(b)] the surface of the reliefs appears to disintegrate, and from these sites product crystallization (nucleation) is initiated. At flow rates in the range 10-30 cm3 min-l the dehydration shows more complicated behaviour which, however, can be considered as an interplay of the reactions characteristic of high and low flow rates. When low (< 10 cm3 min-l) flow rates of dry nitrogen are used the acicular reliefs do not form on the (010) faces and the behaviour of the three cleavage surfaces becomes very similar, showing modifications that are identical to those already described in many rehydrations and termed ‘ orange-peel formation ’.4* However, in contrast to previous reactions where this phenomenon has been observed during rehydration and takes place almost contemporaneously over the whole surface, in the present instance (the second case of ‘orange-peel formation ’ observed in dehydration6) the textural change spreads relatively slowly across the surface, often starting from a crystal edge parallel to the c axis [plates S(a)-(e)] or from some unidentified imperfection on the surface (plate 6).In the latter case the overall shape of the modified area recalls that of a single, vacuum- grown acicular relief. When the ‘ orange peel ’ advancing boundary meets the growing ellipsoidal ‘ nuclei ’ which originate almost contemporaneously, the former almost stop the latter’s growth, as shown in plate 5.When the ‘orange peel’ has entirely covered the surface, the progress of the reaction is revealed only by the resumed growth of the ‘nuclei’ which, however, tend to become round, white and with diffuse borders (plate 7). Rehydra t ions Apart from the presence of acicular reliefs on the (010) surfaces when the preliminary dehydration is performed by strong dehydrating agents, the rehydration behaviour of the three cleavage surfaces is almost identical and is exemplified by the sequence in plate 8 , which shows a partial vacuum dehydration of a (010) face [plates 8(a) and (b)] followed by admission of nitrogen bubbled through a 40 % water solution of sulphuric acid [plates 8(c) and (d)].By comparison of the top left of plate 8(c) with plates 2(g) and (h), it appears that the presence of water vapour enhances the same kind of surface reorganization as previously observed ; comparison with plate 4 ( d ) indicates that surface rehydration also takes place under high nitrogen flow. The rehydration of surfaces partially dehydrated by mild treatments appears to be limited to a weakening of the features observed in dehydration. Thermoanalytical Determinations Classical studiesg-” have indicated that the system sodium thiosulphate-water is complex, and previous thermoanalytical mea~urements~~-~~ have characterized its behaviour. Indeed, we have verified that small changes in experimental conditions may lead to apparently contrasting results.Variable-temperature dehydrations have been briefly described in the experimental section. Is0 thermal Dehydrations Complete decomposition (confirmed by weighing the residue) of sodium thiosulphate pentahydrate was performed in the previously mentioned temperature range either in vacuum or in various dry nitrogen flows. The behaviour of crystal slabs was found to be almost independent of the type of cleavage faces exposed. The overall shape (fig. 1) of the thermal response curve is composed of two main endotherms, whose separation may be complete or incomplete according to the temperature and dehydration conditions. However, at any temperature the separation improves on reduction of theG.G. T. Guarini and S. Piccini 335 1 I I I 1 I 0 5 10 15 20 tlh Fig. 1. Thermal curve of the isothermal ( T = 303 K) dehydration of a (010) doubly cleaved crystal slab of the pentahydrate under a dry nitrogen flow of 20 cm3 min-l. The two main endotherms correspond to the loss of three and two water molecules, respectively. The narrow peak indicated by the arrow has been arbitrarily added to illustrate the initial behaviour of the dehydration under vacuum or high nitrogen flow. dry N, flow. Weight measurements in interrupted experiments and planimetric evaluation of the fraction decomposed (a) corresponding to the separate enthalpic contributions indicate that the two endotherms are due to the loss of three and two water molecules, respectively. This is considered a relevant indication of the stability of a dihydrate.In dry N, dehydrations the mean value of the overall enthalpic change was 281 8 kJ mol-l, almost coincident with that found in the variable-temperature experiments. Because of instrumental characteristic^,'^ lower values (250 f 20 kJ mol-') were recorded in vacuum experiments. When vacuum or high N, flows were employed, an extremely narrow initial small subpeak could be detected (indicated by the arrow in fig. I). This behaviour was never observed using medium or low nitrogen flows; however, by analogy with microscopic examinations, at high nitrogen flows a significant splitting of the initial portion of the dehydration could be achieved and was attributed to the dehydration of the surface layer.16 Also, the time needed to complete the initial subpeak was equivalent to the time necessary, under comparable conditions, to reach the beginning of surface rehydration in microscopy experiments [e.g.plate 2 (41. Isso thermal Reh y drat ions Again no difference was observed in rehydrating the dehydration product prepared using crystals doubly cleaved parallel to one or other of the (OOl), (010) and (111) planes. The amount of water contained in the nitrogen flowing over the dehydrated crystals is obviously a further variable (besides temperature and flow rate) controlling the behaviour of the system. In practice any result, ranging from incomplete rehydration to deliquescence, can be obtained from appropriate changes of these variables. Some examples will illustrate this point.At 293 K, or slightly higher temperatures, the rehydration of dehydrated crystals using nitrogen bubbled through water proceeds to336 Dehydration of Na,S,O, * 5H20 Single Crystals 1 I I I I I I 5 10 15 20 25 tlh Fig. 2. Rehydration exotherm showing the separate recovery of two and three water molecules by the completely dehydrated material under a flow of nitrogen bubbled in a 60% solution of sulphuric acid up to the point marked * and thereafter in a 40% solution. deliquescence. In contrast, at 303 K with the same wet N, flow, water uptake is completed when it corresponds to re-formation of the pentahydrate. At lower partial pressure of water, achieved by bubbling the nitrogen through sulphuric acid solutions, rehydration was generally separated into two contributions (fig.2) : the first corresponded to the incorporation of two water molecules into the salt (and again gave indication of some stability of the dihydrate), while the second was due to the further incorporation of three. Excluding the occurrence of deliquescence, the total enthalpic changes measured were always close to the values determined in dehydration experiments. The influences of variations of nitrogen flow were also tested in a number of experiments, the main result being the confirmation of the relative stability of the dihydrate in various rehydrating conditions. Discussion The results reported above clearly indicate that the dehydration of Na,S20, - 5H,O crystals is a complex process. However, by considering together both optical and thermal results, we believe that some insight into the mechanism of these reactions can be gained.Often the mechanism of a transformation is postulated from the obedience of the kinetic results to a particular model equation. In the present system a mechanism based on microscopic and thermal evidence is first suggested. Support from kinetic data is then discussed. The Mechanism of Dehydration The following interpretative suggestions are based on the formation, as an intermediate, of sodium thiosulphate dihydrate ; its existence, already reported,14 is confirmed here by our thermal measurements during both dehydration and rehydration at constant temperature. Moreover, comparisons between the time needed to reach complete surface coverage by the product (i.e.the completion of experiments by optical microscopy) and the durations of thermal runs in comparable conditions indicate that in the former case dehydration cannot have exceeded the dihydrate stage. Strong Dehydrating Conditions The initial formation of elongated acicular reliefs on the (010) face is the only significant difference observed between the dehydration and rehydration behaviour of crystal slabsG. G. T. Guarini and S . Piccini 337 doubly cleaved parallel to one or other of the three cleavage planes. No differentiation was expected in rehydrations, as the dehydrated product retains only the external appearance of the original crystal. However, in dehydrations the observed lack of dependence on surface structure was surprising. From the above results we are forced to believe that the transformation itself is unaltered by such structural changes.Thus the only observed differentiation (i.e. the formation of elongated acicular reliefs) appears as a consequence of a situation having only limited bearing on the initial stages of the true reaction (‘true reaction’ meaning detachment of the water molecules from their bonding within the structure). It may, however, be effective in controlling the further development of the system. Evidence from studies on other crystal~’~ leads us to identify the above situation with the perfection of the cleaved face. We term perfect a face having the least possible number of both original and artificially introduced defects, including steps, points of emergence of dislocations, scratches, indentations, etc.Indeed we believe that the first event in the dehydration of sodium thiosulphate pentahydrate (and of many other crystal hydrates) is the formation on all the external crystal surfaces (and independent of their orientation) of a fully dehydrated surface layer. This process, now confirmed in some instances1’ by the splitting of the initial dehydration endotherm (fig. l), accounts only for ca. 1 % of the entire reaction6 but may determine its subsequent evolution. Indeed, further transfer to the gas phase of the reaction product must take place through this layer and will be influenced by its continuity and permeability which, in turn, appear to depend more on its perfection than on the orientation of the original cleavage face on which it forms.In the present reactant the most perfect surfaces are obtained by cleavage parallel to (010). As a consequence, the permeability and continuity of the dehydrated layer formed on such surfaces is expected to prevent an easy escape of the inner gaseous product. Subsequently, within the interface between this superficial layer and the pentahydrate crystal bulk, accumulation of water occurs particularly at sites to which lattice defects of the parent crystal permit easy diffusional migration of released water molecules. Within such zones (both defective’’ and containing water that promotes crystallization”) the formation of crystalline dihydrate may take place. An indirect proof of the above is given by the relevant correspondence of acicular reliefs on cleavage- matched halves [plate 3 (a)].Subsequent growth of this intermediate product by concurrent dehydration of the underlying pentahydrate and rehydration of the superposed anhydrous layer proceeds preferentially in the c direction of the penta- hydrate, probably because of the favourable disposition of the water molecules,21 and generates the deformations of the dehydrated layer referred to here as acicular reliefs. This explanation is suggested and supported by analogy with observations concerning the photodimerization of anthracene crystals in air.22 In this reaction [see in particular fig. 4 of ref. (22)] the product (dimer) forms beneath a surface layer which is then deformed and eventually breaks, as in the present reaction [plate 4(b)]. Following the generation and development of the acicular reliefs, the observed behaviour of the (010) cleavage faces is comparable with that found on the (111) and (001) cleavage planes.The observed retexturing is attributed’ to a generalized rehydration of the initially dehydrated surface layer. Because of both the perfection and the thickness of the layer, this process is delayed on the (010) surface with respect to the other imperfect [as a careful inspection of the ‘ pseudo-flat ’ zones of plate 1 ( a ) will show] cleavage planes. However, the suggested comparison of plates 2 ( g ) and (h) and the top left of plate 8(c) shows that the process can be considered to be the same in both cases. In fig. 3 a schematic diagram illustrates the above mechanism. The reacting crystal is seen sectioned along a hypothetical plane perpendicular to the surface submitted to dehydrating conditions.First a dehydrated surface layer of anhydrous salt, thickness 6,338 Dehydration of Na,S,O, - 5H20 Single Crystals /,I- t N a *S *03 - 5 H 2 0 Na2%03 - 2 H 2 0 Na2S203 Fig. 3. Schematic view of the mechanism suggested here for the dehydration of (010) faces of sodium thiosulphate pentahydrate under strong dehydrating conditions. The initial formation of a dehydrated layer of thickness 6 is shown. is formed; subsequently the progress of the system through the intermediate formation of dihydrate to complete dehydration takes place. Poor Dehydrating Conditions Microscopic observations indicate a different dehydration pathway, but the behaviour is common to all the three cleavage surfaces investigated [plates 5 (a)-(e) and 61.Textural changes are strictly similar to the 'orange-peel formation ' observed in rehydrati~n,~? and is attributed to the rehydration (probably to dihydrate) of the initially fully dehydrated surface layer by water molecules derived from the subsurface decomposing pentahydrate. This process might also release the stress present in the dehydrated surface layer because of the removal of one of the lattice constituents. Support for this interpretative hypothesis is provided by the behaviour of the ellipsoidal nuclei which form (at defective positions where there is an ample supply of water molecules) at the same time as the retexturing but stop growing when surrounded by the retextured surface.The subsequent resumption of growth is believed to be due to the later dehydration of the dihydrate to anhydrous salt, as suggested by the change in colour and shape of the nuclei. The Kinetics of Dehydration The dehydration mechanism outlined above is further supported by the results of the kinetic analysis of the constant- temperature thermal dehydrations performed under dry nitrogen flow. According to the suggested mechanism of dehydration, the kinetics of this complex process involving concurrent reactions could not have been described by those rate equations based on nucleation and growth models1 which are strongly influenced by the nucleation step. Assuming that the formation of the dehydrated layer is equivalent to the ' instantaneous nucleation ', discussed by Jacobs and T ~ m p k i n s , ~ ~ we expected contracting-envelope kinetics to apply to the present reaction after initiation across all surfaces.Furthermore, particularly for the escape of water through the almost perfect (010) surface layers, the possibility of diffusional control had to be considered.G. G. T. Guarini and S. Piccini 339 1 .o 0.8 -%. d A 0.6 v I - 0 . 4 0.2 120 240 I 360 tlmin Fig. 4. Fitting of the first dehydration stage by the contracting-circle equation; results recorded under various N, flows ( F in cm3 min-l) and temperatures (T in K) are plotted to illustrate the dispersion of rate data discussed in the text: (a) T = 296, P = 13; (b) T = 303, F = 25; (c) T = 303,F= 1 5 ; ( d ) T = 3 0 3 , F = 2 0 ; ( e ) T = 2 9 8 , F = 2 0 ; C f ) T = 2 9 3 , F = 2 0 ; ( g ) T=303,F= 10; (h) T = 303, F = 15.Initial attempts to perform the kinetic analysis of the overall constant-temperature thermal curves in terms of contracting-envelope and diffusion equations or combinations of these2* did not give an acceptable kinetic fit to our data. However, the shape of the thermal curves and the thermal evidence of the formation of the intermediate dihydrate suggested that an extension of the method suggested by Tang25 for reactions occurring with different mechanisms in different a ranges could be applicable here. The whole a range was then divided into two portions (the first 60% corresponding to the reaction from pentahydrate to dihydrate and the final 40% corresponding to the reaction from dihydrate to anhydrous), and to each portion a new reaction variable a, and a2, respectively) with 0 < ai < 1 ( i = 1,2) was assigned.These data were then tested for obedience to the set of kinetic equations usually applied in rate studies of the reactivity of solids3 for both a1 and a2 ranges. Throughout the whole range of temperatures and nitrogen flows, a contracting-area equation gave a good fit (fig. 4) for the greatest part (ca. 0.08 < a, < 0.80) of the first stage, thus giving the expected support to the suggested dehydration mechanism. The best fit to the second stage (ca. 0.05 d a2 d 0.8) was obtained using an n = 2 Avrami-Erofeev equation (fig. 5). While the latter finding was unexpected, it can be justified [see ref. (24), chap. XI if it is assumed that the second dehydration stage takes place on an extremely fine-grained product of the preceding decomposition, whose crystallites must be closely packed because the external shape of the original parent crystal is preserved to the end of dehydration.Unfortunately, even if the rate laws used are satisfactorily obeyed across the ranges340 Dehydration of Na,S,O, - 5H,O Single Crystals 2.5 2 .o ”, 8 n I 1.5 4 v 4 G I Y 1.0 0.5 I I I 240 480 720 tlmin Fig. 5. Fitting of the second dehydration stage by the n = 2 Avrami-Erofeev equation. Only three representative curves are plotted to illustrate that both temperature (T) and N, flow (F) dependences are much more regular in this case. The fact that the plots do not extrapolate to the origin is probably due to the uncertainty in zero time caused by the incomplete separation of the endotherms corresponding to the two dehydration stages.(a) T = 303, F = 20; (b) T = 303, F = 10; (c) T = 294, F = 10. of temperatures and nitrogen flows employed, no reliable Arrhenius parameters could be evaluated. This is particularly evident for the first (contracting-circle) stage, and is thought to be a consequence both of small but unavoidable handling differences and of the fact that the crystal samples and their surfaces necessarily differ from one another. Also, the transformation mechanism already discussed for low nitrogen flows is probably effective, at least at intermediate flows, and is expected to have a different temperature coefficient with respect to the mechanism suggested for high nitrogen flows.Hence a dispersion of measured rate values was obtained from the various series of experiments on the effects of variations of temperatures and nitrogen flows. The starting material for the second step in reaction was much more homogeneous (i.e. a microcrystalline dihydrate that had evolved through the immediately preceding pentahydrate dehydration), and accordingly the spread of rate values was much lower in the n = 2 Avrami-Erofeev rate process. Consequently estimates of the Arrhenius parameters could be made, giving EA x 60 kJ mol-l, just greater than the heat ofG. G. T. Giiarini and S . Piccini 34 1 reaction per mole of water and for A a mean value of ca. lo7 min-l. Owing to their unreliability, we shall not discuss these data further.One important result of this kinetic analysis is the significant conclusion that despite the initial formation of a dehydrated layer, this rate process does not show diffusional limitation. This can only mean that, after the layer has been converted into dihydrate according to the mechanism suggested, there is no impedance to the escape of the internally formed water, which must therefore diffuse outwards without hindrance across the layer of product microcrystals. Diffusional control (if present) can therefore be effective only at the very beginning of the reaction, but the present experimental methods are not sensitive enough to detect its influence. Conclusions Experimental evidence, from optical microscopy and thermal analysis, indicates that, in the dehydration of sodium thiosulphate pentahydrate the initial event controlling reaction is the formation of a layer of anhydrous salt on the external surfaces of the reactant crystal.However, thermal measurements for both dehydration and rehydration reactions indicate the intermediate formation of a dihydrate. To resolve this apparent inconsistency a mechanism, which also accounts in terms of surface perfection for the only observed differentiation among crystal slabs doubly cleaved parallel to one of three different crystallographic planes, has been suggested. Moreover, kinetic analysis of the thermal curves affords evidence in favour of the mechanism proposed. The present research, in which we report evidence concerning the formation of a dehydrated layer at the initial stage of the dehydration of a crystal hydrate, constitutes further support for our view that in these (and related) reactions nuclei are only generated through crystallization of already reacted material. Consequently the formation and growth of such nuclei bears no correspondence to the true reaction as previously defined, which at least during the initial stages of the transformation has already taken place to result in the formation of the surface layer.Beyond depriving the previous nucleation and growth mechanism of part of its credibility, the above statement suggests that there may be no relation between the kinetic parameters deduced from the microscopic study of formation and growth of nuclei and those deduced from other methods such as gravimetry, etc.In fact, while the former parameters are connected only to the crystallization of a product already formed, the latter are more strictly dependent on the true reaction. This may account for at least part of the spread of values found in the literature for ‘apparently’ the same reaction. For the initial stages of the present reaction we have found evidence that the dehydration does not greatly depend on the nature of the cleavage surface studied. Its perfection, however, appears of some importance. This is considered again as an indication that, even with a different role (no longer as reaction sites,l? l9 but merely as crystallization promoters) the crystal defects can still have a very important role in determining the reactivity of solids. The present findings indicate that the mechanism of a transformation can change as a function of their abundance.Beyond the evidence from optical microscopy, the spread of rate values found in the systematic investigation of the dependence of rate on nitrogen flow and temperature is thought to be proof of the above. We are indebted to Mr Paolo Parri for his careful preparation of drawings and prints : we also thank the referees for their constructive criticisms and helpful suggestions. Financial support from the Italian Minister0 della Pubblica Istruzione (60 %) is greatly acknowledged.342 Dehydration of Na,S,O, - 5H,O Single Crystals References 1 Chemistry of the Solid State, ed. W. E. Garner (Butterworths, London, 1955). 2 D. A. Young, Decomposition of Solids (Pergamon Press, Oxford, 1966). 3 M. E. Brown, D. Dollimore and A. K. Galwey, in Comprehensive Chemical Kinetics, ed. C . H. Bamford 4 A. K. Galwey. R. Spinicci and G. G. T. Guarini, Proc. R. Soc. London, Ser. A, 1981, 378, 477. 5 G. G. T. Guarini and L. Dei, J. Chem. SOC., Faraday Trans. I , 1983, 79, 1599. 6 L. Dei, G. G. T. Guarini and S . Piccini, J . Therm. Anal., 1984, 29, 755. 7 See ref. (I), chap. 8. 8 L. Dei, G. G. T. Guarini and S . Piccini, Mater. Sci. Monogr., 1985, 28 B, 691. 9 S. W. Young and W. E. Burke, J . Am. Chem. SOC., 1904, 26, 1413; 1906, 28, 315. and C . F. H. Tipper (Elsevier, Amsterdam, 1980), vol. 22. 10 M. Picon, C.R. Acad. Sci., 1924, 178, 700. 1 1 H. M. Dawson and C . G. Jackson, J. Chem. SOC., 1907, 552. 12 T. Golgotiu and V. Rotaru, Bul. Inst. Politeh. Zasi, 1972, 18, 37. 13 B. Lorant, Z. Anal. Chem., 1966, 219, 256. 14 B. H. Nirsha, G. M. Serebrennikova, Yu. V. Oboznenko, B. V. Zhadanov, V. I. Safonova and V. A. Olikova, Zh. Neorg. Khim., 1982, 27, 3035. 15 W. W. Wendlandt, Thermal Methods of Analysis (Wiley, New York, 2nd edn, 1974), pp. 150 ff. 16 G. G. T. Guarini and S . Piccini, Proc. 6th National Meeting on Calorimetry and Thermal Analysis, 17 G. G. T. Guarini and M. Rustici, Reactivity ofsolids, 1987, 2, 381. 18 G. G. T. Guarini and M. Rustici, Proc. XXI Congr. Italian Association of Physical Chemistry, Siena, 19 J. M. Thomas, Adv. Catal., 1969, 19, 293. 20 M. Volmer and G. Seydell, Z . Phys. Chem., 1937, 179, 153. 21 P. G. Taylor and C. A. Beevers, Acta Crystallogr., 1952, 5, 341. 22 D. Donati, G. G. T. Guarini and P. Sarti-Fantoni, Mol. Cryst. Liq. Cryst., 1973, 21, 289. 23 P. W. M. Jacobs and F. C. Tompkins, in ref. (l), chap. 7. 24 B. Delmon, Introduction Ci la Cine'tique He'terogPne (Technip, Paris, 1969), chap. XIII. 25 T. B. Tang, Thermochim. Acta, 1980, 41, 133. Naples, 4-7 December 1984, C8. 6-9 October 1986, E7-253. Paper 71545; Received 25 March, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400331

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1988, 84(1), 343-354 A lH Nuclear Magnetic Resonance Chemical-shift Study of Inverted Microemulsions of Aerosol OT in Benzene and Cyclohexane Partitioning of Water between Hydrocarbon and Aqueous Phases Frank Heatley Department of Chemistry, University of Manchester, Manchester M13 9PL Chemical shifts in inverted microemulsions of sodium bis(2-ethyl- hexy1)sulphosuccinate (AOT) in perdeuterobenzene and perdeuterocyclo- hexane have been measured as a function of AOT concentration, H,O/ AOT ratio (R) and temperature. Free water in the hydrocarbon phase has been detected and quantified. The concentration of free water depends on R, and approaches the solubility of water in pure hydrocarbon when R + 1. There is evidence of a specific interaction between the AOT polar head- group and C,D,, leading to deshielding of the protons.The AOT head- group conformation shows little dependence on composition. Diffusion measurements' have shown that inverted microemulsions formed by Aerosol OT [AOT, sodium bis (2-ethylhexyl) sulphosuccinate, 13 in hydrocarbon solvents consist of discrete AOT-water aggregates dispersed in a continuous hydrocarbon phase. The state of water and AOT in the aggregates has been studied previously2-6 using 'H n.m.r. spectroscopy. As far as the water is these studies conclude that the H 2 0 peak moves to higher frequency as the H,O/AOT ratio increases, and at values of this 10 CH, i 1 2 I 9 CH2 Na' -S03-CH-C02 -CH2-CH-CH2-CH,-CH2-CH, * (1) 3 4 5 6 3' 4 ' 5 ' 6 ' 7 ' 8' CHZ-CO, -CH,-CH-CH,-CH,-CH2-CH3 1' 2 ' I I 9' CH, 10' CH3 ratio greatly exceeding unity the chemical shift approaches an asymptotic value characteristic of bulk water.Relatively little attention has been directed towards the possible existence of water in the hydrocarbon phase, although this has been detected using Fourier transform i.r. spectro~copy.~ Such a possibility was also suspected during a recent study6 of the AOT conformation using 'H n.m.r. spectroscopy, where it was observed, although not reported, that the water chemical shift depended not only on the H20/AOT ratio but also on the AOT concentration and temperature. A particularly striking observation in C6D6 was that on reducing the AOT concentration from ca. 0.2 mol dm-3 to ca. 0.02 mol dmP3 the H 2 0 peak moved upfield by some 2.4 ppm, an enormous shift by 'H n.m.r.standards. Shifts of this magnitude have not been reported 343344 N.M.R. Studies of Inverted Microemulsions of AOT previously. In order to investigate this effect further, a more systematic study is reported here of the dependence of the water and AOT n.m.r. parameters on the composition of AOT microemulsions in deuterated benzene and cyclohexane. Experimental ‘H n.m.r. spectra were recorded on a Varian Associates SC-300 spectrometer operating at 300 MHz. For microemulsions in C6D6 and C6D12, the solvent residual proton signal was used as secondary reference. A few representative samples were calibrated against internal tetramethyl silane ; the solvent signal was found to be independent of composition and resonated at 7.17 6 (C,HD,) or 1.38 6 (C,HD,,).For measurements on pure water and on solutions of sodium methyl sulphonate (CH,SO,Na) in D20, the chemical shifts of H,O or residual HDO signals were measured using a trace amount of sodium 2,2- dimethyl-2-silapentane-5-sulphonate (DSS) as internal reference (0.00 6’). The AOT was obtained from Sigma Chemical Company, St Louis, U.S.A. and was used without further purification. There was no evidence of impurities in the n.m.r. spectra, and after drying at elevated temperature under high vacuum, the elemental analysis was in close agreement with the formula (found: C, 54.0; H, 8.5; S, 7.3; Na 5.0%; C,,H,,O,SNa requires C, 54.0; H, 8.3; S, 7.2; Na, 5.2%). There is other evidence’ that AOT from this source is sufficiently pure for this type of study.CH,SO,Na was prepared by neutralisation of an aqueous solution of CH,SO,H with aqueous NaOH to pH 7, followed by freeze drying. Deuterated solvents were obtained from Fluorochem Ltd, Glossop, and were used as received. Results and Discussion During the course of this investigation it became evident that the results obtained conflicted with some previously reported n.m.r. data concerning the assignment of the ‘H spectrum of AOT2 and the values of coupling constants in the head-gro~p.~*, These topics are considered first, followed by discussion of the water chemical shifts. Spectrum Assignment and Chemical Shift of AOT Fig. 1 shows the ‘H n.m.r. spectrum of a C6D12 microemulsion with an H20/AOT molar ratio (R) = 23.7. The assignment given was verified by relative intensity measurements and spin-decoupling experiments.The assignment of H-3 and H-3’ is assumed to follow the same shielding order established6 for AOT in CD,OD. The earlier assignment2 of H-1’ to a peak at 2.02 6 and of €3-3 and H-3’ to a peak at 3.30 6 is incorrect. The spectra of C6D6 solutions were similar in form to those in C6D12, but protons in or near the head-group lay to significantly higher frequencies, as can be seen from table 1. A further difference is that in C6D6, H-4 and H-4’ give distinct resonances, whereas in C6D,, they overlap. At constant R value, the AOT chemical shifts vary little with concentration, but at constant concentration the shifts depend on R, especially H-1 . The significance of these shifts is discussed in conjunction with the water chemical shift below.Headgroup Conformation The H-1 and H-1’ protons form an ABX spin system, with two vicinal coupling constants JAX and JBx. At 300MHz, with the aid of resolution enhancement by Lorentz-Gaussian transformation, lo sufficient resolution was achieved in both the AB and X regions to obtain JAx and J,, with reasonable accuracy; illustrative spectra of H-1 and H-1’ in C6D,, are shown in fig. 2. As the resolution-enhancement time constantsF. Heatley 345 H-3 H-1 PI I I H-3' I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I 1 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 6 (PPm) Fig. 1. 300 MHz 'H n.m.r. spectrum of an AOT inverted microemulsion in C,D,, at 22 "C. AOT concentration 0.231 mol per kg C6DI2, R = 23.7.show, the natural linewidth decreased markedly as the concentration and R increased. N.m.r. parameters obtained using analytical expressions for the line frequenciesll are listed in table 2 for a range of C,D,, solutions. JAx and Jsx vary little with either concentration or R, and are similar in magnitude to values reported previously6 for solutions in CD,OD, D,O, CDCl, and C,D6. In this earlier work it was shown using a combination of lH-lH and 13C-lH coupling constants that the AOT head-group exists principally as conformation (111) in fig. 3, and it must be concluded that this is also true for C6Dl, microemulsions over a wide range of compositions. These results are rather different from data presented by Maitra5 for microemulsions in iso-octane.It was reported there that (JAx+JBx) increased from ca. 12 Hz to ca. 15 Hz as R increased from 0 to 50. Because of this increase it was suggested that at low R, AOT exists predominantly as conformation (I), and is converted into (11) and/or (111) at high R. However in (I) both JAx and J,, are gauche couplings, and from the well established Karplus relationship between vicinal lH-lH coupling constant and dihedral angle,12 a value of only 6 8 Hz would be expected for (JAx+JBx). It is perhaps significant that Maitra's spectra were recorded at 60 MHz on a continuous-wave spectrometer without resolution enhancement. The increase reported in (JAx + JBx) reported by Maitra may therefore only be an apparent increase arising from an improvement in the natural resolution as R increases.A similar comment may be made concerning the increase in (JAx + JBX) with increasing temperature reported by Maitra and Eicke3 for AOT solutions in CDC1, and iso-octane. In CDC1, for example, (JAx+JBx) apparently increased from ca. 12 Hz at 0 "C to ca. 14 Hz at 50 "C. This increase was also interpreted in terms of a preference for conformation (I), but again the magnitude of (.IAx + JBx) is not consistent with this analysis. However, increasing the temperature does improve the resolution, and the reported behaviour may also be artificial.w P m Table 1. AOT 'H chemical shifts (6) at 22 "C in inverted microemulsions ~~ 1 03cAOTa R H- 1 H-lfb H- 3 H-3' H-4 H - 4 C CHZd CH,d 3 500 108 32.2 11.3 477 477 477 477 1 .o 1 .o 1.1 1.6 1.3 2.5 5.3 11.7 4.83 4.86 4.89 4.90 4.80 4.75 4.67 4.62 (a) C,D, microemulsions 3.62 4.39 - 3.64 4.43 4.17 3.65 4.43 4.17 3.65 4.42 4.16 3.61 4.40 - 3.59 4.40 4.15 3.55 4.4 1 4.15 3.53 4.43 4.17 (b) C,D,, microemulsions 1.77 1.79 1.79 1.80 1.78 1.80 1.81 1.84 ca.1.61 ca. 1.63 ca. 1.63 ca. 1.63 ca. 1.63 ca. 1.65 ca. 1.65 ca. 1.66 1.46 1.48 1.48 1.48 1.48 1.48 1 s o 1.52 1.02 1.04 1.04 1.04 1.02 1.02 1.03 1.08 % k 1.3 4.36 3.21 4.15 3.99 1.60 1.32 0.92 279 56.0 1.3 4.35 3.19 4.14 3.98 1.60 1.32 0.9 1 12.6 1.3 4.35 3.19 4.14 3.98 1.60 1.32 0.9 1 2.3 1.3 4.37 3.19 4.13 3.97 1.61 1.31 0.9 1 2 % 23 1 3.4 4.30 3.17 3.99 1.59 1.32 0.90 6' 23 1 7.3 4.25 3.15 4.14 3.98 1.59 1.32 0.92 2 % i$ G- - 23 1 14.6 4.23 3.15 4.15 4.00 1.61 1.33 0.93 23 1 23.7 4.22 3.15 4.15 4.01 1.61 1.33 0.93 b 0 Y a Concentration in mol AOT per kg C,D, or C,D,,.resonances of width ca. 0.3 ppm (CH,) or 0.1 ppm (CH,). Centre of AB part of ABX spectrum. ' Overlapped by CH, peaks in C,D,. Centre of complexF. Heatley 347 I I I I 20 10 0 -10 Hz I I I I I I I I 40 30 20 10 0 -10 -20 -30 Hz Fig. 2. 300 MHz n.m.r. spectra of H-1 (left) and H-1' (right) in AOT inverted microemulsions in C,D,, at 22 "C. AOT concentrations (mol per kg C,D,,) and R values are, respectively, (a) 0.279, 1.3; (b) 0.0225, 1.3; (c) 0.231, 23.7. The spectra have been resolution-enhanced using the Lorentz-Gaussian transformationlo with exponential (7'') and Gaussian ( TG) time constants, respectively, of (a) TE = 0.1, TG = 0.3 s (H- 1); 0.05, 0.15 s (H-1'); (b) 0.08, 0.24 s (H-1); 0.04, 0.12 s (H-1'); (c) 0.15, 0.4 s (H-I); 0.1, 0.3 s (H-1').The frequency zeros are arbitrary. The variation in linewidth with R indicates an increase in mobility of the AOT molecules. The motion of AOT in p-xylene microemulsions has been studied using multifrequency 13C longitudinal and nuclear Overhauser enhancements,13 and the results analysed in terms of two processes. However, it was noted that the 13C and 'H linewidths were also controlled by a third unexplained motion. A full interpretation of the linewidth variation therefore awaits a detailed lH relaxation study including transverse relaxation time measurements.348 N.M.R. Studies of Inverted Microemulsions of AOT Table 2. Chemical shifts and coupling constants for the H-1 +H-1' ABX system in AOT in inverted microemulsions in C6DIza 103cA,,b R A V A B J A R J A X J B X 279 135 23 1 22.5 23 1 23 1 23 1 1.3 10.3 1.3 11.4 1.3 12.2 3.4 16.4 (11.7 7.3 19.1 14.6 20.3 23.7 21.7 (19.4 18.4 18.1 17.9 17.7 17.4 17.6 17.6 17.5 17.3 11.8 11.3 10.8 11.3 10.9 11.6 12.1 11.8 11.5 3.3 3.8 4.1 3.7 4.3)" 3.7 3.3 3.5 3.8)" a All values in Hz at 300 MHz, uncertainty+0.3 Hz.Temperature 22 "C unless indicated otherwise. Concentration in mol AOT per kg C,D,,. " Values at 53.5 "C. so, so, so 3 H H COZR (1) (11) (111) Fig. 3. Head-group conformations of AOT. Water Chemical Shift We consider first the water chemical shift, JW, as a function of AOT concentration with R constant. Results for C,D, and C,D,, microemulsions are presented in fig. 4 and 5, mespectively. All systems behave in a similar fashion in that at higher concentrations, 6 , shows an extensive plateau region, designated but at lower concentrations the water peak moves rapidly to lower frequency with decreasing concentration.The low- frequency shift, clearly demonstrated in fig. 6, begins at a lower concentration in C,D,, than in C,D,. The relationship between and R is shown in fig. 7. As R increases, 6: increases asymptotically. In these microemulsions the AOT and water molecules may exist either freely in the hydrocarbon phase or as aggregated species. Within the aggregates, the water molecules may be regarded either as bound in an ion solvation shell or as unbound. In the case of systems with R % 1 , the unbound state will presumably approximate bulk water. At 37 "C, the critical micelle concentrations (c.m.c.) of AOT in C,H, and C,H,, are 4 x lop4 and 5 x mol kg-', re~pective1y.l~ If the temperature dependence of the c.m.c.is similar to that in CCl,,14 the corresponding values at 22 "C are ca. 1.4 x and 1.7 x mol kg-'. Both values are much lower than any AOT concentration used in this work, so it can be assumed that in all systems studied here, the AOT is essentially fully aggregated. In contrast, the solubility of water in C,H, and C,H,, at 22 "C is 0.0337 and 0.0067 mol kg-l, re~pective1y.l~ These values are greater than the water content of the more- dilute microemulsions studied here, and hence in these solutions dissolution of a significant proportion of free water in the hydrocarbon phase is possible. It is believed that this is the reason for the movement of the water peak to lower frequency at low concentrations, for the following reasons.(i) The shift is in the direction and of theF. Heatley 349 5 4 n 5 a W 3 <cg) 3 2 AOT/mol (kg c6 D6 ) - I Fig. 4. Variation of JaQ with AOT concentration in inverted microemulsions in C,D, at 22 "C. X , R = 1.1kO.l; 0, R = 10.5k1.5. 2.5l I I 0 0.1 0.2 0.3 AOT/mol (kg C6DIZ)-I Fig. 5. Variation of JaQ with AOT concentration in inverted microemulsions in C,D,, at 22 "C. R = 1.3.3 50 Fig. 6. 300 N.M.R. Studies of Inverted Microemulsions of AOT I I I I I 5 4 3 2 1 8 (PPm) MHz 'H n.m.r. spectrum of an AOT inverted microemulsion in C,D, concentration 0.01 13 mol per kg C6D6, R = 1.6. 5.2 4 . 8 n E 0 : W R 4.4 cg 4 .O 3.6 5 10 15 20 25 R at 22 "C.AOT Fig. 7. Variation of dzq with R in inverted AOT microemulsions in C6D6 and C6D,, at 22 "C. 0, C,D6; x, C6D,,'F. Heatley 351 magnitude expected. From spectra of C,D, and C6Dl, saturated with water, the chemical shifts of free water in those solvents are 0.41 and 0.85 6, respectively. (ii) In C,D, microemulsions with R = 1, the water prot-on longitudinal relaxation time at 22 "C increa:es from 0.30 s at 0.48 mol kg-' AOT, (daq = 4.10) to 0.64 s at 1.1 x mol kg-' AOT (da, = 1.76), consistent with the development of a more mobile state. (iii) Free water in the hydrocarbon phase has been detected, although not quantified, in AOT-C6H6 microemulsions using Fourier-transform i.r. spectroscopy. As the AOT concentration increases at fixed R, the hydrocarbon phase becomes saturated with water. Beyond this point the water properties are increasingly dominated by the aggregated state, and the chemical shift reaches an asymptote, 6zq, characteristic of this state.A more detailed consideration of the concentration dependence is postponed until after discussion of 6:q. The variation of a:, with R in fig. 7 can be understood in terms of a variation of the balance between bound and unbound water within the aggregates. At low R, 6: tends towards the chemical shift of the bound state. As R increases, the unbound fraction increases and tends towards an asymptotic value corresponding to the unbound state. The latter is close to the chemical shift of bulk water, allowing for extrapolation uncertainties and differences in reference procedure.Two types of aggregated water in AOT microemulsions have also been detected using calorimetry16 and proton dissociation kinetics of solubilised alcohol^.'^ For a given R, the water peak in C6D, resonates some 0.3 ppm to higher frequency compared with C6D12, resembling the AOT head-group protons described above. If the aggregates are spherical, the difference in magnetic susceptibility between the discrete and continuous phases has no effect on chemical shifts measured using an internal reference in the continuous phase." The high-frequency shift in C6D6 suggests an interaction with the head-group such that the AOT protons lie preferably in the plane of the aromatic ring, thus leading to a reduction in shielding as a result of ring current~.'~ The variation of Szq with R observed here is similar in form to data reported previously for microemulsions in [2H,,]heptane,2 iso-octane and ben~ene,~ and iso- octane and cy~lohexane.~ However, numerical comparison of the shifts is hindered by differences in concentrations, temperatures and reference procedures.Wong et a1.,2 for example, used an external reference. Maitra and c o - ~ o r k e r s ~ * ~ used internal TMS as reference, but an AOT concentration of only 0.1 mol drn-,. This concentration is lower than that required to reach the 6iq plateau, particularly in benzene, and hence the chemical shifts reported are not fully characteristic of the aggregate. The low-frequency shift in the bound water state in the microemulsions is consistent with the low-frequency shift of water in aqueous solutions of sodium methyl sulphonate, CH,SO,Na.We have observed that at 22 "C the HDO chemical shift decreases linearly from 4.816 in neat D,O to 4.62 6 in a solution containing 1.9 mol CH,SO,Na per kg D,O. In these solutions the observed chemical shift of the water, baq, is averaged over bulk water (denoted by w) and water of solvation (s) s,, = PWSW+ps6, where P, is the probability and 6, the chemical shift of state x. If n is the coordination number and rn the molality of the solution (in D,O), then P, = nrn/50. Since Pw + P, = Hence nm 50 1, eqn (1) can be rewritten as ~ 6,, = 6, +-(as - 6,). From the above data, dJaq/drn = 6 for n, one thus obtains a value - 0.1 ppm kg mol-'. Adopting a reasonable value of of ca. -0.9 ppm for 6,-6,, which is commensurate 12 FAR I352 N.M.R.Studies of Inverted Microemulsions of AOT with the approximate range of chemical shifts produced by extrapolating the data in fig. 7 to R = 0 and R = a. Eqn (1) also applies to the equilibrium between bound and unbound states within the aggregate. If 6, and 6, are known, it is a simple matter to obtain 4. This was done by Maitra,' although the source of 6, and 6, was not specified. It is felt by this author that the derivation of 6, and 6, by extrapolation of the data in fig. 7 to R = 0 and R = co results in uncertainties which are too large compared to the total spread of chemical shifts to give useful quantitative conclusions. In addition, properties such as aggregation numbers,20 area per AOT molecule,? 'H and 23Na n.m.r.relaxation times2 and AOT chemical shifts (table 1) vary with R, indicating structural changes in the interface which could invalidate the assumption of a constant 6,. Also, in iso-o$ane microemulsions, d.s.c. rneas~rementsl~ show only bound water for R < 4.5, yet 6, still depends on R. The application of eqn (1) to the equilibrium between aggregated water and free water in the hydrocarbon phase is more amenable to quantitative analysis, since the extreme situations are readily attained. Rewriting eqn (1) as 6,, = P,S,+Pf6f where a and f indicate aggregated and free water, respectively, then 6, is well approximated by the plateau value 6zq, while 6, may be taken as the water chemical shift in the hydrocarbon containing water alone. 6, actually depends on the H,O/AOT ratio in the aggregate, rather than the overall H,O/AOT ratio R in the microemulsion.At lower concentrations the large low-frequency shift indicates that Pp is significant, and the H20/AOT ratio in the aggregates will differ from the overall composition. However, the error introduced by neglecting this consideration is small even for the most extreme case, as can be demonstrated by a hypothetical example. Corpider a C,D, solution with R = 1 , and of such a concentration that the observed shift 6, = 2.26 lies midway between 6, (0.41 6) and 6; (4.1 1 6). If 6, is equated to a:,, eqn (2) then gives 4 = 0.5. However, this implies that in the aggregates H,O/AOT = 0.5, and 6, is better given by 6& for R = 0.5, i.e. 4.01 6. This 'improved' value for 6,, however, then only leads to a small change in P f to the value 0.47, a correctio? which is small compared to the uncertainty arising from errors in the measurement of 6,, 6, and 6;.In the following analysis, 6, has therefore been approximated by 6% for the appropriate value of R. Tables 3 and 4 give the amount of free water in C,D, and C,D,, microemulsions, respectively. Only thlpse systems are included which can be analysed reasonably accurately, i.e. where 6, differs significantly from 6, and 6;. For a given value of R, the concentration of free water in the hydrocarbon is reasonably constant as the AOT concentration varies, consistent with a description in terms of a distribution between two phases. As R increases, the concentration of free water increases, and at R $ 1 it approaches the solubility of pure water in the hydrocarbon" (at 22 "C, 0.061 wt% in C6H6, equivalent to 0.031 mol kg-l in C,D,, and 0.01 1 wt % in C6H12, equivalent to 6.1 x lo-, mol kg-l in C,D,,).This behaviour supports the conclusion from chemical- shift data that at R % 1 the properties of the water pools approach those of bulk water. The temperature dependence of 8,, was studied in the range 20-50 "C as a function of composition. 6, decreased linearly with increasing temperature within experimental error, in agreement with results of Maitra and Eicke3 for CDCl, and iso-octane systems, an,d present results are therefore reported in table 5 as the temperature coefficient, dd,,/dT. The last entry for each hydrocarbon corresponds to water in the hydrocarbon alone at a concentration equal to the solubility limit at 20 "C.A further useful reference value is the temperature coefficient for pure water, viz. - 10.5 ppb K-'. Maitra and Eicke3 attributed the temperature dependence to a change in rotational t Values calculated in ref. (5) using data from ref. (20).F. Heatley 353 Table 3. Water chemical shifts and concentration of free water in the hydrocarbon phase of C,D, microemulsions at 22 "C, (6, = 0.41) 1 03cAOTa R 4 103c,b 182 108 56.2 32.2 17.5 11.3 17.5 15.4 47.2 21.4 11.1 5.56 1 .o 1 .o 1 .o 1.1 1.2 1.6 2.0 4.0 10.9 9.8 9.1 8.6 4.1 1 4.1 1 4.1 1 4.13 4.15 4.22 4.30 4.47 4.82 4.79 4.76 4.74 3.90 3.70 3.38 2.82 2.19 1.76 2.35 2.82 4.60 4.17 3.43 2.30 0.057 0.1 1 0.20 0.35 0.52 0.65 0.50 0.4 1 0.05 1 0.14 0.3 1 0.57 10 12 11 12 11 12 18 25 26 29 31 27 a AOT concentration in mol per kg C6D6. kg C,D,.Uncertainty probably ca. f 20 O h . Concentration of free water in the C6D6 in mol per Table 4. Water chemical shifts and concentration of free water in the hydrocarbon phase of C,D,, microemulsions at 22 "C (6, = 0.85) 56.0 1.3 3.90 3.8 1 0.030 2.2 27.0 1.3 3.90 3.73 0.056 2.0 12.6 1.3 3.90 3.61 0.095 1.6 5.17 1.3 3.90 3.1 1 0.26 1.7 2.29 1.3 3.90 2.6 1 0.42 1.3 5.76 12.3 4.51 4.37 0.038 2.9 a AOT concentration in mol per kg C,D,,. C,D,,. Uncertainty probably ca. & 20 %. Free water concentration in C6Dl, in rnol per kg Table 5. Temperature coefficients of JaQ in C,D, and C,Dl, microemulsions, measured as the average rate of change between 22 and 52 "C .~ _ _ -~ ~ ~- C6D12 - C6D6 ~ -~ ~ (diaq/dT) (''a, /'TI 1 0 3 c ~ ~ ~ a R /PPb K-' 10"4,,Ta R /PPb K-' 500 182 108 56.2 32.2 17.5 11.3 477 477 477 477 1 0 1 .o 1 .o 1 .o 1 .o 1.1 1.2 1.6 1.3 2.5 5.3 1.7 - 13.3 -21 - 26 - 35 - 33 - 28 - 23 - 12.5 - 12.2 - 12.0 - 11.6 0.61 279 135 56.0 27.0 12.6 5.2 2.29 23 1 23 1 23 1 23 I 0 1.3 1.3 I .3 1.3 1.3 1.3 1.3 3.4 7.3 14.6 24 - 8.6 - 9.7 - 13.6 - 20.0 - 28 - 38 - 37 - 8.5 - 9.0 - 8.5 -9.0 - 0.52' 'I Concentration in rnol per kg C,D, or C,D,,. with water at 20 "C. This value is for C,D, or C,D,, alone saturated 12-2354 N.M.R. Studies of Inverted Microemulsions of AOT isomer populations alone, but this interpretation is clearly too simple. Apart from the fact that the rotational isomer distribution does not vary greatly with temperature, three other factors are involved.These are the distribution between free and aggregated water, the distribution between bound and unbound water in the aggregate, and the distribution between hydrogen-bonded states in the aggregate. It is this last factor which is responsible for the temperature shift in pure water, and hence it is not surprising that when R $ 1 and the AOT concentration is within the d:q plateau region, the temperature coefficient is close to that of pure water. In fact, within the dzq plateau, the coefficient is more or less independent of R, suggesting that the temperature coefficients of bound and unbound states in the aggregate are similar. As the AOT concentration falls and the amount of free water becomes perceptible, the temperature coefficient rises to a maximum value some three- to four-fold greater than in pure water.This behaviour is clearly attributable to an increase in the proportion of free water with temperature. References 1 P. Stilbs and B. Lindman, J. Colloid Interface Sci., 1984, 99, 290. 2 M. Wong, J. K. Thomas and T. Nowak, J. Am. Chem. Soc., 1977, 99,4730. 3 A. N. Maitra and H-F. Eicke, J. Phys. Chem., 1981, 85, 2667. 4 A. Maitra, G. Vasta and H-F. Eicke, J. Colloid Interface Sci., 1983, 93, 383. 5 A. Maitra, J. Phys. Chem., 1984, 88, 5122. 6 F. Heatley, J. Chem. Soc., Faraday Trans. I , 1987, 83, 517. 7 H. Kise, K. Iwamoto and M. Seno, Bull. Chem. Soc. Jpn, 1982, 55, 3856. 8 R. J. Abraham and P. Loftus, Proton and Carbon- 13 NMR Spectroscopy : An Integrated Approach 9 A. M. Howe, C. Toprakcioglu, J. C. Dore and B. H. Robinson, J. Chem. Soc., Faraday Trans. I , 1986, (Heyden, London, 1978), p. 14. 82, 241 1. 10 A. G. Ferrige and J. C. Lindon, J. Magn. Reson., 1978, 31, 337. 11 J. A. Pople, W. G. Schneider and H. J. Bernstein, High Resolution Nuclear Magnetic Resonance 12 Ref. (8), p. 45. 13 J. Carnali, B. Lindman, 0. Soderman and H. Walderhaug, Langmuir, 1986, 2, 51. 14 M. Ueno and H. Kishimoto, Bull. Chem. Soc. Jpn, 1977, 50, 1631. 15 Landolt-Borstein, Zahlenwerte und Funktionen (Springer Verlag, 6th edn, 1962), vol. 2, part 2b, 16 C. Boned, J. Peyrelasse and M. Moha-Ouchane, J. Phys. Chem., 1986, 90, 634. 17 M. J. Politi and H. Chaimovich, J. Phys. Chem., 1986, 90, 282. 18 Ref. (ll), p. 80. 19 Ref. (I l), p. 180. 20 R. A. Day, B. H. Robinson, J. H. R. Clarke and J. V. Doherty, J. Chem. Soc., Faraday Trans. I , 1979, (McGraw-Hill, New York, 1959), p. 134. p. 3395. 75, 132. Paper 7/742; Received 24th April, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400343

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1988, 84(1), 355-364 The Effect of the Low-oxidation-state Metal Ion Reagent Tris-picolinatovanadium( 11) Formate on the Surface Morphology and Composition of Crystalline Iron Oxides Geoffrey C. Allen,* Colin Kirby and Robin M. Sellers Central Electricity Generating Board, Berkeley Nuclear Laboratories, Berkeley, Gloucestershire GL13 9PB The LOMI (low-oxidation-state metal ion) process is used widely in the decontamination of water reactor systems. An attempt has been made to elucidate the mechanism of dissolution by reference to a number of single crystals of iron oxides polished on specific crystallographic planes, e.g. magnetite, Franklinite and haematite. These have been studied by a combination of electron-spectroscopic techniques such as X-ray photo- electron spectroscopy, Auger electron spectroscopy and Mossbauer spectroscopy.Electron microscopy and chemical kinetic measurements have also been undertaken. The complex formed between vanadium(I1) and picolinic acid is a strong reducing agent, capable of bringing about the rapid dissolution of a number of oxides rich in iron(111).l-~ These include a-haematite, magnetite and most notably nickel ferrite, which is often difficult to dissolve by other methods such as acid or complexing attack. These oxides are typical of those found on pipework surfaces in power plants, especially water- cooled nuclear reactors. Even at acidities in the pH range 4-5 rapid dissolution of iron(rI1) oxides can be achieved using the tris-picolinatovanadium( 11) complex. These are conditions under which corrosion of the bare metal is minimal, and since a strong reductant is in any case intrinsically non-corrosive, this reagent is exploited commercially for the decontamination of power plant.2 With formate as the counter-ion and in the presence of formic acid as a buffer, the complex forms the basis of the LOMI (low- oxidation-state metal ion) reactor decontamination system.Detailed studies have been undertaken with a number of iron-rich oxide powders, and although these were intended mainly as a means of characterising the kinetics of the dissolution reaction, changes in surface morphology were also investigated. 3, Powders do not readily lend themselves to this kind of work, and it has proved difficult to correlate the morphological changes with other aspects of the dissolution.To overcome such problems single crystals have been used, and the present paper reports the results from these studies. In particular we have attempted to correlate the dissolution kinetics with changes in surface morphology and composition for specific crystallographic planes. The work focuses principally on magnetite (Fe,O,), but a number of other materials have also been investigated. Experiment a1 Crystals of magnetite (Fe,04) were obtained from schists collected from Anglesey, Gwynedd, North Wales. Franklinite crystals (MnZn)Fe,O, were obtained from Sterling Hill near Franklin, New Jersey, U.S.A. and were extracted from a calcite matrix. Both sets of crystals had an octahedral habit reflecting their spinel-type structure and were typically 4 mm across.355356 Eflect of a LOMI Reagent on Iron Oxides (111 1 (110) (100) Fig. 1. Relationship of the various crystallographic planes prepared for the present study. Crystals of haematite a-Fe,03 and also Fe2Ti0, and FeNbO, were obtained from Dr B. Wanklin of Oxford University. The crystals were prepared from high-temperature melts. Their surface area varied from tens of square millimetres to < 1 mm2 for iron niobate. Specific crystallographic planes were prepared by setting the sample in Woods metal (a very-low-melting-point alloy) or Perspex and then cutting approximately the correct plane as shown in fig. 1 with a diamond saw and finally polishing with diamond paste. The accuracy of the technique was tested by taking Laue X-ray diffraction pictures of a number of samples and was found to be within ca.5 O of the specified plane. The crystals were then extracted from the mount and set in gold clasps for ease of handling and to provide an electrically conducting support for use in the X-ray photoelectron spectrometer. X-ray photoelectron spectra were obtained on a Kratos ES200 instrument which incorporates a double-focussing hemispherical electrostatic analyser and an A1 Ka X-ray source separated from the main chamber by an aluminium window. Typical base pressures obtained in the sample region during a run were ca. 5 x Torr.? The spectrometer was operated in the fixed-analyser transmission mode, in which the electrons are pre-retarded to 65 eV, the pass energy of the analyser.A Cambridge Instruments Stereoscan 150 scanning electron microscope was used to monitor changes in surface morphology due to dissolution. This instrument is fitted with a liquid-nitrogen-cooled lithium-drifted silicon crystal detector which energy-analyses the X-ray fluorescence excited by the electron beam and can give relative elemental concentrations. A depth of ca. 1 pm into the sample is excited by the electron beam. The V2+ solutions were prepared from triply distilled water by the electrolytic reduction of 0.2 mol dm-3 VOSO, in 0.5 mol dmP3 H2S0, at a lead cathode. Picolinic acid (Aldrich) and sodium formate (B.D.H.) were recrystallised once from water before use. The total vanadium content was determined spectrophotometrically as the peroxovanadium(v) complex, taking E~~~~~ = 281 dm mol-' cm-'.The amounts of V" and V'" were determined by direct measurements on the stock solutions at 400nm taking &(V3+) = 8.3 dm3 mol-1 cm-' and c(V2+) = 0.9 dm3 mol-1 cm-'. The surfactant Triton X- 100 was B.D.H. G.P.R. grade. The apparatus used for the kinetic runs is similar to that described in the literature and consisted of a constant-temperature all-glass reaction ~ e s s e l . ~ All solutions were deoxygenated by bubbling with high-purity argon (B.O.C.) for ca. 1 h before commencing the reaction by the addition of V2+. A slow stream of argon was passed through the solution throughout the reaction to prevent aerial oxidation of the V(pic), complex. The rate of dissolution was measured by withdrawing 1.5 cm3 aliquots at timed t 1 Torr = 101 325/760 Pa.G. C.Allen, C. Kirby and R . M. Sellers 357 intervals with a syringe. The reaction was quenched by exposure to air [oxidising the V(pic); complex] and cooling below 10 "C by placing the vessel in crushed ice. Before determining the metallic content of each aliquot by atomic absorption spectrometry using a Baird 5 100 spectrometer it was diluted by a factor of ten with 0.1 mol dm-3 HCl (B.D.H. C.V.S. reagent) and filtered through a 0.22 pm filter (Millipore Millex GS) to remove any particulate matter. Kinetic data were obtained for a particular crystal plane by blanking all other faces with a coating of silver applied as a colloid. We had previously shown that this coating was impervious to the etching reagent during a trial with a blanked-off polished face.This face showed no sign of etching after immersion in the reagent over a period of several hours. The surface area of a face was measured photographically from an enlargement. Calibration of the enlargement was obtained by measuring the distance between two points on the crystal with a travelling microscope. This method gave results which were reproducible to within a few percent. Kinetic Measurements A preliminary investigation of the crystallographic surface dependence of the reaction rate of V(pic), with magnetite was undertaken because both geothite (FeOOH)5 and haematite (Fez03)6 exhibit anisotropic etching in acids. In fact the basal (001) plane of a-haematite dissolves at least an order of magnitude faster than the other planes.Early experiments showed regular etch figures on the magnetite crystals again suggesting anisotropic etching. The fabrication of micromechanical devices on silicon wafers relies on similar anisotropic etching behaviour. The propensity for one plane to be etched faster than another in silicon is being exploited to produce devices as complicated as gas chromatographs or accelerometers on a single chip. In our experiments we made the reasonable assumption that the rate of reaction is proportional to the initial surface area, which does not vary during the reaction, thus dm dt ---=-kA where m is the mass of the undissolved crystal and k is a rate constant with units of g m-2 min-l. On integrating eqn (1) we get m0-mm, = kAt. (2) Experimentally it is more convenient to measure the concentration c, (in g dm-3) of the metal (e.g.Fe) in solution. This is given by c, = x(m0 -m,)/ V (3) where x is the weight fraction of the metal in the oxide and Y is the volume of the solution. Substituting eqn (3) into eqn (2) gives C, = kAtx/V. (4) Thus for a constant surface area a plot of c, us. time ( t ) will give a straight line whose slope is proportional to the rate constant k . It has been shown previously for powdered magnetite that the reaction rate is proportional to the changing surface area and a cubic rate law was derived. Preliminary results for the three crystal planes (loo), (1 10) and (1 1 1) are presented in table 1. It can be seen immediately that there is a difference in reaction rate between the planes according to the order (1 11) > (100) 2 (I lo).However, the change is at most by358 Eflect of a LOMI Reagent on Iron Oxides Table 1. Rate of dissolution of magnetite crystals face k / g m-2 min-' PH (100) 0.58 k0. 10 3.56 (111) 0.91 kO.10 3.50 (1 10) 0.34 & 0.10 3.75 i I I I I Fe" 1 I I I I 725 720 71 5 710 705 binding energy/eV Fig. 2. The iron 2p ionizations of magnetite excited by A1 K, X-radiation: (a) polished crystal, (6) etched crystal. a factor of three between the fastest and slowest faces, much less than the changes reported in the acid dissociation of haematite. A similar difference has been reported in the dissolution of yttrium iron garnet by phosphoric acid in the presence of iron(r1) chloride, the fastest dissolution rate being ca. 1.5 times the rate of the slowest dissolving crystallographic plane.Larger variations are observed in pure phosphoric acid, where the difference is a factor of 4.8 Spectroscopic Studies X-Ray Photoelectron Spectroscopy X-Ray photoelectron spectroscopy (X.P.S.) proved very useful in the present study, giving information about the chemical state of the elements present on the surface of the crystals. In particular it was possible to distinguish between iron in its + 3 and +2 oxidation lo Magnetite (Fe,O,) is a mixed valence compound with Fe'I and FeI'I present in the ratio 1 : 2. In fig. 2(a) is shown the Fe 2p ionisations for magnetite. The largest splitting of ca. 14 V is due to the spin-orbit splitting between the 2p3,2 and 2p1,2 components.G. C. Allen, C. Kirby and R. M.Sellers 359 535 530 i I I 535 530 binding energy/eV Fig. 3. The oxygen 1s ionizations of magnetite excited by A1 K, X-radiation: (a) polished crystal, (b) etched crystal. Focussing on the 2p3,, peak one can see a shoulder due to Fe" on the low-energy side of the larger Ferrr peak separated by ca. 1.7 eV. Another clue to the amount of Fe" present in the sample is the degree to which the region between the 2p3,2 and 2p,,, is ' filled in '. In a pure Fe1I1 compound distinct shake-up satellites are visible in this area, but inclusion of Fe" causes these satellites to smear out and thus provide a sensitive probe for the presence of FeI'. It must be remembered that we are monitoring only the surface composition and that the spectra may not be representative of the bulk.Indeed polishing has been shown" to produce a slightly oxidised surface. In fig. 2(b) is shown the Fe 2p region for a magnetite crystal after undergoing treatment in a solution of vanadous picolinate. There is a significant increase in the Fe" contribution. Crude manual deconvolution and area measurements show that the Fe". FerIr ratio is now ca. 1 : 1. This ratio is obtained rapidly (within a few minutes) and was produced in all runs. A similar change has been reported previously" for the (1 11) plane of magnetite. The same change in ratio is now confirmed for other crystallographic planes, e.g. (100) and (1 lo), under similar reaction conditions. Changes in the X.p. spectrum due to etching are also evident in the oxygen Is region. Fig. 3(a) shows the 0 1s peak prior to reacting with the LOMI.It consists of an unresolved doublet. The higher-binding-energy component is due to chemisorbed water producing OH groups on the surface, and the lower-binding-energy component' is due to the 02- anions in the bulk structure. After reaction the spectrum in fig. 3(b) is obtained, showing approximately a twofold increase in the OH- signal compared to 02-. Perhaps this result is not too surprising since the reaction was performed in an aqueous medium, but it does show that any proposed reaction mechanism must take into account the degree of hydration or hydroxylation at the crystal-liquid interface. Photoelectron spectra of haematite (Fe203) and Franklinite [Fe,(Zn, Mn)O,] crystals show only typical FerI1 signals before and even after etching.The significance of this result will be discussed later. Experiments on Fe,TiO, crystals, which have a pseudobrookite structure, yielded some surprising results in that the X.P.S. signal due to iron vanishes completely after360 Eflect of a LOMI Reagent on Iron Oxides etching. This means that the iron has been leached out to a depth of at least 20 A into the crystal. To the naked eye, however, the crystals had appeared unaffected by the reaction, and very little iron was detected in solution. Auger Electron Spectroscopy A combination of Auger electron spectroscopy (A.e.s.) and argon iron sputtering were chosen to determine the depth at which iron reappears in the etched Fe,TiO, crystals, A.e.s. only samples a depth of ca. 1-5 atomic layers and hence can produce better resolved depth profiles than X.P.S.Fig. 4(a) shows the Auger spectra of the crystal after reacting with the LOMI. Note the absence of signal due to iron, confirming the X.P.S. results. The depth profile shown in fig. 4(b) was obtained by measuring the peak-to-peak heights for the oxygen KLL, titanium LMM and iron LMM Auger transitions every 60 s duri5g sputtering. The rate of argon-ion sputtering had been previously calibrated at 50 A min-l on oxides of known thickness. During the first minute of each cycle the adventitious carbon is removed and the Auger transitions show an increase in intensity. There is then a gradual increase in the iron and decrease in the titanium signals until at ca. 35 min. sputter time they reach a constant value corresponding to an Fe/Ti atom ratio of 2: 1 calculated using the relative sensitivity factors Fe = 0.45 and Ti =, 0.21.12 The depth at which this constant bulk composition is reached is ca.1750 A in Fe,TiO,. Obviously this value will only correspond to the particular set of reaction conditions used in this experiment. However, the fact that it is so large is important, because previous mechanisms proposed for the dissolution process in magnetite suggested that the reaction zone is a maximum of 2-3 atomic layers deep. Mossbauer Spectroscopy Mossbauer spectra of magnetite crystals were obtained by the conversion-electron method (c.e.m.). Instead of measuring the absorbance of y-rays (14.4 keV) directly, the sample is placed in a proportional counter where electrons produced by relaxation of excited ,'Fe nuclei are detected.Conversion-electron Mossbauer spectroscopy has the advantage of surface sensitivity, since only electrons produced close to the surface can escape (and be detected) owing to their short mean free path within the solid. In these experiments 25 mCi of 57C0 embedded in a rhodium matrix was used as the y-source. The spectrometer has been described in detail elsewhere. l3 Spectra of magnetite before and after etching in vanadous picolinate are presented in fig. 5(a) and (b). These spectra have been interpreted in terms of two six-line magnetic hyperfine patterns superimposed. l4 Only the low-energy components of the spectra are completely resolved. The spectra were recorded at room temperature, which is above the Verwey transition temperature (1 10-120 K).The Fe'I and FeIII on octahedral B sites are therefore equivalent owing to a fast electron exchange, and hence the two six-line patterns are due to FerI1 on tetrahedral A sites and [Fe'l+Fe'T'] on octahedral B sites. The magnetic field at the A site is larger than at the B site, and hence causes a larger splitting in the spectrum, which is labelled to show the two sites (fig. 5). The spectrum of the etched sample shows a significant depletion of the B site signals with respect to the polished crystal. This demonstrates that V(pic), at least shows a preference for attack at octahedrally coordinated Fe3+ even if it does not attack them exclusively. Scanning Electron Microscopy Scanning electron microscopy (s.e.m.) was used to monitor changes in surface morphology and composition due to reaction with the LOMI reagent.This isG . C. Allen, C. Kirby and R. M. Sellers 6 - 5 - % G . G . 5 3 - 2 - 1 - 0 - 361 0 l t 100 300 500 700 900 kinetic energy/eV I L I I I 0 12 2 1 36 48 60 sputter time/min I 1 I 0 100 300 500 700 900 kinetic energy/eV Fig. 4. (a) Auger electron spectrum of etched Fe,TiO, crystal. (b) Intensity of iron, titanium and oxygen Auger peak intensities plotted us. argon ion sputter time in minutes for an etched Fe,TiO, crystal. ( c ) Auger electron spectrum of the sputtered Fe,TiO, crystal showing the presence of iron.362 QJ m E Y + 00 Efect of a LOMI Reagent on Iron Oxides ( a ) A B A 0 I I. I I I I I I I 50 250 450 c ha nnr 1 11 urn ber Fig.5. Conversion electron Mossbauer spectra of a magnetite crystal (a) before etching and (b) after etching in V(pic);. exemplified by pictures of the (1 11) face of magnetite before and after reaction shown in plate 1. The initial crystal surface is quite smooth at a magnification of 800 x except for the occasional inclusion of silicaceous deposits. After reaction the magnetite surrounding the silicate has been removed and the silicate stands proud from the surface. Away from these areas the magnetite has undergone general surface roughening to form tiny (ca. 1 pm) pyramidal structures (plate 2) and deep etch pits which are bounded by approximate (100) faces which were shown in the previous section to react more slowly than the (1 11) face. Another interesting feature associated with the (1 11) surface is the furrowed or striated sides to the large cracks.The etched (100) face is quite different (plate 3). It remains generally smooth with a slightly ' scalloped ' appearance and small areas of furrows or striations which terminate in pyramidal structures on the walls of deep fissures (plate 4), exactly the reverse of the structures observed on the (1 11) face. The (1 10) face (plate 5) is similar to the (1 11) in that it undergoes general surface roughening with the formation of tiny pyramids ca. 2pm across. Discussion Kinetic measurements have shown that dissolution of magnetite by V(pic); is fastest for the (1 11) plane by at least a factor of two more than the other planes studied, i.e. (1 10) and (100). The (1 1 1) plane presents to the solution a close-packed array of oxygen atoms (possibly as hydroxide) which will be easily protonated under the acidic conditions used in this reaction.It has been shown previously* that the reaction rate increases at low pH, and a mechanism was proposed involving an outer-sphere electron-transfer reaction with the protonated sites. Reduction of Ferrl to Fe" involves an incrcase in ionic radiu?,l5 causing a breakdown of the lattice structure [r(Fe3+) = 0.65 A, r(Fe2+) = 0.75 A]. Dissolution is often thought to occur at kink and ledge sites at the surface,l' where the number of nearest neighbours and hence the binding energy of an atom is lower. These can occur at steps in a regular crystal or at faults in the solid phase such as screw or edgeJ .Chem. SOC., Faraday Trans. 1, Vol. 84, part 1 A 1 mm 20cm 6 1 mm 2 0 ~ m Plate 1 Plate 1. Scanning electron microscope pictures of the (1 1 1) face of magnetite (A) before and (B) after etching in V(pic);. Etch details: 4.8 mmol dm-3 V(~ic)-~, pH 3.7, 2 h, 2.10 % dissolution. G. C. Allen, C . Kirby and R. M. Sellers (Facing p . 362)J . Chem. SOC., Faraday Trans. I , Vol. 84, part I (a) 1 0 p (bl Plate 2 Plate 2. Scanning electron microscope pictures of the (1 11) face of an etched magnetite crystal showing some etch figures. Etch details: 4.8 mmol dm-3 V(pic);, pH 3.7, 2 h, 2.10 %. G. C. Allen, C. Kirby and R. M. SellersJ . Chem. SOC., Faraday Trans. 1, Vol. 84, part 1 A lmm ( a ) 200km (b) 20pm ( c) B 1 mm 2OO4~m 20pm Plate 3 Plate 3.Scanning electron microscope pictures of the (100) face of magnetite (A) before and (B) after etching in V(pic),. Etch details: 4.8 mmol dm-3 V(pic);, pH 3.6, 2 h, 1.24%. G. C. Allen, C. Kirby and R. M. SellersJ. Chem. Soc., Faraday Trans. I , Vol. 04, part 1 10 pm 4 rm Plate 4 Plate 4. Scanning electron microscope pictures of the (100) face of magnetite after etching in V(pic);. Etch details. 4.8 mmol dm-3 V(pic);, pH 3.6, 2 h, 1.24%. G. C. Allen, C. Kirby and R. M. SellersJ. Chem. SOC., Faraday Trans. I , Vol. 84, part 1 1 mm Plate 5 Plate 5. Scanning electron microscope pictures of the (110) face of magnetite after etching in V(pic);. Etch details: 6.5 mmol dm-3 V(pic);, pH 3.9, 3 h, 0.44 %. G. C. Allen, C. Kirby and R. M. SellersJ . Chem. SOC., Faraday Trans.1, Vol. 84, part 1 Plate 6 Plate 6. Scanning electron microscope picture of franklinite after etching in V(pic);. Etch details : 4.8 mmol dm-3 V(pic);, pH 3.6, 1 h, 0.93 YO. G. C. Allen, C. Kirby and R. M. SellersG. C. Allen, C. Kirby and R. M . Sellers 363 dislocations. Plate 6 is an etch pit obtained during dissolution of Franklinite. It shows the typical spiral structure associated with the screw dislocations. The importance of the crystallographic fault density has been demonstrated by the variation in reaction rates of nickel ferrite powders annealed at different temperatures. The magnetite crystals used in this work were naturally formed, and hence their temperature history and fault density are unknown. Surface roughening is seen on each crystal plane after etching indicating that a general attack mechanism of the sort described is operating. This mechanism will be responsible for the variation in dissolution rates at each face if we assume a homogeneous distribution of faults, and it may be related to the density of reaction sites on each face.Confirmation of the variation in dissolution rates is obtained by an analysis of the etch pits found on the (1 11) surface. These are bounded by (100) walls suggesting that the (100) dissolution rate is slower than the (1 11) rate.7 Similar pits have been observed in single crystals of copper etched in 0.1 mol dm-3 perchloric acid.17 The exact morphology was found to be a function of the applied potential and dissolution time. Octahedral pits with (1 11) facets could be formed of truncated pyramids with a (001) bottom facet due to the different dissolution kinetics of the two facets.This was also related to the surface energy of the different planes. The surface energy of magnetite may be easily modified by protonation, and hence the morphology of the pits may be dependent on the reaction conditions. We have shown by Mossbauer spectroscopy that FeIII in octahedral sites is preferentially attacked. The Fe" and FeIII atoms are electronically coupled, so that the possibility exists for effects at the surface to be transmitted deep into the crystal structure. Depletion of iron in octahedral sites was observed at the surface. Franklinite has FeIII only on octahedral sites. No evidence of Fe" was observed after reaction, confirming that iron is rapidly removed after reduction.No Fe" was observed on haematite after reaction, again confirming that iron is rapidly removed into solution. This evidence suggests that magnetite shows increased FeT1 at the surface after reaction, not because of the reduction of FeI" to Fe" but because Fe'II is immediately removed from the surface after reaction, leaving the original Fe". The data directly implicate preferential dissolution of octahedrally coordinated Fe'I', whereas previous studies with the inverse spinels nickel and cobalt ferrite and the normal spinel manganese ferrite indirectly indicated a preference for LOMI dissolution of octahedral sites. It is of interest from a practical standpoint to realise that the oxide lattice in magnetite is completely disrupted by the removal of the FeIII ions in octahedral holes, which comprise only one third of the total cationic complement of the crystal.However, removal from Fe,TiO, of Fe"', which constitutes two thirds of the cations, results in the formation of a stable titanium oxide of unknown composition, probably owing to the insolubility of Ti4+ under these conditions. The reaction in this case is controlled by the diffusion of the iron cations through the oxide lattice. Can these observations throw any light on the Fe/Cr oxide systems commonly found on reactor surfaces? It is known that Cr3+ is only dissolved slowly by the LOMI reagent because it is not reduced by V2+. Hence it is possible that removal of iron will promote a chromium-rich coating on the oxide surface.FelIr would then have to migrate through a slowly dissolving chromium oxide layer before being rapidly removed from the surface. Obviously the exact structure of the chromium layer and the speed at which it forms will depend on the composition of the original spinel. Many useful results could thus be obtained by investigating a series of spinels of the form Fe(3-,) Cr, 0, using X.P.S., A.e.s. and argon-ion sputtering for depth profiling. The work was carried out at the Berkeley Nuclear Laboratories of the Technology364 Eflect of a LOMI Reagent on Iron Oxides Planning and Research Division, and the paper is published with permission of the Central Electricity Generating Board. References 1 New Scientist, 1983, 99, 394. 2 D. Bradbury, M. G. Segal, R. M. Sellers, T. Swan and C. J. Wood, Water Chemistry of Nuclear Reactor Systems 2, (British Nuclear Energy Society, London, 1981), p. 403. 3 M. G. Segal and R. M. Sellers, J. Chem. Soc., Chem. Commun., 1980, 991. 4 M. G. Segal and R. M. Sellers, J. Chem. Soc., Faraday Trans. I , 1982, 78, 1149. 5 R. M. Cornell, A. M. Posner and J. P. Quirk, J. Znorg. Nucl. Chem., 1974, 36, 1937. 6 I. H. Warren, M. D. Bath, A. P. Prosser and J. T. Armstrong, Inst. Min. Metall. Trans., Sect. C, 1969, 7 J. B. Angell, S. C. Terry and P. W. Barth, Sci. Am., 1987 (April), 36. 8 C. P. Klages, Muter. Res. Bull, 1984, 19, 1329. 9 G. C. Allen, M. T. Curtis, A. J. Hooper and P. M. Tucker, J. Chem. SOC., Dalton Trans., 1974, 1525. 10 C. R. Rrundle, T. J. Chuang and K. Wandett, Surf. Sci., 1977, 68, 359. 11 G. C. Allen, R. M. Sellers and P. M. Tucker, Philos. Mag., Sect. B, 1983 48, L5. 12 L. E. Davis, N. C. MacDonald, P. W. Palmberg, G. E. Riach and R. E. Weber, Handbook of Auger 13 G. C. Allen and R. A. Mascall, CEGB Research, 1982, 13, 32. 14 T. Swan, D. Bradbury, M. G. Segal, R. M. Sellers and C. J. Wood, CEGB Research, 1982, 13, 3. 15 F. F. Wells, Structural Inorganic Chemistry (Clarendon Press, Oxford, 4th edn, 1975), p. 259. 16 M. G. Segal and R. M. Sellers, Adv. Znorg. Bioinorg. Mech., 1984, 3, 97. 17 B. Yan, E. Youngs, G. C. Farrington and C. Laird, Corrosion Sci., 1986, 26, 121. 78, 21. Electron Spectroscopy (Physical Electronics Industries Inc., Minnesota, 2nd edn, 1978). Paper 71845; Received 13th May, 1987

ISSN:0300-9599    

DOI:10.1039/F19888400355

出版商:RSC    

年代:1988

数据来源: RSC

摘要:

J. Chem. Suc., Faraday Trans. 1, 1988, 84(1), 365-366 CORRIGENDA Corrigendum to Reactions of Recoil 38Cl Atoms with Dichloroethanes Klara Berei,* Laszlo Vasaros and Istvan Kiss Central Research Institute for Physics of the Hungarian Academy of Sciences, H-1525 Budapest 114, P.O. Box 49, Hungary J . Chem. Soc., Faraday Trans. 1, 1986, 82, 3003-301 1 On p. 3003, the first line of the main text should read: halogenomethanes and not halogenoethanes as printed. Corrigendum to The Role of Solvent Reorganization Dynamics in Homogeneous Self-exchange Reactions Gunter Grampp,* Wolfgang Harrer and Walther Jaenicke Institute of Physical and Theoretical Chemistry, University of Erlangen, Egerlanhtrasse 3, 0-852 Erlangen, Federal Republic of Germany J. Chem. Suc., Faraday Trans. I, 1987, 81, 161-166 By mistake table 1 of our paper contains some new measurements for TCNE' which were not discussed in the text and not given in fig. 3. As is seep in fig. 1 below, from these measurements a second straight line log (kex ~ ~ y - 5 ) as a function of y is found for ethers as solvents. We explain this result by a specific interaction between the C=C double bond of TCNE and the oxygen of the ethers.' -1.54 I I I - 0 o:i 012 0.3 0-4 0.5 0.6 7 Fig. 1. The solvent dependence of the electron self-exchange reaction TCNE/TCNE'- : x , non-ethereal solvents; m, ethereal solvents. W. Harrer, Ph.D. Thesis (University of Erlangen, 1986).

ISSN:0300-9599    

DOI:10.1039/F19888400365

出版商:RSC    

年代:1988

数据来源: RSC