|
11. |
Characterization studies of potassium phosphotungstate glasses and a model of structural units |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 251-267
Ulagaraj Selvaraj,
Preview
|
PDF (1100KB)
|
|
摘要:
J . Chern. Soc., Faraday Trans. I, 1989, 85(2), 251-267 Characterization Studies of Potassium Phosphotungstate Glasses and a Model of Structural Units Ulagaraj Selvaraj, H. G. Kershava Sundar and Kalya J. Rao* Solid State and Structural Chemistry Unit, Indian Institute of Science, Bangalore - 560 012, India K,GWO,-P,O, glasses have been studied over a wide range of com- positions. Their physical, thermal and spectroscopic properties, such as density, molar volume, microhardness, heat capacity, glass transition temperature and infrared spectral changes have been investigated. A structural model is presented to rationalize the experimental observations. According to this model, binary phosphotungstate glasses are built up of [WO,!,] octahedral and [POO,,,] tetrahedral units which share corners.Addition of K,O to binary phosphotungstates breaks the various linkages present in them. Consequently, a number of structural units are formed and the resulting glasses may be characterized by a network of polyhedra with different numbers of unshared corners. These structural units are con- solidated into a structural phase diagram. The colours of these glasses, due to the presence of W5+ ions and their stabilization in the network as [W5+06,,]- and [PO,,,]' structural pairs, are also discussed. In simple oxides, such as SO,, B,O, and P,05, the glass-forming tendency is pronounced, since they form extensive two- and three-dimensional networks consisting of linkages of the type -A-0-B-, where A and B are network-forming cations (i.e. Si4+, B3+ and P5+).1-4 The -A-0-B- bond angles vary over a wide range, which accounts for the loss of periodicity.In addition, the coordination numbers of A and B are small (3 or 4). Glasses are also formed over a wide range of compositions between transition-metal (TM) oxides, such as V205, MOO,, WO,, Fe,O, and CuO and simple oxides such as P,O, and B,0,.5-13 Several crystalline phosphotungstates are built of mixed [WO,,,] and phosphates structural units.14-16 A few studies on the presence of TM ions in octahedral coordination in glassy and amorphous systems have also been reported. Kim and Condrate'' have investigated the structure of W0,-P205 glasses. These studies indicate the presence of WO,, which share corners with the neighbouring phosphate tetrahedra or other WO, octahedra. EXAFS studies of potassium phosphotungstate glasses13 have suggested the presence of tungsten in octahedral coordination in these glasses.Recent e.s.r. investigations of phosphotungstate glasses" further confirm the presence of W" ions in octahedral coordination. Corroborative evidence for the presence of tungsten in octahedral coordination has also been obtained through studies of related glassy and amorphous systems. Based on X-ray diffraction analysis of the structure of K20-2W0, glasses, Imaoka and Hasegawa18 developed a structural model of these glasses. The model indicates that each structural unit consists of a WO, octahedron and a WO, tetrahedron combined with two K+ ions. Recent pulsed neutron diffraction study on 2Na20-3WO, and Na20-2W0, glasse~'~ further confirms the presence of WO, and WO, mixed structural motifs in these systems with rather high oxygen ion activity.The short-range order parameters determined from an analysis of the radial distribution function obtained from the electron-diffraction pattern of amorphous WO, also indicates the presence of WO, octahedra. This has been supported by modelling 25 1252 Pot ass ium Ph osp h o t ungs t a t e Glasses studies.’O All these studies therefore strongly suggest that the structure of phospho- tungstate glasses also possess WO, structural units. It can be seen from the electronegativity difference21 that W-0 bonds are considerably covalent and are comparable to Si-0 or Mo-0 bonds. We therefore propose on the basis of the above experimental facts that the phosphotungstate glasses consist of covalently bonded [WO,,,] octahedral structural units which participate in network formation.P,O, is also a well known covalently bonded oxide and provides [POO,/,] tetrahedral units for the network formation. One of the oxygen atoms in [POO,,,] unit is coordinated to phosphorus through a double bond. The structure of phosphotungstate glasses is therefore built of corner shared [WO,i,] octahedral and [POO,,,] tetrahedral structural units. Even though the adaptability of octahedral and tetrahedral units in a random network is not in accordance with the well known Zachariasen’s rules,, phosphotungstates form glasses over a wide range of compositions. Addition of K,O causes breaking of such a network, and as a result, non-bridging oxygen atoms or octahedra and tetrahedra with unshared corners are created.When the K,O concentration is high the number of unshared corners in octahedral tungsten units may increase, so that discrete anions may be formed with a resultant decrease in the coordination of tungsten from 6 to 4. In this paper, we present a number of physical, thermal and spectroscopic properties of phosphotungstate glasses and discuss the nature and modification of the network structure utilizing the notion of structural units found very appropriate in understanding the structure and properties of phosphomolybdate glasses. 8 * ’* 22-24 Experimental Glass Preparation The potassium phosphotungstate (PPT) glasses were prepared from reagent-grade K,CO,, WO, and (NH,),HPO, mixed according to their molar compositions. Batches to give ca.20 g glass were heated in a platinum crucible in an electric furnace first at 373 K and then at 673 K for several hours in order to remove water, carbon dioxide and ammonia and to sinter, respectively. The product was melted between I173 and 1573 K, depending on its composition for 30 min in air with periodic stirring to ensure homogeneity and attainment of thermal and chemical equilibrium. Glass discs, glass films and glass beads were obtained by the methods described During melting of the mixture, there was some loss of P,05 and WO,. However, chemical analysis25 of these glasses indicated that the loss was within 1 mol O/* of the glass compositions reported in table 1. The glasses prepared were coloured owing to the presence of reduced W5+ ions resulting from the loss of oxygen from the melts.The density, molar volume, heat capacity, glass transition temperature, e.s.r. and i.r. transmission measurements were made according to methods reported e l ~ e w h e r e . ~ ~ ’ * ~ ~ - ~ Since some oxygen is lost during melting, the molar volume of these glasses was calculated from the corrected molecular weight, M’, calculated using the relation 22-24 where x, y and (1 - x - y ) are the mole fraction of K,O, WO,-b and P,O,, respectively; MKBO, MwO,-, and M, o, are their respective molecular weights; 6 is the loss of oxygen, which was calculated From a knowledge of the concentration of W5+ ions determined from e.s.r. studies (see later). The loss of oxygen from WO, results in the conversion of W6+ + W5+ ions. In the i.r.spectra there were no characteristic absorption bands in the range 4000-1 400 cm-l except for shallow undulations resulting from interference effects in the film. This was confirmed by varying the thickness of the film; the positions and distances between two undulation maxima of glass films of different thicknesses were found to be different (fig. 1).glass no. composition (mol %) K,O WO, P,O, colour density / g molar volume /cm3 mol-' microhardness /kg mm-* Cp at (Tg-20)K 3nR T,/K /J mol-' K-' /J m% K-' /J mol-' K-' I 2 3 4 5 6 7 8 9 10 - 10 20 30 40 10 20 30 40 - 80 70 60 50 40 60 50 40 30 20 20 20 20 20 20 40 40 40 40 40 blue blue blue - - blue blue blue blue blue 5.55 5.06 4.71 4.32 3.92 4.68 4.07 3.69 3.38 3.06 38.5 39.5 39.6 39.9 40.4 41.8 44.7 45.6 45.7 45.9 48 7 375 333 288 245 552 41 8 339 293 250 98 I 887 833 78 I 712 772 730 689 652 623 I07 110 106 102 104 125 118 I17 117 114 - 50 60 76 85 36 37 35 36 38 114.7 112.3 109.8 107.3 104.8 129.7 127.2 124.7 122.2 119.7 N wl w254 Potassium Phospho t ungs ta te Glasses 1 1 1 1 1 1 1 1 1 1 1 l 1 1 1 1 1 1 1 1 1 1 1 1 0 1200 1000 800 600 400 20 w avenumber/an-' Fig.2. Infrared spectra of blown xK2@(80-x)W0,-20P20, glass films. Numbers in the figure represent glass numbers in table 1.U. Selvaraj, H . G. K. Sundar and K. J . Rao 255 6 I I I I I I I I I I I 1 1 1 I I I I I I do0 1200 1000 800 600 400 2 oc w avenumber/cm-' Fig 3. Infrared spectra of blown ~ K , ~ ( ~ O - . U ) W O , ~ O P , O , glass films.Numbers in the figure represent glass numbers in table 1. Results and Discussion The characterization data [density, molar volume, microhardness, glass transition temperature (q), heat capacity (C,) at ( T, - 20)K and heat-capacity difference (AC,) between the glass and the supercooled liquid at T,] of PPT glasses are listed in table 1. The i.r. spectra of PPT glasses are presented in fig. 2 and 3. respectively. Frequencies of prominent absorption peaks are listed in table 2 for the PPT glasses. The literature values of known absorption frequencies in related corn pound^'^*^^-^* are listed in table 3. Significant changes occur (new absorption peaks appear and some peak positions shift) in the spectra of both xK20-(80 - x)WO,-20P20, glasses [low-P205 glasses (LPG)] and xK2&(60 - x)W0,40P20, glasses [high-P,05 glasses (HPG)] with higher con- centration of alkali-metal oxide.The composition dependence of the spectra and various physical properties can be explained as a consequence of the formation of new structural units which result during modification of glass structure by K,O. A phase diagram comprising information about such structural units is presented below. A Structural Model of Potassium Phosphotungstate Glasses and a Phase Diagram of Structural Units In the light of the experimental findings summarized in the introduction section, we propose that the structure of binary phosphotungstate (W0,-P,O,) glasses consists of six connected octahedral [WO,,,] units and three connected tetrahedral [POO,,,] units,Table 2.Infrared absorption frequencies of dominant structural elements of K,&WO,-P,O, (PPT) glasses glass no. positions of absorption maxima/cm-Ia - 1 2 - 3 - 4 - 5 - 6 1340(w) 7 - 8 - 9 - 10 - - 1160 (sh) - 1160 (sh) - 1 170 (sh) - 1190(s) 1 170 (s) - 1220 (s) - 1260 (s) - 1240(~) - 1 2 7 0 ( ~ ~ ) - 1 2 6 5 ( ~ ~ ) - - 985 (vs) - 975 (vs) - 950 (vs) 1105 (sh) 930 (vs) 1065 (s) 910 (vs) 980 (sh) 1 1 15 (sh) 990 (vs) - 965 (vs) - 955 (vs) 1095 (s) - - - - 895 (s) - 885 (vs) - 870 (vs) 935(s) - 9 2 0 ( ~ ~ ) - 9lO(vs) - 9 0 5 ( ~ ~ ) - 885 (s) - - - 830 (s) 705 (s) 820 (s) 730 (s) - 740(~) - 755 (m) - 750 (m) - 760(m) - 755(m) - 755 (m) 720 (sh) 640 (s) 645 (s) 645 (m) 640 (m) 635 (m) 630 (w) 630 (w) 630 (w) - - 502 (w) - 500 (w) - 505(sh) 590(m) - 585(m) - - 500 (m) - 490(m) - 485 (m) - 500 (s) - 5 15 (s) - 365 (w) 370 (m) 370 (m) 370 (sh) 380 (w) 380 (w) 380 (m) 380 (m) dominent 2 5 2 s.225 (s) WW,WP 2 structural s units 2. cs, ~- 230 (s) WW,WP 230 (s) WW,WP,W, 230 (s) WP,W, 240 (s) WP,W,,W, 0, 255 (m) W,,W,,W,,P,,P E si 235 (m) WP,W,,W, n s 240 (m) WP,W,,P, E 260(m) PP,W,,P, 2 270 (m) W,,W3,P, 0 x t, a vs, very strong; m, medium; w, weak; sh, shoulder.U. Seluaraj, H . G. K. Sundar and K . J . Rao Table 3. Infrared absorption frequencies of some crystalline and glassy compounds related to K,O-WO,-P,O, glasses positions of absorption maxima compound /cm-'" WO:, (crystal) W,O,(PO,), (crystal) 815(s); 765(m); 390(sh); 375(m) 1225(vs); 1 I62 (m); 1 I25 (w); 1100 (s); 1070(s); 990 (vs); 960 (m); 918 (w); 869 (s); 785 (vs); 720 (sh); 659 (s); 629 (w); 595 (m); 575 (s); 545 (w); 450 (w); 431 (s); 400 (s); 388 (w); 341 (s); 320 (m) 830 (s); 315(s); 295 (m) 1285 (s); 950 (s); 780 (vw); 650 (w); 475 (s) 1 1 10 (vs); 1080 (sh); 1050 (sh); 1025 (s); 980 (w); 930(m); 890 (vs) ; 720 (m) 1300 (s); 1266 (vs); 1 I50 (s); 1100 (s); 851 (vs); 760 (m); 677 (s) K,WO, (crystal) p,o, (glass) K,P,O, (crystal) (KPO,), (crystal) 257 a ~ ~ , very strong; m, medium; w, weak; sh, shoulder.which possess linkages of the type, W-0-W (WOW), W-0-P (WOP) and P-0-P (POP). Every P,O, formula unit gives rise to two [POO,,,] tetrahedral units in which three corners are used for oxygen-sharing (bridging oxygen). Structure of glass with equimolar proportions of WO, and P,O, can be built in such a way that [WO,,,] units are all surrounded by [POO,l,] tetrahedral units and cice versa, a situation similar to a chemically ordered random n e t w ~ r k .~ * > ~ ~ Further, W0,-rich glasses would contain principally (WOW + WOP) linkages, whereas P,O,-rich glasses possess (WOP + POP) linkages. A possible reason for the chemical ordering is that WOP linkages are energetically more favourable than a combination of WOW and POP. The W-0 bond energy34 (in WO,) is 136.82 kJ mol-', whereas the P-0 bond energy,' (in P,O,,) is 351.46 kJ mol-'. The bond energies of various linkages change in the order < E,,,, < Go,. Further, Lor is greater than the mean energy of f&.o,v and E,,,,,,. Therefore, the order of easiness of bond breaking is WOW > WOP > POP, and hence the linkages can be safely assumed to be broken sequentially in the same order when the K,O content in the glass increases.For example, when K,O is added to W0,-rich glasses WOW linkages between two [WO,,,] (= W) octahedra are broken first and this can be represented by the following chemical reaction : 2[W06,,]+K,0 --+ 2[W0,1,0]-+2K'. The units of the type [WO,,,O]-, which have non-bridging oxygen, are the new structural units. [WO,,,O]- is designated as W,. Similarly, various other structural units are formed as more and more bonds are broken or equivalently more and more bridging oxygens are converted into non-bridging oxygens. Some examples of structural modifications and the resulting structural units are given below. [W0,,,]+K20 + ~O,,,O2]'-+2K+; [w0,,20,]2- = W, 2[WO6,,]+ 3K,O + 2[W0,,,0,I3-+6K+; [W03,,03]3- = Wi 2[POO,/,] + K 2 0 + [POO,,,O]- + 2K' ; [POO,,,] = P258 Potassium Phospho t ungs ta te Glasses [POO,,,O]- = P, = PO3- (metaphosphate unit) [POO,,,] + K,O -, [PO0,,,0,]2- + 2K+; [POOl,,0,]2- = P, 2[POOl,,0,]2- = P20:- (pyrophosphate unit) 2[POO,,,] + 3K,O -, 2P0,3-+ 6K+; PO;- = P, (orthophosphate unit).The W, unit can be quite stable provided the two negative charges are present on two diametrically opposite atoms in the octahedron. However, breaking of the next WOW or WOP linkage on a W, centre and thereby creating a Wi centre is not likely because it only helps the conversion of Wi into W, (= WO,"-). The third non-bridging oxygen [ -0-1 in the octahedron can only be present as a close neighbour of the two already existing [ - 0-1 ions, which result in strong electrostatic repulsions.This induces conversion of W;l into W,. The additional oxygen ion is simply given up, which can be expressed as 2[w0,,,0,]2- + 0,- --+ [wo,,203]3-; 2[Wo,,,o,]3- --+ 2w0:- + 02-. Thus W;l is never really created as a stable species. An important aspect of W, --+ W, conversion is breaking and rearrangement of the remaining WOW or WOP linkages of W, (with other polyhedra) so that no uncharged broken linkages (dangling bonds) are retained. We expect that this rearrangement takes place at high temperatures in the molten state. This rearrangement does not require any extra reaction volume, since the superscribed spheres around W, and W, defects have the same radius.Based on the knowledge of (a) the reactions for the formation of various structural units, (b) glass composition and (c) energy hierarchy of WOW, WOP and POP linkages in the network, along with the assumption that the maximum possible number of WOP connections are always present in the glasses, we can determine the structural units present in any binary or ternary composition. The following examples are given to illustrate the nature of structural units in typical glass compositions. It is assumed that all the tungsten ions are present in the W6+ state. (A) 80WO,-20P,O5 glass : 80W03-20P,0, -, 2O[WO6,,I [poo3/212 + 3O[wO6,,l [WO,,,l --+ 20WP+30WW where WP = [WO,,,] [POO,,,], and WW = [WO,,,] [WO,,,]. (B) 20K,0-40W03-40P,05 glass: 40[w06/2] [POO,,,], + 20K20 + 20w, + 20P, -k 20P + 20wP -k 40K+.(C) 40K,0-40W03-20P,0, glass : 2O1WO6,,I [poo3/212 + 1O1WO6,,I WO,,,l + 40K20 -+ 2O[WO6,,I [poo3/212 + 20W, + 40K+ + 20K,O -+ 20W, + 20P, + 20P + 20W, + 80K+. The structural units present in the entire region of K,O-WO,-P,O, glasses are shown in fig. 4 as a ternary phase diagram of structural units. In this structural phase diagram, structural units specific to each boundary are denoted. The proportion of various structural units at any point in the bounded area can be determined using an appropriate lever rule. However, structural units such as W, can be in equilibrium with W, and with W,, which results from the following reactions: 2w,,w, + w w, + W@W, + w . The equilibrium of such interconvertible species may be important at high temperature,U.Selvaraj, H. G. K. Sundar and K . J. Rao 259 20 40 60 80 p2°5 WW+WP WP PP+WP Fig. 4. Structural phase diagram of K,GWO,-P,O, (PPT) glasses. Structural units present in each region are determined by the units indicated on boundaries. Symbols are explained in the text. Closed circles correspond to composition investigated in this work. particularly if the reaction enthalpy is not very high. The boundaries in the structural phase diagram of fig. 4 are therefore an idealized representation. Using the structural phase diagram we can identify structural units such as WW, WP, PP, W,, P, etc. present in any given glass. We have ignored further differentiation of structural units of the type W,P, W,W,, W,P, W,W, etc. in the phase diagram. They are of little significance to the present objectives of this investigation. While a full statistical analysis of the species present is possible in principle, we desist from attempting a further refinement of this model owing to the absence of necessary thermodynamic data.The dominant structural units present in our glasses are listed along with the i.r. data in table 2. The various properties and the i.r. spectra are discussed below in the light of the structural model given above. Density and Molar Volume The variations of density and molar volume of LPG and HPG with composition are presented in fig. 5 and 6, respectively. The molar volumes are significantly high, suggesting a rather open structure. Otherwise, the molar volume would roughly correspond to the total volume of 02- and K+ ions if the glasses were purely ionic, with W6+ and P5+ ions occupying interstices generated from random close packing (RCP) of the only two large 0,- and K+ ions.On this for example, glass of composition 30K,O-50W0,-20P,O, should correspond to a molar volume of 31.3 cm3 mol-’, while the actual molar volume is 39.9 cm3 mol-1 and is much higher (by ca. 27 YO) than the260 Potassium Phosphotungstate Glasses 6 .O 2 I I I 1 0 10 20 30 40 Kz 0 ( ~ 1 % ) Fig. 5. Variation of density with glass composition for (a) xK,O-(80 - x)W0,-20P2O, and (b) xK,0-(60 - x)WO,4OP,O, glasses. 1 I 1 I 0 10 20 30 40 Kz 0 ( ~ 1 % ) Fig. 6. Variation of molar volume with glass composition for (a) xK,O-(80 - .u)W0,-20P20, and (6) xK,&(60 - x)W0,40P20, glasses.RCP volume of the larger ions. Therefore the glass possesses a more open structure. Such higher rnoIar volume (open structure) strongly suggests the presence of a covalent network. The molar volume of LPG is always lower than that of HPG for a given concentra- tion of K,O, since the substitution of WO, by P,O, introduces twice the number ofU . Selvaraj, H . G . K . Sundar and K . J . Rao 26 1 I I 1 1 I I 0 10 20 30 40 Fig. 7. Variation of microhardness with glass composition for (a) xK2w80 - x)WO,-20P20, and (b) xK20-(60 - .u)W0,40P20, glasses. Kz 0 (m~l%) [POO,,,] tetrahedral units for each [w0,,2] octahedral unit. As a rough estimate, for example, it is possible to compare the change in molar volume when one mole of crystal- line P,O, is substituted by crystalline WO,. The increase in weight for the process is 89.91 g, while the volume decreases by 27.01 cm3 mol-l.This suggests the direction of density variation as a function of P,O, concentration even in glasses. The molar volume of LPG remains almost constant, while that of HPG increases during the substitution of WO, by K,O. In the case of HPG substitution of WO, by K,O causes the formation of structural units of the type [POO,/,O]-, which have a slightly higher specific volume compared to other structural units in LPG. Because the [POO,,,O]- (E PO, or metaphosphate) units are only two-connected (to the neighbouring polyhedra), it can rotate eccentrically around the -0-P-0- chain. Hence it pushes the surrounding structural units outwards, resulting in a slight increase of specific volume.Microhardness The microhardness values (H,) of PPT glasses are listed in table 1 and are also shown in fig. 7 as a function of composition. Microhardness of the binary 60W0,40P2O, glass is much higher than that of 80W0,-20P20, glass. The magnitude of microhardness has now been known to be empirically related to bond ene~gies.~*-~O In the present glasses, W-0-W, W-0-P, and P-0-P are the principal covalent linkages which determine the cohesive energies of the network, and the energies of these linkages increase in the same order. Thus the microhardness of HPG (60W0,-40P2O, glass) which consists of larger numbers of stronger covalent W-0-P and (some) P-0-P linkages, is higher than that of LPG (80W0,-20P2O, glass) in which larger proportion of comparatively weaker covalent bonds of the type W-0-W are present.262 Potassium Phosp ho t ungs tate Glasses 200- 220- I 11 7 160- 180- Y 80- 100- I I , I I I I I I I 1 I 700 740 780 I 820 860 900 1 I .I I 1 I I I I I I I 1 800 840 880 920 960 1000 TIK Fig. 8. Temperature dependence of heat capacity for xK20-(80 - x)WO,-20P20, glasses. Numbers in the figure represent glass numbers in table 1. I and I1 in the heat capacity and temperature scales correspond to different glasses. The microhardness value decreases sharply with alkali-metal oxide content as a result of the modification of the glass network. This decrease is expected because the method of Hv determination involves deformation, and the deformation of covalently bonded structure is more difficult than the deformation of modified, more ionic ~ t r u c t u r e s .~ ~ ~ 41 Further, the decrease in microhardness values is more pronounced in HPG as compared to LPG. This is again a consistent observation, because stronger P-0-P and W-0-P linkages are involved in the process of modification in HPG, whereas relatively weaker W-0-W linkages are involved in LPG. Heat Capacity and Glass Transition Temperature Heat-capacity plots for PPT glasses are presented in fig. 8 and 9, respectively. Heat capacities at 20 K below Tg for all the glasses are listed in table 1. Calculated values of the Dulong-Petit heat capacity of these glasses have also been listed. Table 1 shows that C, values near Tg are lower than the Dulong-Petit heat capacity. The low C, values are vital clues to the presence of strongly coordinated groups of atoms rather than isolated atoms.The Dulong-Petit Cp values are lower, in spite of significant contribution to heat capacity from frozen entropies of these glasses. The addition of modifier oxides to PPT affects the magnitude of the heat capacity. It degrades the network structure, so that the melt viscosity decreases. Glasses formed from such systems tend to become more ' ideal ' in the sense of lower magnitude of frozen entropy (or the corresponding heat capacity).42 This effect is borne out by the decreased heat capacity in both LPG and HPG (table 1) upon modification. For example, C, decreases by ca. 22 J mol-' K-' as the K 2 0 concentration in PPT glasses increases from 10 to 40 mol0/o. The expected decrease in Dulong-Petit heat capacity as a result of a decrease in the number of particles due to the substitution of K 2 0 (3 particles) for WO, (4 particles) is very much lower (ca.10 J mo1-I K-'). The variation of ACp in PPT glasses is also interesting (table 1). It is generally much lower for HPG as compared to LPG. We may also note that ACp values in HPG are approximately constant, while ACp increases with increasing network degradation in LPG. The AC, value at Tg is related to the viscosity behaviour of the melts. For example,U. Selvaruj, H . G. K. Sundar and K. J . Rao - 100- 263 I I I I I I I I I I I I I Fig. 9. Temperature dependence of heat capacity for xK,O - (60 - x)W0,40P,05 glasses. Numbers in the figure represent glass numbers in table 1.I and I1 in the heat capacity scales correspond to different glasses. it is very small in glasses whose melts are highly viscous above Tg, such as in silica, and is large for ionic glasses whose viscosities are low above q.43 We can attribute the difference in behaviour of ACp values between HPG and LPG in the present study to the possible differences in viscosity behaviour of the respective melts above Tp. HPG samples contain a greater proportion of stronger W-0-P or P-0-P linkages, whereas in LPG sample there is a greater concentration of weaker W-0-W linkages. The viscosities of HPG and LPG above < are thus expected to be high and low, respectively (at a corresponding temperature above TR; Tg itself is nevertheless an isoviscous temperature). ACp values are, thus low in HPG and high in LPG, in reasonable agreement with a general observation made by Angel1 and Sichina'" about the relation between viscosities and ACp of glass-forming melts close to TR.Further, LPG exhibit narrow glass transition region, while in HPG this region is significantly broad. This feature is again consistent with the possibility that LPG behave more like ionic glasses. The glass transition temperatures (T,) for various glass compositions are listed in table 1 and are also shown in fig. 10. There are two interesting features in the behaviour of q. First, is higher for LPG as compared to HPG for a given concentration of K,O, and secondly, it decreases with increasing K,O content in both systems. It is understandable that Tg decreases with increasing K,O content in both LPG and HPG (fig.lo), since the network structure is weakened by the addition of K,O. In the light of the cluster model of glass transition, discussed e l ~ w h e r e , ~ ~ the decrease in Tg indicates a decrease in the cage-like vibrational frequency of the most mobile species in the tissue portion of the glass (in this case the K+ ion). In order to apply the cluster-tissue description to phosphotungstate glasses, we assume that clusters are the more ordered polyhedral groupings, and the tissue is the disordered and distorted regions of the glass structure. With modification (or addition of K,O), the specific volume increases, particularly with the formation of two-connected species like [POO,,,O]-. The increase in specific volume broadens the energy profile and decreases the energy separations or vibrational excitation energy ; hence it decreases q.values of HPG and LPG in PPT glasses are quite different from those of potassium phosphomolybdate glasses.' In potassium phosphomolybdate glasses, HPG exhibit higher glass transition temperature than LPG for a given concentration of K,O, whereas it is just the opposite in PPT glasses. In fact in binary W0,-P,O, glasses, the magnitude of Tg changes by ca. 210 K for a change of However, the relative264 Po tassium Phospho t ungs ta t e Glasses 1000 c 0 10 20 30 40 Kz 0 (m~l%) Fig. 10. Glass transition temperatures (T,) for (a) xK,&(80 - x)WO3-20P,O, and (6) xK,&(60 - x)W0,-4OP,O5 glasses. composition from 80W0, to 60W0,. Since generally (but not always) scales with the liquidus temperature (TL), the dissimilar behaviour may be due to dissimilar trends in TL of the two systems in the composition range investigated.The melting temperature of P,O, ( T , = 855 K) is close to that of MOO, (T, = 1068 K) but far less than that of WO, (T' = 1746 K). In phase diagrams of systems with strongly dissimilar melting temperatures, like that of W03-Pz05, in the simplest situation the liquidus temperature is expected to rise steeply on the W0,-rich side. On the other hand, the same region may consist of a deep eutectic in the MOO,-P,O, system since the melting temperature of MOO, and P,O, are similar. Thus in the region 2WO mol% P,O,, the phase diagram may consist of liquidus lines with opposite trends in the two systems, W0,-P,O, (increasing liquidus temperature) and MOO,-P,O, (decreasing liquidus temperature), and thereby give rise to opposite trend in Tg.Unfortunately, the phase diagram of these systems are not available to ascertain this explanation. Infrared Spectra In glass 1 (fig. 2) of PPT glasses, the absorption peak at 985 cm-l may be associated with W-0 stretching frequency of [WO,,,] groups (in view of data in table 3) which are simultaneously part of WW and WP structural units. The absorption maximum at 640 cm-' may be assigned to 0-W-0 stretching frequency" of WW or WP structural units. A shoulder at 1160 cm-l and the weak band at 502 cm-' may be associated with P-0 stretching and 0-P-0 bending vibrations,28 respectively, of P units (from WP). Glass 2, in which WO, is substituted by K,O, does not have any additional peaks, owingU .Selvaraj, H . G. K . Sundar and K. J. Rao 265 to the low concentration of W,. However, the presence of W, causes a downward shift of the W-0 stretching frequency as a result of the formation of non-bridging oxygen atoms. The addition of K,O thus weakens the W-0 bond, and the shift of W-0 stretching frequency represents a decrease of the relevant force constant. In glass 3 the concentration of W, is further increased, and additional absorption maxima at 895, 830 and 365 cm-' register their presence. Manifestation of a peak at 705 cm-' may be because of the relatively higher concentration of WP units. In glass 4, the formation of W, is expected, and correspondingly a new absorption at 590 cm-' is observed. Also, the features corresponding to WP units are more resolved, for they are isolated in the structure by the intervention of W, and W, units in the structure; a shoulder near 1105 cm-' due to P-0 stretching in P units (from WP) becomes apparent.Glass 5 contains P, units and exhibits additional features, such as clearly resolved peak in the region 1000-1200 cm-' and an additional shoulder at 980 cm-l. The absorption in the region 1000-1200 cm-l has been assigned to P-0 stretching vibrations.26 Glasses 610 (fig. 3) correspond to HPG, and they contain higher concentrations of WP structural units. In glass 6 the weak absorption band at 1340 cm-l corresponds to P=O stretching of the P units (part of WP). The band has been identified as a P=O stretching mode in phosphosilicate glasses.46 In a study of phosphosilicate films W ~ n g ~ ~ has observed a similar band at 13 10 cm-l.A similar band has also been observed in the i.r. spectra of a number of organophosphorus compounds containing the P-0 Absorption maxima at 1220, I 1 15, 935, 755 and 500 cm-' may all be attributed to WP structural units, while bands at 990, 630, 370 and 225 cm-' originate from [WO,,,] (= W). The spectra of glasses 7 and 8 have the same features as in glass 6 except for: (i) the disappearance of the peak at 1340 cm-l, (ii) a downward shift of the W-0 stretching frequency and (iii) an upward shift of the band at 1220 cm-' (glass 6) to 1260 cm-' (glass 8). The downward shift of the W-0 stretching frequency may be associated with the formation of W,, W, and W,. The band around 1260 cm-' is due to P=O stretching, which is also known to occur in many alkali-metal and alkaline-earth metaphosphate glasses.48 In silver phosphate glasses4' this band appears between 1220 and 1240 cm-'.The change in P=O stretching frequency in many inorganic phosphates is related to the polarizing power of the modifying cations, which alters the bond order and hence the force constant of the P=O unit in the phosphate structure. In glass 8 we expect P, (= metaphosphate groups) also to be present. Consequently, the peak at 1260cm-' is observed. The variation of this frequency may be associated with the concentration variation of the modifying K' cation. The spectrum of glass 9 retains all the essential features of the spectrum of glass 8, except for a higher intensity of the peak at 1270 cm-' and the disappearance of W-0 stretching frequency.Owing to the higher concentration of P, the intensity of the 1270 cm-' peak increases and the W-0 stretching frequency merges with the band associated with P or P,. The spectrum of glass 10 bears out the effect of the dominance of W,, W,, P, and PP structural units. The additional peak at 1095 cm-' and the shoulder at 720 cm-' are due to high concentration P,. Thus the infrared spectra of the whole range of glasses is consistent with the nature of structural units present in them. Colour and Non-stoichiometry of WO, in Relation to the Glass Structure The PPT glasses prepared by quenching the high-temperature melts are coloured owing to the presence of W5+ ions.The colour of various glasses is listed in table 1. It has been noted by other worker~"~ that many TM phosphates tend to lose oxygen in their molten state during the preparation. The charge imbalance created by a the loss of oxygen in the system is compensated by the reduction of TM ions. As a result of the loss of oxygen, stoichiometric WO, becomes non-stoichiometric WO,,-. The charge balance is achieved266 Potassium Phosphotungstate Glasses Table 4. Concentrations of W5+ and W,,,,, and their ratio in K,C&WO,-P,O, (PPT) glasses composition (mol Yo) W5+ in Wtotal in K,O WO, P,O, /g /g W5+/Wtota, 1 mol glass 1 mol glass - 80 20 0.47 1 147.080 0.0032 10 70 20 0.154 128.695 0.0012 - 60 40 2.515 110.310 0.0228 10 50 40 1.875 9 1.925 0.0204 20 40 40 I .743 73.540 0.0237 30 30 40 0.441 55.155 0.0080 by the chemical conversion of W6+ to W5+, which is accompanied by the structural conversion of [POO,,,] to [PO,,,]+ units.The W5’ ions are stabilized in the glass structure as ~5+0,,2]-[P0,,2]+ (= W’P’) coulombic pairs. A combination of W5+O,,,]- ( G W’) and [PO,,,]+ (= P’) units can accommodate oxygen deficiency and achieve electrical neutrality over short distances (maximizing coulombic energy). The Jahn-Teller W5+ ions are stabilized in [W”’O,,,]- units by distorting the octahedra to a lower symmetry (possibly C4). The concentration of W5+ ions of PPT glasses are listed in table 4. The number of W5+ ions is lower than that of HPG. The effect of W5+ ions and their importance in transport properties and electron spin resonance are discussed elsewhere.”*22 Conclusions Potassium phosphotungstate systems form glasses over a wide range of compositions.The binary phosphotungstate [WO,-P,05] glasses are considered to be made up of a random network of corner-shared [WO,,,] octahedral and [POO,,,] tetrahedral structural units. Addition of K,O causes breaking of such network, and as a result non-bridging (unshared) charged oxygen centres (structural units) are created. The resulting structural units are characterized on the basis of glass composition and the relative energy of covalent linkages. PPT glasses are therefore conveniently described using a phase diagram of structural units. We have rationalized the properties of these complex glass systems using the phase diagram of structural units.The colour of the glasses, which is due to the presence of W5+ ions, and their stabilization in the glass structure as ~ 5 + 0 6 , a ] - [POal2]+ structural pairs have also been discussed. References 1 H. Rawson, Inorganic Glass Forming Systems (Academic Press, New York, 1967). 2 W. H. Zdchariasen, J . Am. Chem. Soc., 1932, 54, 3841. 3 S. R. Elliott, Physics qf Amorphous Maferiuls (Longman, London, 1984). 4 U . Selvaraj, K. J. Rao, C. N . R. Rao, J. Klinowski and J . M. Thomas, Chern. Phj3.s. Letf., 1985, 114, 5 G. S. Linseley, A. E. Owen and F. M. Hayatee, J . Non-crysf. Solid.y, 1970, 4, 208. 6 M. Sayer and A. Mansingh, Phys. Reu. B, 1972, 6, 4629. 7 B. Bridge and N. D. Patel, J . Mater. Sci., 1986, 21, 1187. 8 U.Selvaraj and K. J. Rao, J . Non-cryst. Solids, 1985, 72, 315. 9 U. Selvaraj and K. J. Rao, J. Non-cryst. Solids, 1988, 104, 300. 10 R. Braunstein, Solid State Commun., 1978, 28, 839. 1 1 Ch. Ruf, K . Barner and R. Braunstein, Solid State Commun., 1985, 54 1 1 I . 12 C. Y. Kim and R. A. Condrate Sr, J . Phys. Chem. Solids, 1984, 45, 1213. 13 F. Studer, A. Lebail and B. Raveau, J . Solid State Chem., 1986, 63, 414. 14 A. F. Wells, Structural Inorganic Chcrnisfry (Oxford University Press, Oxford, 1975). 24.U. Seltlaraj, H. G. K. Sundar and K. J . Rao 267 15 B. Raveau, in Advances in Solid Sfate Chemistrj., ed. C. N. R. Rao (Indian National Science Academy, 16 M. Lamire, Ph. Labbe, M. Goreaud and B. Raveau, J . Solid State Chem., 1987, 66, 64. 17 U. Selvaraj and K.J. Rao, Chem. Phys., 1988, 123, 141. 18 M. Imaoka and H. Hasegava, Yogyo Kyokai Shi, 1976, 84, 389. 19 S. Kobayashi, M. Misawa and K. Susuki, J . Non-cryst. Solids, 1987, 91, 180. 20 J. Greeneche, J. Teillet and J. M. D. Coey, J . Non-cryst. Solids, 1986, 83, 27. 21 L. Pauling, Nature of the Chemical Bond (Cornel University Press, New York, 1960). 22 U. Selvaraj and K. J. Rao, Philos. Mag., 1988, 58, 203. 23 K. J. Rao, G. Parthasarathy, U. Selvaraj and E. S . R. Gopal, Phys. Chem. Glasses, 1985, 26, 101. 24 U. Selvaraj, Ph.D. Thesis (Indian Institute of Science, Bangalore, 1987). 25 Vogers Textbook of Qualitatice Inorganic Analj~sis, ed. J. Bassett, R. C. Denney, G . H. Jeffery and J. Meudham (Longman, London, 1978). 26 D. E. C. Corbridge and E. J. Lowe, J. Chem. Soc., 1954, 493. 27 D. E. C. Corbridge, Top. Phosphorus Chem., 1969, 6, 235. 28 J. Wong and C. A. Angell, Glass Structure by Spectroscopy (Marcel Dekker, New York, 1976). 29 K. Ohwada, Spectrochim. Acta, Part A , 1970. 26. 1035. New Delhi, 1986). 30 K. Nakamoto, Infrared and Raman Spectra qf Inorganic and Coordination Compounds (Wiley, London, 1978). 31 M. F. Daniel, B. Desbat, J. C. Lessegues, B. Gerand and M. Figlarz, J . Solid State Chem., 1987, 67. 32 G. Lucovsky, Phys. Rer. B, 1977, 15, 5762. 33 K. J. Rao and R. Mohan, J . Phys. Chem., 1980, 84, 1917. 34 R. J. Ackerman and E. G. Raugh, J . Phys. Chem., 1963, 67, 2596. 35 D. E. C. Corbridge, The Structural Chemistry of Phosphorus (Elsevier, Amsterdam, 1974). 36 J. D. Bernal, in Liquid: Sfructure, Properties and Solid Interactions, ed. T. T. Hughel (Elsevier, 37 H. G. K. Sundar and K . J. Rao, J . Chem. Soc., Faraday Trans. I , 1980, 76, 1617. 38 P. S. L. Narasimham and K. J. Rao, J . Non-crj,st. Solids, 1978, 27, 225. 39 M. Yamane and J. D. Mackenzie, J . Non-crjist. Solids, 1974, 15, 153. 40 D. M. Marsh, Proc. R. Soc. London, Ser. A , 1964, 279, 420. 41 J. D. Mackenzie. Mechanical Behaz:iour qf Materials (Society of Materials Science. Japan, 1972). vol. IV, p. 347. 42 K. J. Rao, in Preparation and Characterization of Materials, ed. J. M. Honig and C. N. R. Rao 43 C. A. Angell and W. Sichina, Ann. N. Y . Acud. Sci., 1976, 279, 53. 44 K. J. Rao, Proc. Indian Natl Sci. Acad., 1986, 52, 176. 45 E. A. Corl, S. L. Silverman and Y. S. Kim, Solid Stare Electron., 1966, 9, 1009. 46 J. Wong, J . Non-cryst. Solids, 1976, 20, 83. 47 L. J. Bellamy, The Infrared Spectra of Comples Molecules (Chapman and Hall, London, 1975). 48 C. K. Shih and G . J. Su, Proc. 7th Int. Congr. Glass (Brussels. 1965). 49 R. F. Bartholomew, J . Non-Crjyst. Solids, 1972. 7. 22 1. 235. Amsterdam, 1965). (Academic Press, New York, 1981). Puper 8/0057 I K ; Receired 16th February, I988
ISSN:0300-9599
DOI:10.1039/F19898500251
出版商:RSC
年代:1989
数据来源: RSC
|
12. |
Influence of pretreatment on the properties of Ag/α–Al2O3catalysts containing large(± 1µm) pure and Cs-promoted silver particles. Part 1.—Extent of oxygen and hydrogen sorption and TPD studies |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 269-277
Garmt R. Meima,
Preview
|
PDF (864KB)
|
|
摘要:
J. Chem. SOC., Furaduy Trans 1, 1989, 85(2), 269-277 Influence of Pretreatment on the Properties of Ag/cc-Al,O, Catalysts containing Large (& 1 pm) Pure and Cs-promoted Silver Particles Part 1.-Extent of Oxygen and Hydrogen Sorption and TPD Studies Garmt R. Meima,"? Loek M. Knijff, Ralph J. Vis, Adrianus J. van Dillen, Frederik R. van Buren7 and John W. Geus Department of Inorganic Chemistry, State University of Utrecht, Croesestraat 77", 3522 AD Utrecht, The Netherlands In order to elucidate the influence of the pretreatment on the (catalytic) properties of silver catalysts containing large silver particles (& 1 pm), the interaction of oxygen and hydrogen with reduced and pre-oxidized samples has been investigated. The effect of the presence of caesium on the silver surface has also been investigated.Volumetric (ad)sorption of oxygen and hydrogen, TPD experiments, corn bined with chemisorption measurements, and electron microscopy were used as characterization techniques. The combined results indicate that the penetration of oxygen into these large silver particles does not occur to any significant extent. When the catalysts are saturated with oxygen, only ca. 40% of the monolayer coverage of oxygen atoms is situated in the subsurface layers. This type of pretreatment exhibited small, but distinct, effects on the uptake of oxygen and hydrogen at 443 K. Higher uptake of oxygen was observed after reduction of the catalyst at more elevated temperatures. The presence of caesium lessens considerably the variations in the uptake. Supported silver catalysts are unique in their ability to catalyse the industrial production of ethylene oxide from ethylene.This property has prompted much basic research into oxygen chemistry at silver surfaces. However, there is still no general consensus as to the relative amounts and the various states of chemisorbed oxygen, in spite of a vast number of investigations. The epoxidation of ethylene is generally claimed to be structure-sensitive. Therefore, studies have also been concerned with factors such as the mean particle size2T3 surface morphology, and the effect of silver crystallite ~rientation,**~ many of which are in conflict. Therefore, this and previous s t u d i e P were initiated to gain more insight on the influence exerted by the particle size and the topography of the silver surface with respect to the oxygen-silver chemistry. It has recently been demonstrated that the presence of subsurface oxygen is crucial for a high selectivity in the ethylene epo~idation.~ It is therefore especially important to discriminate between the oxygen present on the surface of the silver particles and that which has diffused into the bulk of the silver. In earlier work reported by our group, particle sizes of 70 and 150 nm were studied.6-8 The pretreatment exerted a strong influence on the activity of these catalysts toward the oxidation of carbon monoxide with molecular oxygen. Moreover, it also influenced the extent of oxygen uptake at 443 K.It was argued that during the various pretreatments, changes took place in the surface structure of the silver particles and that these surface structural changes were responsible for the variation in activity and extent of oxygen sorption.t Present address: Dow Chemical (Nederland) B. V., P.O. 48. 4530 AA Terneuzen, The Netherlands. 10 269 F A R 1270 Agla-Al,03 Catalysts The aim of the present study was to extend the work mentioned above to larger (supported) silver particles. Another goal was to study the influence of caesium (a known selectivity promoter for ethylene epoxidation) in relation to the phenomena discussed above. The relative amounts of adsorbed and absorbed oxygen were studied by combined oxygen chemisorption and TPD measurements, which in previous work6* was shown to be a powerful technique for discriminating between the two species, since penetrated oxygen does not desorb during TPD measurements up to 673 K. Another independent characterization technique that can be used to distinguish between the two species is the measurement of hydrogen sorption.The total uptake of hydrogen relates to the total amount of sorbed oxygen, whilst the amount of hydrogen rapidly adsorbed relates to the amount of surface oxygen. lo In this paper the results are presented on the influence of pretreatment on the uptake of oxygen and hydrogen by large (supported) silver particles. The effect of pretreatment on the activity toward CO oxidation will be discussed in an accompanying paper." Experimental Catalysts The catalysts were prepared according to a described method.12 In short, the preparation procedure was as follows: spheres of a-alumina (A in diameter) with a surface area smaller than 1 m2 g-' were dried under vacuum at 343 K for 30 min.The support was then impregnated (incipient wetness) with a solution containing silver nitrate in demineralized water. The impregnated support was subsequently dried under vacuum at 343 K for 30 min, after which time an amount of dioctyl sebacate was vacuum impregnated. The support was than heated to 573 K for 45 min to reduce the silver and remove the excess dioctyl sebacate. In this case, the above procedure was performed twice in order to obtain high loading, viz. 19.9 wt % Ag. The catalyst was caesium promoted by placing it in a flask containing an amount of Cs2C03 in methanol. This was rotated under vacuum at 443 K until all the methanol had evaporated and was then dried at 383 K for 2 h.The measured caesium content was 190 ppn?. In order to assess the effect of caesium unambiguously, the catalysts with and without caesium must not differ in their structural properties. Therefore, the caesium-free catalyst used in this work was obtained by employing a wash procedure on the caesium- containing catalyst. The procedure was as follows. The promoted catalyst (5.7 g) was placed in a Soxhlet apparatus and washed with water (125 cm3). The wash water was decanted and fresh water (150 cm3) was added after which the catalyst was washed a second time and dried at 343 K. The caesium concentration in the wash water was determined with a Jobin Yvon inductively coupled plasma atomic emission spectrometer JY 38 (ICP-AES). After the first wash procedure, the wash water was found to contain 90 ppm caesium and 15 ppm after the second.Analysis of the catalyst after the wash procedure showed that the catalyst was free of caesium within the detection limits. The silver content of the washed catalyst remained unchanged. The catalysts are designated Ag-Cs and Ag, respectively. Adsorption and TPD Studies Procedures The adsorption studies and the thermal desorption experiments were conducted in a classical Pyrex high vacuum and gas supply system described elsewhere.13 The instrumentation and measuring procedures have also been reported.' All adsorptionG . R. Meima et al. 27 I isotherms were measured at 443 K, usually in the range &5 Torr?.For mean particle- size calculations, the isotherms were extrapolated to zero pressure as is often done. As an alternative, single-pulse experiments were also performed. In this case a single pulse of oxygen was dosed, leading to varying (final) pressures above the catalyst. The total amount of oxygen taken up was registered at various time intervals. TPD (temperature-programmed desorption) experiments were performed at a heating rate of 5 K min-'. The gas phase was continuously analysed by means of a mass spec t rome ter . Unless stated otherwise, pretreatment of the catalyst was performed overnight at various temperatures, either in a 10 YO H,-N, flow or a 100 YO oxygen flow followed by evacuation at the desired temperature (usually the same as the pretreatment temperature) for at least 1-2 h.The gases were purified and dried using standard methods. Oxygen (99.999 */o) and hydrogen (99.998 YO) for the adsorption experiments were obtained from AGA Gas and used without further purification. The preshaped catalyst particles were crushed into smaller particles and a sieve fraction of 0.5-1.0 mm diameter was used in the experiments. The adsorption and TPD measurements were performed with 13.01 g of the Ag-Cs catalyst and 8.47 g of the Ag catalyst. The morphology of the silver particles was examined by a Philips 505 scanning electron microscope which was operated at an accelerating voltage of 30 kV. In order to avoid electrostatic charging a sputtered gold layer was applied. Results and Discussion Extent of Oxygen Uptake at 443 K after Various Pretreatments Fig.1 shows the oxygen adsorption isotherms after various pretreatments recorded for the Ag-Cs catalyst at 443 K. The numbers of the adsorption curves correspond to the pretreatments listed in table 1. Only the amount of chemisorbed oxygen after extrapolation to zero pressure is listed in the table. The results are also given for the Ag- catalyst after an identical pretreatment sequence. It was important to ensure that the sequence was measured after the catalyst had been conditioned such that no carbon dioxide desorption was observed during recording of the oxygen isotherm. In previous studies6* ' we have shown that carbonaceous impurities (which form CO,) can greatly enhance the uptake of oxygen owing to the fact that bulk penetration of oxygen is facilitated by these impurities.During the recording of the first isotherms of both catalysts (after reduction at 523 K), some carbon dioxide was observed in the gas phase (measured after the adsorption experiment). These measurements are not shown in the figure or listed in the table. The results show that the promoted catalyst initially chemisorbed more oxygen than the non-promoted catalyst. However, at the end of the sequence the difference in amount of oxygen taken up between both is much smaller. Also notable is that reduction at higher temperatures (673 K) causes increased uptake of oxygen by the catalysts (except for measurements 3 and 4 for the promoted catalyst). From the results presented in table 1, it also follows that oxidation of the catalysts at 523 K more or less restores the initial (semi-conditioned) state of the catalyst after the first reduction procedure at 523 K.Once again, the initial amount of oxygen is taken up by the catalysts. However, with the promoted catalyst, the amount of sorbed oxygen drops irreversibly following oxidation at 673 K. Comparing both catalysts, it is clear that they both react analogously with respect to the pretreatment procedure. However, the relative variation between the various t 1 Torr = 101 325/760 Pa. 10-2272 Ag/a-Al,O, Catalysts plTorr Fig. 1. Oxygen chemisorption isotherms recorded at 443 K for the Ag-Cs catalyst after various consecutive pretreatments. The numbers of the isotherms relate to the pretreatment listed in table 1.Table 1. Amount of sorbed oxygen at 443 K for the unpromoted and promoted catalyst after various pretreatments 0, uptake Ag-Cs 0, uptake Ag after extrapol. d (Ag-Cs)" after extrapol. d ( 4 0 " pretreatment /cm3 STP g;: IPm /cm3 STP g;: I P ( I ) H, 523 K (2) H, 673 K (3) H, 523 K (4) H, 673 K (5) H, 523 K (6) H, 673 K 0.141 0.99 0.1 10 1.27 0.155 0.90 0.145 0.96 0.143 0.98 0.1 10 1.27 0.143 0.98 0. I38 1.01 0.128 1.09 0.1 17 1.19 0.137 1.02 0.132 1.06 0,523 K 0,673 K a Apparent mean size of silver particles. amounts of oxygen taken up during the sequence is smaller for the Cs-promoted catalyst. In table 1 the apparent mean silver-particle size is also listed as a function of the pretreatment. It was calculated assuming a monolayer coverage of oxygen atoms at 443 K and a total number of ad-atoms of 1.31 x lo1' per m2.14 Estimation of the mean silver-particle size from electron micrographs by measuring several particles, resulted in a mean diameter of 0.9 f 0.2 pm.The diameter has a relatively large error due to the fact that many of the particles are of irregular shape, which hampered an accurate estimation of the surface area. A representative scanning electron micrograph is shown in plate 1. A reasonable agreement thus exists between the two techniques, although oxygen chemisorption measurements at 443 K tend to give rise to a somewhat larger mean particle size, especially for the non-promoted catalyst.J . Chem. SOC., Faraday Trans. I , Vol. 85, part 2 Plate 1 Plate 1. A representative scanning electron micrograph of the Ag-Cs catalyst. G.R. Meima et al. (Facing p . 272)G. R. Meima et al. 273 Table 2. Sorbed amount of oxygen at 443 K and the corresponding area of the oxygen desorption peak after a consecutive TPD experiment for the unpromoted catalyst (for explanation see the text) peak area equilibration vaads 0, TPD peak dosed amount time (cm3 STP) g;: area (arb. uni ts) vads 0 2 C 19 h 0.133 C 15 rnin 0.1 15 b 15 min 0.109 a 2 min 0.07 1 0.465 3.5 0.440 3.8 0.355 3.3 0.060 0.8 Combined Oxygen Chemisorption/TPD Experiments Preliminary measurements indicated that the total amount of oxygen taken up by the catalyst depended not only on the oxygen pressure, but also on the equilibration time. For the combined oxygen chemisorption/TPD experiments, we therefore chose a somewhat different oxygen adsorption procedure, in order to obtain useful and reproducible results.During these experiments single-pulse experiments were performed, the pulse size and the equilibration time of which were varied. Only the results for the non-promoted catalyst will be reported here. Three different pulse sizes were employed : (a) 0.120 cm3 STP (i.e. 0.071 cm3 STP g;:). This amount of oxygen was less than necessary for saturation of the catalyst. The pulse was very quickly (within 2 min) and completely taken up. The final remaining pressure above the catalyst was negligible; (b) 0.277 cm3 STP (i.e. 0.164 cm3 STP g,:). This dose of oxygen was sufficient to completely saturate the catalyst. 'I'he final pressure above the catalyst varied between about 0.5 to 1 .O Torr ; ( c ) 1.023 cm3 STP (i.e.0.607 cm3 STP g,:). This amount of oxygen was much larger than that required for saturation. A final pressure of ca. 4-5 Torr was measured, depending on the adsorption time. Three equilibration times were also chosen: (i) 2 min; (ii) 15 min; (iii) 19 h. We have investigated four pulse size-time interval combinations (see table 2). All experiments were performed after oxidation at 598 K overnight followed by a reduction at 523 K for 2 h. The combined adsorption/TPD experiments were performed in the following manner. The catalyst was dosed with a pulse of oxygen and allowed to equilibrate for the chosen time interval. The catalyst was then cooled to room temperature and evacuated for ca. 30 min (p < Torr), after which time a TPD experiment up to a temperature of 673 K was performed.The desorbing gases were continuously monitored by mass spectrometry. A typical adsorption isotherm in which the amount of chemisorbed oxygen is registered as a function of time [combination (b)-(ii)] is shown in fig. 2 and the corresponding TPD spectrum is shown in fig. 3. The results of all the combinations studied are listed in table 2. As can be seen in the last column, the oxygen desorption peak area correlates nicely with the amount of previously chemisorbed oxygen, with the exception of the small pulse (which is very rapidly adsorbed). It can thus be stated that when the catalyst is dosed with oxygen sufficient for saturation, the amount of oxygen desorbed during a TPD experiment is directly related to the previously chemisorbed amount. Extent of Hydrogen Sorption As stated earlier, hydrogen chemisorption measurements at 433 K were also performed in order to obtain information on the relative amount of adsorbed oxygen, in274 o-0611 Ag/u-Al,O, Catalysts o7 0.1 2 - - 0.10 - 7: - 0.08 - M t% 3.m 0.06- I I I I I 0 3 6 9 12 15 tlmin Fig. 2. Amount of oxygen taken up as a function of time after dosing the Ag-catalyst with 0.164 ml STP gii. For explanation see the text. 373 473 573 673 T/K Fig. 3. TPD spectrum obtained after recording the isotherm of fig. 2. (Ag catalyst; see text). 0, (32) (-); H,O (1 8) (- - - - ). comparison with the absorbed amount. These measurements were always performed immediately after oxygen chemisorption measurements.To that end, after an oxygen chemisorption measurement, the catalyst was first evacuated for ca. 39 min (p < Torr) at 443 K before the hydrogen chemisorption experiments. For a good understanding, it is important to note that during this evacuation procedure hardly any oxygen was desorbed from the catalyst. Renewed oxygen chemisorption measurements after the evacuation procedure showed that the amount taken up was negligible, only very small amounts of oxygen being adsorbed at relatively high pressures (starting at ca. 2 Torr). Even after evacuation overnight only ca. 10% of the initially chemisorbed amount was readsorbed. We therefore conclude that at the pressures and temperatures employed in this study, oxygen is irreversibly adsorbed. Vannice et a1.l'".l5 also observedG. R. Meima et al. 275 Table 3. Amount of sorbed oxygen and consecutive take up of hydrogen at 443 K for the unpromoted catalyst. The total amount of hydrogen, the rapidly adsorbed fraction and the proportions of both as compared with the total amount of sorbed oxygen (for explanation see the text) equili b- 'ads '2 H, tot H, fast dosed ration (cm3 STP) (cm3 STP) (cm3 STP) H2 tot H, fast amount time 'ads '2 'ads '2 C 19 h 0.135 0.21 1 0.157 1.56 1.16 C 15 min 0.1 18 0.193 0.149 1.64 1.26 b 15 min 0.103 0.168 0.138 1.63 1.33 a 2 min 0.07 1 0.1 15 0.107 1.62 I S O only very small amounts of reversibly held oxygen at these temperatures, especially on a conditioned catalyst. Titration measurements of the chemisorbed oxygen by hydrogen are therefore not obscured by this effect.The hydrogen sorption could be considered as two types viz. an initial, rapid adsorption, and a slower uptake. Distinction between the two was made in the following manner. The catalyst was dosed with small pulses of hydrogen until a (remaining) pressure was observed above the catalyst (after an adsorption time of at least 10 min). A larger pulse was then given which resulted in a pressure of ca. 4-5 Torr above the catalyst. The amount of hydrogen taken up after ca. 10 min was designated 'fast'. (The adsorption process at that time was increasingly slow). The total amount of hydrogen taken up was designated as the amount measured after increasing the adsorption time to ca. 19 h and the pressure above the catalyst to ca. 8 Torr. The results of the hydrogen titration measurements at 443 K are listed in table 3. Again, these results apply to the non-promoted catalyst.The oxygen chemisorption procedures were the same as those given in table 2. As can be seen, the oxygen chemisorption measurements reproduce very well. Besides the extent of rapid hydrogen and total hydrogen uptake, the proportions of both as compared with the amount of sorbed oxygen are also listed. As can be seen, for all measurements the total amount of hydrogen taken up is related by a factor of ca. 1.6 to the amount of previously sorbed oxygen. On the other hand, the ratio of the amount of rapidly adsorbed hydrogen shows an increase when short equilibration times and low pressures are employed. The proportions are consistently lower than two, even after an adsorption time of ca.19 h. This implies that not all the chemisorbed oxygen is reacting to form water, and, therefore, that not all the oxygen is present on the surface.10~'5 The amount of oxygen adsorbed on the surface can be calculated if it is assumed that all the rapidly adsorbed hydrogen reacts only with this species to form water. It then follows that the amount of surface oxygen is half the amount of hydrogen rapidly taken up. If the above assumption is correct, it then also follows that the amount of hydrogen which is only very slowly taken up by the catalyst [H,(tot)-H,(fast)], has a ca. 1 : 1 correlation with the remaining amount of (penetrated) oxygen. This correlation indicates that the penetrated oxygen species, most presumably, reacts (slowly) with hydrogen to form a hydroxyl (like) group.In table 4 the results of the calculations are shown. As can be seen, a good correlation exists between the measured amount of oxygen and the amount calculated from the hydrogen titration measurements, especially after adsorption at relatively high oxygen pressures and long equilibration times. Lower pressures and shorter equilibration times presumably lead to a relatively larger amount of surface oxygen as the ratio276 Ag/a-Al,O, Catalysts Table 4. Calculated amounts of oxygen adsorbed on the surface and oxygen dissolved in the bulk from the rapidly adsorbed and slow uptake amounts of hydrogen at 443 K surface/O, bulk/O, calcd uptake : measured uptake : cm3 STP cm3 STP of o2/cm3 STP of o,/Cm3 STP surface hi: hi: gi; g,: 0, (%) 0.079 0.054 0.133 0.135 (c-3) 58.5 0.075 0.044 0.1 19 0.118 (c-2) 63.6 0.069 0.030 0.099 0.103 (b-2) 67.0 0.054 0.008 0.062 0.071 (a- 1) 76.I H,(fast): O,(tot) is higher. However, as the H,(tot) : O,(tot) ratio is constant, it must be that an equilibrium is eventually reached, giving rise to ca. 60% surface bound oxygen in all cases. The combined oxygen adsorption and TPD measurements listed in table 2 suppcrt the above proposition, as a good correlation exists between the amount of sorbed oxygen and the oxygen peak area. An understandable exception is formed when the catalyst is dosed with an amount of oxygen insufficient to saturate the catalyst [combination (a)-(i)]. In this case, the peak area is relatively small owing to the fact that penetration into the bulk takes place during the TPD measurement.A final important result must be mentioned here. After pretreatment in oxygen at 673 K, the total amount of chemisorbed hydrogen was 0.203 ml STP g,:. This amount is comparable to that taken up after chemisorption of oxygen at 443 K if relatively long equilibration times were employed (see table 3). This indicates that even severe oxidation does not give rise to greater bulk penetration of oxygen as compared with an adsorption experiment. Further Discussion and Conclusions As compared with previous the variation in the amount of oxygen and hydrogen taken up as a function of the pretreatment is small. We believe that this is caused mainly by variations in the mean silver-particle size.In this investigation very large silver particles were studied (ca. 1 pm), whereas in the previous studies particle sizes of 70 and I50 nm were studied. The results presented in this paper indicate that excessive penetration of oxygen into these large silver particles does not take place. With the smaller particles, on the other hand, the hydrogen chemisorption measurements demonstrate that vast amounts of oxygen penetrate into the silver lattice, especially after oxidation at elevated temperatures.6 With the larger silver particles, a maximum of (only) ca. 40 % of the monolayer could penetrate into the bulk (or more presumably into the near surface layers). It was further shown, that if the catalysts were dosed with a further amount of oxygen which was sufficient for saturation, the relative surface population (60 %) was independent of the adsorption pressure and equilibration time.It thus seems that the sites at which the penetration occurs are in some way related to the adsorption sites, as the total uptake did vary with the above parameters (time, pressure). A more or less obvious explanation is that the sites on which the adsorption of oxygen takes place, also act as the sites through which the penetration takes place. The relation between the sites will be further discussed in part 2 of this study." The observation that only relatively small amounts of oxygen can penetrate into these large silver particles is in accordance with findings of others on (bulk) silver specimens. The solubility and the diffusion coefficient of oxygen in pure silver is very I O W , ~ " ~ ~G .R . Meima et al. 277 especially under the experimental conditions employed in the work presented here (relatively low temperatures and pressures). The excessive penetration of oxygen thus seems to be restricted to silver particles smaller than ca. 1 ,urn. (i) a somewhat larger uptake of oxygen at 443 K was generally observed after reduction at more elevated temperatures. This is presumably caused by the more complete removal of the penetrated oxygen at these temperatures; (ii) The Cs-promoter renders the catalyst less sensitive towards the various pretreatments with respect to the uptake of oxygen at 443 K. It also enhances the amount of oxygen that can initially be taken up at 443 K. As will be argued in part 2 of this study," we believe that the higher uptake is caused by the formation of stable defects, due to the presence of caesium on the silver surface; (iii) The role of the Cs-promoter became less apparent during the experimental sequence, especially after pretreatment at elevated temperatures (see table 1).This is presumably due to the slow desorption of caesium at these temperatures. As a final conclusion, these results indicate that the influence of the pretreatment on the surface structure is restricted to the near surface layers. Large changes in particle morphology do not seem to take place. Other important features are References 1 (a) M. A. Barteau and R. J. Madix, in The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis, ed.D. P. Woodruff and D. A. King (Elsevier, Amsterdam, 1982), vol. 4, chap. 4; (b) R. W. Clayton and S. V. Norval, Catal., 1980, 3, 70; (c) X. E. Verykios, F. P. Stein and R. W. Coughlin, Catal. Rev. Sci. Eng., 1980, 22, 197; (4 P. A. Kilty and W. M. H. Sachtler, Catal. Reti. Sci. Eng., 1974, 10, 1, and references herein. 2 J. C. Wu and P. Harriott, J. Catal., 1975, 39, 395. 3 X. E. Verykios, F. P. Stein and R. W. Coughlin, J. Catal., 1980, 66, 368. 4 J. W. Woodward, R. G. Lindgren and W. H. Corcoran, J. Catal., 1972, 25, 292. 5 C. T. Campbell, J. Catal., 1985, 94, 436. 6 G. R. Meima, L. M. Knijff, A. J. van Dillen, J. W. Geus, J. E. Bongaarts, F. R. van Buren and K. Delcour, in New Developments in Selectice Oxidation, European Workshop Meeting, 17-1 8 March 1986, Louvain-la-Neuve, ed. B. Delmon and P. Ruiz; Card. Today, 1987, 1, 117. 7 G. R. Meima, R. J. Vis, M. G. J. van Leur, A. J. van Dillen, J. W. Geus and F. R. van Buren, J , Chem. Soc., Faraday Trans. I , 1989, 85, 279. 8 G. R. Meima, L. M. Knijff, A. J. van Dillen, J. W. Geus and F. R. van Buren, J. Chem. Soc., Faraday Trans. I , 1989, 85, 293. 9 C. Backx, J. Moolhuysen, P. Geenen and R. A. van Santen, J. Catal., 1981, 72, 364. 10 (a) S. R. Seyedmonir, D. E. Strohmayer, G. L. Geoffroy, M. A. Vannice, H. W. Young and J. W. Linowski, J. Catal., 1984, 87, 424; (b) S. R. Seyedmonir, D. E. Strohmayer, G. L. Guskey, G. L. Geoffroy and M. A. Vannice, J. Catal., 1985, 93, 288. 11 G. R. Meima, M. Hasselaar, A. J. van Dillen, F. R. van Buren and J. W. Geus. J . Chem. SOC., Faraday Trans. I , 1989, 85, in press. 12 U.S. Patent, no. 4 530 616, September, 1982. 13 J. Joziasse, Ph.D. Thesis (Utrecht, 1978), chap. 5. 14 J. J. F. Scholten, J. A. Konvalinka and F. W. Beekman, J. Catal., 1973, 28, 209. 15 D. E. Stohmayer, G. L. Geoffroy and M. A. Vannice, Appl. Catal., 1983, 7, 189. 16 W. Eichenauer and G. Muller, Z . Metallkd. 1962, 53, 321. 17 H. H. Podgurski and F. N. Davis, Trans. Metal. SOC. AIME, 1964, 731. 18 V. M. Gryaznov, S. G. Gul'yanova and S. Kanizius, Russ. J. Phys. Chem., 1973, 47, Paper 8 1006 1 1 C ; Receioed 1 8 th 517. February, 1988
ISSN:0300-9599
DOI:10.1039/F19898500269
出版商:RSC
年代:1989
数据来源: RSC
|
13. |
The interaction of oxygen with isolated silver particles ofca.70 nm supported onα-alumina. Part 1.—Oxygen sorption and temperature-programmed desorption measurements |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 279-291
Garmt R. Meima,
Preview
|
PDF (1299KB)
|
|
摘要:
J . Chem. SOC., Faraday Trans. I , 1989, 85(2), 279-291 The Interaction of Oxygen with Isolated Silver Particles of ca. 70 nm supported on a-Alumina Part 1 .--Oxygen Sorption and Temperature-programmed Desorption Measurements Garmt R. Meima,"? Ralph J. Vis, Michel G. J. van Leur, Adrianus J. van Dillen and John W. Geus Department of Inorganic Chemistry, State University of Utrecht, Croesestraat 77a, 3522 AD Utrecht, The Netherlands Frederik R. van Buren Dow Chemical (Nederland) B. V . , P.O. Box 48, 4530 A A Terneuzen, The Netherlands The interaction of oxygen with a silver catalyst containing isolated silver particles of ca. 70 nm supported on a-Al,O, has been studied using (combined) volumetric adsorption and temperature-programmed desorp- tion measurements after various pretreatments.The amount of oxygen taken up is greatly influenced by the pretreatment. Considerable amounts of oxygen are able to penetrate into the silver lattice. Impurities present in or on the silver particles strongly affect the amount of oxygen taken up at 443 K. It is argued that penetrated oxygen can facilitate the reorganization of the silver surface giving rise to low-index facets. The catalytic importance of silver, mainy in selective oxidation reactions, has initiated the study of the (chemi)sorption of oxygen by several techniques. However, many of the published results remain contradictory, especially those concerning the various states of oxygen and their relative populations under different conditions.' An important reason, which may (partly) explain many of the contradictory results in the literature, is the variety of (pre)treatment conditions used by the different investigators.Therefore, we have recently studied the influence of the pretreatment on the catalytic properties of a silver catalyst.2 Both the influence of the pretreatment on the uptake of oxygen and on the catalytic activity for the oxidation of carbon monoxide were systematically investigated . It was observed that the activity of the catalyst varied greatly with the pretreatment. The catalyst strongly deactivated after pretreatment in oxygen at elevated temperatures, whereas treatment in hydrogen (re)activated the catalyst. Interestingly, the changes in activity were largely reversible. The reversibility of the changes in activity, and the fact that sintering of the silver particles did not take place to a significant extent, implicates that the surface structure of the silver particles is determining the catalytic activity of silver catalysts.Hence, the precise structure of the silver surfaces exposed after different pretreatment produces is very important. It is therefore interesting to consider whether catalytic reactions or chemisorption on silver are structure-sensitive (or demanding). To establish unambiguously an effect of the surface structure, measurements on monocrystalline surfaces have to be performed. Results from the literature on single-crystal surfaces have shown that ordered chemisorption of oxygen is only observed on the (1 10) Albers et al.5*6 have shown that oxygen does not adsorb at all on a defect-free Ag( 1 1 1) surface.These authors t Present address: Dow Chemical (Nederland) B.V., P.O. Box 48,4530 AA Terneuzen, The Netherlands. 279280 Interaction of Oxygen with Silver Particles on Alumina could only adsorb oxygen after deliberately damaging a (111) surface by ion bombardment, thus creating surface defects. The (1 1 1) plane is the thermodynamically most stable; thus it can be expected to be the most abundant plane in the surface of silver particles. Therefore, the number of defects in this surface can greatly influence both the sorption of oxygen and the catalytic activity. Our previous results2 could thus be explained satisfactorily by ascribing the influence of the pretreatment to : (i) a change in morphology leading to a larger or smaller fraction of (1 11) planes at the surface and (ii) a varying number of defects in Ag( 11 1) surfaces.The oxygen chemisorption measurements performed earlier support the above explanation. After removal of the carbonaceous impurities initially present in the catalyst, an amount of oxygen considerably less than a monolayer of oxygen atoms was adsorbed at 443 K. This is indeed to be expected in view of the finding of Albers et ~ 1 . ~ 9 ~ that an undisturbed (1 11) plane of silver does not adsorb oxygen at all. The aforementioned study2 dealt with alumina-supported silver particles of 150-200 nm diameter. Since many silver particles were making contact (owing to the heavy loading) but had not completely coalesced, the number of grain boundaries in the catalyst was considerable. Since grain boundaries may strongly affect the (ad)sorption of oxygen, the aim of this work is to study a similar catalyst which contains only isolated silver particles supported on a-alumina, thus diminishing the possible formation of grain boundaries.Another important reason for initiating this study is to obtain further insight into the state and the amount of the different oxygen species present after various pretreatments. It is especially important to discriminate between the oxygen present on the surface of the silver particles and the oxygen which has penetrated into the lattice, as it has recently been demonstrated that the presence of subsurface oxygen is crucial for a high selectivity in the ethylene epoxidation reaction.',* In this paper the results will be discussed of the influence of the various pretreatments on the amount of oxygen taken up at 443 K, which is often used to calculate the metallic surface area of silver catalysts.Combined oxygen chemisorption and TPD (temperature- programmed desorption) measurements with continuous monitoring of the composition of the gas phase by means of a mass spectrometer were employed to obtain information on the relative amounts of subsurface oxygen present. First, the effect of the number of reduction cycles at 473 K on the uptake of oxygen will be discussed. Subsequently, the effect of the reduction and evacuation temperature on the extent of oxygen sorption will be dealt with. Finally, the uptake of oxygen combined with subsequent thermal desorption up to 673 K will be reported.Scanning electron microscopy (SEM) was used to study the mean particle size and the morphology of the silver particles. The effect of the pretreatment on the activity for the CO oxidation reaction with molecular oxygen will be discussed in a later paper.g Experimental Catalyst The silver catalyst was prepared by precipitation of silver from an aqueous solution onto a suspended a-alumina support (Fluka, purum). Precipitation was effected by reduction of the Ag(NH,)i complex with formalin at room temperature, following a method described elsewhere in more detail.lO*" The catalyst contained 1.5 wt O h Ag and was dried in air at 393 K for 18 h. The B.E.T. surface area of the a-alumina carrier was measured with a Quantasorb instrument (Quantachrome Corp.) and amounted to 0.80 0.02 m2 g-'.G.R. Meima et al. 28 1 L .O I 4 M n 2 3.0 ui W 1 1 1 1 1 1 2 3 L 5 6 PlTOrr Fig. 1. Oxygen chemisorption isotherms recorded at 443 K. Fresh sample reduced at 473 K (4 times): x , 1st; 0, 2nd; 0, 3rd; A, 4th. Adsorption and TPD Studies The adsorption studies and the thermal desorption experiments were conducted in a classical Pyrex high-vacuum and gas-supply system described elsewhere. l2 Linear heating of the sample and temperature control were performed with a Leeds Northrup (type LN 1 300) temperature programmer and Euro therm controller. Temperature readings were made by means of a Cr-A1 thermocouple placed in a narrow inlet tube which ended in the middle of the catalyst bed.During the TPD experiments a heating rate of ca. 5 K min-l was combined with continuous pumping and analysis of the gas phase by means of a quadrupole mass spectrometer (Leybold Heraeus QM 200). During the adsorption measurements the pressures were monitored and recorded by a pressure transducer (Setra Systems type 236). All adsorption isotherms were measured at 443 K, usually over the range &5 Torr (1 Torr z 133.3 Pa). For calculation of the (apparent) mean particle size the isotherms were extrapolated to zero pressure, as is usually done. With the combined oxygen-chemisorption-TPD measurements, however, the total amount of oxygen taken up at 443 K was used in the calculations. A sample of 11.53 g of the catalyst was used throughout the experiments. The pretreatment of the catalyst (reduction or oxidation) was, unless stated otherwise, performed overnight at various temperatures, either in a 10 O h H,/N, flow or in a 100 YO oxygen flow followed by evacuation at the desired temperature (usually the same as the pretreatment temperature) for at least 1-2 h. The gases were purified and dried using standard procedures.Oxygen (99.998 O h ) for the adsorption experiments was obtained from AGA Gas and was used without further purification. The catalyst was also examined within a Philips 505 scanning electron microscope, which was operated at 30 kV accelerating voltage. Results Extent of Oxygen Uptake at 443 K by the Fresh Catalyst Fig. 1 shows four consecutively measured oxygen adsorption isotherms, which were recorded at an adsorption temperature of 443 K.Prior to the measurements, the catalyst282 Interaction of Oxygen with Silver Particles on Alumina was always reduced overnight at 473 K. An adsorption temperature of 443 K was chosen because at this temperature silver takes up a maximum amount of oxygen.13.14 It is even often stated that at this temperature a monolayer coverage of oxygen atoms is obtained. As can be seen, the uptake of oxygen steadily decreases with each reduction/ adsorption cycle. A reproducible amount of sorbed oxygen was only observed after four consecutive reduction/adsorption cycles. During recording of the first isotherm, another remarkable feature was noted : after a pulse of oxygen had been admitted to the sample, the initial decrease in pressure due to the adsorption of oxygen was followed by a pressure increase.Mass-spectrometric analysis of the gas phase showed that the rise in pressure was caused by the evolution of carbon dioxide, presumably originating from some carbonaceous impurities left in or on the catalyst after the preparation procedure. As can be seen in fig. 1, the desorbing carbon dioxide clearly affected the shape of the first adsorption isotherm. The pressure of the non-adsorbing carbon dioxide causes the adsorption isotherm to rise seemingly less steeply. The drop in the amount of oxygen taken up can certainly not be attributed to the sintering of the silver particles at these relatively low temperatures (473 K), as electron microscopic investigations after the experiments showed no change in the mean silver particle size.Influence of the Reduction Temperature on the Uptake of Oxygen at 443 K Fig. 2 shows the influence of the reduction temperature on the amount of oxygen taken up at 443 K after the conditioning of the catalyst described above. Prior to the recording of the isotherms, the catalyst was always evacuated at the same temperature as the reduction temperature for at least 1 h (p < lop5 Torr). The reduction (time 2 h) was always preceded by an oxidation at 598 K. Reproducible results were obtained only when using this procedure. In particular the previous oxidation treatment was necessary. This is illustrated by the different amounts of oxygen taken up by the catalyst after reduction at 473 K shown in fig. 1. Without previous oxidation, ca.2.0 cm3 (s.t.p.) g;: is eventually taken up by the catalyst (fig. 1). However, after oxidation of the catalyst at 598 K and subsequent reduction at 473 K, an (enhanced) amount of ca. 2.75 cm3 (s.t.p.) g,: is taken up by the catalyst (fig. 2). The results listed in table 1 were all obtained after previous oxidation of the catalyst at 598 K. The amount of oxygen taken up is given per gram of silver. All measurements were reproduced several times and showed some (random) spreading. Thus the mean value is listed, both after extrapolation of the isotherms to zero pressure and at an arbitrary pressure of 5 Torr. The pressure was very near the (final) pressure above the catalyst during all measurements. The extreme values (highest and lowest) of a particular measurement are given in parentheses.Assuming a monolayer coverage of oxygen atoms at 443 K and a total number of 1.3 1 x 1019 atom per m2,13 an (apparent) mean particle size of spherical or cubical silver particles can be calculated. The calculated mean particle sizes (spheres) are also given in table 1. It must be stressed that the results listed in the table 1 were completely reproducible and reversible with respect to the reduction temperature. If, for example, after a reduction at 673 K a relatively small amount of sorbed oxygen had been measured, subsequent reduction at 473 K led to the larger amount, provided the catalyst had been oxidized at 598 K prior to the reduction. Further experiments showed that it was not the reduction temperature that determined the amount of oxygen taken up, but the evacuation temperature after reduction (which was the same as the reduction temperature throughout the measurements in table 1).If,J. Chem. SOC., Faraday Trans. I , Vol. 85, part 2 Plate 1. A representative SEM micrograph of the catalyst. G. R. Meima et al. Plate 1 (Facing p . 283)G. R. Meima et al. 283 0.0 173 573 673 Trcduction /K Fig. 2. Amount of oxygen sorbed at 443 K after reduction at three temperatures (pretreatment: oxidation at 598 K). Ranges measured and mean values are indicated (see text). Table 1. Amount of oxygen sorbed per gram of Ag after reduction at three temperaturesa 0, uptake/cm3 (s.t.p.) gii reduction temperature/K extrapolation at 5 Torr d/nm 473 2.72 (2.66-2.80) 3.27 (3.15-3.41) 51 523 2.1 1 (1.88-2.25) 2.60 (2.37-2.78) 66 673 1.39 (1.33-1.56) 1.59 (1.56-1.91) 100 a The amounts in parentheses relate to the extremes of the measurements.Apparent mean particle size as calculated from the isotherms after extrapolation to zero pressure. for instance, the catalyst was reduced at 423 K and evacuated at 673 K, the amount taken up was the same as that after reduction and evacuation at 673 K. A further investigation of the effect of the reduction time showed that prolonged reduction periods had no measurable influence. A representative SEM micrograph of the catalyst is shown in plate 1 and the particle size distribution as calculated from the SEM micrographs is shown in fig. 3. As can be seen, the majority of the particles have a mean particle size of ca.70 nm. This is in reasonable agreement with the mean particle size as calculated from the amount of oxygen taken up after reduction and evacuation at 523 K. However, evacuation at lower temperatures than 523 K clearly leads to an underestimate of the particle size, whereas evacuation at more elevated temperatures leads to an overestimate. If sintering and redispersion of the silver is ruled out, this implies that the sorption of oxygen at 443 K is more complex than the simple assumption that it leads to a monolayer coverage of oxygen atoms over the complete silver surface. Presumably, simultaneous absorption of oxygen is also taking place after evacuation (and reduction) at temperatures lower than 523 K. We will further comment on sorption of oxygen in the Discussion section.284 Interaction of Oxygen with Silver Particles on Alumina 0 100 200 300 LOO crystallite diameter/nm Fig.3. Particle size distribution as estimated from SEM micrographs (ntota, = 239). Combined Chemisorption and TPD Measurements In a previous study2 we have shown that combined adsorption and TPD measurements offer a promising method to distinguish between the amount of oxygen adsorbed and the amount of oxygen absorbed. The combination of both techniques is based on the fact that the penetrated or subsurface oxygen is not significantly desorbed during a TPD experiment (up to temperatures as high as 673 K). This opens the possibility of desorbing the surface oxygen without desorbing the penetrated oxygen. The amount of desorbed (surface) oxygen can be determined subsequently by renewed adsorption of oxygen at 443 K.The experimental procedure used was as follows. First the catalyst was oxidized at 598 K (as argued earlier, this was necessary to obtain reproducible results), after which it was reduced and evacuated at the desired temperature. Subsequently, an oxygen adsorption isotherm was recorded. Immediately thereafter, the catalyst was cooled to room temperature, evacuated for ca. 1 h (p < lop5 Torr) and a TPD experiment was run. After the TPD measurement, once again an oxygen adsorption isotherm was recorded, followed by a renewed TPD experiment. (Repetition of the sequence can give additional information with respect to the penetration of oxygen into the bulk during a TPD experiment.) During the TPD measurements the gas phase was continuously analysed by mass spectrometry.The results of a typical experiment are shown in fig. 4 and 5. These figures clearly indicate that a large amount of oxygen was retained by the catalyst after the first TPD experiment. Surprisingly, the second oxygen desorption peak was even larger, despite the much smaller amount of oxygen that had previously been taken up. It can thus be stated that (at this stage) there is no direct relation between the amount of oxygen previously taken up and the area of the oxygen peak of the subsequent TPD measurement. It can also be seen in the TPD spectra that relatively large amounts of water were also desorbed, especially during the first TPD run. The desorption of water was undoubtedly associated with a previous reduction procedure ; if TPD spectra were recorded without a previous reduction procedure, the amount of water diminished after each experiment.This effect is clearly demonstrated in fig. 5 ; the amount of desorbed water was much smaller during the second TPD experiment. Water being desorbed at much more elevated temperatures than the reduction temperature is indicative of the fact that the high-temperature peak is not caused by the desorption of physically adsorbed water.G. R. Meima et al. 285 1 2 3 L 5 6 pnorr Fig. 4. Oxygen chemisorption isotherms recorded at 443 K, after reduction at 523 K ( X) and after a consecutive TPD experiments up to 673 K (0). 9 6 n 4 a d k3 v 4 3 0 .-\ / \ / \ / / \ / \ \ I I I I I / / / 1,' I I / I I I I / I I I I I / / / I 2; 273 373 4 3 573 673 TIK Fig.5. A TPD experiment after reduction at 523 K (1) followd by renewed adsorption of oxygen at 443 K and a consecutive TPD experiment (2): (-) 0, (32); (---) H,O (18).286 3.0 2.0 - m I < M n Y 4 s" 1.0 ui W % \ 0.0 Interaction of Oxygen with Silver Particles on Alumina 1.5 I I 1 1 1 0.0 1 2 3 4 5 n 1 I 1 1 2 3 L 5 n Fig. 6. (a) Amount of oxygen sorbed as a function of the adsorption-TPD sequence cycle. (b) Oxygen desorption peak area as a function of the adsorption sequence cycle. Table 2. Amount of oxygen sorbed at 443 K and the area of the oxygen desorption peak after various pretreatmentsa amount of 0, 0, uptake TPD peak peak area in peakb/cm3 (s.t.p.) g; O/O 0, uptake pretreatment /cm3 (s.t.p.) g;: (arb.units) ~~ ~ red. 523 K/ TPD/443 K TPD/443 K TPD/443 K TPD/443 K red. 523 K/ TPD/443 K red. 673 K/ TPD/443 K evac. 523 K evac. 673 K evac. 673 K 2.37 0.96 0.67 0.57 0.48 1.89 1.06 1.88 1.01 0.087 0.098 0.120 0.109 0.1 12 0.194 0.250 0.175 0.256 0.037 0.102 0.179 0.191 0.233 0.103 0.236 0.093 0.253 0.36 0.41 0.50 0.45 0.46 0.8 1 1.04 0.73 1.06 15.1 42.2 74. I 78.5 95.8 42.5 97.8 38.3 104.6 a Oxygen chemisorption-TPD sequence cycle repeated several times. desorbed and the percentage as compared to amount taken up. Amount of oxygen To obtain further information on the amount of surface oxygen relative to the penetrated amount, the adsorption-TPD sequence was repeated several times (after reduction at 423 K). The results are represented in fig. 6. On the left-hand side of the figure the amount of oxygen taken up is plotted as a function of the number of the adsorption-TPD sequence.On the right-hand side of the figure the corresponding area of the oxygen desorption peak is plotted as a function of the cycle number: the amount of oxygen that can be taken up by the catalyst diminishes after each adsorption-TPD measurement. After a steep drop the amount sorbed slowly levels off after a sequence ofG. R. Meima et al. 287 ca. five combined adsorption-TPD cycles. Despite the continuous drop in the total amount of oxygen taken up (at ca. 5 Torr), the oxygen desorption peak area is not, or hardly, affected. Indeed small, but significant, rise is observed during the first measurements, before a more or less constant peak area is obtained.This effect is even more pronounced if the same experiment is performed after reduction and evacuation of the catalyst at 673 K. The peak area is ca. 30% larger after the second sequence, whilst the take-up of oxygen is ca. 45% smaller. The results are summarized in table 2. The amount of oxygen listed in the table corresponds to the total amount of oxygen taken up by the catalyst at 443 K per gram of silver. Note that when two consecutive TPD measurements were performed without an intermediate oxygen sorption experiment, no oxygen desorption was registered during the TPD measurement. A quantitative determination of the amount of oxygen being desorbed during the TPD experiment is possible if it is assumed that at the end of the sequence of the experiment shown in fig.5 the amount of oxygen adsorbed per cycle is completely desorbed. As can be seen in table 2, the proportion of the peak area and the amount of oxygen taken up by the catalyst reaches a value of ca. 0.24 at the end of the experiments. The amount of oxygen desorbed during the TPD measurements was calculated using this proportion. The last column in the table thus refers to the percentage of oxygen that was desorbed, as compared to the total amount of oxygen that was previously taken up by the catalyst. The results show that although the amount of oxygen sorbed after an evacuation (and reduction) at 523 K is larger than after the same pretreatment at 673 K, the amount that is subsequently desorbed is much smaller. The amount of oxygen retained by the catalyst throughout the sequence is also larger after a reduction at 523 K, viz.2.87 and 1.10 cm3 (s.t.p.) g& respectively. Discussion The results clearly demonstrate that the fresh catalyst must first be conditioned before reproducible oxygen sorption results can be obtained. Carbonaceous impurities enhance the amount of oxygen sorbed at 443 K owing to bulk penetration of the oxygen atoms. Other workers have also reported an enhanced uptake of oxygen13* 15, l6 owing to the presence of carbonaceous impurities. Scholten et al.13 inferred bulk penetration from a continuous (almost linear) increase of the amount of oxygen taken up with time. A similar increase was also observed in our study, thus indicating that the same process was taking place (see fig.1). In line with this assumption is the observation that removal of the carbonaceous impurities altered the shape of the adsorption curve. The presence of carbon dioxide in the gas phase produced by interaction with the admitted oxygen affects the shape of the isotherm. It seems difficult to obtain catalysts substantially free of carbonaceous impurities when using organic reducing agents. However, de Jong," using a support with a much larger surface area, was able to produce carbon-free silver particles when using a similar preparation technique as in this study by applying thorough washing procedures. We believe that the mean particle size, which was smaller in de Jong's study (10-20 nm), is of importance. The chance that inclusion or occlusion of carbon takes place in particles of this size is much smaller. During our study it proved impossible to prepare a catalyst containing large silver particles free of these impurities, in spite of thorough washing procedure. Therefore, it is important to take care to remove possible carbonaceous impurities when the free silver surface area is to be determined from the extent of oxygen taken up at 443 K.Carbonaceous impurities undoubtedly lead to an overestimate of the silver surface area. More surprising was the effect that the reduction temperature (or more precisely the evacuation temperature after a reduction) had on the amount of oxygen sorbed at200 Interaction of Oxygen with Silver Particles on Alumina 443 K (see fig. 2). The higher the reduction temperature, the smaller the amount of oxygen taken up by the catalyst.This effect was certainly not caused by trivial effects such as sintering, as the results were completely reversible (provided that the catalyst was first reoxidized at 598 K). From the mean particle size as deduced from the electron micrographs (viz. 70 nm), a monolayer coverage of oxygen atoms would give rise to an amount of oxygen taken up by the catalyst of 2.0 cm3 (s.t.p.) g& As can be seen in table 1, reduction of the catalyst at elevated temperatures, viz. 673 K, leads to a smaller amount of oxygen sorbed than corresponding to a monolayer coverage. Reduction at low temperatures (473 K) gives rise to a larger amount. De JonglO has also reported that the extent of oxygen sorption is enhanced after reduction of silver at low temperatures.He argued that this is caused by the incomplete removal of subsurface oxygen at these temperatures. The remaining oxygen stabilizes the surface in such a state that penetration of oxygen is facilitated during a subsequent chemisorption measurement. The low extent of oxygen sorption after reduction and evacuation at 673 K must be ascribed to the presence of large (111) facets with defect-free surfaces after this treatment . Our combined adsorption-TPD results support this explanation. Despite the amount of oxygen taken up after a reduction at 523 K being larger than that taken up after reduction at 673 K (1.25: l), the corresponding TPD peak area is twice as small (0.50 : 1). As argued earlier, only surface oxygen can be desorbed during a TPD measurement up to 673 K.This implies that after reduction at 523 K more bulk penetration takes place than after reduction at 673 K. According to table 2, the amount that can be desorbed after a reduction procedure at 523 K is only ca. 15 % of the total amount of oxygen previously taken up by the catalyst. It must be emphasized that this does not necessarily imply that only 15 % of the oxygen is actually adsorbed on the surface after the first adsorption experiment. It is plausible that the amount of oxygen adsorbed is initially much larger, but that during the subsequent TPD measurement the oxygen is not only desorbed, but also penetrates into the bulk. This hypothesis is supported by the results listed in table 2. As shown, the TPD peak area becomes larger during the adsorption-TPD sequence.This effect is especially marked after reduction at 673 K. That TPD peak (initially) becomes larger during the sequence implies that a correlation between the amount of oxygen taken up and the TPD peak area is found only after saturation of the bulk with oxygen. As stated earlier, the saturation amount is larger after reduction at 523 K. Clearly, the surface is more ‘transparent’ for oxygen atoms after reduction at 523 K. After a reduction at 523 K a large amount of water was desorbed during the first TPD experiment, the desorption of which was still continuing at relatively high temperatures (see fig. 5). The latter observation indicates that much of the bulk oxygen has been transformed into OH groups during the subsequent reduction procedure.2 We therefore believe that oxygen that has actually penetrated and has subsequently reacted to hydroxyl groups is causing the enhanced transparency of the silver surface.A remaining and intriguing question is the following. Why is the amount of oxygen being desorbed at the end of the sequence after reduction at 523 K smaller than after reduction at 673 K, in spite of the results of TPD measurements up to 673 K? We believe that this effect can only be explained if it is assumed that morphological changes take place during the various pretreatments. From the literature there is conclusive evidence that annealing in oxygen leads to a different morphology than does thermal treatment in hydrogen. Metal particles, including silver particles annealed in an inert or hydrogen-containing atmosphere, have been observed to exhibit a morphology having rounded edges and corners.Sundquistl’ carefully investigated a number of different metals. Annealing silver particles of ca. 1 pm, he obtained a shape in which in addition to flat planes a significant fraction ofG . R. Meima et al. 289 atomically rough surfaces was exposed at the rounded edges and corners. Consequently, a supported silver catalyst being kept in hydrogen at elevated temperatures will contain many surfaces capable of adsorption of oxygen. On all surfaces except the undisturbed (1 1 1) planes, oxygen will be adsorbed dissociatively. Oxygen thus adsorbed on the particles of a catalyst that has been thermally treated in hydrogen will lead to a large oxygen peak during a TPD experiment.Many reports in the literature deal with the faceting of silver surfaces by a thermal treatment in oxygen or air. Shuttleworth et aZ.18 showed that faceting of polycrystalline silver only occurred in the presence of oxygen. Later, Moorel' investigating a large number of thermally faceted silver crystals, concluded that heating in air or oxygen leads to formation of (1 1 1) and (100) facets. The cause of the faceting, which is also displayed by other metals after annealing in oxygen, is still debated. Hondros and Moore2' claimed that the faceting of silver is due to its net evaporation. The latter proceeds from ledges present on low-index planes. With little or no oxygen present, the ledges maintain a steady-state spacing during evaporation. However, preferential adsorption of oxygen unanchors some of the ledges, leading to bunched ledges.As a result, very small areas of bunched ledges are separated by wide areas of flat planes. Hondros and Moore2' found that suppressing the net evaporation of silver prevented faceting. Rhead and Mykura,21 on the other hand, could not completely reproduce the results of Hondros and Moore. These authors assume that the faceting is due to an anisotropic lowering of the surface energy by interaction with oxygen. More recently, Czanderna22 concluded that (1 11) microfacets are formed on the surface of his silver powders during a cycle comprising outgassing, adsorption, outgassing and reduction. is of interest : in their study on the effect of residual gas on the optical properties of silver films, they observed that films deposited at relatively high pressures of oxygen (f lop5 Torr) sometimes showed a more rapid annealing stage. They argued that annealing proceeds more rapidly if a mobile dissolved oxygen species is present in the silver lattice.According to these authors, the mobile oxygen enhances the self-diffusion of silver, causing significant smoothing and grain growth. The effect on the surface structure of exposure to hydrogen at elevated temperatures has not been studied as extensively as that of oxygen. We have already dealt with Sundquist's results. In line with our work, Sandler et have also clearly observed that hydrogen is retained in the silver lattice if penetrated oxygen is present. They speculated that topographic changes may take place owing to the presence of hydrogen.These authors, however, did investigate the matter further. Hondros2' observed that a previously annealed smooth silver surface became uneven and showed blistering after subjecting it to hydrogen at elevated temperatures. Prolonged heating caused complete piercing of the specimen at the grain boundaries. In massive specimens the grain- boundary penetration was sometimes sufficient to allow crystals to be removed bodily from the specimen. Auroux and Gravelle'' observed high heats of interaction of oxygen after reduction of silver catalysts. They postulated that the surface of a freshly reduced silver particle contains a large number of lattice defects. Finally, it has also been observed that water vapour can increase the surface roughness of a silver These observations are important in view of the fact that the single-crystal work has shown that a silver (1 11) plane does not adsorb oxygen at all if it is defect-free.Based on the above observations, we suggest that the previous oxidation of silver particles leads to faceted particles. Evacuation at 523 K is not sufficient to remove the hydroxyl groups generated by the previous reduction and the oxygen adatoms resulting from the reaction of the hydroxyl groups completely. The remaining oxygen and hydroxyl groups can enhance the uptake of oxygen atoms into the bulk. Therefore, the relatively large extent of oxygen sorption after evacuation at 523 K is accompanied by a relatively small oxygen TPD peak.Evacuation at 673 K more completely removes the In particular, a paper by O'Handley et290 Interaction of Oxygen with Silver Particles on Alumina hydroxyl groups and oxygen, resulting in silver particles, displaying the equilibrium shape characteristic of almost clean silver. The larger fraction of atomically rough planes in the surface of the rounded particles brings about a larger oxygen desorption peak, whereas penetration into the bulk does not take place to a significant extent. In conclusion, it can be stated that our results indicate that complex morphological changes of the silver surface are taking place during our experiments. In this respect, not only is the reduction or evacuation temperature of importance, but also the amount of oxygen which penetrates the silver lattice after a certain pretreatment.Another feature pointing towards the importance of the detailed topography of the silver particles is the difference between the present results, obtained with individual, isolated, silver particles, and those of a previous study,2 in which agglomerated silver particles were investigated. The mean particle size was only about a factor of 2 larger than the particles in this study (70-150 nm). The agglomerated silver particles, however, did not exhibit the above variation in the amount of oxygen sorbed at 443 K with the reduction or evacuation temperature. As stated in the introduction, this catalyst was relatively heavily loaded (8 wt %), which led to a very dense population of the silver particles on a-alumina with a low surface area.Moreover, the particles were irregularly shaped and tended to stick together in chain formation. However, as they did not undergo complete coalescence during the subsequent pretreatment procedures, these particles must contain a large number of grain boundaries and surface defects. We believe that the presence of these grain boundaries may greatly influence the oxygen chemisorption properties of silver particles. After removal of the carbonaceous impurities the heavily loaded catalyst of ref. (2) sorbed 0.5 cm3 (s.t.p.) g,:. This amount of oxygen is significantly lower than that one would expect to be taken up from the size of the silver particles seen in the electron microscope. The observed particle size (150-200 nm) suggests an extent of oxygen sorption of 0.9-0.7 cm3 (s.t.p.) g;: for a complete monolayer.The relatively small take up of oxygen must therefore be due to a significant fraction of undisturbed (1 11) planes in the surface of the silver particles. Conclusions The amount of oxygen taken up by silver particles of 70 nm at 443 K is strongly influenced by the pretreatment procedure. Even during the process of adsorption, considerable amounts of oxygen can penetrate the bulk of the silver lattice. The results are also affected by small amounts of impurities such as carbon and oxygen (or hydroxyl groups) in or below the surface. These impurities can dramatically enhance the amount of sorbed oxygen. Measurement of the silver surface area by means of the chemisorption of oxygen at 443 K is thus liable to provide erroneous results.The results of this work indicate that a monolayer coverage is presumably not always obtained during exposure to oxygen at 443 K. A much more plausible explanation is that only certain sites or patches on the silver surface are capable of adsorbing, as well as absorbing oxygen. The net result of both adsorption and penetration, however, often nicely agrees with that calculated for adsorption of a monolayer, even for various mean particle sizes. Nevertheless, the agreement must be fortuitous ; penetration of oxygen into some surfaces compensates for the lack of adsorption on undisturbed (1 1 1) surfaces. It is of interest to compare the excess of oxygen taken up beyond a complete monolayer with the thermodynamic solubility of oxygen into bulk silver.Eichenauer and Muller26 and Gryaznov2' have reliably determined the solubility of oxygen in silver. At 443 K a solubility of (1-2) x cmL2 (s.t.p.) g;; can be calculated from the data of the above authors. The solubility appears to be much lower than the oxygen taken up beyond a monolayer. For the catalyst of the present investigation reduced and evacuated at 473 K an oxygen take-up of 2.7 cm3 (s.t.p.) g;; was measured after removal of the carbonaceous impurities. Since a monolayer of oxygen involves 2.0 cm;* (s.t.p.) g,:, theG. R. Meima et al. 29 1 excess is 0.7 cmi2(s.t.p.) g,:. This amount of oxygen is much larger than that corresponding to the bulk solubility. At some surfaces, therefore, either multilayer adsorption or a higher solubility in the selvedge of silver must be present.Moore2* has dealt with the older literature pointing to the surface of silver thermally treated in an oxygen atmosphere containing an amount of oxygen significantly greater than the solubility of oxygen in bulk silver. The pretreatments also strongly affect the surface structure of the silver particles. Our results, combined with those from the literature, indicate that faceting of the silver particles occurs at elevated temperatures in oxygen atmospheres and that this phenomenon is enhanced by the presence and the amount of dissolved oxygen. Reduction, on the other hand, presumably leads to a roughening of the surface owing to the removal of the penetrated oxygen species. These surface structural effects not only influence the oxygen chemisorption properties, but also strongly determine the catalytic activity of these catalysts for the oxidation of CO with molecular ~ x y g e n .~ ? ~ We therefore believe that when studying the catalytic activity and selectivity of silver catalysts for various reactions, it is important to pay special attention to the pretreatment of the catalysts and the conditions to which the catalyst has previously been subjected. References 1 (a) M. A. Barteau and R. J. Madix, in The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis, ed. D. P. Woodruff and D. A. King (Elsevier, Amsterdam, 1982), vol. 4, chap 4; (6) R. W. Clayton and S. V. Norval, Catalysis, 1980, 3, 70; (c) X. E. Verykios, F. P. Stein and R.W. Coughlin, Catal. Rev. Sci. Eng., 1980,22, 197; ( d ) P. A. Kilty and W. M. H. Sachtler, Catal. Rev. Sci. Eng., 1974, 10, 1 and references herein. 2 G. R. Meima, L. M. Knijff, A. J. van Dillen, J. W. Geus, J. E. Bongaarts, F. R. van Buren and K. Delcour, in New Developments in Selective Oxidation, European Workshop Meeting, 17-1 8 March, 1986, Louvain-la-Neuve, ed. B. Delmon and P. Ruiz, Catal. Today, 1987, 1, 117. 3 H. A. Engelhardt and D. Menzel, Surf. Sci., 1976, 57, 591. 4 M. Bowker, M. A. Barteau and R. J. Madix, Surf. Sci., 1980, 92, 528. 5 H. Albers, J. M. M. Droog and G. A. Bootsma, Surf Sci., 1977, 64, 1. 6 H. Albers, W. J. J. van der Wal and G. A. Bootsma, Surf. Sci., 1977, 68, 47. 7 C. Backx, J. Moolhuysen, P. Geenen and R. A. van Santen, J. Catal., 1981, 72, 364. 8 R. B. Grant and R. M. Lambert, J. Chem. SOC., Chem. Commun., 1983, 662. 9 G. R. Meima, L. M. Knijff, A. J. van Dillen, F. R. van Buren and J. W. Geus, J. Chem. SOC., Faraday Trans. I , 1989, 85, 293. 10 K. P. de Jong, Ph.D. Thesis (University of Utrecht, 1982), chap. 5. 11 M. Jarjoui, B. Moraweck, P. C. Gravelle and S. J. Teichner, J. Chim. Phys., 1978, 75, 1060. 12 J. Joziasse, Ph.D. Thesis (University of Utrecht, 1978), chap. 5. 13 J. J. F. Scholten, J. A. Konvalinka and F. W. Beekman, J. Catal., 1973, 28, 209. 14 K. M. Kholyavenko, Ya. M. Rubanik and N. A. Chernukina, Kinet. Katal., 1964, 5, 505 [Kinet. Catal. 15 A. I. Boronin, V. I. Bukhtiyarov, A. L. Vishnevskii, G. K. Boreskov and V. I. Savchenko, Kinet. 16 A. Auroux and P. C. Gravelle, Thermochim. Acta, 1981, 47, 333. 17 B. E. Sundquist, Acta Metallurg., 1964, 12, 67. 18 R. Shuttleworth, R. King and B. Chalmers, Proc. R. SOC. (London), Ser. A , 1948, 193, 465. 19 A. J. W. Moore, Acta Metallurg., 1958, 6, 293. 20 E. D. Hondros and A. J. W. Moore, Acta Metallurg., 1960, 8, 647. 21 G. E. Rhead and H. Mykura, Acta Metallurg., 1962, 10, 843. 22 (a) A. W. Czanderna, J. Phys. Chem., 1966, 70, 2120; (b) A. W. Czanderna, Thermochim. Acta, 1978, 23 R. C. OHandley, D. K. Burge, S. N. Jasperson and E. J. Ashley, SurJ Sci., 1975, 50, 407. 24 Y. L. Sandler, S. Z. Beer and D. D. Durigon, J. Phys. Chem., 1966, 70, 3881. 25 E. D. Hondros, J. Inst. Met., 1959/60, 88, 275. 26 W. Eichenauer and W. Miiller, Zeit. Metallk., 1962, 53, 321. 27 V. M. Gryaznov, S. G. Gul'yanova and S. Kanizius, Russ. J. Phys. Chem., 1973, 47, 1517. 28 A. J. W. Moore, in Metal Surfaces: Structure Energetics and Kinetics (American SOC. for Metals, Paper 8/010121; Received 14th March, 1988 (Engl. Transl.), 1964, 5, 4371. Katal., 1984, 25, 1510 [Kinet. Catal. (Engl. Transl.), 1984, 25, 13011. 24,359. Metals Park, Ohio, 1963), p. 155.
ISSN:0300-9599
DOI:10.1039/F19898500279
出版商:RSC
年代:1989
数据来源: RSC
|
14. |
The interaction of oxygen with isolated silver particles ofca.70 nm supported onα-alumina. Part 2.—CO oxidation measurements |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 293-304
Garmt R. Meima,
Preview
|
PDF (895KB)
|
|
摘要:
J. Chem. SOC., Furaduy Trans. I , 1989, 85(2), 293-304 The Interaction of Oxygen with Isolated Silver Particles of ca. 70 nm supported on &-Alumina Part 2 . 4 0 Oxidation Measurements Garmt R. Meima,*t Loek M. Knijff, Adrianus J. van Dillen and John W. Geus Department of Inorganic Chemistry, State University of Utrecht, Croesestraat 77a, 3522 AD Utrecht, The Netherlands Frederik R. van Buren Dow Chemical (Nederland) B. V., P.O. Box 48, 4530 A A Terneuzen, The Netherlands The oxidation of carbon monoxide with molecular oxygen over a catalyst containing isolated silver particles of cu. 70 nm supported on a-Al,O, has been studied after various pretreatments. Both the extent of oxygen sorption and the activity were greatly affected by the pretreatment. A rise in activity was observed after reduction and consecutive cooling in hydrogen, whereas a drop followed an oxidizing treatment.A severe deactivation of the catalyst occurred after reduction at 673 K followed by cooling in nitrogen. The differences in activity due to the various pretreatments are discussed in terms of changes in the surface structure of the silver particles. This work deals with the properties of monocrystalline silver particles of ca. 70 nm diameter supported on a-alumina. In part 1 we reported on the interaction of oxygen with these silver particles. The interaction was studied by measuring volumetrically the sorption of oxygen and hydrogen, and by temperature-programmed desorption (TPD), in which the composition of the gas phase was continuously determined mass- spectrometrically.The pretreatment of the catalyst appeared to affect strongly the subsequent interaction with oxygen or hydrogen. The aim of this paper is to relate the interaction with oxygen discussed in part 1 to the activity in the oxidation of carbon monoxide. Since the catalytic properties of silver are complicated, we have chosen a relatively simple catalytic reaction, namely the oxidation of carbon monoxide. In addition to the pretreatment, the size and defect structure of the silver particles can also strongly affect the interaction with oxygen and the catalytic activity. Earlier, we studied the interaction with oxygen and the activity in the oxidation of carbon monoxide by supported silver particles of 150-200 nm diameter, also supported on a-alumina.2 These silver particles had agglomerated and shared grain boundaries.A comparison of the properties of the present catalyst containing essentially monocrystalline silver particles with those of the catalyst studied earlier can thus provide information about the effect of grain boundaries. The catalyst containing silver particles of 150-200 nm diameter, after being freed of carbon taken up during the catalyst preparation, exhibited a relatively large absorption of oxygen into the silver. The extent of interaction with hydrogen at 443 K could be used to assess the amount of oxygen that had penetrated into the silver. The catalytic activity in the oxidation of carbon monoxide was found to decrease strongly on oxidation at or above 473 K. Hydrogen reduction, on the other hand, raised the activity appreciably.Surprisingly, cooling in nitrogen from the reduction temperature to the temperature at which the catalytic reaction was performed resulted in a lower activity than cooling in hydrogen. t Present address: Dow Chemical (Nederland) B.V., P.O. Box 48, 4530 AA Terneuzen, The Netherlands. 293294 Interaction of Oxygen with Silver Particles on Alumina As discussed in the part of this work dealing with silver particles of ca. 70 nm diameter, treatment in hydrogen at elevated temperatures lowered the amount of oxygen subsequently taken up at 443 K. Combination of TPD experiments and volumetric sorption measurements showed that the drop in the take-up of oxygen is due to a lower absorption of oxygen into the silver.However, the extent of oxygen adsorption was found to increase after treatment in hydrogen at elevated temperatures. The effect of pretreatment on the subsequent sorption of oxygen was explained by the observation of Albers et al.3 that an undisturbed clean silver (1 11) surface does not adsorb molecular oxygen. Much evidence has been published that thermal treatment in oxygen leads to faceting of silver particles, and thus to an increase in the fraction of (1 1 1) planes in the The cause of this faceting has still not been elucidated. Some authors believe that evaporation of silver, which proceeds more rapidly in an oxygen atmosphere, leads to faceting. The crystallographic planes having the lowest rate of evaporation finally remain.8 Other authors, however, were not able to reproduce satisfactorily experiments in which a net evaporation of silver was prevented.Faceting was not observed when net evaporation was p r ~ h i b i t e d . ~ The other authors therefore assume that a change in the anisotropy of the surface energy due to interaction with oxygen causes the faceting. Irrespective of the explanation, it has been demonstrated beyond doubt that thermal treatment in oxygen leads to faceted silver particles with a large fraction of (1 11) planes in their surface. As a result, the extent of adsorption of oxygen will be relatively small. Relatively large silver particles contain a significant amount of dissolved oxygen, which was found to desorb only slowly at fairly high temperatures. Since desorption of oxygen atoms adsorbed on the silver surface proceeds much more rapidly, the transport to the surface must determine the rate of the desorption process.The relatively high mobility of oxygen atoms in the silver bulk suggests that the transport of oxygen atoms from a position in or just below the surface to a position where they are adsorbed on the surface is rate-determining. The difficult migration to the adsorption sites ensures that, during cooling after thermal treatment in oxygen, a number of oxygen atoms can reach positions in or just below the silver surface. Oxygen atoms in these positions produce defect sites at the surface where oxygen molecules can dissociate and, more importantly, where the resulting oxygen atoms can easily penetrate the silver surface. After thermal treatment in oxygen, therefore, the silver surface contains many (1 1 1) surfaces, where no adsorption of oxygen can proceed, and a limited number of defect sites acting as portholes for the migration of oxygen atoms into the surface.It is possible that the defect sites are preferably present in other crystallographic planes than the abundant (1 1 1) planes. Thermal treatment in hydrogen, on the other hand, has been observed to lead to metal particles with rounded edges and corners. Rounded silver particles have a fraction of atomically rough crystallographic planes in their surface which is appreciably higher than that of the faceted silver particles resulting from treatment in oxygen. The lower fraction of (1 11) planes means that the extent of oxygen adsorption is higher after hydrogen reduction.The lack of surface defects, on the other hand, ensures that the penetration of oxygen atoms into the surface proceeds to a much smaller extent. The selvedge (surface layer) of the silver particles will lose more oxygen as the thermal treatment in hydrogen is carried out at more elevated temperatures. Consequently, the number of dissolved oxygen atoms that can reach sites in or below the surface drops as the temperature at which the hydrogen reduction or subsequent evacuation has been carried is raised. As a result, the number of portholes where oxygen atoms can penetrate into the silver will decrease with an increase in the temperature of thermal pretreatment. The amount of penetrated oxygen will thus drop rapidly at more elevated temperatures of hydrogen reduction.It is of interest to investigate the effects of pretreatment on the catalytic activity ofG. R. Meima et al. 295 silver catalysts in oxidation reactions. If the activity is governed by the extent of the silver surface where the dissociative adsorption of oxygen can proceed, oxidation of the silver is expected to lead to a lower activity of the catalysts. The smaller extent of oxygen adsorption observed, as dealt with in the preceding paper' after thermal treatment in oxygen, must be reflected in a lower oxidation activity. Thermal treatment in hydrogen, which raises the extent of oxygen adsorption, can be expected to lead to an increase in the oxidation activity. The catalyst containing silver particles of 150-200 nm in diameter, dealt with in ref.(2), exhibited the above variations in activity. The fact that these variations in activity were reversible indicates that the size of the silver particles, and hence sintering, did not affect the activity. The large silver particles contained grain boundaries at which the dissociative adsorption of oxygen can proceed. Moreover, faceting of large silver conglomerates does not occur readily. Therefore, the effects of pretreatment on the catalytic activity may be significantly larger with the present catalyst containing monocrystalline silver particles of ca. 70 nm. In this work we investigated the oxidation of carbon monoxide from 323 to 523 K. Since an excess of oxygen was used, interaction of the silver surface with oxygen during the measurements of the activity must be considered.After thermal treatment in oxygen, the catalyst was always cooled in oxygen. When the catalyst was reduced prior to the catalytic measurements, cooling was carried out in either hydrogen or nitrogen. Experimental Catalyst The procedure for preparation of the catalyst is described in Part 1.l It contained 1.5 wt % of silver supported on a-Al,O,. The mean silver particle size was ca. 70 nm in diameter, as determined by electron microscopy. CO Oxidation Measurements The catalyst powder was pressed into tablets and a sieve fraction of 0.5-1 .O mm particles was used in the experiments. A sample of 2.16 g was placed in a tubular glass reactor of internal diameter 9 mm. The volume of the catalyst bed was 1.65 cm3. The pretreatment of the catalyst was, unless otherwise stated, performed overnight (ca.16 h) at various temperatures, either in a 10% H,-N, flow, a 100% 0, flow or a 100 YO N, flow. Prior to the measurements, the catalyst was generally cooled in the same gas flow. In order to investigate the influence of the gas atmosphere after reduction at 673 K, the catalyst was sometimes flushed with nitrogen before cooling. During the measurements a gas flow of 50 cm3 min-' containing 1 vol% CO, 1 vol YO O,, balance N,, was passed downwards through the catalyst bed. The amount of CO, formed was measured conductometrically . The conversion was calculated by comparing total combustion with the measured conversion. A complete description of this technique and the apparatus used have been published elsewhere.l1 After loading, the catalyst was subjected to various consecutive pretreatments. Owing to irreversible changes taking place during the measuring sequence, the conversion us. temperature curves will be discussed in chronological order. Thereafter, these results will be treated in the form of Arrhenius plots largely depending upon the atmosphere of the pretreatment . As long as the temperature remained below ca. 470 K, stable activities were generally measured over the complete temperature range. Higher temperatures sometimes led to296 Interaction of Oxygen with Silver Particles on Alumina 100 80 60 n 53 Y 40 20 T/K Fig. 1. CO conversion us. temperature after subsequent pretreatments: (1) N,, 473 K (runs 1A and 1B) and first 0,, 473 K (run 1C); (2) 0,, 473 K (run 2B); (3) H,, 523 K; (4) 0,, 598 K (runs 4A and 4B). a decrease in activity, especially after a previous reduction.The drop in activity was caused by the net oxidizing atmosphere in which the measurements were performed. Then non-steady-state kinetics were measured, exhibiting a drop in activity during the measurement. The (parts of the) conversion curves in which the activity was unstable will be mentioned explicitly. All other conversion curves represent the steady-state activity after a particular pretreatment. In fig. 1 and 2 the conversion of CO into CO, is plotted as a function of temperature after the following pretreatment sequence (measurement after each pretreatment, some pretreatments performed twice) : (1) N, 473 K (runs 1A and 1 B) and 0,473 K (run 1 C), (2) 0, 473 K (runs 2A and 2B), (3) H, 523 K, (4) 0, 598 K (runs 4A and 4B), ( 5 ) H, 523 K, (6) H, 673 K and (7) H, 673 K cooled in N, (runs 7A and 7B).As can be seen in fig. 1, the freshly loaded catalyst initially exhibited a relatively elevated activity after pretreatment in nitrogen [curve (l)], The same conversion curve was measured after the first pretreatment in oxygen at 473 K. The second oxidation at 473 K caused a strong deactivation of the catalyst. A similar consecutive pretreatment further deactivated the catalyst only slightly [curve (2)]. For reasons of clarity, only this curve is shown (representing run 2B) because the curve representing run 2A coincides with curve (4). The TPD measurements dealt with in part 1 indicated the presence of carbonaceous impurities in the catalyst.' Evidently the second oxidation treatment exhausted the carbon.The presence of carbon atoms in the silver surface promotes the sorption of oxygen. Subsequent reduction of the catalyst at 523 K more or less restored the initial activity of the fresh catalyst [curve (3)]. However, above ca. 450 K the conversion no longer increased with the temperature. At temperatures higher than ca. 470 K a decrease inG. R. Meirna et al. 297 100 80 60 h e Y 40 20 0 313 4i3 5i3 T/K Fig. 2. CO conversion us. temperature after subsequent pretreatments: (5) H, 523 K ; (6) H,, 673 K ; (7) H,, 673 K cooled in N, (runs 7A and B); (4); see fig. I . activity was observed as a function of time. At these temperatures the catalyst was thus slowly being deactivated during the measurement.Reoxidation of the catalyst at elevated temperatures (598 K) again led to a strong deactivation [curve (4)]. However, the drop in activity was smaller than after the previous oxidation [curve (2)]. A second similar pretreatment did not influence the activity further. Renewed reduction of the catalyst at 523 K [curve (5)] led to an activity higher than the original (active) state. No deactivation of the catalyst was observed over the complete temperature range, in contrast to the behaviour of the catalyst after the first reduction [curve (3)]. As shown, a slow deactivation occurred only at temperatures higher than ca. 470 K [curve (3)]. Since the catalyst being reduced the second time exhibited complete conversion at 470 K, deactivation evident from a decrease in conversion with time cannot be established in this temperature range.The drop in activity is apparently not sufficiently high as to lead to conversions measurably different from 100%. Subsequent reduction at 673 K led to a slight deactivation of the catalyst, as compared with reduction at 523 K [curve (6)]. Above 400 K the conversion was less than that of the fresh loading. At temperatures > 500 K the conversion clearly rose more slowly with temperature. Again non-steady-state activities were measured. After ca. 2 h on stream at 510 K, a conversion curve was measured which was comparable with that measured after oxidation at 598 K [curve (4)]. The arrow in fig. 2 denotes the deactivation of the catalyst as a function of time.The dashed curve represents (the stable) activity after the deactivation process. An interesting phenomenon was observed when, after reduction at 673 K, the catalyst was not cooled in hydrogen, but in nitrogen. Such a pretreatment results in a severe deactivation of the catalyst [curve (7)].298 Interaction of Oxygen with Silver Particles on Alumina 500 400 2, I 0 4 5 -2 -4 Fig. 3. Arrhenius plots for the oxidation of CO over 70 nm diameter silver particles supported on a-alumina. After both hydrogen and oxygen treatment relatively straight plots are found, corresponding to an activation energy of 40 kJ mol-l. The logarithm of the pre-exponential factors is in relative units. x , Run 2B, -8.8; +, run 3, -10.4; 0, run 5, -10.9; 0, run 7B, -8.6.Arrhenius plots of all the above conversion curves were constructed. In all figures the lines are identified by the measurement run number followed by the numerical value of Ink, (in relative units). The plots were obtained assuming a first-order in CO and a zeroth-order in 0, and CO,. Only the orders in CO and 0, were measured: the order in CO, was not. The assumed zeroth order agrees with the literature; however, it is presumably only valid if low concentrations of CO, (< 10 ~ 0 1 % ) are present in the feed.l27 l3 The limited stability of the silver catalyst caused an appreciably larger scatter of the data than brought about by, e.g., variations in the temperature and the accuracy of the determination of the conversion. Nevertheless, a number of runs displayed fairly straight Arrhenius plots.These runs are collected in fig. 3. Apparently, treatment in either hydrogen or oxygen can result in a reasonably straight Arrhenius plot. Minor deviations from linearity indicate a comparatively slight instability. Within experimental error all the runs represented in fig. 3 exhibit an activation energy of 40 kJ mol-'. This activation energy agrees nicely with the value measured previously for 150-200 nm diameter Ag particles. In fig. 4 Arrhenius plots of runs where serious deactivation occurred are represented. As can be seen, deactivation does not affect the activation energy, but involves a drop in the pre-exponential factor. During run 2A the deactivation started at ca. 385 K ; a small number of active sites was lost.The decrease in active sites was much more severe during run 6. The susceptibility of this silver catalyst to deactivation, especially after pretreatment in hydrogen, is also apparent from run 3. As an example of this deactivation, the Arrhenius plot for run 6 is given in fig. 4(b). It is likely that the catalyst did not contain a significant amount of carbon in the runs for which Arrhenius plots are presented in fig. 5. The carbon was nearly completely removed by the previous oxidations. The small difference in k, between runs 2A and 2B can be explained by the removal of residual carbon. The minor variation in activity after pretreatment in oxygen points to relatively stable active sites with the carbon-free catalyst.G. R. Meima et al.-4 299 - I I I I I I I 500 400 2 / * 5 -2. - 4 - > 2 -0 2.5 lo3 KIT Fig. 4. Arrhenius plots measured for the oxidation of CO on 70 nm diameter silver particles supported on a-alumina: (a) measured during run 2A after pretreatment in oxygen at 473 K; ca. 373 K the catalyst lost a number of active sites, which led to deactivation; (b) measured during run 6 after pretreatment in hydrogen at 673 K ; owing to the loss of a larger number of active sites, the deactivation is now more severe. The logarithm of the pre-exponential factors in relative units. As is apparent from fig. 6, where Arrhenius plots for runs measured after pretreatment of the catalyst in hydrogen are represented, the variation in the activity of the reduced catalyst is much larger. As considered above, the catalyst deactivated during run 6, while being cooled in nitrogen, which was done in runs 7A and 7B, resulted in a relatively low pre-exponential factor and thus in a relatively low number of active sites.The larger variation in the number of active sites observed after thermal treatment of the catalyst in hydrogen nicely agrees with the experimental evidence on the take-up of oxygen and hydrogen and temperature-programmed desorption. Thermal treatment in hydrogen leads to silver crystallites with rounded corners and edges. The atomically300 T/K 2< Interaction of Oxygen with Silver Particles on Alumina icles sumor Fig. 5. Arrhenius plots for the oxidation of CO on 70 nm diameter silver par I I ed on a-alumina. The catalyst was pretreated in oxygen at the temperatures indicated.The logarithm of the pre-exponential factor in relative units. 0, Run 2A, -9.3; x , run 2B, -8.8; El, run 4A, -9.3; +, run 4B, -9.3. Fig. 6. Arrhenius plots for the oxidation of CO on 70 nm diameter silver particles supported on a-alumina pretreated in hydrogen at the temperatures indicated. In runs 3, 5 and 6 the catalyst was cooled in hydrogen. In runs 7A and 7B the catalyst was cooled in nitrogen. The logarithm of the pre-exponential factors in relative units. 0, Run 3, - 10.4; a, run 5 , - 10.9; 0, run 6, -10.3/9.25; +, run 7A, -8.1; x , run 7B, -8.6. rough surfaces present at the curved surfaces contain many sites where oxygen can be dissociatively adsorbed. However, the crystallites are liable to faceting to a varying extent.Faceting may be due to interaction with the excess of oxygen used during the CO oxidation. At more elevated temperatures interaction with oxygen may lead to the growth of (1 1 1) facets. Since the mobility of dissolved hydroxyl groups has been foundG. R. Meima et al. 30 1 1111111111 2.5 **' lo3 KIT Fig. 7. Arrhenius plots for the oxidation of CO on 70 nm diameter silver particles supported on a-alumina measured with the fresh catalyst after treatment in nitrogen and oxygen. The logarithm of the pre-treatment factors in relative units. a, Run 1A; 0, run 1B; 0, run 1C. to be higher than that of dissolved oxygen atoms, faceting by oxygen migrating to the surface from the bulk will proceed more readily after pretreatment in hydrogen. Hydroxyl groups arriving at the surface will react to form water, which is desorbed, and oxygen atoms, which are desorbed only at relatively high temperatures.During cooling after the pretreatment in hydrogen, dissolved hydroxyl groups will continue to migrate to the surface. In hydrogen the oxygen atoms remaining at the surface after the desorption of water will be reduced and thus removed from the surface. With cooling in nitrogen, on the other hand, the transport of hydroxyl groups to the surface will continue, but removal of the resulting oxygen atoms as water is not possible. The oxygen atoms remaining at the surface will give rise to faceting. Since the number of oxygen atoms in the selvedge will be smaller than after treatment in oxygen, the number of surface defects at which oxygen can be dissociatively adsorbed will also be small.The result is a smaller number of active sites, and hence an activity smaller than found after treatment in oxygen. Simultaneously with the formation of hydroxyl groups in the silver surface and the subsequent formation of water, some hydroxylation of the a-alumina support may also take place. However, we have reason to believe that this has no influence on the catalytic activity. a-Alumina is known for its relative inertness, and the number of surface hydroxyl groups as compared with y-alumina is small. Moreover, pure a-alumina has no measurable catalytic activity for the oxidation of CO in the studied temperature range. We therefore ascribe the abovementioned effects after reduction to changes in the silver surface and not to the support material.The temperature-programmed desorption experiments described in our previous paper' have demonstrated that the fresh catalyst contained some carbon. Like penetrated oxygen, carbon atoms in the surface layer promote the dissociative adsorption of oxygen. The catalyst containing carbon can therefore be expected to exhibit a relatively high activity. The promoting effect of carbon is evident from the results represented in fig. 7, where Arrhenius plots for runs 1A-1C are shown. The activity of the fresh catalyst after treatment in nitrogen or oxygen is significantly higher than that of the catalyst after more extended oxidation. Moreover, the activity is 11 F A R I302 Interaction of Oxygen with Silver Particles on Alumina stable, as can be concluded from the fact that the plots for the first three runs agree well.From fig. 1, in which the conversion has been plotted as a function of temperature, the promotion by carbon is also evident. Remarkably, all three plots represented in fig. 7 indicate a transition to a reaction characterized by a higher activation energy, viz. 60 kJ mol-'. The transition was at 433 K. Transition to a process exhibiting an activation energy of 60 kJ mol-' was also found with the catalyst containing silver particles of 150-200 nm diameter.2 Measurements on still larger silver particles (ca. 1 pm) to be dealt with in a subsequent paper14 also result in Arrhenius plots containing two straight lines showing a sharp transition corresponding to activation energies of 40 and 60 kJ mol-l.As will be discussed in a subsequent paper, the experimental evidence points to the same active sites being involved in reactions of a mechanism that are different at lower and higher temperatures. The pre-exponential factor of the Arrhenius plot corresponding to an activation energy of 40 kJ mol-' indicates a higher number of active sites than that measured after more extensive oxidation treatment. Discussion In other studies2,'4v'5 we have demonstrated that the activity of silver catalysts for the oxidation of CO with molecular oxygen is due to a relatively small number of sites. We believe that the results obtained in this study can also be explained in the above manner. With a few exceptions, the results closely resemble those of a previous study.2 The deactivation of the catalyst after oxidation is caused by the silver crystallites assuming a shape exposing more (1 11) facets during thermal treatment in There has been published an extensive body of experimental evidence over the years pointing to faceting of silver particles at high temperatures in the presence of of which the results by Melle et al.4 are illustrative.These authors studied the morphology and electron-diffraction patterns of silver particles of ca. 5 mm diameter. The silver particles were derived such that their surfaces were untouched by any physical or chemical preparation. Whereas the freshly prepared particles had the almost spherical shape of a sessile drop, thermal treatment in oxygen led to completely faceted particles exposing mainly (1 1 1) facets.The monocrystalline silver particles of 5 mm diameter exhibited faceting most clearly on pretreatment in oxygen at temperatures near the melting point of silver. It is obvious that the much smaller silver particles investigated in this work exhibit faceting at much lower temperatures. Accordingly, a strong deactivation was found in our work on larger silver particles on interaction with oxygen at temperatures of 598 K or higher. As stated above, the number of defects in the surface, and particularly in the (1 11) planes of the silver particles, will be raised by treatment at high temperatures in oxygen. Consequently, cooling after treatment in hydrogen or nitrogen leads to the presence of a relatively small amount of oxygen in the surface layer.The oxygen results from the reaction of hydroxyl groups to form water and oxygen atoms. The amount of oxygen is sufficient to bring about faceting, but the number of surface defects in the (1 1 1) planes will be very small, which results in a relatively low activity. Incorporation of other foreign atoms in the surface layer of silver can also produce sites on the (1 1 1) planes where oxygen can be dissociatively adsorbed. Consequently the fresh catalyst, which contains some carbon according to previous mass-spectrometry results, exhibits a relatively high activity. The carbon can be removed by oxidation at higher temperatures, which leads to a drop in the activity. We can now consider the effect of the size of the silver particles and of the presence of grain boundaries.Since both silver catalysts have an activation energy of 40 kJ mol-', their activities can be compared by comparing their pre-exponential factors. In table 1 the pre-exponential factors are summarized. It can be seen that the variation in the pre-exponential factors with the method ofG. R. Meima et al. 303 Table 1. Comparison of the activities of 15&200 and 70 nm diameter silver particles supported on a-alumina in the oxidation of carbon monoxide pre-exponential factor/ lo6 min-' g;: particle size/nm minimum maximum ratio 150-200 0.5 70 0.3 10.4 20 5 .O 17 pretreatment is of the same order of magnitude. The fact that the silver catalyst with larger particles of 150-200 nm diameter displayed the largest variation is not expected. Apparently the surfaces of higher curvature are faceted to almost equal extents with the silver particles in the two catalysts by treatment in oxygen or cooling in nitrogen after hydrogen treatment.The absolute values of the pre-exponential factors are of interest. The surface area per g of silver of the catalyst containing silver particles of 150-200 nm diameter is expected to be smaller by a factor of 2-3. If the number of active sites per unit area of silver surface were the same, the catalyst containing silver particles of 70 nm diameter should have a pre-exponential factor exceeding that of the catalyst containing silver particles of 150-200 nm diameter by a factor of 2-3. In fact the pre-exponential factor of the catalyst having silver particles of 70 nm diameter is smaller than or equal to that of the other catalyst.The relatively high value found for silver particles of 150-200 nm diameter may be due to the presence of lattice defects, or to the fact that the larger particles cannot easily take up a shape not containing curved surfaces. Conclusions The most remarkable feature of the silver catalyst studied here is the large variation in activity with the method of pretreatment. The variation in activity is not due to sintering of the silver particles. First, electron microscopy did not indicate significant sintering. More important, however, is the fact that the decrease in activity is reversible. The strong variation in activity must therefore be a attributed to the morphology, the shape of the silver particles and the number of defects in closely packed (1 11) surfaces.The experimental results of this work can be explained completely by combining the well established effect of thermal treatment in oxygen on the shape of the silver particles, and by Albers' observation that a clean, defect-free silver (1 11) surface does not adsorb ~ x y g e n . ~ An effect on the shape of the silver particles even at low temperatures is not unexpected. The Tammann temperature for silver is 541 K (calculated as 0.52 Tm, where T, is the melting temperature). It is usually assumed that the onset of mobility of the lattice atoms is at the Tammann temperature. An effect on the surface topography of small particles at temperatures of the order of the Tammann temperature of silver is therefore to be expected. Another important point is that the results of the (ad)sorption and thermal desorption experiments agree very well with the pattern of catalytic activity. The results for the adsorption and dissolution of oxygen corroborate the data on the catalytic activity.We thank Dow Chemical (Nederland) B.V. for financial support. 11-2304 Interaction of Oxygen with Silver Particles on Alumina References 1 G. R. Meima, R. J. Vis, M. G. J. van Leur, A. J. van Dillen, J. W. Geus and F. R. van Buren, J. Chem. SOC., Faraday Trans. I , 1989, 85, 279. 2 G. R. Meima, L. M. Knijff, A. J. van Dillen, J. W. Geus, J. E. Bongaarts, F. R. van Buren and K. Delcour, in New Developments in Selective Oxidation, European Workshop Meeting, 17-18 March, 1986, Louvain-la-Neuve, ed. B. Delmon and P. Ruiz, Catal. Today, 1987, 1, 117. 3 (a) H. Albers, J. M. M. Droog and G. A. Bootsma, Surf. Sci., 1977, 68, 1 ; (b) H. Albers, W. J. J. van der Wal and G. A. Bootsma, Surf. Sci., 1977, 68, 47. 4 H. Melle, E. Menzel and J. Zaunert, Colloq. In?. Cent. Nut. Rech. Sci., (1970), p. 187. 5 B. E. Sundquist, Acta Metallurg., 1964, 12, 67. 6 B. Chalmers, R. King and R. Shuttleworth, Proc. R. Soc. (London), Ser. A , 1948, 193, 465. 7 A. J. W. Moore, Acta Metallurg., 1960, 8, 647. 8 E. D. Hondros and A. J. W. Moore, Acta Metallurg., 1960, 8, 647. 9 G. E. Rhead and H. Mykura, Acta Metallurg., 1962, 10, 843. 10 (a) A. W. Czanderna, J. Phys. Chem., 1966, 70, 2120; (b) A. W. Czanderna, Thermochim. Ada, 1978, 11 A. J. van Dillen, Ph.D. Thesis (University of Utrecht, 1977), chap. 4. 12 A. F. Benton and R. T. Bell, J. Am. Chem. Soc., 1934, 56, 501. 13 G. W. Keulks and C. C . Chang, J. Phys. Chem., 1970, 74, 2590. 14 G. R. Meima, M. Hasselaar, A. J. van Dillen, F. R. van Buren and J. W. Geus, J. Chem. Soc., Faraday I5 G. R. Meima, M. G. J. van Leur, A. J. van Dillen, J. W. Geus, J. E. Bongaarts, F. R. van Buren, 24, 359. Trans. I , submitted for publication. L. J. G. M. Derks and K. Delcour, Appl. Catal., in press. Paper 8/01013G; Received 14th March, 1988
ISSN:0300-9599
DOI:10.1039/F19898500293
出版商:RSC
年代:1989
数据来源: RSC
|
15. |
Cationic lead(II) halide complexes in molten alkali-metal nitrate. Part 1.—A thermodynamic investigation of the fluoride system |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 305-316
Lars Bengtsson,
Preview
|
PDF (856KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I , 1989, 85(2), 305-316 Cationic Lead(@ Halide Complexes in Molten Alkali-metal Nitrate Part 1 .-A Thermodynamic Investigation of the Fluoride System Lars Bengtsson and Bertil Holmberg" Inorganic Chemistry I , Chemical Center, University of Lund, P.O. Box 124, S-22100 Lund, Sweden The complexation of fluoride ions with lead(1r) ions in molten equimolar (K,Na)NO, has been investigated in the temperature range 240-280 "C. Complexes with formal compositions PbF,, PbF+ and Pb,F3+ were identified and the stability constants for these species were determined at five temperatures from fluoride activity measurements in the systems (K+, Na+, Pb2+)-(NO;, F-) by means of a fluoride ion-selective electrode. AH",, and AF,, for the stepwise formation of Pb,F2,m-n have been estimated, yielding: AK2 = -7.0 kJ mol-', AT, = 38.3 J K-l mol-1 for PbF++F-+ PbF,; AKl = - 15.3 kJ mol-', AT, = 30.0 J K-' mol-' for Pb2++ F--+ PbF'; A%' = - 5.9 kJ mol-', A S l = 1.9 J K-' mol-' for PbF' + Pb2+ -, Pb,F3+.The thermodynamic results deviate from what is predicted by a quasi-lattice ideal configuration. These deviations are assumed to arise from coordination effects between lead(r1) and nitrate and Pb-Pb interactions in Pb,F3+. The solubility of PbF, in excess of Pb(NO,), in (K,Na)NO,(l) was measured at 280 "C. The complexation model derived from the e.m.f. measurements in dilute melts, expressing all activities in mole-fraction units, does accurately describe the solubility of PbF, in the range C,, 2 C,. In melts with an excess of fluoride over lead@) the model fails though, probably because polynuclear, i.e. polyfluoric, species are formed, Cationic complexes M,XY+(m > 1) with anions as coordination centres constitute an interesting class of complex ions.Such structural building elements have been reported in solid compounds involving highly polarizable central anions and d'O or dlos2 metal ions as ligands.' In solution, cationic complexes with silver(I), thallium(I), cadmium(I1) and lead(r1) have been reported in molten salts and molecular solvents (reviewed by Holmberg and co-workers2-8). The complex Ag413+ exhibits the surprising structural characteristic of having covalent Ag-I bonding distances but no detectable specific coordination geometry, both in (K,Na)NO,(l) and ~ a t e r .~ - ~ The complex Hg213+ in aqueous and DMSO solution displays a well defined geometry with a 90" Hg-I-Hg angle.6 Further knowledge of the properties of this kind of coordination compound is obviously needed to explain their particular physical properties. It is of special interest to find out if this ability of forming cationic complexes extends into systems with less polarizable anions. Preliminary measurements of the solubility of PbO and PbF, in molten alkali-metal nitrates with an excess of Pb(NO,), clearly indicated the formation of species with 0 and F as coordination centres. These results have initiated a more extensive thermodynamic study of the lead(n) fluoride association in molten equimolar (K,Na)NO, in the temperature range 240-280 "C. Ionic liquids such as molten salts are expected to promote the formation of M,XY+ 305306 Pb" Halides in Molten Alkali-metal Nitrate ions, since the difference in electrostatic repulsion between coordinated cations is minimized if the process M"+ + Mm-l XY-" -+ M, Xy+ proceeds via a selective exchange of solvent cations in the coordination sphere of X- (in contrast to a replacement of a neutral molecule in a molecular solvent).The most comprehensive class of solid compounds involving lead(r1) as a metal ligand contain the oxide ion as central anion. Many of these compounds may be described as double salts, for example nPbO - Pb(NO,),, nPbO PbAO, (A = S, Mo, W), nPbO * PbX, (X = F, C1, Br, I), nPbO * PbCO,, nPbO - MO, (M = Si, Ge) and nPbO * M,O, (M = Nb, Ta), having structure elements best described as coordination polyhedra of lead(I1) around a central oxide ion.The lead(r1) hydroxide system, both in aqueous solution and in isolated solid phases, shows several examples of extended or discrete cationic complexes. Some of the compounds of this type also involve fluoride as central anion. Several structures with mixed [(Bi, Pb),(O, F)i+], layers (n = 0-4) or even pure [Pb,Fi+], layers have been The structures are often related to pyrochlore or Aurivillius phases or the cubic form of PbF,. These fluoride-containing sheets have atomic arrangements similar to that in tetrahedral PbO. Some phase diagrams on reciprocal systems reveal interesting compounds like Pb,F,(CrO,), that might include cationic lead fluoride structures considering the stoichiometric Pb/F ratio.21 The phase diagrams of mixed lead halides also show the existence of a number of mixed compounds with fluoride.,, Some of these halide structures have atomic arrangements similar to those in the lead oxide fluoride ~ystems.~~-~' Two solid phases with different cationic lead fluoride complexes are especially intriguing.Pb,F,CO, consists of Pb,F7+ tetrahedra linked to a three-dimensional array2' and Pb,RhF, contains discretoe almost planar Pb,Ft+ units with Pb-Pb distances and Pb-F-Pb angles of ca. 4 A and 120°, re~pectively.~' Pure PbF, exists in a low-temperature form with a PbCl, structure and in a high-temperature form with a CaF, s t r ~ c t u r e . ~ , ~ ~ - ~ ~ Mixed PbO-PbF, systems have also been studied in the molten state.Julsrud and Kleppa found a minimum in the enthalpy of mixing at equal composition of fluoride and oxide ascribed to a reduction in Coulomb repulsion between second-nearest neighbours and polarization of the common ion, i.e. lead(rr). The entropies of mixing were shown to agree with the Temkin The PbO-PbF, system forms glasses over a large composition range. X-Ray diffraction studies of such glasses revealed a cationic Pb,O network as the major structure element. The structural model used to describe the radial distribution functions also involves Pb,F entities.35 Mixed halide melts or solutions have been investigated, and their phase diagrams indicated the formation of associated species which are formulated PbF, - PbX, and 4PbF, - PbX, (X = C1, Br, I).36-37 The complexation in aqueous solutions containing lead@) and fluoride has been studied by means of polarography and various e.m.f.methods revealing PbF+, PbF,, PbF,, PbFi- and mixed PbFX complexes, but no polynuclear [referring to lead(~r)] specie^.,^-^' Several papers describe the complexation of lead@) and halide ions other than fluoride in molten nitrate media. These studies will be reviewed in a subsequent paper.8 The vapour composition over PbF,(s) has been determined, using a Knudsen effusion cell in a mass-spectroscopic anal~sis.~O*~~ The vapour consists mainly of PbF, and Pb,F4, but the mass spectrum revealed species like Pbi, Pb2F+, Pb,Fi and Pb,Fi. Pb,F4 was claimed to have C,, symmetry, with two bridging fluorides and Pb-F-Pb angles of 90". Electron-diffraction studies on PbF,(g) have been p5rformed by Demidov et aL5, They present a peculiarly short Pb-F distance of 2.13 A. The present paper reports thermodynamic results pertaining to lead(rr) fluoride association in (K+, Na+, Pb2')--(N0,, F-) melts.The systems have been studied between 240 and 280 "C by means of solubility and potentiometric measurements.L. Bengtsson and B. Holmberg 307 (K,Na)NO,(l) Ag Ag,CrO,(satd) Chemicals (K,Na)NO,( 1) LaF, (K,Na)NO,(l) KF membrane 1 Ag,CrO,(satd) Ag. Pb(N03)2 KF(satd) Experimental KNO,, NaNO,, Pb(NO,), (Merck, p A) and KF (Fluka, p A) were dried at 130 "C for several days before use. No further loss of weight was observed for any of the chemicals when dried in vacuum for 6 h at the same temperature. KF was handled in a glovebox with a dry N, atmosphere.La(NO,), - 6H,O (Merck, p A) was carefully dried in vacuum at increasing temperature until anhydrous La(NO,), remained.53 No deterioration of the nitrate ion was observed. PbF, was prepared by mixing aqueous solutions of Pb(NO,), and KF. The precipitate was washed with large amounts of water and dried at 130 0C.54 Apparatus The furnaces, temperature control, cell construction and measurements have been described elsewhere. 55-57 E.M.F. Measurements A fluoride ion-selective electrode was constructed using a Pyrex glass tube and the LaF, membrane of an Orion 94-09 fluoride electrode. The procedure has been described previously for KSCN 59 This electrode was modified for use in (K,Na)NO,(l), and it works very well up to 300 "C.The internal part of the fluoride-ion electrode was filled with a mixture of (K,Na)NO,, KF and Ag,CrO, to make the melt saturated. The chromate ion also served as a sensitive leakage indicator, revealing any cracks or microscopic channels in the adhesive between the membrane and the glass body. The reference cell consisted of a Pyrex glass tube with a ceramic plug providing contact between the internal and external melts. The reference electrode was filled with (K,Na)NO, saturated in Ag,CrO,. The cell may be described as A typical Nernst plot at 280.0 "C is shown in fig. 1. In the fluoride-ion concentration range 5 x 10-,-7.4 x lo-, mol kg-l (the solubility limit of fluoride) the relation between the e.m.f., E, and log [F-] may be described as (2) E = 4 - k log {[F-]/(mol kg-l)} where E, = -97.1 f0.8 mV and k = 109.5 kO.3 mV, which compares well with the theoretical value [(RT/O In 10Itheor = 109.8 mV. The fluoride electrodes were re- calibrated in every experimental run.The values of k typically varied from 108.5 to 110.3 mV between the calibrations at 280 "C. The same precision and accuracy were obtained at the other temperatures studied. After the calibration was performed, solid Pb(NO,), or stock melt of KN0,-Pb(NO,), was added and the change in e.m.f. was measured. A stable e.m.f. (within f 0,l mV) was recorded in less than 30 min. The total time for calibration and Pb(NO,), titration was 12-18 h. The stability of the fluoride electrode was checked over 24 h and it was stable within k0.5 mV. After each series the electrodes were cooled and washed, and the LaF, membrane was remounted. The total concentrations were 1 x 6 C,/mol kg-' 6 7.4 x lo-, and 0.01 < C,,/mol kg-' < 1 .OO.The LaF, membrane worked very well as a lanthanum(II1)-ion electrode as well,308 Pb" Halides in Molten Alkali-metal Nitrate I I I I O t I I 1 I I 1 -5 -4 -3 -2 -1 log ([F]/mol kg-') Fig. 1. Nernst plot from a typical test run at 280 "C. The straight line was obtained from a least- squares fit of the data for [F-] 2 5 x mol kg-'. yielding k = 36.5 0.5 mV {[(RT/3F') In lOltheor = 36.6 mV> at 280 "C. The concentration range studied was 1.0 x lo-, < C,,/mol kg-l < 1.0 x lo-,. The use of a LaF, membrane for a lanthanum(rI1)-selective electrode has one drawback : the time to reach stable e.m.f.values may extend to 3 h after each addition of La(NO,),. Solubility Measurements Weighed amounts of (K,Na)NO, and Pb(NO,), with solid PbF, were equilibrated in Teflon vessels agitated by Teflon stirrers at 280.0 "C. Only Teflon was in contact with the melt since the presence of solid PbF, may cause corrosion of the Pyrex glass after a long exposure time. Samples were withdrawn from the melts through a glass filter using a preheated pipette. Equilibrium was attained within 72 h. The samples were solidified, weighed and dissolved in an aqueous hexamethylenetetramine solution with an excess of EDTA. The amount of lead@) in the samples was thereafter determined by back- titration with a Pb(NO,), solution using xylenol orange tetrasodium salt as indicator.The method of analysis was tested with samples of known composition and the values obtained differed by < 1 '/o from the known values. The total concentrations used were 0.553 < C,,/mol kg-' < 3.1 12 and 0.875 < C,/mol kg-l < 2.220. Results and Discussion Potentiometric Measurements The fluoride-ion electrode provides very accurate data of the activity of F- (or KF if it is assumed that the K+ activity remains unity during the experiment). Addition of Pb(NO,), to a melt containing fluoride may cause a change in the e.m.f. due to complex formation according to mPb2+ + nF- Pb,F2,m-n. (3)L. Bengtsson and B. Holrnberg 309 Table 1. Overall stability constants for PbmF2R"-" in (K,Na)NO,(l) at different temperatures, using the two activity models described in the text 240.0 122.5+ 1.5 6.16f0.61 57.2k2.6 122.1k1.4 6.10f0.59 46.5k2.6 41 250.0 116.6+1.1 5.05f0.44 52.8f2.1 115.5f1.1 5.15f0.46 43.5f2.3 26 260.0 107.6f0.8 4.95f0.23 47.7f1.4 106.9f0.6 4.90k0.18 38.6f1.2 20 270.0 101.4f1.2 4.67f0.33 43.5k2.3 100.8f1.2 4.65f0.31 34.8k2.2 65 280.0 95.0fl.l 3.99f0.20 40.2f1.9 94.3f1.1 3.82k0.20 31.7f2.0 85 'The standard state is 1 mol kg-'.The standard state is unity mole fractions, but the constants have been transformed into mol kg-l. 'Number of degrees of freedom. The error limits define a 95 % confidence interval. The equilibria may be expressed with either stepwise or overall formation constants (4 a) [ P b, FEm - "3 G n = [Pb2+] [Pb Fim-2-n] m-1 or (4 b) [ Pb, F :m - "1 K m n = [F-] [Pb,FiY;n+l] The complexation may be described as The complexation was mapped by titration with Pb(NO,), at different total fluoride concentrations.The experimental data were initially evaluated graphically in order to provide qualitative information on the system, i.e. to arrive at a correct complexation model that fits the experimental data. Values of k from the calibrations, total concentrations and e.m.f. data were treated using the least-squares program EMFALL. This program minimizes Z(Eexpt, - Eca,cd)2 for the best-fitting complexation model. The results of the computations are displayed in table 1. These results originate from the use of 1 mol kg-l as standard state for the species in equilibrium (3). The experimental data were also evaluated expressing all activities in mole-fraction units, i.e.using unity mole fractions as standard state, since the measurements have been extended to melts having a high concentration of lead(r1). The mole fractions are defined as cationic/anionic mole fractions : A T The mole fractions of uncomplexed lead(I1) and fluoride may be defined in an analogous manner. These definitions make it possible to analyse the complexation using310 Pb" Halides in Molten Alkali-metal Nitrate 50- - - 96 100 50 - 0 - 4.0 -3.0 -2.0 -1 .o log ([Pb2 '1 /mol kg- ' ) Fig. 2. The fraction a,, of F- in different Pb,F2nrn-, complexes for C, -+ 0 (a) and C;, = 63.2 x mol kg-' (b) in molten (K,Na)NO, at 280 "C. the same equations as when expressing activities in mol kg-'. These results are shown in table I also. The formation constants in table 1 reveal only a slight difference between the two activity models, making the choice of activity model (or standard state) a matter of taste.The qualitative results are independent of the activity model used. The stability constants are assumed to describe the complexation at infinite dilution. It is then possible to transform the /3 values between mole-fraction units and rnol kg-I merely by use of a proportional factor, irrespective of the activity model used when deriving these /3 values. Only the complexes PbF,, PbF+ and Pb,F3+ can be detected. No trace of further fluoride-containing species, like PbF;, PbF,2-, Pb,F; or Pb2F,2-, or any cationic polynuclear species, like Pb,Fi+, Pb2F: or Pb,Fz+, are found. Boiko studied the lead(I1) fluoride system in (K, Na)NO,(l).'O The formation of PbF+ Vll = 17 mol-' kg) and PbF, V12 = 425 rnol-, kg2) at 270 "C was established. These reported stabilities of PbF' and PbF, are considerably lower than those derived in the present study (table I).Since only five experimental values were used in the estimation of the stability constants from the polarographic study, they are somewhat dubious. The fractions amn of fluoride in PbmF2,m-n, defined as were evaluated for the limiting case = 63.2 x lop3 mol kg-l, representing the highest fluoride concentration used in the e.m. f. measurements. The distributions of fluoride in different complexes at these two limits are shown in fig. 2. An increase in fluoride concentration causes an increase of PbF, on the expense of PbF+ and F-, while the fraction of fluoride present as Pb2F3+ remains almost unaffected.The complex formation should be formulated as the exchange reactions: -+ 0 mol kg-' and Pbi-,(K/Na)j+, Fi-n+i+m-l +nPb2+ + Pbt(K/Na)j Fi+j-l +m(K/Na)+. (10)L. Bengtsson and B. Holmberg 31 1 Fig. 3. AGin uersus T for stepwise formation of Pb,F2,m-n in molten (K,Na)NO,. The /? values used refer to 1 mol kg-I as standard state, but were transformed into mole-fraction units in the evaluation of the thermodynamic parameters. The coordination of a lead@) ion to fluoride causes the release of one or more solvent cations (K/Na)+ from the positions adjacent to the fluoride ion. An equilibrium expression shows that either of the two activity models used in the evaluation of the stability constants can handle any value of rn and n, as long as the activity coefficients of the solvent cations (K/Na)+ remain constant.We evaluated the stability constants Pll and BZl by use of two different data sets, one pertaining to the concentration interval Gb < 0.2 mol kg-' and the other covering the whole range of qb up to 1.0 mol kg-l. Identical values of Pll and BZl were obtained from the two data sets (although the mean errors are, of course, somewhat larger for the smaller data set), implying that no serious effects of changes in activity coefficients are introduced in the region of high qb. According to Julsrud the main cause of the deviations from ideality in simple molten-salt mixtures with a common anion or cation is differences in size rather than differences in ~ h a r g e ., ~ , ~ ' The crystal radius of lead(r1) is between that of potassium and sodium, making it plausible to anticipate approximate ideal behaviour at the concentrations used. This anticipation is supported by cryoscopic results obtained by van Artsdalen and Doucet et al., showing ideal behaviour (to the Raoult-van't Hoff law) of Pb(NO,), dissolved in NaNO,(l) and KNO,(l) for concentrations lower than 0.20 mol kg-'. Severe deviations from ideality are observed, however, at higher concentrations of Pb(NO,), in pure KNO,(l). 62-64312 Pb" Halides in Molten Alkali-metal Nitrate Table 2. Standard enthalpy and entropy values of the stepwise formation of Pb,F2,"-" in the temperature range 240-280°C, for the two activity models described in the text (the errors quoted are two mean errors) reaction AW,,/kJ mol-Ia AS",,/JK-' mol-la AW,,/kJ mol-lb A%,/ JK-l mol-lb PbF' + F- + PbF, - 7.0 & 4.3 38.3 & 8.0 -9.2k3.9 34.2 f 7.4 29.7 f 0.8 Pb2+ + F- + PbF' - 15.3k0.6 30.0+ 1 .1 - 15.4 f 0.5 PbF' + Pb2+ + Pb,F3+ - 5.9 & 0.4 1.9 f 0.7 - 8.0 f 0.8 -3.9+ 1.6 a The activity model used refers to mol kg-', but the B values were transformed into mole-fraction units when evaluating the thermodynamic parameters. The activity model used refers to ionic mole fractions. The formation constants at different temperatures may be used to derive AH",, and ASO,, for each complexation step. The experimental runs were not designed to give accurate stability constants for PbF,, since our main interest was focused on the cationic part of the system.The constants for PbF, suffer, for that reason, from rather large errors. Fig. 3 shows the variation of the stepwise AGkn with temperature. The standard enthalpy and entropy values estimated for the two activity models are given in table 2, As expected, the formation of PbF+ yields a slightly more negative enthalpy value than the formation of PbF,. The Coulomb attraction between PbF+ and F- ought to be weaker than between Pb2+ (solvated by nitrate ions only) and F-. The enthalpy for formation of Pb2F3+ is slightly less negative, probably because the process involves the combination of two species with 'cationic' characteristics. The fact that A%, is negative at all might suggest that binding forces other than Coulomb interactions are important. ASO,, for formation of PbF+ and PbF, are remarkably large.Comparison with silver halide systems in molten alkali-metal nitrates show marked difference~.~~-~' The silver halide systems have entropy values only slightly larger than those predicted from an ideal configurational entropy change71 where 2 refers to the overall coordination number and j is the number of bound ligands. Assuming a coordination number of 4 for silver gives A$(conf) = 1 1.5 J K-' mol-'. The silver halide systems have experimental A g l = A q = 12-17 J K-' mol-', irrespective of the AKl value. If the coordination number for lead(r1) is assumed to be 4 as well, the experimental A g l and AT, are obviously too large to be explained by the simple statistical model. In the silver halide systems the contribution from other entropy effects, like changes in internal degrees of freedom or changes in number of nearest-neighbour ions, seem to be negligible. In the lead@) fluoride case, on the other hand, they seem to be important. Thomi has investigated several silver oxoanion systems that are not properly described by the purely statistical appr~ach.'~ Apart from the silver nitrite system the oxoanions reveal a trend of increasing negative enthalpies combined with increasing positive entropies.The silver phosphate system shows entropy values of the same size as ATl and A q , in the lead(@ fluoride system. The trend observed suggests that the harder a polyatomic anion is bound to the silver(1) ion the larger is the difference in freedom of vibration and of rotation between coordinated and bulk oxoanion.Stronger interaction between nitrate and lead@) than between nitrate and silver may be an explanation to the high A q l and AT, shown in table 2. The formation of Pb,F3+ in (K,Na)NO,(l) renders almost equal to zero.L. Bengtsson and B. Holmberg 313 Table 3. The total concentrations of lead(rr) and fluoride in mol kg-' and in cationic/anionic mole fractions for melts in equilibrium with solid PbF, at 280 "C. q,/mol kg-' CJmol kg-' lo3 X,, lo3 X, lo8 4 0.553 0.560 0.561 0.568 0.57 1 0.585 0.586 0.594 0.630 0.648 0.698 0.739 0.789 0.858 0.925 0.989 1.052 1.115 1.234 1.41 7 1.534 1.68 1 1.837 1.965 2.117 2.235 2.367 2.549 2.674 2.821 3.1 12 1.106 1..095 1.076 1.05 1 1.042 1.028 1.012 0.991 0.965 0.907 0.894 0.875 0.889 0.927 0.949 0.980 1.019 1.037 1.125 1.229 1.277 1.372 1.475 1.542 1.644 1.687 1.776 1.92 1 1.984 2.061 2.220 48.9 49.5 49.6 50.2 50.5 51.6 51.7 52.4 55.4 56.9 61 .O 64.3 68.4 73.9 79.3 84.3 89.2 94.0 103.0 116.5 124.9 135.3 146.0 154.6 164.6 172.2 180.5 191.7 199.2 207.9 224.5 93.3 92.3 90.7 88.4 87.6 86.3 84.9 83.0 80.4 75.3 73.6 71.6 72.1 74.4 75.3 77.0 79.3 79.9 85.1 90.5 92.4 97.2 102.3 105.1 109.7 110.9 114.7 121.2 123.3 125.8 130.8 8.46 8.33 8.14 7.83 7.72 7.46 7.28 6.98 6.42 5.66 5.19 4.76 4.62 4.60 4.46 4.42 4.45 4.3 1 4.49 4.50 4.39 4.49 4.59 4.56 4.66 4.52 4.60 4.8 1 4.76 4.72 4.66 The solubility product is evaluated in mole-fraction units, using the p-values from table 1, having unity mole fractions as standard state.Assuming a coordination number of 4 for fluoride, the ideal contigurational entropy change is Ag(conf) = 3.4 J K-l mol-', which is of the same magnitude as the experimental value.The experimental ASil is likely to contain large positive entropy contributions other than configurational, of the same type as observed in the formation of PbF' and PbF,. The configurational entropy change for the formation of Pb,F3+ is thus most likely considerably lower than the ideal one. A clearly negative configurational entropy value indicates that the inbinding of the second lead(r1) ion is geometrically restricted in some way. This may be a result of non-ionic interactions between the central fluoride and lead(rr), requiring a specific geometry for maximum orbital overlap. It is also possible that specific interactions between the two ligand lead(r1) ions should be considered.Metal-metal interactions in lead compounds have indeed been manifest in a number of Pbk- c l ~ s t e r s , ~ ~ - ' ~ for instance Pbt- and Phi-. Trinquier and Hoffmann have also shown that Pb-Pb interactions are vital for the structure and physical properties of a number of oxides containing lead(u).'*3 14 Pb" Halides in Molten Alkali-metal Nitrate XPb Fig. 4. The solubility of PbF,(s) in (K,Na)NO,(l) at 280 "C expressed in cationic/anionic mole- fraction units. The solid curve represents the complexation model evaluated from e.m.f. measurements using unity mole fractions as standard state. Solubility Measurements The solubility of PbF, in (K,Na)NO,(l) at 280.0 "C is remarkably high, probably due to complexation processes. The solubility of PbF, in excess of Pb(NO,), is shown in table 3 and fig. 4.The complexation model achieved from e.m.f. measurements in more dilute systems failed to describe the solubility data when using the molality as scale activity model. The high concentrations of both fluoride and lead@) might better be handled by cationic/anionic mole fractions as defined earlier. The complexation model applying mole fractions as activity model does indeed provide a good way of describing the solubility data, at least for melts having Xpb 2 xF (see fig. 4). In table 3 the solubility product for (13) K, = xpb*+(xF-)2 (14) has been evaluated using /3 values from the e.m.f. studies. The calculated solubility product remains almost constant over a large section of the solubility curve, giving an average < K, > = (4.57 & 0.14) x 1C8 in mole-fraction units.In the range XPb < XF the ionic mole fraction complexation model from the e.m.f. measurements cannot describe the solubility curve. Inclusion of species like PbF, and PbF:- improves the situation, but severe deviations remain. The deviations are most likely a result of formation of significant amounts of polynuclear, i.e. polyfluoric, species. Liquid X-ray scattering and Raman spectroscopy measurements, specifically aimed at a clarification of the structure of Pb2F3+ in (K,Na)NO, melts, are at present performed. PbF2(s) t Pb2 + 2F- Dr Arvid Sandell is greatfully acknowledged for his active support in the computations.This work was also supported by a grant from the Swedish Natural Science Research Council.L. Bengtsson and B. Holmberg 315 References 1 G. B. Kauffman, M. Karbassi and G. Bergerhoff, J. Chem. Educ., 1984, 61, 729. 2 B. Holmberg, Acta Chem. Scand. Ser A, 1976, 30, 680. 3 B. Holmberg, Acta Chem. Scand. Ser. A, 1976, 30, 797. 4 B. Holmberg and G. Johansson, Acta Chem. Scand., Ser. A, 1983, 37, 367. 5 T. Yamaguchi, G. Johansson, B. Holmberg, M. Maeda and H. Ohtaki, Acta Chem. Scand., Ser. A, 6 L. Bengtsson, B. Holmberg, A. Iverfeldt and I. Persson, Inorg. Chim. Acta, 1988, 146, 233. 7 B. Holmberg, XXZV Znt. Con$ Coord. Chem. (The Association of Greek Chemists, Athens, 1986), 8 L. Bengtsson and B. Holmberg, J. Chem. SOC., Faraday Trans. I , 1989, 85, 317.9 A. Bystrom, Ark. Kemi Miner. Geol., 1947, 24A. 1984, 38, 437. p. 152. 10 B. Aurivillius, Chem. Scr., 1976, 10, 156. 11 B. Aurivillius, Chem. Scr., 1977, 1, 208. 12 0. Savborg and M. Lundberg, Mat. Res. Bull., 1980, 15, 1433. 13 0. Savborg, Univ. Stockholm, Chem. Commun., 1983, paper no. 6. 14 A. Tairi, J-C. Champarnaud-Mesjard, D. Mercurio and B. Frit, Rev. Chim. Miner., 1984, 21, 680. 15 16 0. Savborg and M. Lundberg, J. Solid State Chem., 1985, 57, 135. 17 0. Savborg, J. Solid State Chem., 1985, 57, 143. 18 0. Savborg, J. Solid State Chem., 1985, 57, 148. 19 0. Savborg, J. Solid State Chem., 1985, 57, 154. 20 0. Savborg, J. Solid State Chem., 1985, 57, 160. 21 I. N. Belyaev, 0. Y. Revina and L. L. Pershina, Zh. Neorg. Khim., 1983, 27, 1558. 22 C. Sandonnini, Gazz. Chim.Ital., 1911, 41, 148; 151; 154. 23 W. Nieuwenkamp and J. M. Bijvoet, Z . Kristallogr., 1932, 81, 469. 24 W. Nieuwenkamp and J. M. Bijvoet, Z. Kristallogr., 1932, 82, 157. 25 A. Rulmont, C. R. Acad. Sci., 1973, 276, 775. 26 B. Aurivillius, Chem. Scr., 1976, 10, 206. 27 B. Aurivillius, Chem. Scr., 1980, 15, 153. 28 J. F. Ackerman, Mat. Res. Bull., 1982, 17, 883. 29 B. Aurivillus, Acta Chem. Scand. Ser. A, 1983, 37, 159. 30 R. Domesle and R. Hoppe, Z. Anorg, Allg. Chem., 1983, 501, 102. 31 J. A. A. Ketelaar, 2. Kristallogr., 1932, 84, 62. 32 P. Boldrini and B. 0. Loopstra, Acta Crystallogr., 1967, 22, 744. 33 I. I. Yamzin, Y. Z. Nozik and N. V. Belov, Sou. Phys. Dokl. (Engl. Transl.), 1961, 6, 370. 34 S. Julsrud and 0. J. Kleppa, Acta Chem. Scand., Ser.A , 1985, 39, 157. 35 B. G. Rao, H. G. K. Sundar and K. J. Rao, J. Chem. Soc., Faraday Trans, I , 1984, 80, 3491. 36 I. I. Ilyasov and A. G. Bergman, J. Gen. Chem. USSR (Engl. Transl.), 1956, 26, 1119. 37 B. S. Podgornova and W. T. Talipov, Uzb, Khim. Zh., 1961, 6, 15. 38 R. E. Connick and A. D. Paul, J. Am. Chem. SOC., 1958, 80, 2069. 39 S. S. Mesaric and D. N. Hume, Inorg. Chem., 1963, 2, 788. 40 E. Bottari and L. Ciavatta, J. Inorg. Nucl. Chem., 1965, 27, 133. 41 A. M. Bond and G. Hefter, Inorg. Chem., 1970, 9, 1021. 42 A. M. Bond and G. Hefter, J. Electroanal. Chem. Interfacial Electrochem., 1971, 31, 477. 43 A. M. Bond, Anal. Chim. Acta, 1971, 53, 159. 44 G. Hefter, J. Electroanal. Chem. Interfacial Electrochem., 1972, 39, 345. 45 A. M. Bond and G. Hefter, J.Electroanal. Chem. Interfacial Electrochem., 1973, 42, 1. 46 H. L. Clever and F. J. Johnston, J. Phys. Chem. Rex Data, 1980, 9, 751. 47 G. T. Hefter, C. B. Chan and N. H. Tioh, Anal. Chem., 1984, 56, 749. 48 S. Shi and H. Zhu, Fenxi Huaxe, 1986, 14, 817. 49 R. Beaudoin and H. Minard, Can. J. Chem., 1987, 65, 528. 50 Y. M. Korenev, A. N. Rykov, S. V. Kuznetsov, A. I. Boltalin and A. V. Novoselova, Zh. Neorg. 51 Y. M. Korenev, A. N. Rykov, S. V. Kuznetsov, A. I. Boltalin and A. V. Novoselova, Zh. Neorg. 52 A. D. Demidov, A. G. Gershikov, E. 2. Zasorin, V. P. Spiridonov and A. A. Ivanov, Zh. Strukt. 53 K . H. Stern, J . Phys. Chem. Re$ Data, 1972, 1, 747. 54 L. Domage, Ann. Chim. (Paris), 1937, 7, 225. 55 B. Holmberg, Acta Chem. Scand., 1973, 27, 875. 56 B. Holmberg, Acta Chem. Scand., Ser. A, 1976, 30, 641. L. Soubeyroux, S. F. Matar, J. M. Reau and P. Hagenmuller, Solid State Ionics, 1984, 14, 337. Khim., 1986, 31, 1832. Khim., 1986, 31, 2195. Khim., 1983, 24, 9.316 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 Pb" Halides in Molten Alkali-metal Nitrate B. Holmberg, Acta Chem. Scand., Ser. A, 1974, 28, 284. B. Holmberg and K. Jarring, J. Electroanal. Chem. Interfacial Electrochem., 1983, 146, 447. B. Holmberg and K. Jarring, Inorg. Chem., 1987, 26, 1713. K. M. Boiko, Ukr. Khim. Zh., 1969, 35, 596. S . Julsrud, Thesis (University of Trondheim, NTH, 1983). E. R. van Artsdalen, J. Phys. Chem., 1956, 60, 172. Y. Doucet and C. Janot, C. R. Acad. Sci., 1957, 244, 1166. Y. Doucet and C. Janot, C. R. Acad. Sci., 1957, 245, 1898. G. G. Bombi, G. A. Sacchetto and C. Macca, J. Electroanal. Chem. Interfacial Electrochem., 1973, 42, 373. D. G. Hill, J. Braunstein and M. Blander, J. Phys. Chem., 1969, 64, 1038. D. G. Hill and M. Blander, J. Phys. Chem., 1961, 65, 1866. A. Alvarez-Funes, J. Braunstein and M. Blander, J. Am. Chem. SOC., 1962, 84, 1538. 1. Elding and I. Leden, Acta Chem. Scand., 1969, 23, 2430. J. Braunstein and R. E. Hagman, J. Phys. Chem., 1963, 67, 2881. M. Blander, Molten Salt Chemistry, ed. M. Blander (Interscience, New York, 1964), p. 127. G. Thome, Thesis (University of Lund, 1980). J. D. Corbett and P. A. Edwards, J. Chem. SOC., Chem. Commun., 1975, 984. P. A, Edwards and J. D. Corbett, Znorg. Chem., 1977, 16, 903. R. W. Rudolph, W. L. Wilson, F. Parker, R. C. Taylor and D. C. Young, J. Am. Chem. SOC., 1978, 100, 4629. B. S. Pons, D. J. Santure, R. C. Taylor and R. W. Rudolph, Spectrochim. Acta, 1981, 26, 365. J. D. Corbett, Chemistry for the Future, ed. H. Grunewald (Pergamon, Oxford, 1983), p. 125. G. Trinquier and R. Hoffmann, J. Phys. Chem., 1984, 88, 6696. Paper 8/00871J; Received 4th March, 1988
ISSN:0300-9599
DOI:10.1039/F19898500305
出版商:RSC
年代:1989
数据来源: RSC
|
16. |
Cationic lead(II) halide complexes in molten alkali-metal nitrate. Part 2.—A thermodynamic investigation of the chloride, bromide and iodide systems |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 317-329
Lars Bengtsson,
Preview
|
PDF (857KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I, 1989, 85(2), 317-329 Cationic Lead(@ Halide Complexes in Molten Alkali-metal Nitrate Part 2.-A Thermodynamic Investigation of the Chloride, Bromide and Iodide Systems Lars Bengtsson and Bertil Holmberg" Inorganic Chemistry I , Chemical Center, University of Lund, P.O. Box 124, S-221 00 Lund, Sweden The complex formation between lead@) ions and chloride, bromide and iodide ions in molten equimolar (K,Na)NO, has been studied. The systems were investigated at four temperatures between 240 and 300 "C. The iodide system was only studied at one temperature, since I- is oxidized to I, in melts rich in Pb(NO,),. Halide-ion activities were measured in the systems (K+, Na+, Pb2+)-(NO;, X-), X-= C1-, Br- and I-, by means of Ag/AgX electrodes. The formation of PbX,, PbX+ and Pb,X3+ was observed.Special interest was focused on the formation of cationic complexes with anions as coordination centres. The thermodynamic parameters AH",, and Arm,, have been evaluated for the chloride and bromide systems. The results are: AK, = -7.9 kJ mol-I, AT, = 26.3 J K-' mo1-1 for Pb2++C1- + PbCl'; A T l = -4.6 kJ mo1-', AT, = 35.2 J K-' mol-' for Pb2++ Br- + PbBr+; A%, = 1.5 kJ mol-I, A%, = 16.1 J K-l mol-' for PbCI' +Pb2+ + Pb,C13+; AGl = - 7.1 kJ mol-', A q l = 1.9 J K-I mol-' for PbBr+ + Pb2+ + Pb,Br3+. The parameters are compared with previous literature data and recent results obtained from the lead@) fluoride system in (K,Na)NO,(l). The formation of Pb,F3+ in alkali-metal nitrate based melts has been reported in the previous paper.' The thermodynamic results indicate that Pb,F3+ might be stabilized by P b P b interactions.An investigation of the influence of increasing covalent character of the lead@) halide bond, from fluoride to iodide, may contribute to the understanding of this interesting class of complexes with anions as coordination centres., Lead(1r) oxide and hydroxide systems as well as PbO * PbX, mixtures exhibit a variety of discrete and extended cationic complexes in the solid state. Most of these mixtures are described as layers or chains of cationic [PbOH+], or [PbO], entities surrounded by halide ions, but the coordination of halide ions by cations, i.e. lead@), is certainly also an important structural feature of several compounds. 3-15 A number of mixed lead(rr) halide sulphide and selenide compounds with similar structures have also been reported.16 Many investigations on the structure of PbCl, have been performed."-24 The stable low-temperature modification is orthorhombic and the coordination of chloride ions may be described in terms of Pb,Cl tetrahedra and Pb3Cl pyramids. In PbBr, two crystallographically different bromide ions are surrounded by four and five lead(r1) ions, re~pectively.~~ The PbBr, structure is geometrically similar to the layered structure of PbI,.The PbI, structure is complicated by structural polytypism, first described by Mitchell26 and recently reviewed by Palosz and co-w~rkers.~' The coordination of chloride in molten PbCl, has been interpreted in terms of distorted Pb,Cl tetrahedra, explaining the observation of different Pb-Pb distances in the radial distribution function, obtained from liquid X-ray scattering measurements.317318 Pb" Halides in Molten Alkali-metal Nitrate Rao and co-workers investigated the coordination in mixed PbO-PbC1, glasses using X-ray scattering, XANES and EXAFS.29-31 Structure elements such as Pb40 tetrahedra and various Pb,C1 entities were used to describe the results. Fused PbC1,-MCl systems have been examined both thermodynamically and structurally. In liquid PbC1,-LiC1 mixtures Pb-Pb interactions at distances comparable to those in pure molten PbCl,,, have been observed, indicating related structure elements in the melts.28 The PbBr,-MBr and Pb1,-MI systems have also been subject to interest, but only a few brief reports on polynuclear species have appeared in the literat~re.~,-,~ Phase diagrams of mixed lead(n) halides reveal a few mixed solid halides.For some of these the crystal structures have been s o l ~ e d . ~ ~ - ~ ' Electron diffraction and Raman spectroscopy studies on gas-phase and matrix-isolated PbX, and mixed PbX,-PbY , halides show bent structure elements with X-Pb-X or X-Pb-Y angles of ca. 94°.40-42 The complexation of lead@) with halide ions in molecular solvents, has been summarized recently.43 A few reports on the formation of ' polynuclear ' complexes Pb,X2,n-n have Yatsimirskii et al. and Fedorov et al. claim that M,X3+ ions (M = Pb, Cd; X = C1, Br, I or SCN) are formed in aqueous s o l u t i o n ~ , ~ ~ , ~ ~ the cadmium(n) complexes being stronger than the corresponding lead(rr) complexes.An increasing stability with increasing size of the halide ion is also observed. Magnetic measurements on mixed lead(1r) halides have indicated the presence of Pb,13+ complexes in the PbI, - 3PbCl,, PbI, - 3PbBr,, 2Pb1, * 3PbC1, and 2Pb1, - 3PbBr, phases.48 The lead(n) halide systems have been extensively investigated in molten nitrate media. Most experiments have been performed with an excess of halide over lead@), though. The experimental data from such studies only reveal PbX+, PbX, and anionic species. Many experimental methods have been used involving e.m.f. measurements with Ag/ AgX e l e c t r ~ d e s , ~ ~ - ~ ~ Pd/PdO/PbO e l e c t r ~ d e s ~ ~ * ~ ~ and Pb electrodes,60 solubility of AgX(s)'l and of pbCrO,(~),~'-~~ freezing-point 66 and polarography. 67-70 The results are summarized in table 1.The use of metallic lead as a lead(1r) indicator electr~de~~~*O is bound to cause erroneous results owing to severe oxidation of the metal in nitrate melts, especially in the presence of such a good oxide-ion acceptor as lead(~~).~l Gaur and co-workers reported the use of a Pd/PdO/PbO electrode for measuring the lead@) activities in molten KNO, - Ba(N03), mixtures,58* 59 by analogy with the Cd" electrode used by Inman.72 The electrode function was apparently puzzling because the experimental Nernst slope was considerably larger than the expected theoretical one. The calculated theoretical slope was based on the assumption that PbO is the solid phase containing lead(Ir) in equilibrium with the nitrate melt.The formation of the solid 5Pb0.Pb(N03), is, however, of great importance in molten alkali-metal nitrate systems containing PbO and Pb(N0,),.73*74 The assumption that the solid containing lead(Ir) is 5Pb0 Pb(NO,), would give theoretical values which are 6 / 5 larger than those considered by the authors. The corrected theoretical slopes coincide almost perfectly with the experimental ones. The electrode thus seems to work well in molten nitrate media, but it obviously suffers from the drawback that it cannot be used at high lead(I1) concentrations owing to the extensive solubility of PbO and/or 5Pb0 - Pb(NO,), in such melts7' The present study is devoted to the complexation of lead(I1) with chloride, bromide and iodide in molten equimolar (K,Na)NO,.The emphasis has been placed on the formation of cationic complexes with the halides as coordination centres for lead(I1) ions. Hence, we have chosen to measure the changes in halide activity as a function of lead@) concentration using Ag/AgX electrodes and applying the proper corrections for the changes in silver halide solubility and complexation. The measurements are extended over a temperature interval 240-300 "C in order that AW,, and AP,, for the association processes be estimated. These thermodynamic parameters are of great value for the understanding of the fundamental coordination chemistry of this class of complexes with anions as coordination centres.L. Bengtsson and B. Holmberg 319 Table 1. Overall stability constants of PbX2,-" in molten nitrate media physical P14 ref.solvent melt halide method T/OC @,, @I2 p13 (K0.25, Na0.75)N03 6 0 . 5 9 Na0.5)N03 NH4N0,. 1 .5H20 NH4N03 * 3H,O NH4N03*2H20 NH,N03 * l.5H2O NH4N03. 3H,O (K0.876, Ba0.062)N03 Br c1 C1 Br c1 Br Br c1 Br c1 Br c1 Br I c1 AgIAgX electrode AgIAgX Agl AgX electrode electrode Agl AgX electrode AgIAgX electrode AgIAgX electrode electrode electrode electrode Ag/AgX AgIAgX Pd/PdO/ PbO electrode AgIAgX electrode Pb electrode 255 303 306 319 325 250 275 300 280 300 240 280 300 280 300 I60 180 200 160 200 275 300 230 250 230 250 40 55 70 70 70 40 55 70 70 70 315 335 335 295 315 335 295 315 335 280 198 153 121 1.02 x 104 49 75 50 175 173 143 180 1 . 4 0 ~ lo4 250 3 . 1 2 ~ lo4 190 1 . 6 2 ~ lo4 170 1 . 1 9 ~ lo4 175 1 . 1 7 ~ lo4 250 2 .7 5 ~ lo4 230 2 . 1 4 ~ lo4 205 1 . 8 4 ~ lo4 990 3 . 9 6 ~ lo5 730 2 . 1 9 ~ lo5 215 204 160 1 . 1 2 ~ 104 200 1 . 6 4 ~ 104 51 52 53 54 500 7 . 4 6 ~ lo4 3 . 2 4 ~ lo6 7.21 x lo7 55 400 4.82 x lo4 1.46 x lo6 2.09 x lo7 526 8 . 4 9 ~ lo4 426 5 . 1 3 ~ lo4 72 3 . 1 0 ~ lo3 72 2 . 7 4 ~ lo3 71 2.41 x lo3 89 2 . 9 4 ~ lo3 54 2.11 x 103 123 3 . 9 4 ~ 103 110 5.06x 103 101 4.85x 103 121 1 . 0 0 ~ 104 77 4.47 x 103 79 2.61 x 103 73 1 . 9 0 ~ 103 69 1 . 4 5 ~ lo3 169 1 . 4 2 ~ lo4 152 1 . 1 6 ~ lo4 137 0 . 9 5 ~ lo4 3687 5.42 x lo6 2900 2 . 8 4 ~ lo6 2383 1.93 x lo6 226 1.46 x lo4 56 57 58, 59 60320 Pb" Halides in Molten Alkali-metal Nitrate Table 1. (conr.) physical solvent melt halide method T/OC p,, P,, P13 P,, refa KNO, c1 Br (K0.59 Na0.5)N03 c1 Br LiClO, c1 (K0.57 Na0.5)N03 c1 NaNO, c1 NaNO, c1 (K0.614, Li0.386)N03 c1 Br (K0.57 Na0.5)N03 c1 (K0.53' Na0.17, Li0.30)N03 c1 Br AgX PbCrO, solubility solubility PbCrO, solubility PbCrO, solubility cry0 scopy cryoscopy polaro- graphy polaro- graphy polaro- graphy polaro- graphy 480 450 250 275 300 250 275 300 275 300 275 275 306 306 180 280 145 145 160 179 207 25 4.95 x lo2 66 4.95 x lo2 193 4.16 x lo3 8.94 x lo4 86 2.77 x lo3 2.98 x lo4 64 2.08 x lo3 2.23 x lo4 140 0.30 x lo4 0.32 x lo5 282 2.39 x lo4 1.12 x lo6 470 4.42 x lo4 1.87 x lo6 140 9.01 x lo3 3 .8 7 ~ lo5 193 1.04 x 104 1.12 x 105 118 0.25 x 104 0.55 x 105 1 18 5.08 x 103 2.73 x 105 4.20 x 103 494 i.74x 104 706 226 7 . 2 8 ~ lo3 247 3 . 2 0 ~ lo4 870 2.46 x lo5 659 376 212 61 62 63 64 65 66 67 68 69 70 All constants are expressed in mole-fraction units.Experimental Chemicals KNO,, NaNO,, Pb(NO,),, KCl, NaCl, KBr, KI, NaI (all Merck, PA) and NaBr (J. T, Baker, Analyzed Reagent) were dried at 130 "C during at least one week before use and were stored over drying agents. The drying procedure was checked by heating the chemicals in vacuum for 6 h, revealing no further loss of weight. The AgCl (Riedel de Haen, 99.6 %), AgBr (Merck, 99 O/O) and AgI (Venton, 99.9 O h ) were dehydrated over anhydrous Mg(ClO,),. Apparatus The furnaces, temperature control, cell construction and measurements have been described previously. 75-77 E.M.F. Measurements The use of silver electrode^?^ in combination with solid silver halide is a well established method for measuring the halide-ion activity in molten ~alts.~'.~'.In the experiments a platinum rod with a fresh silver layer [electrolysed in an aqueous dicyanoargentate(1)L. Bengtsson and B. Holmberg 32 1 Ag (K,Na)NO,(l) (K,Na)NO3(1) AgX(s) AgX(s) Ag (K,Na)X (K,Na)X 1 Pb(N03), Results and Discussion The Ag/AgX electrode provides accurate and consistent information about changes in the halide-ion activity. Additions of Pb(NO,), to the (K+, Na+)--(NO,, X-) melts cause changes in halide activities which may be ascribed to complex formation. The e.m.f., however, involves effects from changes in the silver halide solution chemistry as well. Corrections for the solubility of AgX(s) and complex formation between silver(1) and the halides have to be performed. The influence from mixed lead(r1) silver halide complexes is assumed to be negligible.Before describing the corrections made for the silver halide systems some definitions have to be made. The equilibria AgX(s) + Ag+ + X- K t g = [Ag+][X-] mAg+ + nX- Ag, Xr-"322 Pb" Halides in Molten Alkali-metal Nitrate define the silver halide speciation. Complexation by lead(@ is expressed in the same way mPb2+ + nX- Pb, X:m-n When titrating the test melt with Pb(NO,), the solubility of AgX(s) and relative abundance of different Agm Xr-" complexes are changed due to changes in the activity of free halide. The activity of free halide ions is, in turn, changed owing to complex formation between lead@) and the halides. The e.m.f. measurements may then be split into two separate cases. The sums used below involve all complex ions, i.e.the starting value in the summations is 1 for both m and n. (1) No added Pb(NO,), : where Cx = concentration of halide calculated from added amount of (K,Na)X and From a knowledge of the Ktg and PAg values of the AgX ~ y ~ t e m ~ ~ ~ , ~ ~ . ~ the free halide ion concentration, [X-I0, when no Pb(NO,), has been added, can be evaluated. An e.m.f. E(Cp, = 0) is measured: E(cpb = 0) = E, - k log ([X-],/mOl kg-l). ( 5 ) (2) Pb(NO,), is added: = c"X + SAgX * (6) Assuming a theoretical one-electron slope, case (1) provides E,. The e.m.f. E(cpb) at each melt composition now gives the free halide ion concentration m-1. For the lead@) halide complexation, the sum = [x-] (1 + n/3,,[Pb2+]"[X-]"-') = [X-IY m, n is of fundamental interest.This sum may be evaluated taking the difference where (7)L. Bengtsson and B. Holmberg 323 Table 2. Overall stability constants for PbmX:mln-n in (K,Na)NO,(l) at different temperatures halide T/"C &/kg mol-la PI2/kg2 mo1-2a &, / kg2 mo1-2a no.' c1 240.0 260.0 280.0 300.0 Br 240.0 260.0 280.0 300.0 I 280.0 14.13 f 0.40 12.85f0.16 12.49f0.10 11.50f 0.11 18.43k0.35 18.36 f 0.32 17.31 f0.26 16.61 f 0.47 85.4 f 9.8 71 .O f 2.8 6.43 f 0.80 67.4 f 3.2 5.79 f 0.38 60.0+ 1.1 5.73 k0.34 50.1 & 1.4 5.42 _+ 0.49 197.6 & 6.5 11.46f 1.86 117.1 f 5.3 10.50f1.19 112.8 & 3.4 9.47 f 1.76 79.1 + 5.8 8.63 f 2.66 (1.27k0.22) x lo5 (2.9+ 1.8) x lo2* 35 23 49 27 33 50 46 27 44 a The standard state is 1 mol kg-', but the formation constants may be expressed in mole-fraction units after multiplication with (10.747 mol kg-l)m+n-l, * The value is estimated from a few points at high lead@) concentrations and merely represents the order of magnitude of the stability of Pb213+.cNumber of degrees of freedom. The error limits define a 95 % confidence interval. The lead(rr) halide complexation may be described either by overall stability constants, eqn (3b), or by stepwise constants Kmn according to or [ P b ,XEm -7 Kmn = [ Pb2 + J [ P b, - X :m- 2-n J [Pb,X im-7 K n n = [F-] [PbmX2,r2;n+1] ' The correction term E,,,, may be calculated for each melt composition from eqn (9), since fi-3 and the equilibrium constants for the silver halide systems are known. Corrected experimental data were fitted to eqn (7) and (8) by use of a least-squares program EMFALL, as described previously.' The results obtained at four different temperatures are displayed in table 2.Only the complexes PbX,, PbX' and Pb2X3+ were detected for all three halides, analogous to the lead(rr) fluoride system. No higher anionic or cationic complexes, nor any polynuclear, i.e. polyhalide, species (apart from PbX,) were found. The fraction amn of the different species is defined as and may be used to elucidate the distribution of halide ions on various complexes as a function of lead(rr) concentration. Such curves are displayed in fig. 1, using typical total halide concentrations for each system. The experiments were designed to provide detailed information of the cationic metal complexes. This resulted in inaccurate values of the formation constants for PbX,.These constants more or less served as correction factors in the computations of the other stability constants. Since lower concentrations of Pb(NO,), have been used in the chloride, bromide and especially in the iodide systems than in the fluoride case,l the values of the stability constants are expressed in molality units. The use of mole-fraction units hardly changed the results, apart from a proportional factor.324 Pb" Halides in Molten Alkali-metal Nitrate % log ([Pb2']/mol kg-') Fig. 1. The distribution amn of X- in the complexes PbmXtm-n in molten (K,Na)NO, at 280 "C. (a) The chloride system, C,, = 0.1 mol kg-' ; (b) The bromide system, C,, = 0.1 mol kg-'; (c) The iodide system, C, = 1 x mol kg-'. The stability constants obtained for PbX' and PbX, agree very well with the literature values for (K,Na)NO,, melts in table 1.The formation constants in (K,Li)NO,(l) are somewhat higher. In molten KN0,-Ba(NO,), the complex formation is more pronounced than in (K,Na)NO,(l) in spite of the higher temperatures used. The stronger complexation in these molten nitrates might be ascribed to the reciprocal Coulomb effe~t.~'.~* The complexation in aqueous NH,NO, melts is weak, most likely due to a strong solvation of lead(n) by water molecules. The thermodynamic parameters AH",, and AS:, for the chloride and bromide complexation reactions have been estimated from the temperature variation of AGin = -RTln Kmn. From fig. 2 and 3 it is seen that AG;l and AG;, are good linear functions of T for both systems.The standard enthalpy and entropy changes have been estimated by linear-regression analysis as temperature independent constants in the actual temperature range. The results are given in table 3. Selected values from table 1 have been used to evaluate the enthalpies and entropies for the formation of PbX+ in table 4. The parameters of table 4 and our results all show large differences. The results obtained by PbCrO, solubility seem somewhat strange, since both A g , and A%, haveL. Bengtsson and B. Holmberg 325 I I I I -20- - - u a -23- I 1 I I I I 1 I -6 - - & - j - 7 - - - 8 - 0- 8 - 9 - - I I I 1 510 530 550 570 T / K Fig. 2. AG:,, uersus T for stepwise formation of PbCl+ and Pb2C13+ in molten (K, Na)NO, in the temperature range 240-300 "C.The standard state is unity mole fraction. 510 530 550 570 T/K Fig. 3. AG:,, uersus T for stepwise formation of PbBr' and Pb2Br3+ in molten (K, Na)N03 in the temperature range 240-300 "C. The standard state is unity mole fraction. large negative values.62 The reason for this deviation might arise from the experimental difficulties of using a spectrophotometric method. The chloride system explored using Ag/AgX electrode e.m.f. measurements agree very well with our data,51 but the bromide system deviates markedly.52 The deviating parameters for the bromide system may be due to the neglection of corrections for the AgBr complexation and solubility or the use of a complexation model disregarding the formation of Pb2Br3+. Both these effects are far more important in the bromide case than for the chloride.The AISomn values found for the formation of PbF' and PbF, were interpreted as a326 Pb" Halides in Molten Alkali-metal Nitrate Table 3. Standard enthalpy and entropy values of the stepwise formation of Pb,XE"-" in molten (K,Na)NO, between 240 and 300 "C reaction AW,,/kJ mol-l AP,,/J K-' mol-la Pb2+ + CI- + PbCl' -7.9f 1.0 26.3 & 1.9 PbCI+ + Pb2+ + Pb,C13+ 1.5f0.8 16.1 f 1.4 Pb2+ + Br- 4 PbBr+ -4.6f 1.0 35.2 1.9 PbBr' + Pb2+ + Pb2Br3+ -7.1 f0.6 1.9f 1.1 "The standard state refers to unity mole fraction. The errors quoted are two mean errors obtained from linear regression. Table 4. Standard enthalpy and entropy values of the formation of PbX+ calculated for different molten nitrate solvents AKJJ K-l mol-la ref.temperature A%'/ solvent melt halide range/"C kJ mol-' (K,Na)NO, C1 Br 25&300 250-300 240-300 250-300 16&200 160-200 3 15-355 295-335 295-33 5 40-70 40-70 - 10.3 f 5.4 - 54.4 f 13.7 - 15.9k0.3 -24.5k3.9 -8.5f 1.0 - 13.0b - 10.4 f 0.8 - 15.1 f0.2 -31.3f 1.2 - 0.4 & 0.3 - 5.9 f 0.3 23.6 k9.9 15.0f0.6 26.4 f 2.2 27.4O 18.7k1.3 16.1 k0.4 13.1 f2.0 34.2 k 0.8 21.3 f 0.9 - 60.9 f 24.9 - 3.2 f 7.2 51 62 52 62 53 53 58, 59 58, 59 58, 59 57 57 a The standard state is unity mole fraction for all nitrate solvents. Only values at two temperatures are available. The errors quoted are two mean errors obtained from linear regression. result of a significant entropy contribution from nitrate i0ns.l These contributions were attributed to differences in rotational and vibrational degrees of freedom between nitrate ions coordinated to lead(1rj and to solvent cations.This interpretation is supported by the high A g l values which are also found for the formation of PbCl+ and PbBr+ in molten (K,Na)NO,. The presence of lithium cations and to a lesser extent barium cations would be expected to cause a lower A q l , since both ions probably interact more strongly with nitrate ions than sodium and potassium ions. A smaller difference in degrees of freedom between bulk nitrates and nitrates coordinated to lead(n) would yield a smaller contribution to A g l . Lower values of A g l are indeed found in (K,LijNO,(l) and to a lesser extent in KN0,-Ba(NO,),(l). The ammonium ion, on the other hand, is often compared with the potassium ion regarding polarizability and size.High A g l values are thus expected and found. The stabilities of the PbX+ complexes, X = F, C1 and Br, exhibit a minimum for X = C1. The cause of this minimum is the TAT, contribution, while the - A w l values decrease in the order F > C1 > Br. This observation is not surprising though, bearing inL. Bengtsson and B. Holmberg 327 mind the unique physical properties of the fluoride ion compared to the other halides, for which a more continuous change is observed. The thermodynamic parameters for the process PbX+ + Pb2+ e Pb2X3+ are sometimes uncertain owing to the rather high relative errors in the stability constants. It is obvious, though, that the coordination of a second lead@) ion by bromide is energetically more favourable than by chloride, as indicated by the AGl values.A very low A%, value (of the same magnitude as A%, for Pb,F3+) is found for Pb2Br3+, indicating the coordination of the second lead(r1) to be geometrically restricted. Since the bromide ion is far more polarizable than the fluoride ion, increased covalency of the Pb-X bonds might stabilize the Pb2Br3+ ion. The low A%l value of the fluoride system was interpreted in terms of P b P b interactions between coordinated lead(1r) ions stabilizing the Pb,F3+ ion. Pb2C13+, on the other hand, shows a larger A%, and a positive AG,. The positive thermodynamic parameters for Pb2C13+ might be due to a maximum in electrostatic Pb2+-Pb2+ repulsion. In Pb2F3+ the repulsion may be reduced without a large loss of stabilizing energy, merely by increasing the Pb-Pb distance, since the Pb-F bond is essentially electrostatic.In Pb2Br3+ the repulsion may be offset by a larger donation of electron density to the coordinated lead(I1) ions, since the Pb-Br bond is essentially covalent. The chloride system is intermediate to the fluoride and bromide system, and the thermodynamic parameters may reflect the inadequacy of either mechanism for reduction of the electrostatic Pb2+-Pb2+ repulsion. These hypotheses about the origins of the thermodynamic patterns are of course at the present stage of rather a speculative nature. The ideas are at the moment tested in structural investigations on the lead@) halide systems in molten (K,Na)NO, at 280 "C by means of liquid X-ray techniques and Raman spectroscopy.Dr Arvid Sandell is greatfully acknowledged for his extensive support in the computations. Mr Filip Frostemark is also acknowledged for performing part of the experimental work and treatment of data. This work was supported by a grant from the Swedish Natural Science Research Council. References 1 L. Bengtsson and B. Holmberg, J. Chem. Soc., Faraday Trans. I , 1989, 85, 305. 2 G. B. Kauffman, M. Karbassi and G. Bergerhoff, J. Chem. Educ., 1984, 61, 729. 3 N. Demassieux, C . R. Acad. Sci., 1933, 196, 1683. 4 0. Gabrielson, Arkiv. Miner. Geol., 1957, 2, 299. 5 H. P. Julien and R. E. Ogilvie, J. Am. Chem. Soc., 1960, 82, 293. 6 S. S. Malcic and M. S. Zivadinovic, Bull. Inst. Nucl. Sci. Boris Kidrich, 1960, 10, 47. 7 M. Gasperin, Bull. Soc. Fr. Miner.Crist., 1964, 87, 278. 8 R. Nasanen, P. Merilainen, R. Uggla and M. Hyle, Suom. Kemi., 1964, B37, 45. 9 R. Nasanen, Suom. Kemi., 1966, B39, 105. 10 C. K. Msller, Mat. Fys. Medd. Dan. Vid. Selsk., 1966, 35, no. 5. 1 1 G. Perrault, H. Vincent and M. Renaud, Bull. SOC. Fr. Miner. Crist., 1971, 94, 108. 12 H. Vincent and G. Perrault, Bull. SOC. Fr. Miner. Crist., 1971, 94, 323. 13 P. Tsai and R. P. Cooney, J . Chem. Soc., Dalton Trans., 1976, 1631. 14 B. Aurivillius, Chem. Scr., 1982, 19, 97. 15 V. Kramer and E. Post, Mater. Res. Bull., 1985, 20, 407. 16 B. Krebs, 2. Anorg. Allg. Chem., 1973, 396, 137. 17 H. Brrekken, 2. Kristallogr., 1932, 83, 222. 18 M. Straumanis and J. Sauka, 2. Phys. Chem., 1942, B51, 219. 19 E. Mooser and W. B. Pearson, Acta Crystallogr., 1959, 12, 1015.20 K. Sahl and J. Zemann, Naturwissenschaften, 1961, 48, 641. 21 R. L. Sass, E. B. Brackett and T. E. Brackett, J . Phys. Chem., 1963, 67, 2863.328 Pb" Halides in Molten Alkali-metal Nitrate 22 K. Sahl, Beitrage Miner. Petrogr., 1963, 9, 11 1. 23 Y. Z. Nozik and L. E. Fykin, Geokhimiya, 1975, 948. 24 Y. Z. Nozik, L. E. Fykin and L. A. Muradyan, Kristallograjiya, 1976, 21, 38. 25 W. Nieuwenkamp and J. M. Bijvoet, Z . Kristallogr., 1932, 49, 4. 26 R. S. Mitchell, 2. Kristallogr., 1959, 111, 372. 27 B. Palosz, S. Gierlotka, B. Wiktorowska and D. Dziag, Acta Crystallogr., Sect. C , 1985, 41, 1407. 28 H. Morikawa, Y. Miyake, Y. Takagi and F. Marumo, J. Chem. Soc., Faraday Trans. 1, 1981, 77, 29 B. G. Rao and K. J. Rao, Phys. Chem. Glasses, 1984, 25, 11.30 K. J. Rao, J. Wong and B. G. Rao, Phys. Chem. Glasses, 1984, 25, 57. 31 K. J. Rao and J. Wong, J. Chem. Phys., 1984, 81, 4832. 32 H. Ohno, M. Yoroki, K. Furukawa, Y. Takagi and T. Nakamura, J. Chem. Soc., Faraday Trans. 1, 33 M. F. Lantratov and T. N. Shevlyakova, Russ. J. Znorg. Chem. (Engl. Transl.), 1959, 4, 523. 34 M. F. Lantratov and 0. F. Moiseeva, Russ. J. Phys. Chem. (Engl. Transl.), 1960, 34, 171. 35 I. G. Murgulescu and S. Zuca, Rev. Roum. Chim., 1965, 10, 129. 36 I. G. Murgulescu, S. Sternberg and M. Terzi, Electrochim. Acta, 1967, 12, 1121. 37 I. I. Ilyasov and A. G. Bergman, Russ. J. Znorg. Chem. (Engl. Transl.), 1961, 6, 1094. 38 H. Hirsch, J , Znorg. Nucl. Chem., 1978, 40, 1623. 39 L. H. Brixner, H-Y. Chen and C. M. Foris, J. Solid State Chem., 1981, 40, 336.40 M. W. Lister and L. E. Sutton, Trans. Faraday Soc., 1941, 37, 406. 41 G. A. Ozin and A. V. Voet, J. Chem. Phys., 1972, 56, 4768. 42 I. Hargittai, J. Tremmel, E. Vajda, A. A. Ishchenko, A. A. Ivanov, L. S. Ivashkevich and V. P. 43 H. L. Clever and F. J. Johnston, J. Phys. Chem. ReJ Data, 1980, 9, 751. 44 V. E. Mironov, Russ. J. Inorg. Chem. (Engl. Transl.), 1961, 6, 458. 45 E. A. Gyunner and N. D. Yakhkind, Ukr. Khim. Zh., 1970,36, 147. 46 K. B. Yatsimirskii and A. A. Shutov, Zh. Fiz. Khim., 1953, 27, 782. 47 V. A. Fedorov, N. P. Samsonova and V. E. Mironov, Russ. J , Znorg. Chem. (Engl. Transl.), 1970, 15, 48 I. Guillaume, Bull. Soc. Chim. Fr., 1960, 494. 49 F. R. Duke and H. M. Garfinkel, J. Phys. Chem., 1961, 65, 1627. 50 C. Sinistri, Ric. Sci., 1962, 2, 638.51 M. Rolla and C. Sinistri, Ann. Chim. (Ztal.), 53, 535. 52 D. L. Manning, M. Blander and J. Braunstein, Znorg. Chem., 1963, 2, 345. 53 J. Braunstein and A. S. Minano, Znorg. Chem., 1964, 3, 218. 54 C. Sinistri and E. Pezzati, Ric. Sci., 1965, 8, 979. 55 F. F. Grigorenko and A. A. Suprunenko, Vestn. Kiev Univ., Ser. Khim., 1969, 9, 67. 56 F. F. Grigorenko and A. A. Suprunenko, Vestn. Kiev Univ., Ser. Khim., 1969, 10, 73. 57 R. M. Nikolic and I. J. Gal, J. Chem. Soc., Dalton Trans., 1972, 162. 58 R. K. Gupta and H. C. Gaur, J. Chem. Soc., Faraday Trans. I , 1982, 78, 7. 59 H. C. Gaur, R. K. Gupta and K. W. D.'Verma, Electrokhimiya, 1983, 19, 67. 60 Y. K. Delimarskii and V. F. Grishchyenko, Ukr. Khim. Zh., 1963, 29, 507. 61 J. H. Kennedy, J. Phys. Chem., 1963, 67, 1432.62 F. R. Duke and M. L. Iverson, J. Phys. Chem., 1958, 62, 417. 63 F. R. Duke and W. W. Lawrence, J. Phys. Chem., 1959, 63, 2087. 64 J. D. van Norman and R. A. Osteryoung, J. Phys. Chem., 1962, 66, 1565. 65 E. R. van Artsdalen, J. Phys. Chem., 1956, 60, 172. 66 J. Braunstein, M. Blander and R. M. Lindgren, J. Am. Chem. SOC., 1962, 84, 1529. 67 J. H. Christie and R. A. Osteryoung, J. Am. Chem. Soc., 1960, 82, 1841. 68 Y. K. Delimarskii and V. F. Grishchyenko, Ukr. Khim. Zh., 1963, 29, 502. 69 D. Inman, D. G. Lovering and R. Narayan, Trans. Faraday Soc., 1968, 64, 2476. 70 M. F. Francini and S. Martini, Polurography, ed. G. J. Hills (Macmillan, London, 1966). 71 L. Bengtsson and B.Holmberg, to be published. 72 D. Inman, Electrochim. Acta, 1965, 10, 11. 73 M.-J. Salaha, J. Gilbert and P. Claes, J. Chim. Phys. Chim. Biol., 1982, 79, 63. 74 P. P. Pope1 and S. N. Khoroshevskaya, Russ. J. Znorg. Chem. (Engl. Transl.), 1985, 30, 109. 75 B. Holmberg, Acta Chem. Scand., 1973, 27, 875. 76 B. Holmberg, Acta Chem. Scand., Ser. A, 1976, 30, 641. 77 B. Holmberg, Acta Chem. Scand., Ser. A, 1974, 28, 284. 78 M. Bonnemay and R. Pineaux, C . R. Acad. Sci., 1955, 240, 1774. 79 J. Braunstein, R. M. Lindgren, J. Am. Chem. Soc., 1962, 84, 1534. 80 C. Sinistri and E. Pezzati, Gazz. Chim. Ztal., 1967, 97, 1116. 81 R. Cigen and N. Mannerstrand, Acta Chem. Scand., 1964, 18, 2203. 1967. 1978, 74, 1861. Spiridonov, J. Mol. Struct., 1977, 42, 147. 1325.L. Bengtsson and B. Holmberg 329 82 B. Holmberg, Acta Chem. Scand., Ser. A , 1976, 30, 680. 83 I. Elding and I. Leden, Acta Chem. Scand., 1969, 23, 2430. 84 B. Holmberg, Acta Chem. Scand., 1973, 27, 3550. 85 B. Holmberg, Acta Chem. Scand., 1973, 27, 3657. 86 B. Holmberg, Acta Chem. Scand., Ser. A, 1973, 30, 797. 87 D. L. Manning, R. C. Bansal, J. Braunstein and M. Blander, J. Am. Chem. Soc., 1962, 84, 2028. 88 J. Braunstein, Ionic Interactions, ed. S . Petrucci (Academic Press, New York, 1971), vol. 11, p. 179. Paper 8/00922H; Received 7th March, 1988
ISSN:0300-9599
DOI:10.1039/F19898500317
出版商:RSC
年代:1989
数据来源: RSC
|
17. |
Statistical surface thermodynamics of quaternary liquid systems |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 331-341
Jata D. Pandey,
Preview
|
PDF (669KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I, 1989, 85(2), 331-341 Statistical Surface Thermodynamics of Quaternary Liquid Systems Jata D. Pandey,* Rishi D. Rai and Rajiv K. Shukla Department of Chemistry, University of Allahabad, Allahabad 21 I 002, India Measurements of surface tension were carried out for pentane-hexane- cyclohexane-benzene and pentane-hexane-benzene-toluene at 298.15 K by the differential capillary rise method. The surface tensions of these quaternary liquid systems were also computed theoretically using various empirical and statistical theories of the liquid state. The agreement between theory and the experiment has been found to be fairly good. An attempt has been made to explain the nature of the molecular interactions involved, in the light of excess surface tension whose sign and magnitude depend upon the chain length of the component liquids.A better understanding of the surface thermodynamics of a multicomponent system is of considerable physico-chemical interest and is essential in designing calculations involving separations, heat transfer, mass transfer and fluid flow. A substantial amount of work has been done on binary'-5 and ternary6-12 liquid mixtures and is still in progress. The various physico-chemical properties of a multicomponent system can be studied using the Bertrand-Acree-Bruchfield 13-14 (BAB) equation as well as the statistical-mechanical concept of Flory. 15-18 In Flory theory, the properties of a multicomponent system are based on the properties of their pure component liquids. Experimental data on the surface tension and their molecular interpretation for quaternary liquid mixtures are very rare except for some measurements on viscosity and excess v ~ l u m e , ' ~ refractive index,20 heat capacity2' and liquid-liquid equilibria.22 To our knowledge no study has been made on the surface tension of liquid quaternary mixtures.In this paper, we present the results of surface tension measurements of two quaternary liquid systems : pentane-n- hexanexyclohexane-benzene ( 1 ) and pen tane-hexane- benzene-toluene (2) at 298.15 0.01 K. The main objective is to develop theories for the quaternary liquid system which are already available for binary systems and to test their validity. A comparative study and its correlation with molecular interactions has also been made in the present context.The surface tension has been evaluated from various statistical, empirical and semiempirical relations, viz. Flory-Patterson theory, volume- fraction statistics, the Sanchez method and the Brock-Bird relation. It is hoped that the accumulation of accurate experimental information will ultimately lead to better understanding of the molecular configuration and molecular interactions of liquid quaternary mixtures. Experimental AnalaR liquids were obtained from BDH. They were purified according to procedures in the l i t e r a t ~ r e . ~ ~ Density measurements were carried out in a calibrated bicapillary pycnometer with an accuracy of f0.3 kg mP3. Surface tension was measured by the differential capillary rise method with an accuracy of & 7.3 x N m-I.The results are listed in table 1 together with the literature values24 for comparison. The diameters of the precision-bore capillaries were confirmed at several points along the length of each capillary by mercury weighing. The diameters of the capillaries were found to be 0.01 33 1332 Thermodynamics of Quaternary Liquid Systems Table 1. Densities (p) and surface tensions (c) for the component liquids at 298.15 K P l k a/ 1 0-3 N m-l components obsd lit." obsd lit." pentane 621.6 621.3 15.56 15.50 hexane 655.2 655.0 17.99 17.91 benzene 873.3 873.6 28.99 28.26 toluene 862.6 862.5 27.24 27.96 cyclohexane 773.6 773.8 24.01 24.40 a Ref. (24). and 0.02 cm. The apparatus was cleaned with acidified permanganate and distilled water before being dried in an oven.Mixtures were placed in the cell and their compositions were determined by weighing. The whole apparatus was immersed in a water-filled thermostatic bath whose temperature remained constant within & 0.01 "C. At equilibrium, the difference in the level of the menisci in both the capillaries, h, was constant within the precision of the cathetometer, +0.001 cm, for ca. 1 h. The surface tension of the mixture, 0, was calculated using the relation where rl and r2 are the radii of the capillaries, g is the gravitational acceleration and p the density of the mixture. The angle of contact was assumed to be zero, which was supported by visual observations. Theore tical The statistical-mechanical concept of Flory15-16 yields the following expression for the surface tension of a pure liquid and its mixture: 0 = 6*6(ij) (2) where 0, o* and 6 are the surface tension, characteristic surface tension and reduced surface tension.Flory theory is closely connected with the corresponding-states theory25 of Prigogine, which employs a simple cell model of the liquid state. Patterson and Rastogi,17 in their extension of the corresponding-states theory, dealt with the surface tension in terms of the reduction parameters, (3) called the characteristic surface tension of the liquid mixture. Starting from the work of Prigogine and Saraga,26 they derived a reduced surface tension equation in the case of a van der Waals liquid, which can be written as &4 = k i p*j T*f All the symbols in the above equations have their usual meanings.The most suitable value17 of A4 is 0.29, and this is used in our calculations. Here it is assumed that a quaternary liquid mixture can be considered to be made up of six binary mixtures. The molecules of the four components are divided into equal segments so that V: = V: = Vf = V:. Assuming the additivity of core volumes of theJ . D. Pandey, R . D. Rai and R. K . Shukla 333 components, and adopting the same procedure employed in the case of binary mixtures, it is possible to evaluate the characteristic parameters of a quaternary liquid system. Let All, Az2, A3, A4, A12, 43, A4, 41, 44 and A 31 represent the number of contact pairs between the respective species and let q i i / v be the energies associated with each contact pair, then the intermolecular energy of the quaternary liquid mixture can be represented as - E O = [i i-1 ~ i ~ i i + ~ 1 2 ~ 1 2 + 4 ~ ~ 2 ~ + ~ 4 ~ 3 4 + ~ 1 ~ 4 1 + ~ 4 ~ ~ 4 + ( 5 ) In the light of the above assumption and the definitions set out in the equation It is assumed that random mixing of the four components takes place and that a species of i neighbours at any given site is equal to its site fraction Bi which is defined here as On this basis %j can be defined if %j represent the six possible binary combinations, (10) Applying the assumption (4/4) = (ri/rj)-f = ( V ; / Z J ; ) - ~ , we have defined the segment fraction as i.e.4 2 , 4 3 , 4 4 , 41, 4 4 and 41 di = &riN,e, = 4 r j 4 S i . ; i#j (1 1) (ri 411,. . 4 - Y ( L . 4 , 1-4. . . 1 - - Wi,, ...a - FN K(l,,,,,+ c Y<T/Vi*) i - 1 .. . 4 4 4 4 F = C r i N , / N and S = C $ r i 4 / J N = C t y i , ! $ i-1 i - 1 i-1 where 12 F A R 1334 Thermodynamics of Quaternary Liquid Systems By adopting the familiar Berthelot relationship, qij = (qii qn)t and assuming Tj = 4i, we obtain T j = P,*[l -(P*/P$(q/qy which in turn yields the final expression for the intermolecular energy after substituting the characteristic pressure in eqn (12) : By analogy with the energy of pure components, we define for a quaternary liquid mixture, where 4 4 C = C cri 4 / J N = C wi c. i-1 i- 1 On comparing eqn (14) and (15), the characteristic pressure of a quaternary liquid system can be obtained as Eqn (1 5)--(17) yield the following expression for the characteristic temperature : Assuming the volume reduction parameter of a quaternary liquid system to be linear in mole fractions of the components, we have Thus, it is possible to calculate the surface tension of a quaternary liquid system using eqn (2) in conjunction with eqn (17)--(19).Goldsack and Sarvas2' used mole-fraction and volume-fraction statistics to obtain the expression for the surface tension of non-electrolyte solutions and applied it to a variety of binary polar-polar, non-polar-non-polar and non-polar-polar organic liquid mixtures. Here, this relation has been extended for computing the surface tension of a quaternary liquid system with the assumption of ideal chemical potential equations and molar surface areas obtained from bulk densities. The starting point is to use the mere general equations assuming that the surface phase is thermodynamically separated from 4 4 d i the bulk phase.c y,s = C T,,exp(o-oi)- = I i-1 i - 1 RT where, T,s and &,B are, respectively, the mole fractions of the ith component in the surface phase and the bulk phase. di is the molar surface area of the ith component. This equation can be rearranged to obtain the concentration dependence of surface tension in the case of a quaternary liquid system. Since, the sum of the mole fractions in the bulkJ . D. Pandey, R. D. Rai and R. K. Shukla 335 phase is unity and also d4 = d3 = d2 = dl = d, an approximation often used in the literature, one obtains d can be obtained from the relation where NA is Avogadro's number and pressibility, &, and mass density, p, of a liquid: is the molar volume of the ith component.Sanchez2' obtained a relationship between surface tension, a, isothermal com- CB,/p)i = d!. (23) This has been applied successfully to binary liquid mixtures by A~ree.~' If $T and po are, respectively, the isothermal compressibility and mass density of the pure components, the following approximations can be used for the extension of the relationship to a quaternary liquid system : 4 PT(1...4) = C #ipoT, (24) i-1 where #i is the volume fraction of the ith component. Eqn (23), (24) and (25) yield the following expression for the surface tension in the case of quaternary liquid systems: a = (iI, C&df ')[ 4 # . P . ] a i-1 # i pTi where di is obtained by di = 4 W T i / P i ) * (27) The Brock-Bird3' relationship is purely empirical and it is often used to predict the surface tension with the aid of critical constants.The Brock-Bird relation, applicable for a quaternary liquid mixture, can be written as where, pC,,,, c,,,, &," and T,:, are, respectively, the critical pressure, critical temperature, critical compressibility factor and reduced temperature of a quaternary liquid mixture. The values of these parameters are obtained by taking mole-fraction averages31 as 4 4 4 p c , m = C Y p c , i , < , m = C 4 K,i, i-1 i- 1 i- 1 T,rn = C XTc,i, and 12-2w w m Table 2. Measured surface tension (aexp), excess surface tension (a,) and theoretical surface tension (acalcd) from various statistical and empirical theories of liquid state and their percentage deviations Y gcalcd N m-' percentage deviation (YO) % z 2- x, 4 4 c~,,,/lO-~ N m-' o , / ~ O - ~ N m-I a b C d a b C d 0.0488 0.0658 0.0813 0.1006 0.1 180 0.1243 0.1410 0.1560 0.1285 0.1537 0.1649 0.1368 0.09 10 0.0649 0.1810 0.1238 0.1078 0.0934 0.0778 0.0629 0.0466 0.1304 0.1262 0.1 192 0.0925 0.1013 0.1258 0.1721 0.1378 0.1656 0.1831 0.2036 0.2238 0.2430 0.2615 0.2842 0.3129 0.1513 0.5888 0.1685 0.5177 0.1507 0.6137 0.1 103 0.2970 pentane (Tkhexane (qhyclohexane (4)-benzene (4) (1) 25.5 -0.5158 23.52 26.01 24.61 24.08 7.76 25.1 - 0.760 1 23.41 25.86 24.44 24.04 6.73 24.9 -0.81 12 23.31 25.71 24.27 23.99 6.38 24.8 - 0.7296 23.18 25.53 24.05 23.92 6.53 24.7 - 0.6663 23.06 25.37 23.88 23.86 6.64 24.7 -0.651 7 23.09 25.35 23.91 23.94 6.51 23.7 -0.3717 22.07 24.07 24.07 22.49 6.87 23.9 .- 0.8078 22.27 24.71 22.82 22.94 6.82 22.8 -0.2064 21.74 23.01 21.94 22.13 4.65 24.8 -0.2253 22.57 25.02 23.20 23.37 8.99 22.9 -0.1620 21.64 23.06 21.83 22.13 5.50 24.0 - 0.9745 22.50 24.97 23.16 23.16 6.25 22.0 - 0.8070 21.66 22.80 21.84 21.90 1.54 25.5 - 0.4994 23.35 26.00 24.62 23.97 8.43 23.1 -0.1330 21.37 23.23 21.26 21.79 7.48 average percentage deviations : & 6.46 - 2.00 - 3.03 - 3.25 - 2.94 -2.71 - 2.63 - 1.56 -3.38 - 0.92 -0.88 - 0.69 - 4.04 - 3.63 - 1.96 - 0.56 & 2.27 6 3.49 5.56 Q 2.62 4.22 q 2.53 3.65 3.02 3.54 3.31 3.40 3.19 3.40 5.10 4.85 4 4.51 4.01 3.77 2.94 6.45 5.76 3 4.67 3.36 3.53 3.50 0.72 0.45 5 3.45 6.00 2 7.96 5.67 k3.88 k4.01 s.0.0943 0.1300 0.1278 0.1450 0.1492 0.1843 0.1823 0.1819 0.1250 0.1691 0.1866 0.1372 0.0660 0.0524 0.1568 0.09 18 0.1373 0.1288 0.1291 0.1384 0.1484 0.1640 0.1601 0.1665 0.2041 0.0826 0.1579 0.1053 0.1434 0.0468 0.4587 0.2974 0.3589 0.3376 0.3421 0.271 1 0.361 3 0.3842 0.2455 0.22 18 0.1250 0.5548 0.7033 0.420 1 0.4582 25.9 24.6 24.9 24.5 24.4 24.3 24.1 24.0 23.9 23.1 24.3 24.6 25.8 25.5 25.1 pentane (X,thexane (X,)-benzene (X,)-toluene (4) (2) - 0.1573 24.05 26.06 24.59 25.19 7.14 - 0.3484 23.25 24.95 23.29 24.42 5.48 -0.2561 23.31 25.15 23.47 24.42 6.38 -0.4100 23.10 24.91 23.19 24.25 5.71 - 0.3900 22.95 24.71 23.02 24.06 5.94 -0.1400 22.48 24.16 22.35 23.67 7.48 - 0.1300 22.27 24.20 22.28 23.25 7.59 - 0.2862 22.29 24.27 22.36 23.25 7.12 - 0.7495 23.07 24.65 22.98 24.22 3.47 - 0.6468 22.18 23.74 21.96 23.26 3.98 - 0.2000 23.25 24.50 22.88 24.94 4.32 -0.5000 22.69 25.10 23.17 23.48 7.76 - 0.8700 24.07 26.67 25.18 25.02 6.70 -0.5000 24.09 26.00 24.62 25.01 5.53 - 0.6400 23.70 25.74 24.08 25.08 5.57 average percentage deviations : f 6.01 -0.61 - 1.42 - 1.00 - 1.67 - 2.41 0.57 -0.41 - 1.12 -3.14 -2.81 -0.32 - 2.03 - 3.37 - 1.96 -2.54 f 1.69 5.05 2.74 5.32 0.73 5.74 1.92 5.34 1.02 5.65 1.39 8.02 2.59 7.55 3.52 6.83 3.12 3.85 - 1.34 4.93 -0.69 5.84 -2.63 5.81 4.55 2.40 3.02 3.45 1.92 4.06 0.07 k5.32 f 1.77 ' Flory theory.Volume-fraction statistics. Sanchez method. Brock-Bird relationship. Table 3. Various parameters of the pure liquid components - P ij/cm3 u*/cm3 a/ 10-3 N K/cm3 component a"/ l O3 K-I &/T Pa-' p/kg m-3 mol-I mol-' P*/GPa-' T*/K m-' TJK P,/GPa mo1-I pentane 1.6226 2123.4 621.6 1.3628 85.1754 0.423 19 4144.84 15.56 469.60 0.0336 304.0 hexane 1.3897 1709.0 655.2 1.3225 99.4560 0.424 03 4432.14 17.99 507.40 0.0297 370.0 benzene 1.2265 967.0 873.2 1.2924 69.2198 0.631 62 4702.41 28.92 562.09 0.0489 259.0 toluene 1.0740 921.5 862.7 1.2627 84.3467 0.554 03 5032.94 27.24 591.70 0.0412 3 16.0 cyclohexane 1.2 1 50 1 140.0 773.4 1.2902 84.2920 0.528 95 4724.33 24.01 553.40 0.0407 308.0 w w 4 ' See ref.(24), (31) and (33).338 Thermodynamics of Quaternary Liquid Systems Results and Discussion The measured surface tension, 0, excess surface tension, oE, theoretical surface tension from various models and their percentage deviations for systems 1 and 2 are recorded in table 2 while all the values for pure component liquids are listed in table 3.Numerical evaluation of characteristic parameters for the pure component liquids were carried out according to the procedure adopted by Flory. The excess surface tension was deduced from the expression A careful perusal of table 2 reveals that volume-fraction statistics provide the best agreement, followed by the Sanchez method, the Brock-Bird relationship and Flory's statistical theory for system 1, whereas volume fraction-statistics provide the best result followed by the Brock-Bird relationship, the Sanchez method and Flory's statistical theory for system 2. The results obtained from Flory's statistical theory can be improved further by considering three- and four-body effects also. In defining the segment and site fractions, a spherical shape of the molecule, i.e.the minimum area of contact has been assumed. The possibility of only two-body interactions has been considered during the extension of the theory. However, there is every possibility of three- and four-body interactions also, and these have been ignored in order to simplify the theoretical procedure. Although three- and four-body interactions contribute very little to the energy of the system they probably can not be ignored in spite of the spherical nature of the molecules. The possibility of three- and four-body collisions increases as the chain length, i.e. area of contact, increases. Therefore, in order to obtain a comparable result, a correction term is needed to include three- and four-body effects in the evaluation of characteristic and interchange energy parameters.If molecules of larger contact area constitute the multicomponent system, the expression for the characteristic pressure of a quaternary liquid system will be of the form: where va, Oa, X123, 434, X341, 412 and XI234 are, respectively, the segment fraction, site fraction and interchange energy parameter average of their possible contibutory liquid components. This relation can ultimately be used to evaluate the characteristic temperature of the multicomponent liquid mixture. Finally, the correct form of the equations for the characteristic pressure and temperature may be used for the computation of the surface tension of a quaternary liquid mixture with a larger area of contact. The application of eqn (2) to quaternary liquid mixtures based on the assumption that they are equivalent to single-component liquids effectively ignores the differences in the concentration occurring at the surface of the mixture.The Gibbs enrichment of a mixture's surface by the component with the lower surface tension is well recognized. That is why normal results show an elevation surface tension of the mixture, resulting in a positive deviation from being a linear function of bulk mole fraction. Hence, there is a tendency of our theoretical values to be higher than the experimental values in most cases. Since the molecules in the present case are not too large, we have considered here only two-body interactions and have obtained good agreement between theory and experiment. The n-alkanes and their derivatives form random coils owing to internal rotation about C-C bonds, and this tendency increases with increasing the chain length.In aJ . D. Pandey, R. D. Rai and R. K . ShukIa 339 1 . 1 0.7 'E ? 2 0.3 zi 0.0 3 -0.1 5 .9 3 -0.5 c -0.9 .e 3 -1.3 2 3 v 3 - -1.7 2 -2.1 -2.5 0 X A 0 0 0 oo 0 0 B 0 0 0 X AX OA 0 0 0 A A X X A X X X 0 8e0 X 0 0 0 0 0 22.0 23 .O 24.0 25.0 measured surface tension/N m-' Fig. 1. An = (ocalcd - oexptl) plotted against measured surface tension oexpt, for the system pentane-hexane-cyclohexane-benzene at 298.1 5 K. 0, Flory theory; 0, volume fraction statistics; A, Sanchez method; x , Brock-Bird relation. quaternary n-alkane system, surface layer of the liquid is enriched with the component of the lower surface tension, thereby minimizing the surface tension of the mixture.The volume-fraction statistics of Goldsac and Sarvas are able to account for the concentration dependence of surface tension of a quaternary liquid mixture with components having a wide range of polarity. The average percentage deviations for the system pentane-hexane-benzene-toluene and pentane-hexane-cyclohexane-benzene are, respectively, - 1.65 and - 2.27 %. This is most probably due to surface orientation effects which would require molar surface area values (ati) very different (generally smaller) than those calculated using the assumption that the bulk density gives a good first estimate of these molar surface areas. Activity coefficient effects could be invoked to account for this, possibly by using regular solution behaviour. The ideal-solution approximation coupled with the bulk densities for calculating the molar surface area has been shown to be useful.The Brock-Bird relationship provides excellent agreement with the experimental values. It is simpler to operate and only the knowledge of critical constants is needed in the calculation. Several mixing rules3' have been proposed to evaluate the critical constants of the mixture using the mole-fraction average from the values of their pure components. This may account for part of the discrepancies in this method. Verification of our experimental measurements has been backed up by the fig. 1 and 2. Plots of [ c T ~ ~ , ~ ~ - ~ ~ ~ ~ ~ ~ ] = ACT against oexp show a nearly linear correlation with the340 Thermodynamics of Quaternary Liquid Systems 1.5 1 . 1 'E z ?? 0.7 .8 $ 0.3 3 E 3 OeO -0.3 I .r.8 H B s $ -0.7 c) 2 -1.1 -1.5 v -1.9 -2.1 0 5 X 0 X 0 xo 0 0 0 0 0 0 X o x X X 0 0 0 0 a A o X X AX a A 0 8 8 0 0 0 23.0 24.0 25.0 26.0 measured surface tension/N m-' Fig. 2. Ao = (ocalcl -oexpt,) plotted against measured surface tension oexpt, for the system pentane-hexane-benzene-toluene at 298,15 K. 0, Flory theory; 0, volume fraction statistics ; A, Sanchez method ; x , Brock-Bird relation. measured values. A close similarity between measured and calculated values of surface tension is seen. At some places in the plots the points deviate much more possibly due to specific interactions or approximations and assumptions of the theories used. Thus it is difficult to state clearly the mode and extent of interactions in quaternary liquid systems.In conclusion, volume-fraction statistics provide excellent agreement for system (I), as does the Brock-Bird relationship for system (2), followed by the other theories. The results throw some light on the relative validity of the theories and the weakening of interactions by the addition of the third and fourth liquid components with similar characteristics, as detailed by R a ~ t o g i ~ ~ during his discussion of excess thermodynamic properties, i.e. excess volumes of multicomponent systems. The authors are extremely grateful to Mr R. C. Soni for his kind help and C.S.I.R. for financial assistance during the work. References 1 B. Edmonds and I. A. McLure, J . Chem. SOC., Faraday Trans. I , 1982,78, 3319. 2 J. D. Pandey and U.Gupta, J. Phys. Chem., 1982, 86, 5235. 3 J. D. Pandey and R. L. Mishra, Chem. Scr., 1972, 11, 1 17. 4 B. S. Carey, L. E. Sariven and H. T. Davis, AICHE J., 1980, 26, 705.J. D. Pandey, R. D. Rai and R. K. Shukla 34 1 5 M. S. Dhilon and H. S. Chough, 2. Phys. Chem. (Leipzig), 1979, 270, 497. 6 T. Fujisawa, T. Uligard and J. M. Toguri, Can. J. Chem., 1985, 63, 1132. 7 P. P. Pugachevich and A. I. Cherkasskaya, Zh. Fiz. Khim., 1980, 54, 2335. 8 P. P. Pugachevich, V. A. Dozorov and M. G. Ioanidi, Zh. Fiz. Khim., 1981, 56, 1479. 9 K. Ridgway and P. A. Butler, J. Chem. Eng. Data, 1967, 12, 509. 10 E. M. Beglyarov, P. P. Pugachevich and R. M. Kamalyan, Kolloidn. Zh., 1974, 36, 14. 1 1 J. D. Pandey and N. Pant, J. Am. Chem. Soc., 1982, 104, 3299. 12 J. D, Pandey, A. K. Shukla, R. K. Shukla and R. D. Rai, J. Chem. Soc., Faraday Trans. I , 1988, 84, 13 G. L. Bertrand, W. E. Acree Jr and T. E. Bruchfield, J. Solution Chem., 1983, 12, 327. 14 W. E. Acree Jr and G. L. Bertrand, J. Solution Chem., 1983, 12, 755. 15 P. J. Flory, J. Am. Chem. SOC., 1965, 87, 1833. 16 A. Abe and P. J. Flory, J. Am. Chem. Soc., 1965, 87, 1838. 17 D. Patterson and A. K. Rastogi, J. Phys. Chem., 1970, 48, 1067. 18 D. Patterson and G. Delmas, Trans. Faraday Soc., 1969, 65, 708. 19 E. L. Heric and J. G. Brewer, J. Chem. Eng. Data, 1970, 15, 379. 20 E. L. Heric and J. G. Brewer, J. Chem. Eng. Data, 1971, 16, 317. 21 T. Wakabayashi and R. Tanaka, J. Chem. Thermodyn., 1986, 18, 175. 22 S. J. Ashcroft, A. D. Clayton and R. B. Shearn, J. Chem. Eng. Data, 1982, 27, 148. 23 Vogel, Text Book of Practical Organic Chemistry (E.L.B.S., London, 4th edn, 1978). 24 J. Timmermans, Physico-chemical Constants of Pure Organic Compounds (Elsevier, New York, 25 I. Prigogine, A. Bellemans and V. Mathod, Molecular Theory of Solutions (North-Holland, 26 I. Prigogine and L. Saraga, J. Chim. Phys., 1952, 49, 399. 27 D. E. Goldsack and C. D. Sarvas, Can. J. Chem., 1981, 59, 2968. 28 I. C. Sanchez, J. Chem. Phys., 1984, 79, 405. 29 W. E. Acree, J. Colloid Interface Sci., 1984, 101, 575. 30 J. R. Brock and R. B. Bird, AZCHE J., 1955, 1, 174. 31 R. C. Reid, J. M. Prausnitz and T. K. Sherwood, The Properties of Gases and Liquids (McGraw Hill, 32 R. P. Rastogi, J. Sci. Znd. Res., 1980, 39, 480. 33 C.R.C. Handbook of Chemistry and Physics (CRC Press, Boca Raton, Florida, 1979). 1853. 1950). Amsterdam, 1957). New York, 1970). Paper 8/00927I; Received 7th March, 1988
ISSN:0300-9599
DOI:10.1039/F19898500331
出版商:RSC
年代:1989
数据来源: RSC
|
18. |
Retention of rod-like solutes in a non-ideal multicomponent mixed solvent |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 343-348
Małgorzata Borówko,
Preview
|
PDF (439KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. 1, 1989, 85(2), 343-348 Retention of Rod-like Solutes in a Non-ideal Multicomponent Mixed Solvent Malgorzata Bor6wko Department of Theoretical Chemistry, M. Curie-Sklodowska University, P1. M . Curie-Skiodowskiej 3, 20-031 Lublin, Poland A quasi-chemical method for calculating the distribution ratio of rod-like solutes proposed previously (M. Borowko, J. Chem. SOC., Faraday Trans. I, 1988, 84, 1961) is extended to a non-ideal multicomponent mobile phase. The effects of molecular interactions in the mixed solvent on the solute retention and structure of the adsorbed layer are discussed. Liquid-solid chromatography (LSC) is one of the most popular methods used in analytical laboratories. However, the molecular mechanism of the LSC process is still unsatisfactorily understood because of its great complexity.Liquid-solid chroma- tography with a mixed mobile phase is affected by many factors which should be taken into account in a theoretical description of this process. Recently a quasi-chemical theory of liquid-solid chromatography has been proposed.l* This treatment, starting with the displacement model developed by Snyder,3* incorporates non-ideality of the mobile and stationary phase, surface heterogeneity, association effects and other factor^.^-'^ Most papers concerning liquid chromatography with a mixed mobile phase disregard the geometrical structure of the solute. It is usually assumed that all molecules in the liquid mixture are spherical.2 The role of solute structure in the chromatographic process has been analysed by Snyder.3 His approach has a semi-empirical character and assumes that solute molecules can be adsorbed in a ‘flat’ or ‘vertical’ configuration on the adsorbent surface.He has also discussed some specific techniques which can be used to test the adsorbate configuration, Recently Martire .and Boehm16-18 presented a statistical-mechanical description of the retention of rod-like molecules from a binary regular solvent. However, their method is relatively complicated and time-consuming. Therefore, in a previous paper a simple method for calculating distribution coefficients for solutes possessing complex shapes has been formulated in terms of the quasi- chemical theory of LSC.lg This treatment involves solute orientation effects in the chromatographic process.These effects have been widely discussed for chromatographic systems with ideal mixed ~olvents.’~ The aim of this paper is to present a simple method for describing solute retention from a non-ideal multicomponent mobile phase which may be used in experimental studies. A discrete model of solute orientation is assumed because this seems to be convenient for estimating the structure of adsorbed layers in real systems. As a consequence we used a cubic lattice model of the liquid mixture, which allows us to define the notion ‘orientation of a rod-like solute with respect to the surface’ in a way which may be practically useful. In many papers concerning liquid adsorption chromatography with a mixed mobile phase the regular solution approximation is applied.l.2.8.9v16 This model seems to be sufficiently precise for determining the most important features of retention from a multicomponent eluant, and for estimating the influence of non-ideality of the liquid mixture on the process.More advanced theories of solutions do not lead to simple expressions for the distribution coefficients and they involve parameters which are difficult to extract from experimental measurements. 343344 Rod-like Solutes in a Non-ideal Mixed Solvent Let us thus consider the retention of rod-like solutes, e.g. rigid rods and L-shaped or T-shaped species, in a multicomponent mobile phase. All solvents are monomeric. A cubic lattice model is assumed for the whole liquid. The lattice layer immediately adjacent to the solid surface is called the stationary phase.The adsorbed rod-like solute molecules have at least one segment in physical contact with the surface. Interactions between these molecules and the adsorbent are dependent on orientation. For each rod- like molecule one can distinguish several different orientations relative to the surface, e.g. for a linear, homogeneous molecule we have two orientations : parallel and normal to the surface.lg The configurational problem is explained with full details in ref. (1 9). The distribution coefficient may be expressed as follows : k, = CP,kf) 2 where pZ is the configurational probability that a particular molecular orientation appears in the mobile phase, kr) is the distribution factor characterizing the ith adsorption mode, defined as where g(i) and x:(~) are the mole fractions of solute molecules having the ith orientation in the stationary and mobile phases, respectively.The summation in eqn (2) is over all possible solute orientations. According to the quasi-chemical theory of LSC competitive adsorption of the solute and solvents may be described by the following series of phase-exchange reactions : kf) = x, O ( i ) / x y ) (2) (3) (l)l+(j)ue(l)u+(j)' for j = 2,3, ..., n (4) where the symbols ( S ( ~ ) ~ , ( S ( ~ ) ) ' denote the solute molecules having the ith type of orientation in the stationary and mobile phases, respectively; however (j)', (j)' are molecules of thejth solvent in these phases; r(i) is the number of solute segments in the stationary phase. It is easy to prove that by means of reactions (3) and (4) all possible phase-exchange processes may be described.1*2*20 The first solvent is assumed to be the strongest eluant.In the theory of an infinitely low concentration of the solute in the mobile phase is assumed. Then the thermodynamic constants for the above reactions are given by (p)' + # i ) ( I>" (S(i))a + # i ) ( 1)' and KIj = (G y;/xi y:) (x; y;/x; y;) for j = 2,3, . . ., n (6) where gci) and x:(~) are the mole fractions of solute molecules having the ith orientation in the stationary and mobile phases, respectively; however, $ and x; denote the mole fractions of thejth solvent in these phases; yz(i), Y ; ( ~ ) , y; and y; are activity coefficients of the particular species. The mole fractions of solvents in the stationary phase fulfill the following condition : n Z$Z1 (7) j - 1 because g + 0.isotherm of the jth solvent :',' Combining eqn ( 6 ) and (7) we obtain the general expression for the adsorption where Kjn = Kln/Klj, and it is the constant characterizing adsorption in the binary solution (j, n).M . Bordwko 345 However, from eqn (2) and ( 5 ) we have k!) = K;;)(y;(i)/y;(i)) (4 y ; / x ; y;)r('). (9) The constant Ki;) may be expressed as a function of the adsorption energies of all solute segments by using standard statistical-mechanical calculations.2.8. 16, 1 9 , 2 1 F or example, when the solute is a homogeneous r-mer we have: K;;) = (ksl)r(i) for k,, = exp [(E, - E l ) / k , T ] (10) where Es and E, denote the adsorption energies of the solute segment and a molecule of the first solvent, respectively ; k , is Boltzmann's constant.Now, let us consider solute retention from a multicomponent regular mixed so1vent.2,8 When solute-solvent molecular interactions may be neglected eqn (9) takes the following form: (11) k, = Kii) y r ( i ) where and the function +in is given by2,' (13) where mij is a well known parameter characterizing the non-ideality of the solution ( i , ~ ] . In the case under consideration the ois parameters are equal to zero for j = 1,2, , . . , n. The adsorption isotherm of the jth solvent [eqn (S)] may be rewritten as When a binary mobile phase is used eqn (14) is identical with that derived by Everett.22 The above treatment permits a calculation of the probability that the adsorbed solute molecule has the ith orientation (y).From eqn (2) and the definition of the distribution coefficient we have = x;'"'/g = k!) XL'i'/k, X : = pZ kr'/k,. (15) This equation may be used as a base for studies of the surface layer structure in LSC Illustrative model calculations have been performed for a binary regular solvent and several rod-like molecules. The following equations are used : k, = (0+20')/3, 0 = k , , Y (16) (17) (18) for homogeneous rigid rods, k, = (2m4 + a3 + 30)/6 for homogeneous T-shaped tetramers, k, = (0 +0* + 2mrp1@*)/6, 0* = k,*, Y for a heterogeneous rmer with a terminal functional group (the asterisk refers to terms connected with the functional group) and (19) for a T-shaped heterogeneous tetramer. The function Y was obtained from eqn (12H14) for n = 2.In all calculations it was assumed that each segment of the solute was preferentially adsorbed from both solvents, e.g. k,, > 1 and k,, > 1 . k , = (2Q30* + (D3 + 2 0 + 0 * ) / 6346 5 k, 3 Rod-like Solutes in a Non-ideal Mixed Solvent I I 0 0.5 0 0.5 x: x: Fig, 1. Influence of the mobile-phase composition on the distribution ratio of homogeneous linear trimers (a) and the probability of parallel adsorption of this solute (b). The parameters are as follows: k,, = 1.25, k,, = 2.0, CO~, = - 1.5, - 1.0,0,1.0,1.5. First, the influence of solvent-solvent molecular interactions on the distribution coefficient was studied. These effects are especially important when the adsorbent is relatively inert to all components of the liquid mixture. Model calculations were performed for such systems.Fig. l(a) shows the influence of the composition of the mobile phase on the distribution coefficient for linear trimers and different parameters characterizing the non-ideality of the solvent. When the solvent is ideal the function k, us. xi decreases monotonically for all investigated molecule^.'^ A similar curve is observed for the negative deviation from Raoult’s law in the solution (I, 2). However, for highly positive values of w,, one minimum is observed in the plot of k, us. x:. For high values of xi an increase in the parameter w,, brings about a decrease in the distribution coefficient. A more complicated relationship is observed for low concentrations of the first solvent. The shape of the curve of k, us. xi plotted for a solute of different structure is similar to that calculated for a rigid rod. Next, the structure of the surface layer is discussed.The probabilities of particular solute orientations in the stationary phase have been calculated from eqn (1) and ( 1 1 )-( 19). Fig. 1 (b) shows the probability of parallel solute adsorption for homogeneous trimers. The configurational probabilities of parallel (P,) and normal (P,) orientations are 1/3 and 2/3, respectively. It follows from fig. 1 (b) that the probability of parallel solute adsorption is a decreasing function of the concentration of the strongest eluant for q, < 0. When the positive deviation from Raoult’s law in the mixed solvent (1,2) is sufficiently high a minimum appears in the plot of 6 us. x:. In fig. 2 the probabilities of particular adsorption modes for T-shaped homogeneous and heterogeneous tetramers are plotted as functions of the mobile-phase composition.In the case under consideration, e.g. when all solute segments are preferentially adsorbed from both solvents, parallel adsorption is favourable over the whole concentration region, 6 > e. A separation based on shape selectivity is predictable by this model of LSC.16,19 TheM. Boro'wko 347 x: x: Fig. 2. Influence of the mobile-phase composition on the probability of a particular adsorption mode for T-shaped homogeneous and heterogeneous tetramers for w,, = - 1 .O and (a) k,, = 1.25, k,, = 2.0; (b) k,, = 1.25, k,, = 2.0, kz, = 1.875, k,*, = 3.0. The numbers in parentheses denote configurational probabilities for particular orientations.0.61 w- 0 0.5 x: Fig. 3. Influence of the mobile-phase composition on the separation factors a for T-shaped homogeneous tetramer and linear homogeneous tetramer. The parameters are as follows : k,, = 1.25, k,, = 2.0 and a,, = - 1.5,0,1.5. separation factor (a) is the ratio of the distribution coefficient of the irregularly shaped isomer to the distribution coefficient of the normal one. Fig. 3 shows the influence of the mobile-phase composition on the separation factor for T-shaped and linear homo- geneous tetramers. Plots of am. xi are obtained for different values of the solvent-solvent interaction parameter co,,. Fig. 3 shows the influence of the mobile- phase composition on this separation factor for different values of 04,.The isomer separation is best in pure solvent 2 for any value of m,,. Moreover, for high values of this parameter the function ct(x:) has a maximum, and the separation may be considerably worse than in both pure solvents. Similar calculations may be performed for other solutes.348 Rod-like Solutes in a Non-ideal Mixed Solvent The general conclusion which can be drawn is that molecular interactions in a mixed solvent play an important role in the distribution of rod-like solutes between the mobile and stationary phases and affect the mode of solute adsorption. It follows from numerous numerical simulations that interactions with the adsorbent surface have a predominant influence on the value of the distribution coefficient ; However, non-ideality of the mixed eluant controls the dependence of the distribution coefficient on the mobile-phase composition.The method presented above facilitates a mathematical description of chroma- tographic systems with a non-ideal multicomponent solvent and solutes of various geometries and structures. Application of a more advanced theory of liquid mixtures than that discussed above should not lead to other conclusions connected with basic properties of the systems considered, and is less useful in practical chromatographic studies. I thank Professor M. Jaroniec for helpful discussions. References 1 M. Borowko and M. Jaroniec, Ado. Colloid Interface Sci., 1983, 19, 137. 2 M. Jaroniec, D. E. Martire and M. Borowko, Ado. Colloid Interface Sci., 1985, 22, 177. 3 L. R. Snyder, Principles of Adsorption Chromatography (Marcel Dekker, New York, 1968).4 L. R. Snyder, Anal. Chem., 1974, 46, 1384. 5 M. Jaroniec, J. Narkiewicz and M. Borowko, Chromatographia, 1978, 11, 581. 6 M. Jaroniec and A. Patrykiejew, J. Chem. SOC., Faraday Trans. 1 , 1980, 76, 2468. 7 M. Jaroniec and B. Oicik-Mendyk, J. Chem. SOC., Faraday Trans. I , 1981, 77, 1277. 8 M. Borhwko, J. Colloid Interface Sci., 1984, 102, 519. 9 M. Borowko and M. Jaroniec, Chromatographia, 1979, 12, 672. 10 M. Jaroniec, J. K. Roiylo and B. Okik-Mendyk, J. Chromatogr., 1979, 179, 237. 1 I M. Jaroniec and J. A. Jaroniec, J. Liquid Chromatogr., 198 1, 4, 2121. 12 M. Jaroniec and J. A. Jaroniec, J. Chromatogr., 1984, 295, 377. 13 M. Borowko and M. Jaroniec, J. Chem. SOC., Faraday Trans, 1 , 1983, 79, 363. 14 L. R. Snyder, J. L. Glajch and J. J. Kirkland, J. Chromatogr., 1981, 218, 299. 15 L. R. Snyder and J. L. Glajch, J. Chromatogr., 1981, 214, 1. 16 R. E. Boehm and D. E. Martire, J. Phys. Chem., 1980, 84, 3620. 17 D. E. Martire and R. E. Boehm, J. Liquid Chromatogr., 1985, 8, 1363. 18 D. E. Martire and R. E. Boehm, J , Phys. Chem., 1983, 87, 1045. 19 M. Borowko, J. Chem. SOC., Faraday Trans. I , 1988, 84, 1961. 20 M. Borowko, M. Jaroniec and W. Rudzinski, Monatsh. Chem., 1981, 112, 59. 2 1 R. Fowler and E. A. Guggenheim, Statistical Thermodynamics (Cambridge University Press, London, 1952). 22 D. H. Everett, J. Chem. SOC., Faraday Trans. I , 1965, 61, 2476. Paper 8/01078A; Received 15th March, 1988
ISSN:0300-9599
DOI:10.1039/F19898500343
出版商:RSC
年代:1989
数据来源: RSC
|
19. |
Photoinduced electron-transfer reactions of micelle-forming surfactant ruthenium(II) bipyridyl derivatives |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 349-362
Farid M. el Torki,
Preview
|
PDF (1011KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I, 1989, 85(2), 349-362 Photoinduced Electron-transfer Reactions of Micelle-forming Surfactant Ruthenium@) Bipyridyl Derivatives Farid M. el Torki and Russell H. Schmehl* Department of Chemistry, Tulane University, New Orleans, Louisiana 70118, U.S.A. Wayne F. Reed Department of Physics, Tulane University, New Orleans, Louisiana 70118, U.S.A. A series of surfactant ruthenium(I1) bipyridyl derivatives have been prepared having either one or two long alkyl chains on a single bipyridine. The critical micelle concentrations of the chloride salts of the complexes were determined using surface tensiometry, and aggregaton numbers were obtained from static light-scattering measurements. In addition, photophysical properties of the complexes were examined as a function of the degree of aggregation.The emission maximum for complexes of the type [(b~y)~Ru(L-Ll~+ L-L = surfactant bipyridine, exhibits a bathochromic shift of ca. 15 nm upon micellization. Luminescence quenching of one complex, [(bpy),Ru(MCl 7)2+ (MC 17 = 4-methyl-4’-heptadecyl-2,2’-bipyridine), with methyl viologen (MV2+) was examined as a function of the surfactant complex concentration by both static and dynamic luminescence measurements. Quenching of micellized complex was found to be at least a factor of 20 slower than quenching of the free surfactant complex. The photoreduction of Mv2+ by the complex in the presence of a sacrificial electron donor, EDTA, was also examined as a function of the concentration of [(bpy,Ru(MC17)I2+. Quantum yields for MV2+ photoreduction were determined and compared with values obtained using [Ru(bpy),I2+ under similar conditions.In the recent past there has been significant interest in the preparation of surfactant transition-metal complexes to serve as photochemical sensitizers in the investigation of interfacial photoreactions. There have been several reports of photochemical and photophysical properties of surfactant ruthenium(r1) bipyridyl complexes incorporated into micellar solutions,2 as films at air/water interfaces and as organized molecular assemblies adsorbed onto solid support^.^. Most of the surfactant ruthenium bipyridyl complexes prepared are sparingly soluble in water; however, a few of these complexes self-aggregate to form micelles in aqueous solutions.In fact, Gratzel has determined that the complex [Ru(bpy),(DC 13)] (ClO,),(DC 13 = 4,4’-tridecyl-2,2’-bipyridine) has a critical micelle concentration of ca. 7 x lo-’ mol dm-3.5 Further, micelles of this chromophore solubilize the electron donor N-methylphenothiazine (MPTH). Excitation of micelles of the ruthenium complex containing MPTH results in the formation of reduced ruthenium and MPTH+. Recombination of the MPTH+ with the micellized reduced ruthenium centres is significantly slower than the same reaction in homogeneous solutions because of the large coulombic repulsion of the system. Surfactant ruthenium bipyridyl complexes of this type have also been used as sensitizers adsorbed onto semiconductor surfaces ;6 enhanced anodic photocurren ts have been observed for ruthenium complexes adsorbed onto SnO, electrodes.The principal photochemistry of ruthenium(I1) bipyridyl complexes is electron transfer with a suitable One of the most thoroughly studied reactions 3493 50 Electron Transfer in Ruthenium(rr) Bipyridyl Derivatives of this type involves reduction of methyl viologen, MV2+, by excited tris(2,2’- bipyridyl)ruthenium(n) (1) [Ru(bpy),12+* + MV2+ 2 [Ru(bpy),13+ + MV+. The oxidized ruthenium and reduced viologen recombine in homogeneous solution at near-diffusion-limited rates’O to regenerate the [Ru(bpy),12+ and Mv2+ : [Ru(bpy),I3++MV+ 2 [Ru(bpy),I2++MV2+. (2) [Ru(bpy),I3++EDTA 2 [Ru(bpy),12++(EDTA)+ (3) However, addition of ethylenediaminetetra-acetic acid (EDTA) as a sacrificial electron donor : results in accumulation of the deep blue MV+ in solution.The MV’ can be used to reduce a variety of substrates, including H+ (in the presence of catalysts).13* l4 This work describes the preparation of a series of surfactant long-chain alkyl- bipyridines (1) and the corresponding [Ru(bpy),(L-L)I2+ complexes as chloride salts. Critical micelle concentrations for the water-soluble complexes were determined, as well as average molecular weights, by static light scattering. For the micelle-forming chromophores the photochemical reactivity with MV2+ was examined as a function of the surfactant chromophore concentration to assess the magnitude of coulombic factors resulting from micellization of the chromophoric complex. Experimental Ligand Syntheses Over the past few years several methods have appeared for the preparation of alkyl bipyridines.15-18 The simplest approach involves catalytic coupling of two alkyl bipyridine~,’~ although terpyridines are formed as significant side-products. More recent methods involve use of 4,4’-dimethyl-2,2’-bipyridine (dmb). Sasse and coworkers prepared a mono- or bis-ethenyl derivative upon condensation of dmb with 5-alkylthophen-2-carbonyldehydes ;17 desulphurization of the ethenyl derivative with W5 Raney nickel resulted in mono- and di-alkylbipyridines with chain lengths > 6. Other methods involve deprotonation of dmb with sodamide in liquid NH316 or with lithium di-isopropylamide in THF followed by coupling with an alkyl halide.18 The preparations given here represent a variation on the latter method.A general method for the preparation of long-chain alkylbipyridines is given below. Alkylbipyridines (General) To a clean, dry, three-necked flask under N, purge and equipped with a condenser, dropping funnel and septum, freshly distilled THF (100 cm3 from Na/benzophenone) is added. To this solution lithium di-isopropylamide (LDA, 1.5 mol dmA3 in hexanes (1 equiv.) is added and the solution is cooled with dry ice. To this mixture a solution of 4,4’-dimethyl-2,2’-bipyridine (0.5 or I equiv.) in a minimum amount of THF is addedF. M . el Torki, R. H . Schmehl and W. F. Reed 35 1 dropwise. The resulting deep blue solution is stirred for 1 h, after which the N,-purged alkyl bromide (in excess) is added. The solution is stirred for 12 h and the temperature is allowed to rise to room temperature slowly as the dry ice sublimes.The resulting pale- yellow solution is quenched with water and evaporated to dryness. All the derivatives discussed here (n > 12) are solids; purification was achieved by recrystallization from ethanol. The monoalkylmethylbipyridine is much more soluble in ethanol than the dialkylbipyridine. Separation of the two derivatives is achieved by dissolving the crude product in a minimum amount of hot ethanol, cooling to room temperature, filtering off the dialkylbipyridine, cooling further in a freezer (- 4 "C) and collecting the monoalkyl derivative by filtration. 4-Methyl-4'-tridecyl-2,2'-bipyridine (MC 13) and 4,4'-Ditridecyl-2,2'-bipyridine (DC13) The two ligands were prepared following the procedure above using 5.7 cm3 of di-isopropylamine (0.041 mol), 11.5 cm3 of n-butyl lithium (0.027 mol), 3.7 g of 4,4'-dimethylbipyridine (0.020 mol) and 6.7 g of n-dodecyl bromide (0.027 mol).Both products were obtained in the reaction and were separated by fractional crystallization from ethanol. The yield of the combined MC13 and DC13 is 65 % based upon the starting bipyridine. Analytical data : MC 13 : m.p. 76-77 "C. Analysis for C,,H,,N, : found (C, 81.83; H, 10.17; N, 7.91); calculated (C, 81.82; H, 10.23, N, 7.95). 'H n.m.r. (S, 3.1 H), 0.85-1.7 (m, 25.5 H). DC13: m.p. 64-65 "C. Analysis for C,,H,,,N,: found (C, 82.91; H, 11.58; N, 5.36); calculated (C, 83.08; H, 11.54; N, 5.38). 'H n.m.r. (m, 50.8 H). (CDCl,): 5.85 (d-d, 2.0 H), 7.1 (d-d, 2.0 H), 8.2 (S, 2.0 H), 2.68 (t, 2.0 H), 2.41 (CDCl3): 6 8.5 (d-d, 1.9 H), 7.1 (d-d, 1.8 H), 8.2 (s, 1.9 H), 2.6 (t, 3.8 H), 0.88-1.6 4-Methyl-4'-heptadecyZ-2,2'-bipyridine (MC 1 7) The ligand was prepared following the procedure above using 5.7 cm3 di-isopropylamine (0.41 mol), 11.5 cm3 n-butyl lithium (2.36 mol dm-3, 0.027 mol), 4.97 g dmb (0.027 mol) and 8.24 g hexadecyl bromide (0.027 mol).Yield 50% based on dmb. m.p. 74-75 "C. Analysis for C,,H,,N,: found (C, 82.25; H, 10.55; N, 6.68); calculated (C, 82.35; H, 10.78; N, 6.86). 'H n.m.r. (CDCl,): 6 8.5 (d-d, 1.8 H), 8.2 (s, 1.9 H), 7.1 (d-d, 1.7 H), 2.96 (t, 1.8 H), 2.42 (s, 2.7 H), 0.88-1.75 (m, 34.8). Metal Complexes Bis( 2,2'- bipyridyZ)cischZororuthenium( 11) [ Ru( bp y) 2C12] This was prepared by reaction of [Ru(drns~),Cl,]'~ (1 g, 2.07 x lop3 mol) with 2,2'- bipyridine (0.64 g, 4.1 x mol) in 10 cm3 of DMF.The solution was refluxed under an Ar atmosphere for 2 h, cooled to room temperature, filtered through a medium frit, and the deep purple precipitate was washed with H,O (2 x 10 cm3). The solid collected was dried under vacuum (45 "C) and yielded 0.65 g (65%) of the cischloro complex. Cyclic voltammetry of the complex in CH,CN yielded a single oxidative wave E" = 0.3 1 V us. SSCE. The above filtrate was diluted with H,O (1 00 cm3), and saturated aqueous NH,PF, ( 5 cm3) was added to precipitate [Ru(bpy),] (PF,),. Bis(2,2'-bipyridine) (4-Methyl-4'-heptadecyl-2,2'-bipyridine)ruthenium~r1) dichloride [Ru(MC17)]Cl2 A solution containing MC17 (0.25 g, 6.2 x mol), [Ru(bpy),Cl,] (0.30 g, 6.2 x lo-, mol) and ethanol (30 cm3) was refluxed for 6 h under an Ar blanket.H,O352 Electron Transfer in Ruthenium(rr) Bipyridyl Derivatives (20 cm3) was added to the hot, deep-orange solution and the solution was cooled in an ice bath for 1 h. The chilled solution was filtered through a fine frit to remove unreacted [Ru(bpy),Cl,], evaporated to dryness, taken up in a minimum of CH3CN and added dropwise to a rapidly stirred solution of ether (100 cm3). The orange solid was filtered and dried overnight in vacuo to give 0.50 g (91 %) [Ru(bpy),(MC17)]C12. Analysis for C4,H,,Cl,Ru~5H,0: found (C, 58.49; H, 6.99; N, 8.58); calculated: (C, 58.70; H, 7.14; N, 8.56). U.V. in H,O: A, = 290 nm ( E = 6.7 x lo4 dm3 mol-1 cm-’), 456 nm ( E = 1.3 x lo4 dm3 mol-1 cm-l).Bis(2,2’- bipyridyl) (4-Methyl-4’- tridecyl-2,2’- bipyridyl )ruthenium(rr) Dichloride [Ru(MC13)]C12 This was prepared in a manner analogous to that for [Ru(MC17)]C12. U.V. in H,O: A, = 295 nm ( E = 6.0 x lo4 dm3 mol-’ cm-l), 458 nm ( E = 1.25 x lo4 dm3 mol-1 cm-l). Bis(2,2’-bipyridyl) (4,4’-Bistridecyl-2,2’ )bipyridyl )ruthenium(ri) Dichloride [Ru(DC 13)]C1, This was prepared as above for [Ru(MC17)]C12, except that a reflux time of 12 h was required. U.V. in H,O: A, = 290 nm ( E = 6.5 x lo4 dm3 mol-’ cm-l), 458 nm ( E = 1.2 x lo4 dm3 mol-’ cm-l). Apparatus Critical micelle concentrations were measured by three independent methods : chloride- ion-selective electrode potentiometry (ISE), surface tensiometry and conductivity. All c.m.c. measurements were made at 25 “C using deionized, distilled water.A Corning model 125 pH meter equipped with a Cl--selective electrode (Orion) and SCE reference (Orion) was used to measure potential as a function of added complex. For KCl the measured potential varied linearly over the range of C1- concentrations in the c.m.c. determination. Surface Tension2’ A Fischer model 20 tensiometer was employed, using the du Nouy ring method with a Pt-Ir ring (d = 1.71 cm) suspended from a counter-balanced lever arm. A precision of f 0.05 dyn cm-’ is obtained with this apparatus. Plots of surface tension us. log [complex] are used in the c.m.c. determination (vide infra). Conduct iuity Measurements 22 An Industrial Instruments Inc. conductivity bridge (model RC 16 B2) was used in conjunction with a 10 cm3 conductivity cell having two platinized Pt electrodes separated by 1 cm.Plots of equivalent conductance ( K / C ) us. C’I2 exhibit a sharp deviation from linearity in the micelle-forming concentration region. The method for c.m.c. determination was analogous to that used by Mysels and coworkers for micellization of SDS.,lF. M. el Torki, R. H . Schmehl and W. F, Reed 353 Molecular-weight Determination Static light-scattering measurements were made using the vertically polarized 632 nm line from a 5 mW He-Ne laser (Spectra-Physics) and detecting scattered light from a 2.5 cm diameter cylindrical cell at automatically stepped angles between 80 and 120' relative to the laser's incident wavevector with a Thorn EM1 9863 PMT. The custom- built cell compartment was thermostatted to f 0.1 "C.A Brice-Phoenix differential refractometer was used for determining the dn/dc refractive-index increments. Filtered toluene was used as the calibration solvent for scaling all relative scattering values to absolute Rayleigh ratios. Toluene's Rayleigh ratio for vertically polarized incident light scattered in the horizontal scattering plane with unpolarized detection was taken to be 3.96 x at 488 nm.23 The inverse fourth-power wavelength Rayleigh scattering law yields a value of 1.4 x The accuracy of the system at 488 and 632 nm was established against NBS polystyrene molecular-weight standards of 37400 Da, (NBS no. 1478), 170900 Da (NBS no. 705) and 1.05 x lo6 Da (NBS no. 1479). At both wavelengths the molecular weights agreed to within 3 % of the NBS values.Two corrections to the raw data were made in calculating the micelles' molecular at 632 nm. weights. First, the c.m.c. plotting where c is solute concentration in gcrnp3 was subtracted from the solute concentration in Kc/R = 1/M+2Bc minus the c.m.c. in g ~ m - ~ , (4) ( 5 ) R is the Rayleigh ratio of the scattered light, M is the weight-averaged molecule weight and B is the second virial coefficient. Secondly, the scattering intensities were corrected for the residual absorbance at 632 nm. The quantum efficiency for luminescence is low enough, and the detection optical band-pass filter narrow enough, that further corrections for detected luminescence were deemed unnecessary. Since there is no angular dependence of scattered light from micelles, the data from the nine angles used, from 80 to 120°, were used to obtain a larger statistical base for each experiment.A more complete description of the scattering apparatus will be presented elsewhere. Luminescence Lifetime Measurements Luminescence lifetime measurements were made using an apparatus previously described.24 In brief, the system employs excitation with a PRA LN 100 N, laser and detection of emitted light at right angles to excitation. The emitted light was dispersed with a GCA/McPherson Ev-700 monochromator and detected with a Hammamatsu R928 PMT. The transient decay was obtained using either a Tektronix 360 AD transient digitizer (60 MHz) or a PAR model 162 boxcar averager equipped with a model 166 gated integrator.A Hewlett-Packard 9826 microcomputer was used for data acquisition and control. All steady-state luminescence measurements were made using a Spex Industries model 1 1 1 C photon-counting spectrofluorimeter equipped with a thermostatted cell-holder and cooled PMT housing. All quenching experiments were carried out in 0.50 mol dm-3 aqueous phosphate buffer (pH 7.0). With methyl viologen, MV2+, as quencher, concentrations were varied between 1.5 and 30 mmol dm-3. Steady-state photolyses were performed using the above spectrofluorimeter equipped with a 450 W Xe arc lamp and employing a slit width of 30 nm (typically I, = 5 x lo-' mol (photon) s-l). In a typical experiment, solutions containing the3 54 Electron Transfer in Ruthenium(I1) Bipyridyl Derivatives ruthenium complex (5-1 pmol dm-3), ethylenediaminetetra-acetic acid (EDTA, 10 mol dm-3), MV2+ (5 mol dm-3) and Na,SO, (to adjust the ionic strength to 0.08) were degassed by passing a rapid stream of water-saturated N, over the stirred solutions.Solutions were photolysed at 460 f 15 nm with intensities averaging 4.5 x mol s-’ as determined by ferrioxalate actinometry. The formation of MV+ as a function of irradiation time was measured using a HP 845 1 diode-array spectrophotometer, monitoring the absorbance of MV+ at 605 nm ( E ~ ~ ~ = 1.37 x lo4 dm3 mol-1 cm-l). Results Syntheses The long-chain alkylbipyridine ligands were prepared by a method which represents a significant simplification of earlier methods.15-17 Formation of the mono- or di-anion of dimethylbipyridine is readily accomplished in THF at low temperatures using lithium di-isopropylamide : The resulting anion may be used in the preparation of surfactant bipyridines via addition of an excess of a long-chain alkyl bromide (vide supra).Thus the alkylbipyridine may be prepared in a one-pot reaction requiring only recrystallization for purification. The anion has also been used in the preparation of &unsaturated alky1bipyridinesl8 and covalently linked bipyridines. 25 The ruthenium complexes of the general formula [(bpy),Ru(L-L)]Cl,, where L-L is a surfactant bipyridine, were prepared by well established literature method^.^ Micelle Formation Self-aggregation of the surfactant ruthenium complexes was examined by three independent methods of demonstrated utility for c.m.c.determination : surface tension,,’ conductivity 22 and chloride-speci fic ion po ten tiomet ry . 2o Surface tensiometry proved to give the most precise results with the apparatus used. From plots of surface tension us. log[complex], the c.m.c. is estimated to be the intersection point of the linear pre- micellar and post-micellization regions. Only small changes in surface tension are observed above the c.m.c., since the activity of the surfactant in solution remains virtually constant. Results obtained for the complexes studied are given in table 1. The c.m.c. of the complexes [Ru(MCl7)] and [Ru(DC13)] were also determined by equivalent conductance and specific ion potentiometry. In all of the methods used the behaviour of the non-surfactant complex [Ru(dmb),12+ (dmb = 4,4’-dimethyl-2,2’- bipyridine) was examined and was found to be analogous to that of strong electrolytes.In conductance measurements, plots of equivalent conductance us. the square root of the surfactant complex concentration exhibit a decrease in the equivalent conductance in the micelle-forming concentration range. Specific chloride-ion electrode potentiometry of the surfactant complexes exhibits two linear regions in plots of the measured electrode potential us. log[Cl-1. The c.m.c. was taken to be the point of intersection of the two linear regions; this approach was used in the determination of the c.m.c. of cetyltrimethylammonium chloride.20 Results obtained from conductance and potentiometric methods are given in table 1.F. M. el Torki, R.H. Schmehl and W. F. Reed 355 Table 1. C.m.c. values obtained for surfactant ruthenium(I1) complexes at 25 "C c.m.c./mol dm-3 c o m p 1 ex surface tension potentiometry conductance [buffer]"/mol dm-3 [Ru(MC17)] 7 k 2 x 10-4 5+2 x 10-4 6 f 2 x 104 0 0.4 0.6 1 .o 0 6k2 x 10-5 - - 3+2 x 10-5 - 3 f 2 x 10-5 - - [Ru(MC13)] 2f0.3 x - [Ru(DC13)] I ko.5 x 10-5 3+ 1 x 10-5 5 f 3 x 10-5 0 - - "Solutions buffered to pH 7.0 with phosphate buffers. Table 2. Micellar aggregation properties determined by static light scattering complex dn/dc Mwu/g mo1-1 Hwb B"/cm3 g-' [Ru(MC13)] 0.3182f0.003 21 800f3500 26.0 f 4.0 ca. 5 x [Ru(MC17)] 0.2948 & 0.003 49 300 f 5900 55.3 f 6.6 ca. 5 x [Ru(DCl3)] 0.2804 f 0.003 12 500 f 1000 12.3 f 1 .O ca. 3 x a Weight-averaged molecular weight. Weight-averaged aggregation number.Second virial coefficient . For [Ru(MC17)] the c.m.c. was determined for pH 7 phosphate-buffered solutions at high ionic strength by surface tensiometry. The c.m.c. values measured were a factor of ten or more lower than those determined by the same method in unbuffered aqueous solution. Ionic strength effects of this magnitude have been observed for simple ionic surfactants. 26 Micelle Molecular Weights Table 2 summarizes the molecular weight, weight-averaged aggregation number, second virial coefficient and dn/dc determinations for RuMC17, RuMC13 and RuDC13. The bulky ruthenium complex in each monomer contributes a large polarizability, and so the dn/dc values in table 2 between 0.2804 and 0.3182 are significantly higher than those of typical light-headgroup surfactants, whose major weight fraction is saturated hydro- carbon.Sodium dodecyl sulphate is such an example, and our determination of dn/dc gave 0.123. For [Ru(bpy),]Cl, dn/dc was 0.361 1 (&- 1 %); (i-e. [Ru(bpy),]Cl, in water scatters ca. 8.6 times per unit mass more than alkyl chains). The effect on dn/dc of attaching alkyl chains to [Ru(bpy),I2+ is given accurately by dn dc - = 0.361 1 (% Ru)+0.123 (YO AC) where Oh Ru and % AC represent the mass fractions per monomer of Ru(bpy), and alkyl chains, respectively. The molecular weights and aggregation numbers of the micelles follow a qualitatively plausible trend. [Ru(MC17)] has a long single chain and forms micelles with a weight- averaged aggregation number of ca.55. The large, divalent headgroup occupies a large surface area because of both steric and electrostatic constraints, and so the aggregate3 56 634 - 630- 626- J 622- 61 8' 614- 610- !2 \ Electron Transfer in Ruthenium( 11) Bipyridyl Derivatives FO-o-o-o-/ P P P o/o I I I I I I I I should have a smaller aggregation number than a micelle composed of a surfactant with similar alkyl chain length but a small headgroup of lesser or no charge. For example, an alkyl sulphate surfactant in pure water has N = 143 for C,, and N = 67 for C,3.27 The shorter-chain [Ru(MC13)] has N = 26, whereas the dialkyl chain complex [Ru(DC13)] has an aggregation number of only 12. From packing considerations alone the [Ru(DC13)] doubles the hydrocarbon volume of [Ru(MC13)], so that for a fixed headgroup area and micelle radius it is reasonable to expect N for [Ru(DC13)] to be roughly half the [Ru(MC13)] value.The error margin in molecular weight (table 2) primarily reflects the uncertainty in the c.m.c., and to a lesser degree the absorbance correction to the scattering. The virial coefficients in table 2 are quoted only in rough terms. Corrections for solute absorbance at 632 nm led to large uncertainties in these values. Nonetheless, the virial coefficients are positive and relatively large, indicating strong electrostatic repulsion and excluded volume effects between micelles. Absorption and Emission Spectra At concentrations above and below the c.m.c. the complexes studied all exhibit a Ru-bpy metal-to-ligand charge-transfer (MLCT) absorption at 456 f 2 nm ( E = 13 000 200 dm3 mol-' cm-').8 Others have examined association of [Ru(bpy)J2+ with SDS, and the absorption characteristics also do not vary with surfactant concentration.2a* '* In fact, only small differences in the MLCT absorption maximum of [Ru(bpy),12+ are observed in solvents of widely varying dielectric constant.2s The complexes all exhibit emission from the MLCT state characteristic of the parent complex, [Ru(b~y)~]~+.For the surfactant complexes in aqueous phosphate buffer the emission maximum shifts from 616 to 630 nm upon micellization of the complex. Fig, 1 shows the dependence of the emission wavelength of the MLCT emission of [Ru(MC17)] on the complex concentration in 0.05 mol dm-3 phosphate buffer. The emission maximum reaches a plateau at 630 nm at concentrations above 0.25 mmol dm-3.Given the c.m.c. values of table 1 it is clear that the observed spectral shift occurs over the micellization range. If the midpoint of the observed spectral-change region is taken as a measure of the c.m.c., a value of (3 1) x lop5 mol dm-3 is obtained. The valueF. M. el Torki, R. H. Schmehl and W. F. Reed Table 3. Emission properties of ruthenium bipyridyl complexes in the presence and absence of micelles" [complex] complex /mmol dm-3 Amax/nm [Ru(MC13)] 5 x 614 [Ru(DC13)] 1 x 615 [Ru(MC17)] 5 x lop6 616 [Ru(DC19)] 5 x lop5 625 (SDS)b 5~ 10-3 630 2~ 10-5 632 1 x 10-3 630 t,,/ns ref. 610f6 this work 615+8 this work - this work - this work 608+4 this work 618f6 this work 666f10 2c,d 357 a All emission spectra in unbuffered H,O unless otherwise stated.In 10 mmol dm-3 sodium dodecyl sulphate. is slightly lower than that obtained from surface-tension measurements for phosphate buffered solutions (0.05 mol dm-3). A similar bathochromic shift in the emission of [Ru(bpy),I2+ occurs upon association of the complex with SDS micelles.21 The emission shift was attributed to an increase in the hydrophobicity of the microenvironment of the complex; spectra of [Ru(bpy),12+ in solutions of linear alcohols of varying chain length exhibit bathochromic shifts in the emission with increasing alkyl chain length.21 Table 3 lists emission maxima of a series of surfactant ruthenium bipyridyl complexes in the pre-micellar and post-micellar concentration regions. Quenching with Methyl Viologen (MV2+) Luminescence quenching of [Ru(MC17)] with MV2+ in 0.50 mol dm-3 phosphate buffer was examined as a function of the concentration of the surfactant complex.Fig. 2 shows Stern-Volmer quenching behaviour for solutions having varying concentrations of [Ru(MC17)]. At a complex concentration of 7.5 pmol dm-3 steady-state emission quenching reaches a plateau at concentrations of MV2+ above 10 mmol dm-3. As the complex concentration increases, the degree of quenching at the plateau decreases and the onset of the plateau region occurs at lower concentrations of MV2+. At sufficiently high complex concentrations (> 1 mmol dm-3) essentially no quenching by MV2+ is observed. The data of fig. 2 were fitted by assuming that the quenching was composed of separate components for free and micellized surfactant and that the emission quantum yield of the complex does not change upon micellization.Using this approximation the quenching may be fitted to the following expression: where a is the fraction of free surfactant in solution and K z and K,":" are the Stern-Volmer quenching constants for quenching of free and micellized surfactant, respectively. Using a Simplex algorithm to fit the data in which a and Kg are variable parameters and Kg;' is fixed, values for the c.m.c. (free surfactant concentration) and quenching rate in the aqueous phase were determined. Reasonable fits were obtained for [Ru(MCl7)] by assuming KPi" < 1 dm3 mol-1 (ktic < 2 x lo6 dm3 mol-l). The average aqueous quenching rate, KE9,, for [Ru(MCl7)] from the fits is 300 f 20 dm3 mol-'.From a the average c.m.c. was determined to be (1.2k0.5) x dm3 mol-', less than half the c.m.c. determined in 0.5 mol dmP3 phosphate buffer by surface tension measurements.358 Electron Transfer in Ruthenium( ii) Bipyridyl Derivatives 4 I I 1 I I 1 I I 0 5 10 15 20 25 30 3'5 [Mv2+]/mmol dm3 Fig. 2. Stern-Volmer quenching of [Ru(MC17)] by MV2+ in 0.50 mol dm-3 phosphate buffer at different complex concentrations (in rnol dm-3): 0, 7.5 x A, 1 x 0, 5 x 0, 1 x 10-3. tlns Fig. 3. Emission decay of [Ru(MC17)] in a solution containing 7.5 x mol dm-3 complex and 10 mmol dm-3 MV2+ at pH 7 (0.50 mol dm-3 phosphate buffer). Solid line represents fit to a double exponential with a baseline. Luminescence lifetime measurements were also used to examine the quenching behaviour.In the absence of MV2+ the emission decay of [Ru(MC17)] is single exponential even at complex concentrations significantly above the c.m.c., where self-quenching might be expected. 29 Upon addition of MV2+ to solutions containing < 50 pmol dm-3 [Ru(MC17)], the emission decay is clearly non-exponential. Fig. 3 shows the observed emission decay of a solution having 10 pmol dmP3 [Ru(MC17)] and 10 mmol dm-3 MV2+; the decay fits a double exponential cleanly, indicating that exchange of monomers out of the micelle is slow relative to the luminescence lifetime.30 Lifetime Stern-Volmer plots for quenching of [Ru(MCl7)] by MV2+ are shown in fig. 4 for both the fast and slow components of the decay. The Stern-VolmerF. M.el Torki, R. H. Schmehl and W. F. Reed 359 I I I I I 15 20 25 0 5 10 [Mv2 '3 /rmnol dm-3 Fig. 4. ifetime Stern-Volmer quenching of [Ru(MC17)] (7.5 x mol ~ ~ n - ~ ) by MV2+ in 0.50 rnol dm-3 phosphate buffer: A, rapidly decaying emission; 0, slowly decaying emission. quenching constants obtained are 400 25 dm3 mol-' and 18 5 dm3 mol-' for the fast and slow components, respectively. Photoreduction of MV2+ Upon addition of a sacrificial electron donor to solutions containing [Ru(bpy),I2+ and MV2+, the [Ru(bpy),13+ formed in the excited-state quenching event is reduced and MV+ accumulates in solution [eqn (3)]. This reaction has been thoroughly investigated by Hoffman and coworkers.11T12 The initial cation radical of EDTA formed upon reduction of the Ru3+ species [eqn (3)] is rapidly deprotonated to yield a strongly reducing radical, EDTA : capable of reducing MV2+.Thus the limiting quantum yield for MV+ formation is 2. Measured values of the limiting yield are < 2 because of inefficient charge separation from the geminate ion pair formed upon quenching of [Ru(bpy),I2+ by MV2+ (qcs < 0.25 at pH 4.7 in 0.10 mol dm-, EDTA).12 For the complex [Ru(MC17)], quenching of the micellized complex would be expected to produce higher yields of solvent-separated ions because of local electrostatic effects in the Stern layer of the micelle. Table 4 lists quantum yields for MV' formation obtained for photolysis of solutions having fixed concentrations of MV2+ (0.005 mol drn-,), EDTA (0.01 mol dm-3), fixed ionic strength ( I = 0.08) and varying concentrations of [Ru(MC17)].In aqueous solutions of [Ru(bpy),I2+ the yield of MV+ formation increases with increasing [Ru(bpy),I2+ concentration. The effect is attributed to an increase in the charge separation efficiency as the fraction of [Ru(bpy)J2+ ion-paired to EDTA2- (at pH 11) decreases upon increasing the chromophore concentration.12 The effect results in an increase of q5MV+ from 0.15 at 5 pmol dm-, [Ru(bpy),I2+ to 0.36 at 500 pmol dm-, [Ru(bpy),12+. For the complex [Ru(MV17)] the opposite trend is observed (table 4). For this complex, micelle formation results in a decrease in the fraction of excited chromophores which are quenched, which leads to a decrease in the yield of reduced MV2+. Fig. 4 shows both q&+ and the Stern-Volmer quenching ratio, I,,/I, under fixed photolysis conditions in which the [Ru(MC17)] concentration is varied at pH 8.7.The decrease in both emission EDTA+ -+ EDTA'+H' (8)360 Electron Transfer in Rutheniurn(ir) Bipyridyl Derivatives Table 4. Quantum yields for MV+ formation upon photolysis of solutions containing EDTA" [complex] complex /mmoldm-3 q3MVt I o / I q3E{+c [ Ru( M C 1 7)] 0.005 0.073 2.30 0.13 0.10 0.053 1.77 0.12 0.50 0.035 1.37 0.13 2.0 0.031 1.32 0.13 - - [RU(bPY),I2+ 0.005 0.15b a Photolysis conditions : Airr = 450 & 15 nm, [MV2+] = 0.005 mol drn-,, [EDTA] = 0.01 mol dm-,, pH 8.7, I = 0.08. All yields k0.005. * From ref. (12). "Yield of MV+ formation corrected for the fraction of excited complex that is quenched: %+ = h v + / [ ( l - (I/Io)I- 0.0 8 1 i O.O 0.0 4- Yf il 4- - 2.0 4 - / Yf 1- '1.5 16' / 0' 0 0-0' 1 I I I 1 3 5 6 7 -log ([Ru]/~IIo~ dm3) Fig.5. Dependence of #MV+ (0) and relative emission intensity, I o / I (0) on the [Ru(MC17)] concentration in pH 8.7 buffered solutions having 0.01 mol dm-3 EDTA and 5 mmol dm-, MV2+. quenching and MV' yield are parallel. The change occurs at [Ru(MCl7)] concentrations greater than the micellization region determined at pH 7 by surface tension methods; however, the presence of EDTA and MV2+ may significantly affect the c.m.c. (specific ion effects on the c.m.c. of simple surfactants have been observed).26 If the observed quantum yields for MV+ formation are corrected for the degree of quenching of the photoexcited [Ru(MC17)], the yield of MV+ does not vary with changes in the [Ru(MC 1 7)] concentration (table 4), suggesting that only non-micellized complex is involved in the photoreaction.Further, the corrected quantum yields are nearly equivalent to the yield measured for [Ru(bpy),12+ at low complex concentration (5 mol dm-3). Conclusions Surfactant-photoactive ruthenium(r1) bipyridyl derivatives aggregate in aqueous solutions to form small micelles. The complexes exhibit changes in luminescence maximaF. M. el Torki, R. H . Schmehl and W. F. Reed 36 1 which correlate with the hydrophobicity of the environment. Quenching of [Ru(MC17)] by MV2+ consists of two components related to quenching of micellized and aqueous complex. Dynamic luminescence measurements in the presence of MV2+ suggest that exchange of chromophores between aqueous and micellar microenvironments is slow relative to the luminescence decay rate. Photoreduction of MV2+ in the presence of EDTA2- and [Ru(MCl7)] is sensitized exclusively by the complex in the aqueous phase; thus increased charge separation yields expected for quenching in the micellar Stern layer are unimportant in influencing net photochemistry, since essentially no quenching of micellized complex occurs.W. F. R. and R. H. S. acknowledge the Louisiana Educational Quality Support Fund for partial support of this work. References 1 See (a) D. G. Whitten et al., in Interfacial Photoprocesses: Energy Conversion and Synthesis, ed. M. S . Wrighton, A h . Chem. Ser. no. 184 (Am. Chem. SOC., Washington, D.C., 1980), pp. 47-68; (b) Inorganic Reactions in Organized Media, ed. S .L. Holt, ACS Symp. Ser. no. 177 (Am. Chem. SOC., Washington, D.C., 1982). 2 (a) G. L. Gaines, Inorg. Chem., 1980, 19, 1710; (6) F. Griesen and G. G. Warr, Chem. Phys. Lett., 1985, 116, 505; (c) R. H. Schmehl and D. G. Whitten, J. Am. Chem. SOC., 1980, 102, 1938; ( d ) R. H. Schmehl, L. G. Whitesell and D. G. Whitten, J. Am. Chem. SOC., 1981, 103, 3761; (e) S. J. Atherton, J. H. Baxendale and B. M. Hoey, J. Chem. SOC., Faraday Trans. I , 1982, 78, 2167. 3 (a) K. P. Seefeld, D. Mobius and H. Kuhn, Helv. Chim. Acta, 1977, 60, 2608; (b) G. Sprintschnik, H. W. Sprintschnit, P. P. Kirsch and D. G. Whitten, J. Am. Chem. SOC., 1977, 99, 4947. 4 (a) S. J. Valenty, in Interfacial Photoprocesses: Energy Conversion and Synthesis, ed.M. S . Wrighton, A h . Chem. Ser. no. 184 (Am. Chem. SOC., Washington, D.C., 1980), pp. 69-95; (b) W. H. F. Sasse and D. N. Furbng, Colloids Surf., 1983, 7 , 115. 5 M. Gratzel, in Energy Resources through Photochemistry and Catalysis, ed. M. Gratzel (Academic Press, New York, 1983), pp. 71-98. 6 (a) H. Matsuda, H. Daifuku, K. Aoki and K. Tokuda, J. Elecrroanal. Chem., Interfacial Electrochem., 1982, 140, 179; (b) 1985, 183, 1. 7 V. Balzani, Coord. Chem. Rev., in press. 8 N. Sutin and C. Creutz, J. Chem. Educ., 1983, 60, 809. 9 (a) D. G. Whitten, Acc. Chem. Res., 1980, 13, 83; (b) T. J. Meyer, Ace. Chem. Res., 1978, 11, 94. 10 (a) C. R. Bock, T. J. Meyer and D. G. Whitten, J. Am. Chem. SOC., 1974, 96, 4710; (6) C. R. Bock, J. A. Connor, A. R. Gutierrez, T.J. Meyer, D. G. Whitten, B. P. Sullivan and J. K. Nagle, J. Am. Chem. SOC., 1979, 101,4815. 11 (a) M. Z. Hoffman, D. R. Prasad, G. Jones and M. Vincent, J. Am. Chem. SOC., 1983, 105, 6360; (b) K. Mandal and M. Z. Hoffman, J. Phys. Chem., 1984,88,185; (c) D. R. Prasad and M. Z. Hoffman, J. Phys. Chem., 1984, 88, 5660. 12 D. R. Prasad, K. Mandal and M. Z. Hoffman, Coord. Chem. Rev., 1985, 64, 175. 13 (a) J-M. Lehn and J. Sauvage, Nouv. J. Chim., 1977, 1, 449; (b) J-M. Lehn and R. Ziessel, Nouv. J. Chim., 1980, 4, 355. 14 J. Kiwi, in Energy Resources through Photochemistry and Catalysis, ed. M. Gratzel (Academic Press, New York, 1983), pp. 297-331. 15 (a) P. E. Rosevear and W. H. F. Sasse, J. Heterocycl. Chem., 1971, 8, 483; (b) W. H. F. Sasse and C. P. Whittle, J. Chem. SOC., 1961, 1347; (c) G. M. Badger and W. H. F. Sasse, Adv. Heterocycl. Chem., 1963, 2, 179. 16 K. D. Bos, J. G. Kraaijkamp and J. G. Noltes, Synth. Commun., 1979, 9, 497. 17 0. Johansen, C. Kowala, A. W-H. Mau and W. H. F. Sasse, Aust. J. Chem., 1979,32, 1453. 18 (a) P. K. Ghosh and T. G. Spiro, J. Am. Chem. SOC., 1980, 102, 5543; (b) H. D. Abruna, A. I. Breikas 19 I. P. Evans, A. Spencer and G. Wilkinson, J. Chem. Soc., Dalton Trans., 1973, 204. 20 L. Shedbusky, C. W. Jakob and M. P. Epstein, J. Phys. Chem., 1963, 67, 2075. 21 K. Shinoda, T. Yamanaka and K. Kinoshita, J. Phys. Chem., 1959, 63, 648. 22 (a) P. Mukerjee, K. J. Mysels and C. I. Dulin, J. Phys. Chem., 1958, 62, 1390; (b) K. J. Mysels and 23 T. M. Bender, R. J. Lewis and R. Pecom, Macromolecules, 1986, 19, 244. 24 W. F. Wacholtz, R. H. Auerbach and R. H. Schmehl, Inorg. Chem., 1986, 25, 227. and D. B. Collun, Inorg. Chem., 1985, 24, 988. P. Kapanan, J. Colloid Sci., 1961, 16, 481.362 Electron Transfer in Rutheniurn(ii) Beyridyl Derivatives 25 (a) R. H. Schmehl, R. A. Auerbach, W. F. Wacholtz, C. M. Elliott, R. Freitag and J. W. Merkert, Inorg. Chem., 1986,2!5,2440; (6) C. M. Elliott and R. A. Freitag, J. Chem. Soc., Chem. Commun., 1985, 156. 26 (a) K. Mysels and L. H. Princen, J. Phys. Chem., 1959,63, 1696; (6) P. Mukerjee, Adv. Colloid Interface Sci., 1967, 1, 241; (c) M. F. Emerson and A. Holtzer, J. Phys. Chem., 1967, 71, 1898. 27 K. S. Birdi, in Micellization, Solubilization and Microemulsions (Plenum Press, New York, 1977), vol. 1. 28 J. V. Caspar and T. J. Meyer, J. Am. Chem. Soc., 1983, 105, 5583. 29 (a) J. H. Baxendale and M. A. J. Rodgers, Chem. Phys. Lett., 1980, 72, 424; (6) V. Lachish, 30 A. Yeketa, M. Aikawa and N. J. Turro, Chem. Phys. Left., 1979, 63, 543. M. Ottolenghi and J. Rabani, J. Am. Chem. Soc., 1977, 99, 8062. Paper 8/01 162A; Received 21st March, 1988
ISSN:0300-9599
DOI:10.1039/F19898500349
出版商:RSC
年代:1989
数据来源: RSC
|
20. |
Cobalt–manganese oxide water-gas shift catalysts. A kinetic and mechanistic study |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 85,
Issue 2,
1989,
Page 363-371
Graham J. Hutchings,
Preview
|
PDF (532KB)
|
|
摘要:
J. Chem. Soc., Faraday Trans. I , 1989, 85(2), 363-371 Cobalt-Manganese Oxide Wa ter-gas Shift Catalysts A Kinetic and Mechanistic Study Graham J. Hutchings,*t Frank Gottschalk, Roger Hunter and S. Walter Orchard Catalysis Research Programme, Department of Chemistry, University of the Witwatersrand, PO Wits, Johannesburg 2050, South Africa An investigation of the mechanism of the water-gas shift reaction over a cobalt-manganese oxide catalyst (Co : Mn = 1 : 1) is described using both kinetic and model reagent studies. The reaction of methanol, ethanol and formic acid are discussed, and these reagents are used as model reactants to probe the reaction mechanism. Based on these kinetic and model reactant studies a mechanism for the water-gas shift reaction is proposed for this catalyst system.It is considered that the mechanism proceeds according to the following steps : (i) water dissociatively adsorbs to form surface hydroxy and hydride species, (ii) subsequent CO adsorption and oxidative addition of the surface hydroxy species lead to the formation of a surface formate species, (iii) decomposition of the formate species via 8-elimination to give gas-phase CO, and surface hydride and (iv) combination of two surface hydride species giving gas-phase H,. The water-gas shift reaction has been the subject of many theoretical and mechanistic studies'-* which are due, in part, to its industrial importance. The reaction is utilised commercially in a number of chemical processes, particularly for the production of ammonia and hydr~gen.~ At present, when the production of high-purity hydrogen is required (e.g.for ammonia synthesis) the process is carried out in two distinct stages, using catalysts that are specifically designed for each stage. The initial process stage involves a high-temperature conversion (300-400 "C) using an ironxhromium spinel catalyst3v5 which reduces the exit CO concentration to ca. 3 mol %. Conversion of the remaining CO is carried out in a lower-temperature process stage (ca. 200-250 "C) using a high-activity copper-zinc oxide catalyst.6* ' Both of these catalysts have been extensively re~earched.~, 5-7 and have been specifically designed for reaction of CO/H, feedstocks derived from the steam-reforming of natural gas or naphtha. Unfortunately these catalysts are not as effective for coal-derived CO/H,, owing to the higher levels of CO in such feedstocks as well as the presence of significant concentrations of sulphur compounds (COS and H,S).We have recently showns that cobalt-manganese oxide catalysts are more active than the present commercial ironxhromium oxide catalyst in the temperature range 350-400 "C, and that this formulation is potentially more useful for conversion of coal-derived CO/H,. In this paper we discuss the mechanism of the water-gas shift reaction over this catalyst using both kinetic and model reagent studies. Experimental The preparation of the cobalt-manganese oxide catalyst has been described pre- vio~sly.~, lo Prior to use, catalysts were calcined (500 "C, 24 h) and reduced in situ in the t Present address : Leverhulme Centre for Innovative Catalysis, Department of Chemistry, University of Liverpool. PO Box 147, Liverpool L69 3BX.363364 Water-gas Shift Catalysts 4.5 4.0 - 'm 3.5 'bD e d 3.0 YI 5 !i g 2.0 8 \ C 2.5 .9 0 1.5 0 E 1.0 0.5 0 200 250 300 350 400 TI "C Fig. 1. CO conversion and rate of CH, formation as a function of temperature, CO GHSV = 510 h-l, CO: N,: H,O = 1 : 1 :4.5. A, Co : Mn = 1 : 1 ; 0, Co : Mn = 1 : 1 + 1 YO K. catalytic reactor with hydrogen (400 "C, 16 h, GHSV = 200 h-'). All catalysts were then evaluated for the water-gas shift reaction using a laboratory microreactor described previously." Catalysts were evaluated for the reaction of model reagents using an all- glass microreactor (i.d. = 14 mm). The reagents were fed to the reactor using controlled vaporisation in nitrogen carrier gas, as described in previous studies.', All products, both liquid and gaseous, were analysed using gas chromatography, and satisfactory mass balances were obtained for all data quoted.Results and Discussion Effect of Temperature and CO/H, Flow Rate The catalytic activity of a cobalt-manganese catalyst formulation (Co : Mn = 1 : 1) was examined and the effect of temperature on the rate of CO conversion and the rate of CH, by-product formation is shown in fig. 1. It is apparent that the catalyst only demonstrates significant activity for the water-gas shift reaction at temperatures 2 250 "C, and the formation of the unwanted by-product CH, also becomes apparent at this temperature. Addition of potassium in the form of potassium carbonate enhances the rate of CO conversion at temperatures 2 250 "C and also markedly decreases the rate of formation of CH,.At temperatures 2 300 "C high CO conversions are readily achievable with this catalyst formulation (e.g. CO conversion > 98%, 350 "C, CO GHSV = 510 h-l, CO:N,: H,O = 1 : 1 :4.5 mol ratio), hence this catalyst can be considered to be a highly effective high-temperature shift catalyst.G. J . Hutchings et al. 365 Table 1. Pseudo-first-order rate constants for Co : Mn catalyst" T/"C k,/lO-' mol g-' atm-1 s-'* 200 225 250 275 300 350 400 450 4.1 6.6 14.9 61.5 110 133 158 20 1 aCo:Mn = 1:1, CO GHSV = 510 h-l, CO:N,: H,O = 1 : 1 :4.5. 1 atm = 101 325 Pa. Pseudo-first-order rate constants for the non-potassium-promoted formulation were calculated (table 1) using the expression F 1 k , =-ln - rnP (1-x) where F is the molar flow rate of CO, m is the mass of catalyst, P is the pressure and XA is the fractional conversion of CO.For temperatures < 300 "C the apparent activation energy for CO conversion was 89 kJ mol-'. A value of this magnitude is indicative that diffusion limitations are not evident in this temperature region. To confirm this a series of further tests was ~0nducted.l~ In particular the rate of produc! formation was linearly dependent on the reactant feedrate and a plot of rate versus TZ was found to be non-linear. At temperatures > 300 "C the apparent activatiyn energy was considerably lower (13 kJ mol-') and the rate was now proportional to TZ, hence it was concluded that at temperatures > 300 "C diffusion limitations were apparent.Consequently all subsequent kinetic and model reagent studies were carried out at temperatures < 300 "C. Effect of H,O and CO Pressure The effect of the partial pressure of H,O on the rate of H,O conversion was investigated for the Co : Mn = 1 : 1 catalyst and the results are shown in fig. 2. These experiments were carried out at 275 "C and 28.2 kPa partial pressure CO. The results demonstrate that the rate of H,O conversion is linearly dependent on the pressure of H,O, indicating that the reaction is first-order in water concentration. The effect of CO partial pressure on the rate of CO conversion was investigated for the Co:Mn = 1 : 1 catalyst and the results are given in fig.3. It is apparent that the rate of CO conversion initially increases with increasing CO partial pressure but then declines to a steady value. The observation of a maximum rate with increasing CO partial pressure is possibly indicative of a Langmuir-Hinshelwood reaction mechanism which has a rate-determining step involving the reaction between two surface-adsorbed intermediates derived from gas-phase CO and H,O. At constant H,O partial pressure the variation in rco with Pco for such a mechanism is given by rco = aPco/( 1 + bP,J2. Although such an expression predicts that a maximum for rco should be observed, it does not fit the observed rate data particularly well. For example the observed rco 13 FAR 1366 1.4 1.2 & I - ei '.O- % 0.8 .! 0.6- 0, E ; 0.4 d 0.2 Water-gas ShiJt Catalysts - - - - - 0 30 40 50 60 sIZo /@a Fig.2. Rate of H,O conversion as a function of PHz0 for the Co: Mn = 1 : 1 catalyst. 7 = 275 O C , CO GHSV = 510 h-l. Fig. 3. 0 5 10 15 20 25 Pco H a at 275 "C. H,O feed rate = 1.36 g(H,O) g(cat)-'. Effect of CO partial pressure on the rate of CO conversion for the Co : Mn = 1 : 1 catalystG. J . Hutchings et al. 367 Table 2. Reaction of model reagents selectivity of gaseous organic phase (mass %) time on line/min" CH, HC0,H CH,OCH, CH,OH HCO,CH, formic acid - - - 100 100 I00 100 - 10 54 134 260 - - - - - - - - - - - - methanol 9 2.68 1.53 81.8 14.0 24 0.63 4.17 67.7 27.5 41 0.3 1 - 2.22 81.8 15.6 59 0.67 3.72 95.6 90 0.14 - 0.6 1 99.2 - a Catalyst temperature = I50 "C. - - - - increases more rapidly than predicted by the above expression.In addition, the observed rco decreases rapidly to a steady value, whereas the above expression would predict a gradual and steady decrease. The observed variation in rco with Pco is therefore not wholly consistent with a simple Langmuir-Hinshelwood bimolecular reaction mech- anism occurring with this catalyst. It is possible that a number of reaction pathways occur, the dominance of one particular mechanism being dependent on Pco, and, for example, at high Pco a regenerative mechanism could occur in which the reaction proceeds in two steps: H,O(g) + * = H, + O(a) CO(g) + O(a) = CO,(g) + * where * is a vacant surface site. Such a mechanism would be consistent with the observed constant rco at high PcO.l4 This mechanism was first proposed by Kul'kova and Temkin" for the Fe,O,/Cr,O, high-temperature shift catalyst and has been shown by a number of workers16.17 to fit experimental results well for these catalysts.More recently16 the regenerative mechanism has also been proposed to operate for the Cu/ZnO/Al,O, low-temperature shift catalyst. Reaction of Model Reagents A number of mechanisms have been proposed to account for the formation of CO, and H, from the reaction of CO and H,O.'-, A central intermediate in a number of these proposals is a surface formate, and to test this possibility for the cobalt-manganese catalyst the reaction of formic acid was investigated. Formic acid was initially distilled and rigorously dried and was then reacted over a cobalt-manganese oxide catalyst (Co: Mn = 1 : 1) at temperatures below which formic acid thermally decomposes (i.e.< 200 "C) and the results are given in table 2. Under all conditions tested formic acid did not react over the cobalt-manganese oxide catalyst. These results are in direct contrast with those obtained for the Cu/ZnO low-temperature shift catalyst,". l9 when formic acid was shown to decompose to CO, and H, at 150 "C. It is therefore apparent that the cobalt-manganese oxide and copper-zinc oxide catalysts operate via distinctly 13-2368 Water-gas Shgt Catalysts L CH30H M M CH30CH Act% I 0 I c=o I M Scheme 1. H J I M H O I II M M different reaction mechanisms, particularly with respect to the structure of the surface intermediate responsible for CO, and H, formation. Formic acid is known to adsorb to form an intermediate of the type: 0 II M-0-C-H by insertion of the metal atom, or surface site, into the OH bond of formic a~id.~.*O Based on these studies it must be concluded that this species is inactive on the cobalt-manganese oxide catalyst.To probe the reaction mechanism of the water-gas shift reaction with this catalyst, methanol and ethanol were individually reacted with CO under a range of experimental conditions and the results for methanol are given in table 2. Ethanol-CO mixtures were not particularly reactive and the only products observed at all conditions were ethene and ethanal. However, for CH,OH-CO mixtures methyl formate was observed as a significant product, albeit at very low CO conversion. The production of methyl formate was not particularly long-lived (ca.1.5 h), but it was always observed as the major reaction product. The formation of methyl formate from reaction of CH,OH-CO is well known for other catalyst systems and particularly at higher reaction pressures.21 The product is considered to be formed via an oxidative addition mechanism according to scheme 1. In this mechanism CH,OH is dissociatively adsorbed at the surface to giveG. J . Hutchings et al. 369 OH M M / L Scheme 3. a methoxy group. A similar pathway could be available for ethanol adsorption, as shown in scheme 2, but now j3-elimination to yield ethene, or a rearrangement to ethanal are the more facile reaction pathways compared to a subsequent CO insertion, accounting for the absence of ethyl formate among the products. Reaction Mechanism The experiments carried out with CH,OH and CH,CH,OH as model reagents are considered to be highly significant with respect to the mechanism of the water-gas shift reaction eperating with cobalt-manganese oxide catalysts.These reagents of the type ROH can be considered as models for water (HOH) in which an alkyl group replaces hydrogen. For methanol the major product observed was that from the oxidative addition of a surface methoxy intermediate to a surface CO. For ethanol, elimination of ethene was observed, which was most likely due to j3-elimination from a surface ethoxy group. Additional mechanistic evidence is obtained from the inactivity of formic acid, demonstrating that the surface intermediate responsible for the formation of CO, and H, is not of the form M-0-CO-H.Based on these model studies we propose a reaction mechanism for the water-gas shift reaction for this catalyst which is shown in scheme 3. It is proposed that water adsorbs dissociatively onto a surface site with subsequent adsorption of CO onto an adjacent surface site, followed by oxidative addition of the surface-bonded hydroxy group to CO. The reaction of CH,OH-CO to form methyl formate is direct evidence for this proposed step and also oxidative addition reactions are well known in organometallic chemistry. 22 Subsequent reaction would occur via j3-elimination to give gas-phase CO, and surface hydride, with combination of surface hydrides, leading to hydrogen formation. Such decomposition of a metallocarboxylic acid to yield CO, is well supported in the reactions of numerous organometallic c o m p l e ~ e s ~ ~ - ~ ~ since it was initially demonstrated for IrCl,(CO,H) (CO) (PMe2Ph),.26 The kinetic data obtained in this study are in support of such a reaction scheme. The key surface species derived from water is the surface hydroxy and the observed linear dependence of the reaction rate on PHz0 is consistent with this mechanistic step.The observed maximum in rco when Pco is varied is also consistent with a rate-determining step involving the reaction between two surface adsorbed species namely CO and OH. The observation that rco becomes constant at high Pco could be due to CO successfully competing with H,O for adsorption sites such that the concentration of surface hydroxy groups, a key intermediate, becomes limited.However, as stated previously at high Pco, a regenerative mechanism'^ l5-I7 may become a dominant reaction pathway owing to the370 Wa ter-gas Sh $t Catalysts CH3 I 0 I M H (d 0 M Scheme 4. previously discussed competitive adsorption. It is therefore possible that at low Pco the proposed surface formate mechanistic pathway is dominant, but at high Pco the regenerative mechanism may also operate competitively. The proposed mechanism based on surface formate is consistent with a number of recent organometallic studies concerning the reactivity of the proposed surface intermediates.2. Recently Chinchen et a1.6 have studied the mechanism of the water-gas shift reaction over both the industrial low-temperature copper-zinc oxide catalyst and the high-temperature iron-chromium oxide catalyst.They concluded that over this catalyst surface formate species of the type MO-CO-H were not involved, on the basis that since such an intermediate was pivotal for the methanol synthesis reaction it was considered improbable that such an intermediate was also involved in the water-gas shift reaction. Such a deduction is not strictly correct since the formate intermediate could be involved in both processes, but its subsequent reaction depends on the relative concentration of surface H in the presence and absence of an excess of water vapour (scheme 4). High surface H concentration would lead to the subsequent formation of a surface methoxy group which on further hydrogenation yields methanol, whereas if the surface H concentration is low then decarboxylation of the formate intermediate could be a preferred reaction.Since van Herwijnen et aZ.lS* l9 have provided some evidence for the involvement of a formate intermediate, of the type M-0-CO-H with the copper-zinc oxide catalyst, it is considered that further mechanistic studies, to clarify the involvement of surface formate intermediates are required for both the iron<hromium oxide and the copper-zinc oxide catalysts. We thank the University of the Witwatersrand and AECI Ltd, Johannesburg, for financial support. References 1 M. I. Temkin, A h . Catal., 1979, 28, 263. 2 P. C. Ford, Acc. Chem. Res., 1981, 14, 31. 3 D. S. Newsome, Catal. Rev. Sci. Eng., 1980, 21, 275. 4 R. M. Laine and R.Wilson, in Aspects of Homogeneous Catalysis, ed. R. Ugo (Reidel, Dordrecht, 5 G. C. Chinchen, R. H. Logan and M. S . Spencer, Appl. Catal., 1984, 12, 69; 89; 97. 6 G. C. Chinchen, M. S. Spencer, K. C. Waugh and D. A. Whan, J . Chem. SOC., Faraday Trans. I , 1987, 7 G. Petrini, F. Montino, A. Bossi and F. Garbassi, Preparation of Catalysts 111, ed. G . Poncelet, P. 8 F. M. Gottschalk and G. J. Hutchings, J. Chem. SOC., Chem. Commun., 1988, 123. 9 M. van der Riet, G. J. Hutchings and R. G. Copperthwaite, J . Chem. Soc., Chem. Commun., 1986, 1985), vol. 5, p. 217. 83, 2193. Grange and P. A. Jacobs (Elsevier, Amsterdam, 1983), p. 835. 798.G. J . Hutchings et al. 37 1 10 R. G. Copperthwaite, G. J. Hutchings, M. van der Riet and J. Woodhouse, Ind. Eng. Chem. Res., 1987, 11 F. M. Gottschalk, R. G. Copperthwaite, M. van der Riet and G. J. Hutchings, Appl. Catal., 1987, 38, 12 G. J. Hutchings, M. V. M. Hall, F. M. Gottschalk and R. Hunter, J. Chem. Soc., Faraday Trans. 1, 13 G . C. Bond, Heterogeneous Catalysis: Principles and Applications (Clarendon Press, Oxford, 1987), 14 K. J. Laidler, Catalysis; Fundamental Principles, ed. P. H. Emmett (Reinhold, New York, 1954), 15 N. V. Kul'kova and M. I. Temkin, Zh. Fiz. Chim., 1949, 23, 695. 16 G. K. Boreskov, T. M. Yureva and A. S. Sergeeva, Kinet. Katal., 1970, 11, 1476. 17 J. E. Kubsh and J. A. Dumesic, AIChE J., 1982, 28, 793. 18 T. van Herwijnen and W. A. de Jong, J. Catal., 1980, 63, 83. 19 T. van Herwijnen, R. T. Guczalski and W. A. de Jong, J. Catal., 1980, 63, 94. 20 R. S. Paonessa and W. C. Trogler, J . Am. Chem. Soc., 1982, 104, 3529. 21 A. Peltzman, Oil Gas J., 1981, 79, 103. 22 D. F. Shiver, ACSSymp. Ser., 1981, 152, 1. 23 N. Grice, S. C. Kao and R. Pettit, J. Am. Chem. Soc., 1979, 101, 1627. 24 M. Catellani and J. Halpern, J. Inorg. Chem., 1980, 19, 566. 25 J. R. Sweet and W. A. G. Graham, Organometallics, 1982, 1, 982. 26 A. J. Deeming and B. L. Shaw, J. Chem. SOC. A , 1969, 443. 26, 869. 103. 1987, 83, 571. p. 58. vol. 1, p. 155. Paper 8/01167B; Received 21st March, 1988
ISSN:0300-9599
DOI:10.1039/F19898500363
出版商:RSC
年代:1989
数据来源: RSC
|
|