|
11. |
Carbon-13 and nitrogen-15 nuclear magnetic resonance and infrared spectroscopic investigations of pyridine adsorbed on silica-gel surfaces |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 761-769
Thomas Bernstein,
Preview
|
PDF (556KB)
|
|
摘要:
J. Chem. Soc., Faraday Trans. I, 1982, 78, 761-769 Carbon4 3 and Nitrogen- 15 Nuclear Magnetic Resonance and Infrared Spectroscopic Investigations of Pyridine Adsorbed on Silica-gel Surfaces BY THOMAS BERNSTEIN,* LEONID K I T A E V , ~ DIETER MICHEL AND HARRY PFEIFER Karl-Marx-Universitat, Sektion Physik, DDR-7010 Leipzig, Linnkstrasse 5, German Democratic Republic AND PETER FINK Friedrich-Schiller-Universitat, Sektion Chemie, DDR-6900 Jena, Lessingstrasse 10, German Democratic Republic Received 23rd March, 1980 Carbon-13 and nitrogen-1 5 high-resolution n.m.r. measurements have been combined with i.r. spectro- scopic investigations to study the interaction of pyridine molecules with OH groups on a partially dehydroxylated and phosphorus-modified silica-gel surface. On partially dehydroxylated silica gel strong hydrogen bonds are formed between the nitrogen atoms and the protons of the hydroxyl groups.The number of active sites is ca. 1.2 nm-2. The formation of pyridinium ions was observed in the case of adsorption on phosphorus-modified silica gel. The number of ions was estimated to be 0.05 nm-*. The adsorption of pyridine on silica surfaces has been investigated in numerous papers. 1-6 There is spectroscopic evidence that a strong hydrogen-bonding type interaction between the pyridine molecules and the surface hydroxyl groups occurs. The wavenumber of the i.r. stretching-vibration band of the OH groups is lowered by ca. 800 cm-l and the n.m.r. line of the free hydroxyl protons is shifted by 7 ppm downfield under the influence of the adsorbed pyridine molecule^.^ Since the formation of a pyridinium ion was not observed the structure of the adsorption complex is of special interest.Infrared spectroscopic investigations of pyridine molecules adsorbed on silica surfaces containing POH surface hydroxyl groups indicated that a stronger interaction occurs compared with the case of pure silica^.^*^ Pohle and Fink3 detected the formation of pyridinium ions at such a modified aerosil surface. In earlier work by other authors, however, no evidence for the protonation of the pyridine molecule was found.5 In this paper we discuss the results of high-resolution n.m.r. spectra and i.r. spectra of adsorbed pyridine molecules in order to derive more detailed conclusions about the nature of the interaction and the structure of the adsorption complex.EXPERIMENTAL MATERIALS For the n.m.r. investigations pyridine molecules were used which were enriched with 95% 15N nuclei (Isocommerz GmbH, Leipzig, G.D.R.). Before adsorption the pyridine was dried t Permanent address: Moscow State University, Faculty of Chemistry, SU-I 17234 Moscow, U.S.S.R. 76 1762 PYRIDINE ADSORBED O N SILICA GEL over zeolite 3A. For the i.r. measurements the pyridine was of spectroscopic grade and contained no water impurities. The unmodified silica gel was Kieselgel according to Stahl (HR 60, Merck, Darmstadt, F.R.G.) with a specific surface area (B.E.T., N,, 77 K) of 320 mz g-l. To calculate the amount of pyridine required for one statistical monolayer a molecular area of 0.4 nm2 was taken.Before adsorption the unmodified silica-gel samples were partially dehydroxylated at 673 K (pressure loW3 Pa). The samples used for modification with PC1, at 673 K [as described in ref. (2)] were pretreated by the same procedure. The specific surface area of the modified samples (SO,-P) was 226m2g-'. The number of POH groups was ca. 50% of the total. Before adsorption the Si0,-P samples were once more activated at 673 K (pressure Pa). METHODS 13C n.m.r. spectra and I5N n.m.r. spectra were taken at 22.63 MHz using a Bruker HX 90 Fourier-transform spectrometer and at 9.12 MHz using a Bruker HX 90R Fourier-transform spectrometer, respectively. The experimental error for the shifts is kO.5 ppm and that for the linewidths & 10 Hz, unless otherwise stated (cf.table 3). The chemical shifts were referred to the neat liquid. Negative shifts are to higher field. 1.r. spectra were recorded with a UR-20 spectrometer (VEB Carl Zeiss, Jena, G.D.R.). Thin tablets (0.d. 2 cm, weight 7 mg ern-,) were placed in a quartz cell with CaF, windows in which dehydroxylation, modification, sample preparation and measurements were carried out. RESULTS AND DISCUSSION ADSORPTION ON PARTIALLY DEHYDROXYLATED SILICA GEL The coverage dependence of the observed 13C and 15N resonance shifts and of the widths of the 15N resonance lines are summarized in table 1. The spectra (cf. fig. 1) were recorded at room temperature. The influence of adsorption on the 13C n.m.r. frequencies is only slight. They do not change as a function of coverage and the shifts are of the same order of magnitude as in the case of non-specific interactions.On the other hand, the large and TABLE 1 .-CARBON- 13 AND NITROGEN- 1 5 RESONANCE SHIFTS S[C(i)] AND LINEWIDTHS A q FOR PYRIDINE MOLECULES ADSORBED ON PARTIALLY DEHYDROXYLATED SILICA GEL (AT ROOM TEMPERA- TURE; 6 IN ppm REFERRED TO THE LIQUID STATE, Avt IN Hz) 13C n.m.r.a I5N n.m.r. coverage (monolayers) S[C(2)] m(3)1 4c(4)1 6") A V; 3.79 1.30 0.79 0.62 0.41 0.16 0.08 0.04 gaseous state7 - 2.0 - 2.6 - 2.6 -2.5 - 2.6 - 2.9 - 2.6 - 2.9 - 0.7 - 0.9 0,3 - 0.3 - 0.3 - 0.2 - 0.6 - 1.7 -0.5 - 0.9 - 0.3 - 0.2 0.3 0.2 -0.1 b - - 6.2 - 15.6 -21.8 - 22.7 - 24.9 - 25.4 - 25.4 - 25.7 6.3 ~~ 20 30 35 35 30 30 30 45 C(4) / \ a Assignment of the carbon nuclei: yo i::: ; signal-to-noise ratio not sufficient.C(2) 'N'BERNSTEIN, KITAEV, MYCHEL, PFEIFER AND FINK 100 Hz H 763 FIG. 1 .-Nitrogen-15 n.m.r spectra of pyridine adsorbed on partially dehydroxylated silica gel [(a) and (b)] and on phosphorus-modified silica gel [(c) and (41; 0 is the coverage in monolayers and N is the number of scans: (a) I9 = 1.35, N = 200; (6) 19 = 0.04, N = 200000; (c) I9 = 1.69, N = 1000; ( d ) 6 = 0.33, N = 187000. coverage-dependent 15N resonance shifts to higher field (cf. also fig. 2) indicate appreciable influence by the adsorption sites on the electron density at the nitrogen atom. Moreover, the shifts are constant for coverages lower than ca. 0.5 statistical monolayers. This behaviour indicates the occurrence of a strong complex. The 15N n.m.r. shift remains constant as long as the number, N , of adsorbed molecules is smaller than the number, N,, of adsorption sites.For N > N,, a fast exchange between adsorbed and free molecules occurs giving rise to a decreasing shift.* From the bending point of the curve at ca. 0.5 statistical monolayers in fig. 2 the number of the adsorption sites is derived as ca. 1.2 nm-2. This value is in good agreement with the results obtained from the investigation of acetone adsorptions (1.4 nm-2). The results of these measurements and the order of magnitude of the 15N resonance shifts suggest that a strong hydrogen-bonding interaction between the lone-pair764 PYRIDINE ADSORBED ON SILICA GEL electrons of the nitrogen atom and the proton of the surface hydroxyl group occurs. The very low values for the resonance shifts of the 13C nuclei which are not in the vicinity of the nitrogen atom are consistent with this explanation because a specific interaction between the n-electron system of the molecule and the surface should give rise to considerable 13C resonance shifts of all nuclei in the ring.A strong interaction with Lewis-acid sites in unlikely to occur since such sites were 8 (monoiayers) FIG. 2.-Nitrogen-1 5 resonance shifts for pyridine molecules adsorbed on partially dehydroxylated ( x ) and phosphorus-modified (0) silica gel. Observed shift 6 (ppm) plotted against coverage B (monolayers). not observed for silicas pretreated at 673 K2 (cf. also the i.r. results described below). In the i.r. spectra of pyridine adsorbed on the partially dehydroxylated silica gel we observed bands which are typical of the hydrogen-bonded molecule in the region 1400-1700 cm-l (cf.fig. 3). The different complexes formed with adsorbed pyridine molecules [hydrogen bonds (HPy), interaction with Brarnsted-acid sites (pyridinium ion, BPy) and with Lewis-acid sites (LPy)] can be identified with the aid of the typical vibrations vSa, vlga and vlgb. Table 2 shows the values of these vibrations for the various complexes. The experimental results for adsorbed pyridine are listed in table 3. With increasing number of adsorbed pyridine molecules the extinction of the HPy bands also increases (bands at 1446, 1490 and 1595 cm-l, cf. fig. 3). Simultaneously the extinction of the stretching vibration band, vOH, of the free silanol groups decreases and a shifted and broadened stretching-vibration band appears (cf.fig. 3). In our measurements its shift is ca. 770 cm-l, which is in accordance with the values observed by other authors. This value is typical of a strong hydrogen-bonding interaction between the pyridine molecules and the surface hydroxyl groups. The species HPy was already desorbed at room temperature, and at 373 K the desorption was complete. The broad band at ca. 3680 cm-l is due to the vicinal silanol groups, which are hydrogen-bonded to each other. This band partially overlaps the intense stretching band of the free OH groups.BERNSTEIN, KITAEV, MICHEL, PFEIFER AND FINK J I 1 ~ 3747 6 5 3680 765 wavenumberlcm-' FIG. 3.--I.r. spectra of pyridine adsorbed on partially dehydroxylated silica gel: ( 1 ) silica gel pretreated at 673 K ; pyridine pressure: (2) 70, (3) 270 and (4) 660 Pa; ( 5 ) evacuated at room temperature; (6) evacuated at 373 K.The occurrence of complexes of types LPy and BPy could not be inferred from the spectra of unmodified specimens. ADSORPTION ON Si0,-P Resonance shifts and linewidths for the n.m.r. measurements are summarized in table 3. Owing to the decreasing intensity and the increasing linewidth, 13C resonance lines could not be detected for coverages c 1.2 statistical monolayers at room temperature. Therefore we recorded the 13C spectra at 373 K. However, 15N n.m.r. measurements could still be performed at room temperature. The 15N resonance lines (cf. fig. 1) are more strongly shifted to higher field (cf.also fig. 2) and the linewidths are larger (cf. tables 1 and 3) than for unmodified silica gel. The enlargement of the linewidths occurring, especially for low coverages, indicates the smaller mobility (by a factor > 2) of the pyridine molecules owing to the stronger interaction with the OH groups on the modified silica-gel surface. Infrared spectroscopic investigations of phosphorus-modified silicas1' 9 l8 indicated a larger acidity of surface POH groups compared with SiOH groups on unmodified silica. Pohle and Fink3 observed a stronger decrease in the wavenumber of the vOH band of the POH groups under the influence of pyridine molecules adsorbed on phosphorus-modified aerosil. The conversion of a proportion of pyridine molecules to protonated species leads to drastic changes in the n.m.r.spectrum (cf. the last row in table 3). We observe a strong 15N resonance shift in that direction where in general the resonance lines of the pyridinium ion appear, but the shift is still much less than the value for the ion (SI = - 115 pprnl6* 19). The changes in the 13C n.m.r. spectrum (recorded at a higher temperature) are small.766 PYRIDINE ADSORBED O N SILICA GEL TABLE 2.-cHARACTERISTIC I.R. VIBRATIONS OF PYRIDINE IN VARIOUS COMPLEXES” 3* ’-12 AND WAVENUMBERS OF ADSORBED PYRIDINE FOR 930 Pa PYRIDINE PRESSURE (a) AND AFTER 4 h DESORPTION AT ROOM TEMPERATURE (b) neat liquid hydrogen-bonding interaction (HPy) pyridinium ion (BPy) interaction with Lewis-acid sites (LPy) adsorbed on partially dehydroxylated silica gel adsorbed on Si0,P 1439 1440- 1448 1535-1 550 1445- 1460 (a) 1446s (b) 1448 w (a) 1448 w 1535 m 1552 m (b) 1535 m 1\552 m 1482 1580 1 48 5- 1 490 1 58 5- 1 600 1484- 1490 1638- 1640 1488-1500 1605-1635 1490 w 1595 s 1580 (shoulder) 1597 w 1493s 1598w 1630 w 1640 w 1493s 1630w - 1640 w - - - a Intensities: s, strong, m, medium, w, weak.Assignment of the characteristic vibrations follows ref. (1 3) and (14). TABLE CARBON-^^ AND NITROGEN-15 RESONANCE SHIFTS AND LINEWIDTHS FOR PYRIDINE MOLECULES ADSORBED ON Si0,-P (13C N.M.R: 373 K, 16N N.M.R.:ROOM TEMPERATURE; FOR DEFINITION OF SYMBOLS Cf. TABLE 1) coverage 13C n.m.r 15N n.m.r (mono- layers) 4C(2)1 m 3 ) 1 4c(4)1 W) A V+ 5.08 - 0.9 -0.2 0.1 - 2.7 29 1.68 - 2.2 0.4 0.5 - 18.8 47 1.19 - 2.2 0.5 0.4 - 26.6 40 0.8 1 - 3.7 - 0.6 0.5 - 26.6 40 0.61 - 3 k l 0+1 o+ 1 -28f1 55f 15 0.31 - 4 f 2 Of2 2 f 2 -31f2 140+20 pyridinium - 7.7 5.0 12.4 -115.1 ion in solution, ref.(15) and (16) These data show that the basis for the interpretation of the resonance shifts is a fast exchange between the adsorbed pyridine molecules (resonance shift SM) and their protonated form (pyridinium ion, resonance shift dl, relative number pI). Hence, the observed resonance shift dabs is given by Sobs = PI s* + (1 --PI) 6 M .BERNSTEIN, KITAEV, MICHEL, PFEIFER A N D FINK 36\45 l J ' 2 767 1 1 1 I 1400 1600 wavenumber/cm -' FIG. 4 . 4 . r . spectra of pyridine adsorbed on Si0,-P: (1) silica gel after modification and activation at 673 K; pyridine pressure: (2) 130, (3) 400, (4) 660 and (5) 930 Pa; spectra taken after evacuation at (6) room temperature, (7) 473 and (8) 573 K.The formation of pyridinium ions in decationated zeolites was investigated by means of 13C n.m.r. and 15N n.m.r. spectroscopy in ref. (6) and (19), respectively. With decreasing pore-filling factor the observed resonance shift, dabs, approaches a value which is typical for the pyridinium ion. However, in contrast to the measurements of Gay,20 who studied the formation of pyridinium ions on pure silicas by the addition of a known number of HCl molecules (up to a molar ratio of pyridine to HCl > l), in ref. (6) and (1 9) a very strong broadening of the observed carbon- 1 3 and nitrogen- 1 5 resonance lines occurred in the region of p I values where all molecules were protonated. This broadening is due to the much lower mobility of the protonated species.,' The broadening of the 15N resonance line of pyridine adsorbed on Si0,-P surfaces also indicates a reduction in the molecular mobility.Owing to the large widths and the smaller surface areas of the Si0,-P specimens, measurements at coverages 0.3 monolayers could not be performed in spite of the high enrichment with nitrogen-1 5 nuclei. Taking the resonance shift dobs = -31 ppm observed for a coverage of 0.3 statistical monolayers, together with the values dM = -26 ppm (for adsorbed pyridine molecules in unmodified silica-gel samples) and = - 115 ppm, we derive the number, N,, of pyridinium ions formed on Si0,-P surfaces as NI < 0.05 per nm2. In fig. 4 i.r. spectra for the adsorption on Si0,-P are shown.Besides the OH stretching vibration band of the free SiOH groups (qOH = 3747 cm-l), a band due to the free POH groups appears (FOH = 3668 em-l). The additional weak band at 3645 cm-l may be attributed either to the occurrence of another type of POH group [e.g.768 PYRIDINE ADSORBED ON SILICA GEL geminal groups, P(OH),, hydroxyl groups on clusters of phosphorus oxide] or to species of the type - formed after the oxidation of trivalent phosphorus. According to the i.r. spectra the concentration of POH surface groups is of the same order of magnitude as the concentration of free SiOH groups. If pyridine molecules are adsorbed on Si0,-P there are clear differences relative to pure silica. Even at the lowest adsorbate pressures we observed bands (at 1552 and 1490 cm-l) characteristic of the pyridinium ion (cf.fig. 4 and table 2). Their appearance is accompanied by a strong decrease in the extinction of vOH of the free POH groups but by only a small change in the wavenumber, FoH, for the free SiOH stretching vibration. With increasing pyridine pressure we obtain the band characteristic of LPy complexes (1630 cm-l) and for a pressure > 530 Pa the bands characteristic of the HPy species appear. The appearance of a further band at 1535 cm-l also points to protonated pyridine molecules. The difference in wavenumbers (1 535 and 1550 cm-l) could be due to the different types of POH groups already mentioned. Desorption at room temperature removed the HPy species completely. The BPy and LPy species could be removed completely only at a temperature of 573 K (cf.fig. 4). Simultaneously, the extinction of the vOH band for the free POH groups increases, which indicates the participation of the POH groups in the formation of the pyridinium ions. Low et aL5 did not observe protonated molecules for Cabosil pretreated at 873 K. Since pyridinium ions can be detected unambiguously under our conditions this result5 suggests that Brsnsted-acid sites are removed after pretreatment at 873 K. CONCLUSIONS The specific interaction between pyridine molecules and adsorption sites on partially dehydroxylated silica gel occurs via the formation of hydrogen bonds. These bonds are formed between a proton of the hydroxyl group and the lone-pair electrons of the nitrogen atom and causes a strong nitrogen-15 resonance shift to higher field.The only small changes observed in the carbon-13 n.m.r. spectra indicate that no specific interaction between the n-electron system and surface sites occurs. In the i.r. spectrum the characteristic bands of the hydrogen bond appear. The formation of pyridinium ions was observed in the case of phosphorus-modified silica gel. From the strong resonance shift of the nitrogen-15 lines a concentration of 0.05 protonated molecules per nm2 was derived. The i.r. spectra reveal that the protonation proceeds on POH groups. G. Curthoys, V. Ya. Davydov, A. V. Kiselev, A. S. Kiselev and B. V. Kuznetsov, J. Colloid Interface Sci., 1974, 48, 58. W. Pohle and P. Fink, 2. Phys. Chem. (Frankfurt am Main), 1978, 109, 77. W. Pohle and P. Fink, 2. Phys. Chem. (Frankfurt am Main), 1978, 109, 205. Th. Bernstein, H. Emst, D. Freude, I. Junger, J. Sauer and B. Staudte, 2. Phys. Chem. (Leipzig), 1981, 262, in press. M. J. D. Low and P. Ramamurthy, J . Phys. Chem., 1968, 72, 3161. H-J. Rauscher, D. Michel, D. Deininger and D. Geschke, J . Mol. Catal., 1980, 9, 369. R. D. Duthaler and J. D. Roberts, J. Am. Chem. SOC., 1978, 100,4969. Th. Bernstein, P. Fink, D. Michel and H. Pfeifer, J . Colloid Interface Sci., 1981, in press.BERNSTEIN, KITAEV, MICHEL, PFEIFER A N D FINK 769 H. Winde, P. Fink and A. Kohler, 2. Chem., 1977, 17, 41. H. Knozinger, Fortschrittsber. Kolloide Polym., 1971, 55, 16. l1 E. H. Parry, J. Catal., 1963, 2, 371. l2 T. R. Hughes and H. M. White, J. Phys. Chem. 1967,71, 2192. l3 L. Corrsin, B. J. Fax and R. C. Lord, J. Chem. Phys., 1953, 21, 1170. l4 D. Cook, Can. J. Chem., 1961, 39, 2009. I5 E. Breitmaier, G. Haas and W. Voelter, Atlas of Carbon-13NMR Data (Heyden, London, 1975). l6 I. Witanowski, L. Stefaniak, S. Szymanski and H. Januszewski, J. Magn. Reson., 1977, 28, 217. P. Fink, W. Pohle and A. Kohler, 2. Chem. 1972, 12, 117. M. L. Hair and W. Hertl, J. Phys. Chem., 1970, 74, 91. D. Michel, A. Germanus and H. Pfeifer, J. Chem. SOC., Faraday Trans. I , 1982, 78, issue 1. I. D. Gay, J. Catal., 1977, 48, 430. H-J. Rauscher, D. Michel and H. Pfeifer, J. Mol. Catal., 1981, in press. (PAPER 1 /476)
ISSN:0300-9599
DOI:10.1039/F19827800761
出版商:RSC
年代:1982
数据来源: RSC
|
12. |
Thermal behaviour of 9-cyanoanthracene photodimer (9-CNAD) |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 771-784
Donato Donati,
Preview
|
PDF (1220KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. 1, 1982, 78, 771-784 Thermal Behaviour of 9-Cyanoanthracene Photodimer (9-CNAD) BY DONATO DONATI AND PIERO SARTI-FANTONI Centro di Studio del CNR sulla Chimica e la Struttura dei Composti Eterociclici e lor0 Applicazioni, c/o Istituto di Chimica Organica, Universita di Firenze, Via Gino Capponi 9, 50121 Firenze, Italy AND GIULIO G. T. GUARINI* Istituto di Chimica Fisica, Universita di Firenze, Via Gino Capponi 9, 50121 Firenze, Italy Received 25th March, 1981 The monomerization reaction of 9-cyanoanthracene photodimer (9-CNAD) in the solid state has been reinvestigated both in isothermal and in linearly increasing temperature experiments using thermal methods and optical microscopy. From both kinds of calorigrams the heat of monomerization has been deduced in agreement with previous findings; the heat of melting of the crystalline monomer (9-CNA) formed during the monomerization reaction has also been redetermined.With supporting evidence from optical microscopy studies, a three-stage mechanism is proposed for the monomerization reaction ; agreement with recent reports is found only for the induction period. Our isothermal experiments show that, in the temperature range of interest, the solid-state monomerization reaction of 9-CNAD requires times far longer than those recently reported by other authors. Additional evidence is presented supporting our previous interpretation of the nature of the initial exothermic peak typical of the monomerization thermal curves. Upon U.V. irradiation, anthracene and some anthracene derivatives give photodimers both in solution and in the solid state.l? As the photodimers are thermally reverted to the monomers, the importance of these compounds for the storage of solar energy has also been c ~ n s i d e r e d .~ ~ ~ In addition the thermal properties of some monomers and dimers have been inve~tigated.~ In particular it was found that the dimers studied were converted back to the corresponding monomers during melting point determinations; indeed Calas and Lalandel clearly pointed out that it is impossible to define exactly the melting point of the dimers owing to the simultaneous monomerization reaction. Then it appeared impossible to study the kinetics of the monomerization reaction by dynamic thermal methods, particularly for those dimers having the monomerization temperature above the monomer melting point, because of the simultaneity of the two thermal events.However, the literatures states that 9-cyano- 10-acetoxyanthracene dimer (9-CN- 10-ACAD) monomerizes at ca. 423 K, i.e. at a temperature lower than the melting point of the monomer (472-473 K). In addition 9-CNAD was reported to melt instantaneously with depolymerization near 478 K;' the same compound was also described as monomerizing under prolonged heating at temperatures much lower than the melting point of the monomer.' In a previous paper5 we investigated the thermal behaviour of anthracene and of some anthracene derivative photodimers in a reproducible way by using a Perkin-Elmer DSC-1 b differential scanning calorimeter.This study showed that 9-CNAD mono- merized before the melting point of the monomer if low scan speeds were used. It was thus possible to separate the exothermal monomerization reaction from the 77 1772 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER endothermic melting of the produced monomer, and we thus found it interesting to study the kinetics of monomerization of 9-CNAD as well as that of 9-CN- 1 0-ACAD, by using dynamic d.s.c. results.* Recently9 the thermal behaviour of 9-CNAD has been reconsidered by using both dynamic and isothermal data obtained by differential scanning calorimetry or by isothermal spectrophotometry ; in this work9 a different interpretation is reported for the initial peak typical of dynamic5? 8 y and isothermallo experiments performed by thermal methods; moreover in the same paper9 the reaction is reported as rather fast (ca.5 min for completion at 407 K by isothermal calorimetry) and relatively slow (ca. 240 min for 70% conversion at 400 K by isothermal spectrophotometry). Results differing from those reported in ref. (9) for the isothermal monomerization reaction by thermal methods have already been reported.1° From a kinetic point of view, zero order has recently been attributed to a reaction period defined as accelerat~ry;~~ l1 this appeared to deserve further consideration. According to the shape of the reported cal~rigrams,~~ l1 the attribution of zero order to the overall reaction in dynamic runsg is also questionable. In fact, a zero-order transformation results in thermal curves mainly parallel to the baseline in isothermal runs or increasing in dynamic experiments, the reaction rate being dependent only on temperature.The melting of 9-CNAD was reported as taking place at 480 K9 and later described by the same authors’l as occurring ‘ occasionally in crystalline samples following incomplete monomerization’; such melting of the dimer was never reported by us in previous s t ~ d i e s . ~ ~ 8 * lo The recently reported data prompted us to check our previous findings on the monomerization of 9-CNAD by renewed dynamic and isothermal experimentation by thermal methods. To improve the accuracy, the new determinations have been performed by a Mettler TA 2000 thermal analyser; this apparatus has a higher sensitivity than the Perkin-Elmer DSC-1 b.The thermal determinations were also supplemented by optical microscopic examinations of the progress of the reaction in 9-CNAD single crystals. EXPERIMENTAL 9-CNA prepared according to the literature12 was dissolved in benzene and irradiated with the glass filtered light of a 250 W G & C medium pressure mercury lamp to obtain the dimer. 9-CNAD so precipitated was filtered and recrystallized several times from THF. By this procedure crystals of different sizes may be grown. For the present work a batch of small crystals of ca. 40pm average size was mainly used. Large (millimetre-size) crystals for use with the optical microscope were grown by the same procedure just preventing a rapid evaporation of the solvent. An old batch of 9-CNAD crystals of ca.30 pm, prepared in 1970, was also used and found to give thermal curves indistinguishable from those obtained using freshly prepared material, thus showing that the dimer remains unaltered even after long storage at room temperature. Both isothermal and dynamic thermal curves were obtained by a Mettler TA 2000 apparatus. Because of the sublimation of 9-CNA, sealed aluminium pans both for the sample and for the reference were used ; excluding the experiments with millimetre-size single crystals, the samples were always gently pressed into the sample pans by means of the flat section of a glass rod of suitable diameter. Weighing was also performed before and after each experiment to check for eventual sample losses. In dynamic determinations the sample was heated using the fast heating rate (100 K min-l) of the apparatus up to 390 K.After a short period at this temperature (necessary to equilibrate the apparatus at the selected sensitivity) the run was started at the chosen scan speed. For the isothermal determinations the chosen temperature was reached again using the fast heating rate, the sensitivity of the apparatus was then increased up to the predetermined value, and the apparatus was allowed to equilibrate.D. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI 773 The whole procedure took ca. 3-5 min. Unless otherwise stated, the recorder paper was allowed to advance at 0.5 cm min-l. Area measurements for heat determinations were performed by polar planimetry and found to be in good agreement with the results of a Simpson integration procedure used mainly for kinetic purposes. The temperature scales of the plots referring to dynamic determinations have been corrected for At,,, according to the manufacturer's instructions.Corrections for the change in sensitivity of the thermocouples with temperature were disregarded for the monomerization reaction because in the temperature range of interest (403-443 K) the sensitivity does not change appreciably from the unit value. Conversely, this type of correction was applied to heat determinations concerning melting of the monomer formed. After each isotherm as well as after low-scan-speed dynamic experiments, the samples were heated dynamically up to 510 K to determine heats of melting. Occasionally, for the sake of comparison, the Perkin-Elmer DSC-lb was also used.Fluorescence properties of 9-CNAD crystals were checked by a Perkin-Elmer MPF 44A spectrophotofluorimeter, equipped with a solid-state device, using 365 nm exciting radiation. The fluorescence spectra show only two very-low-intensity peaks at 420 and 442 nm attributed" to very small amounts of monomer in the dimer matrix. RESULTS A. DYNAMIC DETERMINATIONS The use of the TA 2000 apparatus has confirmed our previous observations and in particular the dependence of the shape of the thermal curves on the scan speed (fig. 1). From the dynamic runs at low scan speeds (1 and 2 K min-l), in which the exothermal portion is well-separated from the endothermic melting of the formed monomer, values of the heats of monomerization (crystalline dimer + crystalline monomer) and of monomer melting were determined, being AH,,, = - 74.4 f 2.5 kJ (mol dimer)-l and AH,,,, = 26.9 f 0.2 kJ (mol monomer)-l, respectively.The superimposition of monomerization and melting is evident at intermediate scan speeds (8-24 K min-l), whereas at higher scan speeds (29 K min-l) only an exothermal event is recorded. In the latter case the heat evolved is due to the transformation crystalline dimer --* melt monomer and, within experimental error, is the algebraic sum of the above-reported heats (on a unit-weight base). Moreover the use of the more sensitive Mettler TA 2000 instrument allows a better definition of a series of peaks, beyond the first, which are superimposed (see fig. 1 for low-scan-speed curves) on the broad exotherm. These peaks have been seen previously8, lo in thermal experiments using the DSC-lb.The shapes and positions of these peaks are strictly reproducible using samples of the same batch under the same experimental conditions. The position of the whole exotherm and therefore the position of the first and subsequent peaks is shifted towards higher temperatures on increasing the scan speed. Thus the temperature of the first peak can be defined only if the corresponding scan speed and temperature at which the scan begins are specified. Similar behaviour has been described previously5 when the change in temperature range (AT) covering the whole decompo- sition was reported for 9-CNAD. The kinetic analysis of the dynamic thermal curves has also been performed for low-scan-speed runs by standard procedures,13 and an Arrhenius diagram in which the data are plotted considering the whole monomerization reaction as either a first-order or zero-order process is shown in fig.2. The first-order plot is exactly like the one already reported,8 being clearly formed by two rectilinear portions of different slope. The values deduced from the slopes for the activation energies differ slightly from those previously reported,8 being higher for the low-temperature portion (ca. 450 kJ mol-1 instead of ca. 360 kJ mol-l) and lower for the high-temperature portion (ca. 152 kJ mol-1 instead of ca. 176 kJ mol-l). Incidentally we note that in ref. (9) the774 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER FIG. 1 .-Dynamic thermal curves for the monomerization of crystalline 9-CNAD.For illustrative purposes the heights of the original calorigrams were multiplied by the factors 'f' listed below. (a) 5.61 mg, scan speed = 1 K min-l, f = 0.25; (b) 3.31 mg, scan speed = 2 K min-l, f = 0.5; (c) 5.17 mg, scan speed = 8 K rnin-I, f = 0.5; ( d ) 3.52 mg, scan speed = 24 K min-l, f = 0.7; (e) 4.68 mg, scan speed = 29 K min-l, f = 0.5. values for the activation energies quoted from our previous paper have been reversed : that referred as the low-temperature value is the high-temperature one and vice versa. Fig. 2 also shows that, as expected, a zero-order process is almost indistinguishable from a first-order one as long as the fraction decomposed is small (the low-temperature portion). As far as the disputed nature of the first peak is concerned,8v9p11 additional information has previously been obtained14 by d.s.c. on quickly cooling to room temperature a sample previously heated to just beyond the first peak.When the sample so treated was heated again under the same conditions, the typical first peak was absent from the broad exotherm corresponding to the monomerization (fig. 3). In slightly different experiments, in which the sample was quickly cooled to room temperatureD. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI 775 0 -2 -Y E: - -4 -6 I I 2.3 2.4 103 K / T I 2.5 FIG. 2.-Arrhenius plot deduced from dynamic experiments at scan speeds of 1 and 2 K min-'. The rate constants were computed using first-order (0 for scan speed = 1 and 0 for scan speed = 2) and zero-order (0 for scan speed = 1 and for scan speed = 2) rate equations.just before the sharp rise of the first peak, the usual initial peak followed by the broad exotherm was obtained upon heating again under the same conditions. The same behaviour has now been found in isothermal experiments (vide infra). B. ISOTHERMAL EXPERIMENTS Typical monomerization isotherms, obtained using the TA 2000 apparatus and shown in fig. 4, indicate that, in agreement with previous reports,lO the beginning of the exotherm and therefore the position of the first peak is progressively shifted776 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER 410 400 430 460 TI K FIG. 3.-Dynamic thermal curve, obtained by d.s.c., showing the effect of interrupting the experiment near the end of the first peak (see text).50 100 tlrnin I 150 FIG. 4.-Typical isothermal calorigrams for the monomerization of 9-CNAD. For illustrative purposes the heights of the original curves were multiplied by the factors ‘f’ listed below. (1) 6.99 mg, T = 404 K, f = 0.5; (2) 5.16 mg, T = 408 K,f= 0.5; (3) 6.16 mg, T = 410 K,f= 0.4; (4) 5.88 mg, T = 415 K,f= 0.3. The first and subsequent peaks have been lettered progressively. towards the origin on increasing the temperature of the isothermal experiments. Again, besides the first peak, several other peaks are superimposed on the rising portion of the broad exotherm. Thus isothermal and dynamic monomerization calorigrams are strictly similar provided that suitable values for temperature and scan speed are chosen. After completion of every monomerization reaction the melting of the formed monomer was determined dynamically (scan speed = 4 K min-l).The heat determinations were also performed (by polar planimetry) both for monomerization and melting, giving mean values and standard deviations as follows: AH,,, = - 77.3 _+ 4.7 kJ (mol dimer)-l and AH,,,, = 26.7 f 0.6 kJ (mol monomer)-l.D. DONATI, P. SARTI-FANTONI A N D G. G. T. GUARINI 777 The values found for the heat of melting of the monomer formed assured us that at the temperatures investigated the monomerization is complete when the apparatus is allowed to reach the final, strictly horizontal base-line. The times required for the completion of the monomerization reaction are longer than those reported in ref. (9), but are in agreement with those expected on the basis of our previously reported dynamic runs at low scan ~peeds.~ We want to point out here that, in our melting determinations following the isothermal monomerizations and the dynamic runs at low scan speed, no peak at 480 K attributable to the melting of the dimer 9*11 has ever been observed.Also, in isothermal experiments the effect of interrupting runs at times just after and just before the appearance of the first peak was investigated. In fig. 5(a) and 50 100 t/min 150 FIG. 5.-Isothermal heat curves showing the effect of interrupting the experiments just beyond (a) and just before (b) the first peak. For illustrative purposes the heights of the original curves are multiplied by 0.7; the first and subsequent peaks are lettered progressively.The vertical broken lines indicate the times at which the temperature of the experiment was resumed after interruption and cooling to room temperature. (b) the corresponding isotherms are reported showing that the first peak is no longer present [fig. 5(a)] when the sample has been previously heated to just beyond the first peak, while the small subsequent peaks remain. Conversely, if the sample had been previously heated to just before the onset of the first peak, the peak in question appears [fig. 5(6)] and the whole exotherm shows the usual shape apparently unaltered even if characterized by a shorter induction period (vide infra). For samples of the same batch, the ratio of the area under the first peak to the area of the whole exotherm is nearly constant (ca.773, at least in the temperature range investigated. However, the shapes of the peaks are strongly dependent on crystal dimensions and crystallization procedures. Owing to the initial stabilization of the TA 2000 instrument at the predetermined temperatures (this problem of stabilization period also occurs for d.s.c.), there are uncertainties in the precise determination of the time origins; thus the kinetic analysis of the decay portion of the isothermal calorigrams has been performed as described previo~s1y.l~ In agreement with dynamic runs (the high-temperature portion), in the temperature range investigated (400-41 8 K) the decay portion of the monomerization reaction of 9-CNAD is well-fitted by a first-order law (fig. 6). An Arrhenius plot of 26 FAR 1778 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER i - 1 -s! c I C FIG.6.-First-order plot of the decay portion of an isothermal calorigram showing the excellent agreement. (h, is the distance of the curve from the base-line at the time considered and is proportional to the reaction rate.) -2.5 .y c rn -3.5 \. \ \ + \ \* \ \. \. \. * \ \ \ \ * \. * \. \. * ' \ \ \ '. \ \ 24 2: 5 103 KIT FIG. 7.-Arrhenius plot for the first-order decay period of the isothermal monomenzation of 9-CNAD.D. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI 779 the calculated first-order rate constants is shown in fig. 7 from which the following activation parameters have been deduced: E, = 134.3 kJ mol-l and A = 7.2 x 1015. We also attempted a kinetic analysis of the first peak by allowing the recorder paper to advance at a rate of 5 cm min-l.The result was that up to a’ = 0.5 (where a’ is the fractional decomposition referred to the first peak only) plots of In a’ against time were linear, thus giving evidence of the existence of an exponential rate law for the rising portion of the first peak. Even if, owing to the previously described stabilization uncertainties, the zero time is not precisely known, we have determined the time needed, at several constant temperatures, to reach the top of the first peak from the instant at which the digital display of the TA 2000 instrument first reached the temperature chosen for the isotherm. These time intervals, including the undetermined stabilization period and a small portion (< 5%) of the reaction, are nevertheless believed to be proportional io (and will be referred to as) induction periods.An Arrhenius plot of these induction periods is reported in fig. 8 showing a good alignment of the experimental points. From -1 h - c .- E -2 T, \ w c - - 3 2.4 2.5 1 0 3 KIT FIG. 8.-Arrhenius plot for the induction periods of the isothermal monomerization of 9-CNAD. the slope a value for the activation energy (E, = 143.5 kJ mol-l) has been deduced, in strict agreement with previously reported data.g Moreover, plots of the fraction decomposed (a) were computed from the isothermal calorigrams by a Simpson integration procedure. These plots are collected in fig. 9 for a number of isotherms. c. OPTICAL MICROSCOPY STUDIES We have already described8 optical microscopy investigations performed under progressive heating in a temperature range covering the first peak.Plate 1 shows a series of micrographs obtained in this way by a Leitz Panphot microscope equipped with a Nikon camera, which were the basis of our previous description.8 Our isothermal calorigrams [ref. (10) and this paper], apart from showing the same general shape as the dynamic thermal curves at low scan speeds, allowed a determination of 26-2780 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER I I 2 1 t l h FIG. 9 . - a against t plots for a number of isothermal monomerizations of 9-CNAD; the discontinuities evident in the initial portions of the curves correspond to the first peaks. 0, 8.22 mg, 402 K; A, 6.99 mg, 404 K; H, 5.16 mg, 408 K; 0, 5.76 mg, 412 K; A, 5.88 mg, 415 K. the time necessary to reach the end of the first peak at a given temperature. Thus we have now performed isothermally the same type of experiment by optical microscopy.In plate 2 a series of micrographs from an isothermal experiment at 408 f 1 K is shown. Plates 1 and 2 clearly demonstrate that at both increasing and constant temperatures the same types of phenomena occur, namely: (i) after a certain period black spots oriented along preferred directions appear suddenly [plates 1 (c) and 2 (6) and (c)] ; (ii) alternatively and contemporaneously but again suddenly the blackening spreads over relatively large zones of the crystal [plates l(6) and 2(d) and ( e ) ] ; (iii) the two previously described features then grow darker and spread fairly regularly until complete darkening of the crystal occurs.In isothermal optical determinations the time required to reach the complete darkening of the crystal is in reasonable agreement with the time taken (deduced from the calorigrams at the same temperature) to reach the end of the first peak. Crystals heated isothermally on the hot stage of the microscope up to complete darkening in transmitted light were collected and used to perform isothermal (408 K) determinations by the TA 2000 instrument in order to evaluate the degree of monomerization reached in optical microscopy experiments. Fig. 10(a) refers to the isothermal calorigram so obtained and is to be compared with fig. l0(6), in which the isothermal decomposition at the same temperature of approximately the same weight of crystals from the same batch as those used in microscopy is reported.The run in fig. lO(b) was stopped after the beginning of the decay period, the sample pan was allowed to cool to room temperature, and was then opened. The crystals so treated were collected and examined in the microscope in transmitted light between crossed nicol prisms. Plate 3 shows that, after the above treatment, the crystals possess small crystallites, extinguishing upon rotation, which protrude from the edges. These are thought to be crystals of 9-CNA. When a crystal has reached complete darkening in transmitted light between crossed nicol prisms, it remains dark even if one or both polarizers are eliminated. Switching to reflected light, the crystal appears yellow and strongly diffuses light from the apparently unaltered surfaces.J .Chem. SOC., Faraday Trans. 1 , Vol. 78, part 3 Plate 1 PLATE 1 .-Progressive darkening of a crystal of 9-CNAD observed between crossed nicol prisms while the temperature of the hot stage was slowly increased (ca. 0.3 K min-l) from the melting point of sebacic acid (404-406 K) to the melting point of 2-benzal-4-phenylpseudo-oxazolone-5 (410 K). The spread of the reaction over relatively large zones is evident in the low part of frame (b) while fast reaction along preferred directions is evident in frame (c). Magnification: x 34. D. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI (Facing p . 780)J . Chem. SOC., Faraday Trans. 1, Vol. 78, part 3 Plate 2 PLATE 2.-Progressive darkening of a crystal of 9-CNAD observed between crossed nicol prisms in an isothermal experiment at 408 K.(a) t = 0; (b) t = 2; (c) t = 3; ( d ) t = 4; (e) t = 5 ; cf) t = 6 min. Spreading of the reaction along preferred transverse directions is evident in frames (b) and (c) while a spread over large areas is clear in frames ( d ) and (e). Magnification: x 43. D. DONATI. P. SARTI-FANTONI AND G. G. T. GUARINIJ . Chem. SOC., Faraday Trans. 1, Vol. 78, part 3 Plate 3 PLATE 3.-Transparent crossed-nicol-prism micrographs of small needle-like monomer crystals formed on the surfaces of 9-CNAD crystals partially decomposed inside the sample pans of the thermal apparatus. Extinction upon rotation by ca. 29' is evident. Magnification: x 62. D. DONATI, P. SARTI-FANTONI AND G. G.T. GUARINID. DONATI, P. SARTI-FANTONI AND G. G. T. GUARINI r- 12 24 36 tlmin L8 FIG. 10.-Isothermal (408 K) calorigrams of a sample of large crystals previously used in optical-microscopy determinations (a) and of approximately the same weight of unreacted crystals of the same batch (b). Curve (a) has been shifted to the right by an amount corresponding to the time needed to reach complete darkening of the crystal in optical-microscopy experiments at the same temperature (ca. 8 min). Base-lines were not drawn but a comparison with fig. 4, curve 2, shows that for curve (b) the run was interrupted after the beginning of the decay portion. DISCUSSION The aim of this work is to gain additional insight into, and to meet some of the discrepancies and different interpretations of, the thermal monomerization of 9-CNAD.5*8-1' A.HEAT DETERMINATIONS The heat data in this paper agree well with those recently redetermined by d.s.c.lo and reasonably well with results previously rep~rted.~? According to the present determinations, the ratio between the area of the melting peak and the area of the broad monomerization exotherm in the same experiment should be, on a weight basis, ca. 1 : 1.5 if no changes in sensitivity have been made between the two thermal events. Except for the top part of fig. 1 in ref. (9), in which the base-line cuts the endothermal melting peak, the same ratio can be deduced from the thermal curves in ref. (9) and (1 1)- The shape of the monomer melting peak also needs some discussion; indeed this has recently been reportedll as consisting of two peaks that the authors attribute to the subsequent melting of a crystalline and an amorphous phase of the 9-CNA formed in the monomerization.While we strongly doubt that an amorphous phase can exhibit a sharp melting point, particularly in a temperature range so near the melting point of the corresponding crystalline phase, the existence of a double maximum in the region of the melting peak, even if not as evident as that reported in ref. (1 l), has sometimes been observed in our d.s.c. experiments (but never in TA 2000 runs). In782 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER our case we attribute the splitting of the melting peak to the slightly retarded melting of crystalline 9-CNA sublimed on the inner part of the cover of the sample pan which, in d.s.c.but not with the TA 2000, may have a slightly lower temperature. Indeed, on carefully opening the d.s.c. sample pans after double-peak melting, a number of small needle-like crystals and/or small drop-like aggregates of crystals of 9-CNA have been observed, by microscopy, adhering to the inside of the aluminium pan covers. Conversely these features have not been observed on the inside of the TA 2000 pan covers. Indirectly a confirmation of this behaviour may be deduced from plate 3, in which needle-like crystals of 9-CNA are observed growing outwards from the partially decomposed parent 9-CNAD crystal. B. EVALUATION OF KINETIC PARAMETERS AND PROPOSED MECHANISM The kinetic analysis of the dynamic calorigrams performed as if the whole reaction were first or zero order (fig.2) gave us the same results as those previously reported8 for the first-order treatment, while excluding the possibility of an overall zero-order process. The existence of a first-order decay portion is confirmed by the kinetic analysis of the constant-temperature data. However, according to a better definition of the shape of the thermal curves (the number of peaks beyond the first one) the fitting of a first-order process to the low-temperature (high-activation-energy) portion of the dynamic thermal runs is probably an artifact and the corresponding activation energy, if meaningful, cannot be related to a single process. With the supporting evidence from optical microscopy, we believe that our isothermal and dynamic thermal determinations indicate the following sequence of events.(1) During the induction period monomer molecules begin to be formed (probably frdtn incipient dimer pairs) probably at defects randomly distributed within the parent dimer crystal and thus simulating a homogeneous distribution (yellowish appearance of the crystals). Very small submicroscopic nuclei of crystalline monomer are thus formed at the end of the induction period (assumed now not to include part of the first peak). (2) Three types of growth of nuclei are evidenced by microscopy: (i) very fast and along preferred directions; (ii) very fast and on particular planes; (iii) regular (i.e. not sudden or spasmodic) spread in all directions perpendicular to the line of observation. We assume that type (iii) is responsible for the general rising portion of the calorimetric (either isothermal or dynamic) exotherms, while the superimposed peaks (first and subsequent) are due to mechanisms of types (i) and (ii) which, because of their exothermicity and high rates, generate heat pulses.We tentatively attribute the fast stages (i) and (ii) to the progress of the reaction along dislocation cores and adjoining (elastically distorted) regions and to the spread of the reaction on slipped planes, respectively. However, if what we have named ‘directions’ are the traces of planes almost parallel to the direction of observation, growth mechanisms (i) and (ii) unify. The subsequent (and probably stress-assisted16) reaction of different planes characterized by progressively lower extension and/or reactivity can explain the presence of various peaks superimposed on the rising portion of the main broad exotherm.The fact that in isothermal and dynamic thermal experiments the shape and position of the peaks is strictly reproducible (if samples of the same batch are used and the experimental conditions replicated), as well as the observed fact that the shape of the peaks changes with crystal dimensions and crystallization procedures, are thought to be in agreement with our interpretation (uide infra under section C). (3) At the end of the acceleratory period, the situation is such that the original dimer crystal, even if preserving its primitive external shape, is so heavily internally cut byD. DONATI, P. SARTI-FANTONI AND G.G. T. GUARINI 783 planes of already formed crystalline monomer that it may be considered as consisting of small residual crystallites of dimer surrounded by Crystalline monomer. Thus we have the situation typically described1', for the first-order decay mechanism observed. When process (3) is also complete, the crystals usually preserve the shape of the original parent dimer crystals but must be considered as aggregates of extremely small particles of crystalline monomer. (The fact that decomposed crystals strongly diffuse light supports this interpretation.) The dimensions of monomer crystallites which constitute the aggregate simulating the original dimer crystal could not be determined by optical microscopy and are thought to be such as to simulate an amorphous phase when X-ray diffraction is used. Indeed the sharp melting observed on further heating the reaction product ensures that a crystalline monomer phase is formed during the reaction.C. NATURE OF THE FIRST (AND SUBSEQUENT) PEAKS The general shape of the thermal curves is the same in thermal experiments for both constant and linearly increasing temperatures (obviously curves of suitable temperatures and scan speeds must be compared). Thus, as expected, identical phenomena are taking place in the two cases. This is confirmed by microscopic examinations, at least for the initial portions of the monomerization reaction. Since our interpretation of the first peak in dynamic runs8 (exactly the same as is given here) was q~estioned,~~ l1 we wish to add some considerations to support our point of view.(1) If the first peak was due to crystallization of monomer formed during the induction period (< 5 7 3 , in our runs the absolute value of the heat of crystallization deduced from the first peak would be at least double the corresponding value for the heat of melting. (2) Again, assuming that the first and subsequent peaks are due to repeated crystallizations, it is not clear why the monomer formed in the decay period should not crystallize but rather form an amorphous phase, especially in the presence of previously formed crystalline monomer. The possibility that an amorphous phase is formed (in analogy to the case for 9-CN-10-ACAD8) is ruled out here by the melting experiments performed after each isothermal run. (3) The shape of the first and subsequent peaks changes with the crystal dimensions and with the crystallization procedures used to obtain the crystalline dimer.We believe that these changes in peak shape are better explained by our interpretation of the peaks than by the hypothesis based on the crystallization of the monomer. (4) The crystallization hypothesis could not explain the fact that, for a given sample, the first peak does not appear again on reheating after quick cooling (see Results sections A and B), while the subsequent peaks remain unaltered. Instead we believe that when a certain family of planes has reacted it will not react again on repeated heating. (5) The absence of the first peak in powdered materialsll is easily accounted for by the present interpretation.In fact, owing to the reported preparation,ll the powdered material is apt to be very poorly crystalline or not crystalline at all. This prevents the fast spreading of the reaction on preferred planes. CONCLUSIONS Our thermal and microscopic results confirm that the overall monomerization reaction of 9-CNAD consists of three subsequent stages as recently s~ggested.~9 l1 As to the mechanism of the three stages, our suggestion, based on our data, differs from784 THERMAL BEHAVIOUR OF 9-CNAD PHOTODIMER what has previously been indicated for both the second and third stages. This disagreement is due mainly to the impossibility that the dynamic thermal curves may be interpreted by an overall zero-order proce~s.~ However, in agreement with the above-mentioned r e p o r t ~ , ~ ~ l1 the first stage (induction period) of the reaction is attributed to the formation of submicroscopic monomer nuclei probably originating from incipient dimer pairs.Our microscopic results then show that the second (acceleratory) period of the reaction is of a complex nature and comprises different kinds of nucleus growth. Thus its interpretation in terms of one of the known kinetic laws is unsafe and further studies are needed for a definitive clarification. At the end of the second stage, the dimer crystal is so heavily intersected by crystalline monomer formed on preferential planes that the remaining portion of the reaction takes place by a first-order consumption of residual dimer entities surrounded by the crystalline monomer already formed.The first and subsequent exothermic peaks evidenced in both dynamic and isothermal experiments are due to fast spreading of the exothermal reaction on preferred planes or directions with formation of crystalline monomer. No endothermic peak at 480 K attributable to the melting of residual crystalline 9-CNAD was observed under our experimental conditions. On the grounds of the present and recently reported1* isothermal results, our previous report, based on d.s.c. dynamic experiments,s is confirmed. The product of the monomerization reaction is crystalline, i.e. the crystallization front is very near the reaction front, and no evidence of amorphous phases is found, since the reaction product (9-CNA) was always found to have a sharp melting point. Our findings show that thermal methods do give reliable results in the study of the monomerization of anthracene and anthracene derivative photodimers. We have also found that 9-CNAD crystals are stable at room temperature over a period of 10 years. This information may be of some interest in view of the possible utilization of 9-CNAD crystals for energy storage. R. Calas and R. Lalande, Bull. SOC. Chim. Fr., 1959, 763. R. Lalande and R. Calas, Bull. SOC. Chim. Fr., 1960, 144. W. R. Bergmark, G. Jones 11, T. E. Reinhardt and A. M. Halpern, J. Am. Chem. SOC., 1978,100,6665. T. Laird, Chem. Ind., 1978, 186. G. Guarini and P. Sarti-Fantoni, Mol. Cryst. Liq. Cryst., 1970, 6, 423. C. Dufraisse and J. Mathieu, Bull. SOC. Chim. Fr., 1947, 307. R. Calas and R. Lalande, Bull. SOC. Chim. Fr., 1952, 434. D. Donati, G. Guarini and P. Sarti-Fantoni, Mol. Cryst. Liq. Cryst., 1972, 17, 187. E. M. Ebeid, S. E. Morsi and J. 0. Williams, J. Chem. SOC., Faraday Trans. I , 1979, 75, 11 1 1. lo D. Donati, G. G. T. Guarini and P. Sarti-Fantoni, Mol. Cryst. Liq. Cryst., 1981, 65, 147. l1 E. M. Ebeid, S. E. Morsi and J. 0. Williams, J. Chem. SOC., Faraday Trans. I , 1980, 76, 2170. l2 L. F. Fieser and J. L. Hartwell, J. Am. Chem. SOC., 1938, 60, 2555. l3 A. K. Galwey and M. E. Brown, Thermochim. Acta, 1979,29, 129; see also G. G. T. Guarini and R. l4 D. Donati, G. G. T. Guarini and P. Sarti-Fantoni, unpublished results. l5 G. G. T. Guarini, R. Spinicci, F. M. Carlini and D. Donati, J. Thermal Anal., 1973, 5, 307. l6 A. K. Galwey and G. G. T. Guarini, J. Chem. SOC., Chem. Commun., 1978, 273. l7 D. A. Young, Decomposition of Solids (Pergamon Press, Oxford, 1966). M. E. Brown, D. Dollimore and A. K. Galwey, in Comprehensive Chemical Kinetics, ed. C . H. Bamford and C. F. H. Tipper (Elsevier, Amsterdam, 1980), vol. 22. Spinicci, J. Thermal Anal., 1972, 4, 435. (PAPER 1 /484)
ISSN:0300-9599
DOI:10.1039/F19827800771
出版商:RSC
年代:1982
数据来源: RSC
|
13. |
Reactions involving electron transfer at semiconductor surfaces. Part 11.—Oxygen isotope exchangeviaphotoinitiatedR1,R0and place exchange processes on ZnO and TiO2 |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 785-801
Joseph Cunningham,
Preview
|
PDF (1298KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I , 1982, 78, 785-801 Reactions Involving Electron Transfer at Semiconductor Surfaces Part 1 1 .--Oxygen Isotope Exchange via Photoinitiated R,, R, and Place Exchange Processes on ZnO and TiO, BY JOSEPH CUNNINGHAM* AND EDWARD L. GOOLD Chemistry Department, University College, Cork, Ireland AND JOSE L. G. FIERRO Instituto de Catalysis y Petroleoquimica, C.S.I.C., Madrid, Spain Received 31st March, 1981 Comparisons are made between the changes in isotopic composition of isotopically pre-equilibrated (i.p.eq.) and isotopically non-equilibrated (i.n.eq.) gaseous oxygen in contact with prereduced or preoxidised samples of ZnO and TiO, at room temperature. In the absence of illumination a place exchange (P.x.) process predominated in the oxygen isotope exchange (0.i.e.) detectable at low pressures, ca.8 x Torr, of i.p.eq. 0, upon contact with preoxidised samples, whereas a homophase, &-type 0.i.e. process predominated for i.n.eq. (l60, + '*O,) contacted with prereduced samples at pressures of ca. lo-' Torr. The latter R,, activity was removed by preoxidation but light restored it with quantum efficiency > 6 for pure ZnO and > 30 for lithium-doped ZnO. A mechanism is described to account for the correspondingly high turnover achieved on each site photoactivated by light. For i.p.eq. 0, exposed to intense illumination in contact with prereduced or preoxidised ZnO, a heterophase &-type process, accompanied by a faster &-type process, predominated and reasons for this are considered. Residual hydroxyls affect the ratio of 0.i.e.processes on TiO, surfaces. Several unusual and interesting heterogeneous photoreactions, which imply the photoassisted cleavage or formation of the oxygen-oxygen bond in molecular 0, at relatively low temperatures, have recently been reported as occurring on u.v.-illumi- nated surfaces of metal oxides. These include involvement of molecular 0, as reactant or product in : (i) selective photo-oxidation of hydro~arbons;l-~ (ii) photoassisted splitting of water over Ti0,;4-s (iii) conversion of CO, plus H20 vapour to methane and oxygen over SrTiO,' and (iv) heterophase oxygen isotope exchange over several metal Such photoassisted oxygen-handling properties of these illuminated gas/metal oxide interfaces can often be studied at room temperature and so can be distinguished with ease from the more conventional thermally-assisted oxygen- handling properties, which only manifest themselves at much higher temperatures.The distinction is particularly clear in the case of the &-type oxygen isotope exchange (0.i.e.) process represented by eqn (1 a) which has been reported by Pichat et al. to occur over ZnO, TiO,, SnO, and ZrO, exposed to intense u.v.-illumination at room temperature,8 but which frequently requires temperatures in excess of 0.4 T, (where T,, the Tamman temperature, is the melting point in K) for detectable operation under thermal-, rather than photo-, activation.1°-12 R,-type : ls0(s) + l8O,(g) + l*O(s) + l60l8O(g). (1 a) Transfer to the gas phase of a monatomic oxygen species, ls0(s), originally associated with the solid surface is required for operation of &-type exchange as per eqn (I a), 785786 PHOTOINITIATED OXYGEN ISOTOPE EXCHANGE whether under thermal- or photo-activation of the surface.This has been directly and conveniently studied in the present and previous work through the use of mass spectrometry to follow the progressive enrichment of oxygen- 16 (and an equivalent depletion of oxygen-18) in a gas phase initially consisting of an isotopically pre- equilibrated (i.p.eq.) mixture of (160, + l60l8O + 180,). An asterisk will be used throughout the text to indicate illuminated O,*/metal oxide interfaces and photoassisted process thereon, e.g. R;" denotes a photoassisted process with the stoichiometry and 0.i.e.effect of eqn (1 a). The approach adopted in the present study is, first, to establish experimentally the principal features of photoassisted 0.i.e. processes over u.v.- illuminated ZnO and TiO, surfaces at room temperature and, secondly, to examine the extent to which such features may be understood in terms of R,*, R;" and R,*-type 0.i.e. processes, i.e. in terms of photoassisted analogues of the principal routes to 0.i.e. delineated by published work on thermally-assisted 0.i.e. over metal oxides at elevated temperatures. Thermally-assisted 0.i.e. processes at oxygen/metal oxide interfaces have been categorised as homophase when only the isotopic composition of one phase (usually the gas phase) is thereby changed, or as heterophase when the isotopic composition in both phases is affected.1°-12 It has also been proposed that all possible 0.i.e.processes can be categorised as R,, R, or R, in type, where the subscripts indicate the number of monatomic oxygen species of the solid surface which are exchanged into the gas phase per 0.i.e. event. Thus eqn (1 b) represents homophase o.i.e., which is also R, in type since it involves zero transfer of oxygen from metal oxide surface to gas phase (1 b) A convenient experimental test of a pure R, activity is that it involves no change in atom fraction of l60 or l80 in the gas phase. Eqn (1 a) represented one of the two heterophase processes usually distnguished in a thermally-assisted 0.i.e. process, the second being represented in eqn (1 c ) Ro-type : l6O,(g) + l80,(g) G= 2160180(g).&-type: 2l6O(s) + 180,(g) f 2l8O(s) + l6O,(g). This envisages the exchange of one gaseous 1802 molecule during a single stay on the surface for two monatomic oxygen species associated with the metal oxide surface. The detailed mechanism of this R,-type process has not been resolved but the importance of a high concentration of surface defects, such as can readily be expected close to the Tamman point, has been recognised.1° It has also been argued that a convenient experimental test for large contributions by R,-type activity to overall 0.i.e. observed under thermal activation is the occurrence of a minimum in plots of [P:4/(P32 denotes the partial pressure of component with m/z = 34, i.e. l60l8O). Table 1 illustrates that, by employing as reactant an i.p.eq.mixture containing (25% 160, + 50% l60l8O + 25% l8O0,), pure R, and R, processes should be readily distinguishable by their differing effects upon isotopic composition of the gas phase. A few w o r k e r ~ ~ ~ ~ l4 have pointed to place exchange [P.x. cf. eqn (2a) and (2b)l as another heterophase process for cases where the metal oxide carried an appreciable surface coverage by 0, chemisorbed non-dissociatively onto its surface during surface preoxidation, i.e. prior to the admission of isotopically enriched gas to study 0.i.e. Changes possibly arising in isotopic composition of the gas phase by this process may be represented as in eqn (2a) and (2b) for cases where the preadsorbed oxygen is (2 a) 160,(chem) 160,(chem)+160180(g) -+ 160180(chem)+160,(g).(2 b) against time of exchange l60,(chem) + 180,(g) + 180,(chem) + 1602(g)TABLE 1 .-SURFACE EXCHANGE PROCESSES CAPABLE OF CHANGING (OR LEAVING UNCHANGED) THE ISOTOPIC COMPOSITION OF AN INITIALLY I.P.EQ. GASEOUS OXYGEN MIXTURE (25 % l6O2, 50 % 160180, 25 % 1802) CONTACTED WITH AN OXIDE SURFACE exchange process reaction net overall effects at m/e = relative effects at m/e = 32("O2(g)) 34(160180(g)) 36(1802(g)) 32(1602(g)) 34(l6Ol80(g)) 36(I8O2(g)) Ro l60,(g) + 1802(g) * l60'80(g) 0 l60-y1) + '802(g) e 160180(g) + 180-yl) 0 160-n(l) + '602(g) * '60,(g) + 16O-yl) 0(+1-1) 1 6 0 - y l ) + 160'8O(g) * '602(g) + 18O-yl) + I (1) + 1802(g) * '602(g) + 2180--n(l) + 1 (1) + 1602(g) e '602(g) + 2160--n(1) 2160-y1) + '60'80(g) * '602(g) + 16O-yl) + '80-"(1) (+ 1 160Fn(ads) + "o2(g) 1602(g) + 180,n(ads) +1 "Oi"(adS) + 1602(g) 1602(g) + 160in(ads) 0(+1-1) 2160-?l 2160-7t R1 { O(+ 1 - 1) R2 l60Fn(ads) + l60l8O(g) e l1602(g) + 160180-n(adS) (+ 1 0 +1 0 - 1 0 0 - 1 0 0 - 1 0 0 0 0 - 1 0 -: 0 1 - + l -: ) - + 3 -2 - 1 -A ) - + 3 -2 - 1 0) x 2 0) x 2 4 00 4788 PHOTOINITIATED OXYGEN ISOTOPE EXCHANGE Such place exchange processes can be expected to be of particular interest in relation to photoassisted 0.i.e.processes, since photosorption of molecular oxygen has been widely re~0rted.l~- l6 Together these may offer a lower temperature route to 0.i.e. than the generation of surface defects needed for conventional R,-type activity. Reference to table 1 shows that predicted effects of pure R,-type or p.x.-type processes operating in isolation would be indistinguishable.However, if the monatomic surface oxygen species required for R,-type 0.i.e. (but not for P.x.) can also become involved in &-type o.i.e., detectable differences in overall effect upon isotopic composition of the gas phase may develop between p.x. and (R, plus Rl). EXPERIMENTAL MATERIALS Two types of additive-free TiO, powder were studied: TiO, (MR-l28), a chloride-free rutile sample of surface area 5 m2 g-l, supplied by New Jersey Zinc Co., and TiO,(A), an anatase sample of surface area 7 m2 g-l, supplied by Tioxide Ltd. Powdered zinc oxide of surface area 4 m2 g-l, supplied by New Jersey Zinc Co. (code ZnO-SP 500) as being chloride-free and of high purity, was used for most experiments. E.s.r. examination confirmed the absence of transition metal ions down to ppm levels.Doped zinc oxides, derived from this material by admixing Li,O or In,O,, were also obtained by courtesty of New Jersey Zinc Co. and are designated as Li-ZnO (446 ppm Li) and In-ZnO (1 10 ppm In). Metal oxides were used without further purification, except for vacuum outgassing to remove adsorbates. Samples whose vacuum pretreatment prior to an experiment consisted of heating to 650 K for 16 h under continuous evacuation, followed by cooling to 295 K in UCLCUO, are referred to as ‘prereduced’ or ‘oxygen-deficient’, MO,,-,, in view of evidence1’ for loss of oxygen from such surfaces in vacuo at 673 K. Large differencesemerge in this study between samples with such oxygen-deficient surfaces and the same sample with its surface in a ‘preoxidised’ or excess-oxygen condition, i.e.MO,,,,! with the magnitude of 6 increasing in proportion to the amount of chemisorbed oxygen retained by the surface. Preoxidation to differing values of 6 was sought by : first, heating the metal oxide sample to 673 K for various times and under various pressures of lS0,, secondly, cooling samples, protected by a liquid-N, trap, to 295 K in the residual 0, and, finally, evacuating the sample briefly (usually 6 min) to < Ton at 295 K prior to introducing lsO-enriched 0,. This was utilised in two forms, both supplied by Norsk Hydro : an isotopically non-equilibrated (i.n.eq.) equimolar mixture of + l80,) and an isotopically pre-equilibrated (i.p.eq.) mixture, ca. (25% lsOZ + 50% ls0l8O + 25 % ls0,), with approximately equal atom fractions of and ls0.Differing pressures at the O,/MO interfaces were found to be convenient for experimental study of R,, R, and p.x. processes. Thus an aliquot of (i.p.eq.) 180-enriched gas, sufficient to produce a system pressure of between 5 x and 5 x lo-, Torr, was found convenient for study of the p.x. process. However, several aliquots, producing system pressures of 5 x lo-, to 5 x 10-1 Torr, were convenient for the study of R:. Provision was made for continuous sampling of the oxygen gas to a mass spectrometer at a rate which did not significantly deplete the system oxygen pressure during several hours in the dark. A Pirani gauge on the reactor volume allowed observation of any changes in system pressure due to photosorption processes, but these were not found to be significant on the extensively dehydroxylated surfaces. ILLUMINATION A N D KINETIC ANALYSIS When intense illumination was required, metal oxide powder distributed over a 10 cm length of the quartz photoreactor was uniformly and continuously irradiated with the output of a 15 cm long, 500 W medium pressure mercury arc lamp (Hanovia 6744).Normally this quartz lamp was surrounded by a Pyrex water-cooled jacket to minimise sample heating and to transmit to the sample only photons of 1 2 300 nm. Further restriction on the wavelength range transmitted to the sample was achieved by various Wratten gelatin filters (no. 38A for 340 < 1/nm Q 640 or no. for 460 Q R/nm d 640). In some studies of an extremely light-sen- sitive R,*-type exchange, low intensity illumination was obtained by employing normal roomJ.CUNNINGHAM, E. L. GOOLD AND J. L. G. FIERRO 789 light (300 d A/nm ,< 800) or room light filtered by the no. 38A filter alone or by the no. 40 filter alone. RESULTS R:-TYPE PROCESS ON OXYGEN-DEFICIENT SURFACES EXPOSED TO INTENSE ILLUMINATION AT Po2 x 0.1 Torr Non-illuminated, oxygen-deficient surfaces of pure and doped zinc oxides were shown in recent work from this laboratory to promote a rapid &-type equilibration in an i.n.eq. (1602 + 1802) l8 The initial pre-illumination section of fig. 1 (a) shows, however, through the absence of any detectable changes in isotopic composition, that 'dark' operation of this rapid A,, process on the oxygen-deficient surface produced no detectable changes in the i.p.eq.mixture (25 % 1602 + 50 % l60l8O + 25 % 1802). Data in fig. 1 (a) (taken during illumination of this i.p.eq. O,*/ZnO(,-& interface by 0.21 I 1 I 1 I 0 1 2 3 4 time/h 0.50 0.47 0.3 0.2 0.1 2 3 1 4 x;,z FIG. 1 .-Variation of isotopic composition following introduction of i.p.eq. (l60, + 1s0180 + *80,) at 0.18 Torr to a prereduced ZnO surface at 300 K. (a) Constancy of mole fractions of 1 6 0 z , 1s0180 and l80, in the dark, followed by changes with duration of intense u.v.-illumination (0 = Xa6; = X3,; 0 = X34). (b) (i) Linearization of photoassisted changes by plotting X3,/X3, against X;:; (ii) similar plot for preoxidised ZnO. intense U.V. light similar to that used by Courbon et aL8) illustrate the onset of a photoassisted 0.i.e.process leading initially to a decrease in mole fraction of 1802, (- AX36), and an equivalent increase in 1 6 0 2 , (+ AX32). Note also in fig. 1 (a) that, during the first hour of illumination, X34 remained unaffected. These observations agree well with what would be predicted for a photoassisted R;'-type process, since ( - AX3& would flow from eqn (1 a), whilst an equivalent (+ AX32) would flow from eqn (1 a') 160(s) + 160'80(g) + 180(s) + '602(g). (1 a')790 PHOTOINITIATED OXYGEN ISOTOPE EXCHANGE It is easily shown that, with the composition of the i.p.eq. mixture used in this work, these two processes initially cause equal but opposite effects upon X34, leading to the observed initial constancy of Xa4. Previous workers who studied a photoassisted 0.i.e.process on u.v.-illuminated TiO,, ZrO,, SnO, and ZnO, employed as a test of the &-character of their data8 the linearity of plots equivalent ot X34/X36 against X;$. Fig. l(b) illustrates that our data are well linearised by this plot, but the following paragraph casts further light on the significance of this test as a criterion for a pure Rf-type process. Two consequences flow from continuing operation of the Rf-type 0.i.e. illustrated in fig. 1 (a); (i) the rate of eqn ( 1 a) will fall below that of eqn ( 1 a’) because P36 drops below whereupon rate of consumption of 160180 by eqn ( 1 a’) exceeds that of its formation by eqn (1 a); and (ii) the gas phase experiences a tendency to be driven away from its initial position of equilibrium with respect to eqn (1 b).That equilibrium may be described by KeQ = (Pi4/P32P36) = 4, and the extent to which any departure therefrom is realised will depend upon the relative rate at which the R,* process tends to drive the system away from equilibration against the rate at which an R, process maintains it. Fortunately we know from comparison of the rate shown in fig. 1 (a) for the R,* process, with rates previously evaluated for R, on oxygen-deficient zinc oxides,18 that R,* is two orders of magnitude smaller than R,. Hence it may be concluded that R, should serve as a fast pre-equilibrium over the oxygen-deficient zinc oxide surfaces, so.that it should be possible at all times to calculate P34 from d(&,qP32P36). Data points obtained experimentally were found to agree well with predicted values obtained from the above calculation and account for the observed slight decrease in the mole fraction of l60l8O(g) [calculated values indicated by the dotted line in fig.l(a)]. Data for prereduced ZnO may thus be described more accurately as an R,* process with an accompanying fast R, preequilibrium. PLACE-EXCHANGE-TYPE O.I.E. AT PRESSURES -c lo-, Torr ON EXTENSIVELY PREOXIDISED METAL OXIDES IN THE DARK It was demonstrated in work recently published from these laboratories,18 not only that extensive preoxidation of zinc oxide surfaces effectively destroyed any R,-type activity in the absence of illumination, but also that the ease with which R,-type activity could be restored to such surfaces by thermal treatment in uacuo decreased as the severity of surface preoxidation increased.Experiments directed towards possible identification of 0.i.e. taking place without the intervention of an R,-type process were therefore made with zinc oxide surfaces extensively preoxidised through being held at 673 K under 100 Torr of 160, and cooled to 295 K in that gas. Contrary to expectations that only slow I?,*- or R,*-type 0.i.e. processes might be possible at such i.p.eq. O,/M O(n+6) interfaces in the absence of thermal activation, the data summarised in fig. 2(a i) show that for extensively preoxidised zinc oxide surfaces, readily detectable and relatively rapid changes in X32, X34 and X36 occurred when an aliquot of i.p.eq. (160, + 160180 + 180,) was admitted to such surfaces in the dark to produce a pressure of 8 x low3 Torr immediately after brief (6 min) evacuation to a residual pressure < Torr.The variations of X32, X34 and X36 in fig. 2(a i) should not contain contributions from R,-type o.i.e., in view of the fact that the gas phase was already isotopically pre-equilibrated. Furthermore, the oxidising pretreatment FIG. 2.- Vanations of mole fractions X32, X3, and X36 with time following introduction of i.p.eq. (160, + 160180 + Torr to extensively dehydroxylated surfaces of ZnO and TiO, at 300 K. (a) Changes over preoxidised ZnO: (i) in the dark; (ii) under intense u.v.-illumination. (b) Changes over prereduced ZnO: (i) in the dark; (ii) under intense u.v.-illumination. (c) Changes on preoxidised TiO, (anatase): (i) in the dark; (ii) under intense u.v.-illumination. at 5 xJ.CUNNINGHAM, E. L. GOOLD A N D J. L. G. FIERRO 79 1 0.11 I,,,,, 0 10 20 30 4 0 50 time/min 0.7 0.5 E .,-I + CrJ a & I 0.3 0.1 I - p x 3 6 0 5 10 15 time/min 0 10 20 " 0 20 40 60 time/min (cii) illumination -- 0.5 1 I I 0 10 20 0 20 40 60 time/min FIG. 2.-See facing page.792 PHOTOINITIATED OXYGEN ISOTOPE EXCHANGE was such as to destroy sites capable of R,-type activity or reaction with i.p.eq. 0,. Absence of R, activity was demonstrated in a separate experiment by introducing the i.n.eq. mixture of ( W , + 180,) over an ZnO,,, surface to a pressure of 8 x lo-, Torr in the dark and observing that no formation of.160180 occurred over a period of hours in the dark. Data plotted on the left of fig. 2(b i) demonstrate that the non-R,-type process which was responsible for the changes in fig.2(a i) did not occur in the dark upon introducing an identical aliquot of the i.p.eq. mixture to a prereduced ZnO,l-d, surface, thereby indicating that a ‘surface excess’ of oxygen was essential for this dark, 0.i.e. activity of the ZnO surface at 8 x Torr. However, fig. 2(c) shows that measurable effects were obtained with TiO,, provided that the anatase surface was preoxidised similarly to ZnO (6 h in 100 Torr lSO, at 673 K, cooled and evacuated to lo-, Torr for 6 min) and then exposed to an aliquot of i.p.eq. mixture which brought the system pressure to 8 x Torr in the dark. A criterion suggested by other workers as a means of testing for involvement of an R,-type process [cf. eqn (lc)] was that the ratio (Pi4/f?32f?36) should show a minimum.However, none of the data here obtained in conditions similar to fig. 2(a i) or (c i) yielded any indication of a minimum when plotted in this way and an R,-type process could therefore be discounted at these preoxidised interfaces in the dark. In the absence of any R, process, which could not occur on preoxidised surfaces, an &-type process could be discounted on the basis that X3, decreased (cf. table 1) and did so without any induction period (see below). A p.x. process was the remaining possibility and close inspection of fig. 2(a i) and (c i) showed that the decreases, (- AX38) and (-AX3,), in mole fraction of lSO, and lS0l8O in the gas phase together equalled (+ AX32), the increase of 1 8 0 , in the gas phase.Furthermore, the ratio ( - AX3,)/( - AX36) equalled two throughout these observations on 0.i.e. in the dark over MO,,,, under initial partial pressures, 4, x 2 x Torr. Operation of the p.x. processes described by eqn (2a) and (2b) would account fully for these observations. The lack of any such processes on prereduced zinc oxide in the absence of illumination [cf. fig. 2(b i)] may be understood on the basis that surface coverage by 1s02(chem) prior to admitting i.p.eq. 0, was very much less. This idea that pre-existing coverage by readily replaceable lS02 determined extent and rate of the p.x. process received further support from experiments which showed that, upon admitting successive aliquots of the i.p.eq. 0, over the same preoxidised metal oxide surface without reoxidation by lSO, between runs, rate of the p.x.process in the dark at 8 x Torr decreased progressively. An estimate of 0.7% monolayer coverage of the preoxidised ZnO surfaces by readily replaceable 1s02(chem) was arrived at on the basis of an approximately 50% reduction in rate of the p.x.-type process from the first to the second aliquot. Since BO,(chem) thereby estimated greatly resembles limiting surface coverages by 0; (calculated for depletive chemisorption by collective electron models19), the symbol O,(chem) in eqn (2a) and (2b) should be understood as including possible involvement of O;(chem). A corollary of these considerations, borne out by experiment, was that the p.x. process, here detected readily at pressures of ca. 8 x Torr, becomes difficult to detect at much larger total pressures of i.p.eq.0, (e.g. > 0.2 Torr), since the number of gas-phase 0, species greatly exceeds BO,(chem) in those conditions. P3, z 4 x and 4 6 % 2 x PHOTOASSISTED O.I.E. EFFECTS O N PREOXIDISED METAL OXIDE SURFACES The preceding sections described results in special conditions which, respectively, favoured detection of an R,* plus an R,-type process (on prereduced and intensely- illuminated surfaces, MO(,-,, under i.p.eq. 0, at ca. 0.1 Torr) or of a p.x. process (preoxidised surface in the dark under ca. 8 x Torr of i.p.eq. 0,). In consideringJ. CUNNINGHAM, E. L. GOOLD A N D J. L. G. FIERRO 793 what photoassisted processes may contribute to 0.i.e. upon illumination of i.p.eq. 0, over preoxidised surfaces, it became clear, therefore, that attention should be given not only to an R;'-type process but also to possibilities: (i) that illumination may enhance, through photosorption effects, the p.x.process noted in the dark; and (ii) that illumination may restore, through photoelectronic or surface photolysis effects,20-22 the Ro-type activity destroyed by prior surface preoxidation in the dark. Fig. 2 (a ii) demonstrates the enhanced rate of change of isotopic composition which occurred upon illumination of an i.p.eq. O,/ZnO(l+a, interface under the low pressures thought to favour detection of a p.x. process. The greatly enhanced initial rates of depletion of X32 and of X36 under intense illumination by the water-cooled 500 W lamp, allied to the enhanced initial rate of increase of X32, appeared consistent with enhancement through 0, photosorption events of a p.x.process, i.e. p.x.*, on ZnO surfaces preoxidised in lSO,. Note from fig. 2 (c) that comparison of rates of dark and photoassisted processes over i.p.eq. O,/TiO~(,+a, interfaces yielded no comparable evidence for a p.x.* process on TiO, at the low pressure employed (8 x Torr). At such pressures, other workers have that dissociative chemisorpion of oxygen predominates on pure TiO,. Dissociatively chemisorbed oxygen would favour R, or R, rather than p.x. processes. Changes brought about in the 0.i.e. activity of preoxidised ZnO surfaces by intense illumination under 0.1 Torr of i.n.eq. (lag, + 180,) are summarised in fig. 3. Plot (i) 0.5 0.25 0 (ii) I I I 30 60 90 time/min FIG.3.-Variations in isotopic composition upon intense illumination of i.n.eq. (lSO2 + lSO2) over an extensively preoxidised ZnO surface at 300 K : (0) Photoinitiated changes in X,, ; (0) photoassisted changes in atom fraction of leO. of fig. 3 illustrates the rapid changes in molecular composition of the gas phase observed upon commencing illumination of the i.n.eq. 02/ZnO(1+6) interface, which had been totally inactive for &-type 0.i.e. in the dark. Two photoassisted 0.i.e. effects can be distinguished in fig. 3 : (a) an initial fast process bringing the gas-phase mole fractions effectively into isotopic equilibration with X34 x 0.50 within ca. 15 min illumination and (b) a slower process causing to drop off from 0.50 at longer794 PHOTOINITIATED OXYGEN ISOTOPE EXCHANGE illumination times.The former process is almost certainly an R,*-type process originating through restoration of &-type activity to the ZnO(,+a, surface by illumination, witness the strong similarities of rate and kinetics to those observed upon introduction of i.n.eq. (1s02+1s02) to prereduced ZnO surfaces in the dark. The second longer-term photoassisted process could have contributions not only from continuation of R,* but also from one or more of the processes p.x.*, R,* and R,*, witness the fall-off in atom fraction of la0 in the gas phase illustrated by plot (ii) of fig. 3 and recall that an R, process would have left this unchanged. The latter point was confirmed experimentally as follows: introduction of an aliquot of the i.n.eq.(lSO2 + l80,) mixture to a prereduced ZnO surface; kinetic observations upon the rapid R, exchange to equilibrium w.r.t. eqn (1 b); conversiofi of the initially varying values of X32, X3a and X36 to atom fractions of l 6 0 or lSO; demonstration that these atom fractions remained invariant. It was also demonstrated that the ratio x36/x32 remained invariant throughout the R, exchange (cf. data on the topmost horizontal plot of fig. 4). Consequently, this ratio was seen to provide a method for representation of the data, which was sensitive to effects of R:, R: and p.x.* processes but which was not affected by any accompanying R,* process. It is used in the following section to differentiate between R,* and those other photoassisted processes. 1.0 0. a I_ x36 0.6 x 3 2 0 .4 0.2 0 O\ 0 10 time/min 20 FIG. 4.-Variations of the X3,/X3, ratio for 0.18 Torr of i.p.eq. (lS0, + lSO1sO+ 1802) over zinc oxide surfaces: (i) a, zero variation in ratio as observed over preoxidised or prereduced ZnO in the dark or under intense illumination by photons of II > 460 nm; (ii) 0, large variations over preoxidised ZnO upon intense illumination by photons of 1 > 300 nm; (iii) 0, moderate variations over prereduced ZnO upon intense illumination by photons of II > 300 nm.J. CUNNINGHAM, E. L. GOOLD AND J. L. G. FIERRO 795 DIFFERENCES I N SPECTRAL RESPONSE AND QUANTUM EFFICIENCY BETWEEN R,* AND OTHER PHOTOASSISTED O.I.E. PROCESSES The topmost horizontal plot of fig. 4 shows inter a h that no change in the x36/x36 ratio occurred when the i.n.eq.(160, + 180,)/Zn0,1+a interface was illuminated by photons of 13. 2 460 nm. This may be contrasted with the readily detectable changes in the X36/X32 ratio which are shown on (ii), the lowest plot of fig. 4, as resulting from illumination of the same interface by photons of 13. > 300 nm at comparable photon flux. This marked wavelength dependence pointed to the need for photons close to the band edge of ZnO (band gap = 3.2eV = 385 nm), and thence to the probable involvement of photogenerated electron-hole pairs, in the initiation of this 0.i.e. Approximate values for the apparent quantum efficiency of this 0.i.e. effect were obtained from the ratio of the experimentally observed initial rate of disappearance of 180, (or the rate of growth of l6O0,) to the rate of absorption of photons of 13.= 300-385 nm. A uranyl oxalate actinometer solution, which absorbed effectively at 300 nm but was largely transparent above 370 nm, was used to measure the total flux of such photons on the sample position from the water-cooled 500 W lamp as 6 x 1017 s-l. Apparent quantum efficiencies of 5 x for TiO,,+d, (rutile MR128) were then estimated on the basis that all such photons were absorbed (i.e. without any corrections for losses by reflection/scattering, since those seemed likely24 to be less than 50%) in the wavelength range 300-370 nm. Restoration of R,-type 0.i.e. activity to ZnO or TiO, surfaces, which had been rendered inactive through preoxidation of the surface by molecular O,, exhibited markedly different wavelength dependences and quantum efficiencies from those just described for other photoassisted 0.i.e. processes.Thus photons qf 13. 2 460 nm, which were shown to be ineffective in promoting those other processes [cf. fig. 4(i)], produced a rapid R,-type 0.i.e. in aliquots of i.n.eq. (1602+1602) brought into contact with preoxidised ZnO or TiO, surfaces which were inactive in the absence of illumination. The efficiency of a low flux of photons of 13. 2 460 nm is illustrated in fig. 5(a) for lithium-doped zinc oxide. Similar observations were made with pure and indium-doped zinc oxide, thereby suggesting that the process originated from zinc oxide rather than dopant or impurity. A second new feature associated with photoregeneration of R,-type activity in preoxidised ZnO or TiO, surfaces was the efficiency of low light intensities.As already rep~rted,~ room light sufficed for this purpose and this striking result is illustrated for zinc oxide in fig. 5 (b) and for TiO, in fig. 5 (c). Attempts to establish apparent quantum efficiencies of this R,*-type process were therefore possible at photon fluxes three to four orders of magnitude lower than used above for the other photoassisted processes. The photon flux transmitted from room light onto the sample position via various gelatine filters was evaluated using potassium ferrioxalate actinometer solutions, which absorbed strongly at A < 510 nm. An apparent quantum efficiency of 5.5 was calculated for ZnO for the R,* process on the basis of a measured number of 5 x lo1' molecules of 1s0180 produced in 3 min illumination during which the integrated flux of photons of 13.= 300-510 nm was 9 x 10l6. This method of calculation merely places a lower bound to the apparent quantum efficiency since any corrections for scattering or reflection by ZnO, which would be very significant above 400 nm, must push 4 to higher values. The validity of values larger than unity for the photoassisted R,* process can be demonstrated yet more conclusively for the data in fig. 5 (a) for lithium-doped ZnO exposed to photons of 13. > 460 nm. The potassium ferrioxalate actinometer established the integrated flux of such photons as 10l6 in 3 min, during which time 3 x 1017 molecules of lsOl8O were produced. The lower bounds of 6 for ZnO and 30 for ZnO,,,, and 2 x796 PHOTOINITIATED OXYGEN ISOTOPE EXCHANGE 0 0.5 0.4 0.3 x34 0.2 0.1 5 10 time/min 15 1 0 7 0 (ii) ( i ) I , start of illurn.0 10 20 30 40 50 timelrnin FIG. 5.-Photoinitiated restoration of &-type activity illustrated by comparing plots of mole fraction of 1s0180 against time for illuminated and non-illuminated systems. (a) For 0.07 Torr of i.n.eq. (l8O, + l80,) contacted at 300 K with a preoxidised surface of lithium doped zinc oxide: (i) in the dark, (ii) under illumination by photons of I 2 460 nm at a flux of 5 x 1013 s-l. (b) For 0.12 Tom of i.n.eq. (160,+1802) over preoxidised zinc oxide: (i) in the dark, (ii) upon exposure to room light. (c) For an initially prereduced surface of TiO, (l), which was then progressively inactivated by exposure to aliquots of (160,+180,) in the dark (2, 3, 4), and finally reactivated by exposure to room light prior to the 5th aliquot (5).J.CUNNINGHAM, E. L. GOOLD AND J. L. G. FIERRO 797 0.3 x34 0.1 0 4 time/min FIG. 5(c). 8 for Li-ZnO, thereby placed on the apparent overall quantum efficiency of the R,* process could imply a photoinitiated chain reaction in the i.n.eq. (1602 + 180,)/ZnO~,+d, system. A parallel possibility was that each photoactivated surface location was capable of correspondingly high turnover achieved in the A,-type process [represented by the forward direction of eqn (1 b)]. Evidence in support of the latter possibility came first, from the fact that several aliquots of (lSO2 + 1802) could be exchanged on a photoreactivated surface after illumination was removed and secondly from the long times for which activity restored by illumination persisted after illumination ceased.Data for zinc oxides in fig. 6(a) demonstrate partial persistence of such photogenerated R,-type activity for times up to 15 h after illumination. These observations would be understandable in terms of photogenerated surface locations on zinc oxide, each of which remained capable of initiating several &-type events long after illumination. The efficiency of very low light intensities and of photons outside the band edge of ZnO argued against a need for photolysis of the ZnO surface in the photogeneration of these active sites. Rather they suggested that photoregenerated R, activity resulted from redistribution of electrons onto surface locations on ZnO surfaces which could trap and hold separated charge carriers for long times after ill~mination.~~ It had been noted in a previous publication that although illumination also led to a slight ‘memory-effect’ of TiO, surfaces for R, exchange, this decayed exponentially with time so that its persistence was very much shorter than on the zinc oxides; this is illustrated in fig.6(b) showing half-time for decay of photoactivation as ca. 5 min on TiO,.PHOTOINITIATED OXYGEN ISOTOPE EXCHANGE 1 I I 5 10 15 delay time/h I I I I 0 20 40 60 delay time/min FIG. 6.-Data comparing ‘memory’ effects of previously illuminated zinc oxide and TiO, surfaces and illustrating the extent to which photoregenerated activity for &-type exchange persisted whenever introduction of an aliquot of i.n.eq.(l60, + l8O0,) was delayed for the indicated times. Initial rates of &-type 0.i.e. (arbitrary units) against indicated post-illumination decay times for: (a) Li-ZnO, following illumination by Iz =- 500 nm from room light, 0 ; zinc oxide, following illumination by I > 300 nm from room light, 0. (b) TiO, after exposure to I > 300 nm from room light. DISCUSSION Particular interest attaches to the foregoing demonstration of quantum efficiencies greatly exceeding unity for the R,* process on zinc oxide surfaces under room light illumination, since they indicate the achievement, albeit in a system going downhill rather than uphill in free energy, of two of the objectives underlying world-wide attempts to use semiconductor surfaces for solar energy conversion, viz: (i) efficacy of light in the visible part of the spectrum and (ii) a genuine ‘multiplier effect’ whereby light modifies electron-hole distribution at the semiconductor surface in such a way and for a sufficiently long time that each absorbed photon ultimately becomes responsible for activation and conversion of many reactant molecules to products.Some consideration of the origins of this multiplier effect (i.e. turnover 9 1 achieved for each photoactivated site) in the i.n.eq. (1s02+1802)/ZnOT,+d, system may be of value in providing pointers to possible realization of similar multiplier effects in other systems. Evidence and a model have been presented elsewhere18 by two of the present authors relating &-type activity on ZnO surfaces in the absence of illumination to the availability of composite e- (0-Zncus) surface locations distinguished by : (a) the localization thereon of an electron which otherwise would be delocalised into the conduction band or valence band, i.e.each such composite site provides a surface-state energy level lying in the band gap, and (b) high coordinative unsaturation of the zinc part in the site. Activation of 1 8 0 2 (or l8O2) molecules visiting such sites from the gas phase involves: (a) partial donation of the electron into n* antibonding orbitals of the oxygen and (b) further weakening of the l80- l 8 0 bond through bonding interactions between the -Zncus part and one l80- . Depending on the strength ofJ. CUNNINGHAM, E. L. GOOLD AND J.L. G. FIERRO 799 this latter interaction, the ls0-l8O bond may be broken (presumably on the locations featuring zinc with the highest degree of coordinative unsaturation and/or at high temperatures) or merely weakened to varying degrees (likely to predominate at low temperatures and on sites of lower degrees of coordinative unsaturation). In either case the original 180, molecule is activated by the e-(0-Zn,,,) location to yield thereby a primary surface intermediate, which in turn reacts with 1602 (or 180,) species to cause the first of a series of 0.i.e. events which propogate at that location. A possible chain-reaction-type 0.i.e. process based on an initial dissociative interaction has been described e1~ewhere.l~ Another based on weakening rather than breakage of the bond may be depicted as follows and would equally well account for the reversible second-order kinetics already reported.It may be appreciated from this scheme that once an electron on the e-(0-Zn,,,,) location, and for as long as the electron has become localised remains so localised, conditions exist for turnover of several molecules of (1602+180,) to lSO1*O on the same site. In our view, room light suffices to promote electrons onto the (0-Zn) locations, although we do not yet know whether such electrons originate from the valence band (which would seem favoured by the high quantum efficiency) or from extrinsic filled surface states in the band gap. An aspect of the results which is of general interest in relation to oxygen liability and reactivity at illuminated metal oxide surfaces is the wide range of photoassisted 0.i.e.processes at O,/ZnO* and O,/TiO,* interfaces. Thus the results served to identify conditions favouring the predominance of R,* or R;" processes under illumination and of a p.x. process at the dark preoxidised interfaces. In view of considerable evidence from e.s.r. and conductivity measurements for retention of l60; radical ions on TiO, and ZnO surfaces preoxidised in our conditions and then evacuated at room temperature,17 it is reasonable to conclude that chemisorbed l60, was one species playing an important role in the 'dark' p.x. process, which exhibited the stoichiometry of eqn (2a) and (2b), both for ZnO(,+& and for TiO(,+@. Likely rate-limiting factors of a dark p.x.process involving 0; were: (i) the low concentration of holes (needed to neutralise l60; in the first step) in the n-type semiconductors and (ii) a slow rate of desorption of l60, (possibly needed to vacate an active site onto which lSO, or 160180 could adsorb). One way in which the incidence of intense u.v.-illumination altered conditions at the interface was in producing a large enhance- ment of minority carriers (holes). Consequently, it was tempting to assign the large increase in rate of 0.i.e. depicted in fig. 2(a) for the preoxidised ZnO surface to an enhancement of the dark p.x. process due to the greater availability of holes for neutralization of l60;. However, the failure, illustrated in fig. 2(c), of illumination to enhance 0.i.e. on a TiO, surface exposed to the same low pressure provided one argument against too facile an acceptance of minority carrier concentration for800 PHOTOINITIATED OXYGEN ISOTOPE EXCHANGE neutralization of l60; as the only rate-limiting factor.Another and more basic argument is that the alterations brought about by intense u.v.-illumination incident onto the O,/ZnO(,,g and O,/TiO,,+, interfaces can be much more extensive than merely enhancing lsO;(chem) + h+ + ls02(ads) -+ lSO,(g). One additional possibility which emerges clearly from recent e.s.r. studies is that other molecular oxygen anions, such as Oy, O;, O:-, may be present at significant concentration on the interfaces in some conditions.2s, 27 Another and closely related possibility is that molecular oxygen anion species pre-existing on the dark interfaces may readily be photodissociated by illumination to yield inter alia the monatomic anion radical 0-, e.g.Lunsford has pointed to the probability of the process 0; + hv + 0 + 0- for 0, on MgO surface^.^ It follows from these considerations that monatomic rather than molecular oxygen anion species may play the dominant role at intensely illuminated O,/MO interfaces. Supporting evidence for this viewpoint comes from recent studies by Pichat et al. on photoconductance, o*, at O,/ZnO and O,/TiO, interfaces and its dependence upon oxygen pressures.28 For ZnO surfaces those workers found (6 log a*/S log POL) NN -4 for a wide range of 0, pressures, including those employed in collecting the data summarised in fig. 2(a). Pichat et al.interpreted this inverse half-order dependence on oxygen pressure as indicating that 0- species controlled the photoelectronic equilibrium at u.v.-illuminated O,/ZnO interfaces.28 Application of this idea to our data for illuminated ZnO in fig. 2(a) would suggest that 0.i.e. should be controlled by 0- species and by their involvement in R,*- or Rs-type processes, [as per eqn (1 a), (1 b) or (1 a’)], rather than by 0, or other molecular oxygen anions and their involvement in p.x.*. Plot (ii) of fig. l(b) provides some support for this idea, since it demonstrates good linearity of data for illumination-induced 0.i.e. over a preoxidised ZnO surface when plotted in the manner recommended for testing agreement with an &-type 0.i.e. process. The situation in respect of u.v.-illuminated O,/TiO, interfaces appears not yet to be resolved to a degree comparable with that just described for ZnO.Thus, Pichat et al. deduced from their photoconductance studies that (6 log o*/S log Pop) equalled - 2 at low pressures but came to equal - 1 at higher pressures. Nevertheless, fig. 2( c) illustrates that no photoassisted 0.i.e. process resulted upon illumination of an i.p.eq. O,/TiO, interface under the low pressures thought to favour 0- species. This difference was no doubt contributed to by different degrees of surface dehydroxylation of the TiO, samples used by Pichat et al. for photoconductance measurements (outgassed 15 h at 473 K) and by us for 0.i.e. studies (outgassed 15 h at 673 K), witness our ability to observe photoassisted 0.i.e.at low pressures when our TiO, surfaces were outgassed at only 383 K. Kinetic data for this latter process with i.p.eq. 0, on hydroxylated TiO, surfaces at low pressures did not fit the R,-type analysis, which was shown in fig. 1 (b) to be successful for dehydroxylated ZnO surfaces. Failure of photoassisted 0.i.e. over TiO, surfaces to fit a more simple kinetic behaviour exhibited by ZnO surfaces had also been noted previously19 in cases where a photoassisted R; process predominated, i.e. when i.n.eq. + 180,) was present over ZnO or TiO, surfaces under room light illumination. Possible origins of the more complex kinetic behaviour on TiO, surfaces, and in particular their relation to the concentration and mobility of surface hydroxyls and oxygen photoadsorption are under continuing investigation.Valuable technical assistance from J. Caffrey in actinometry and other aspects of this study is acknowledged. We also thank the Department of Education of the Irish Government for a maintenance grant (to E.L.G.) and Instituto de Catalysis y Petroleoquimica C.S.I.C., Madrid for sabbatical leave (to J. L. G. F.).J. CUNNINGHAM, E. L. GOOLD AND J. L. G. FIERRO 80 1 M. Formenti, H. Courbon, F. Juillet, A. Lissetchenko, J. R. Martin, P. Meriaudeau and S. J. Teichner, J. Vac. Sci. Technol., 1972, 9, 947. R. I. Bickley and F. S. Stone, J. Catal., 1973, 31, 398. M. Iwamoto and J. H. Lunsford, J . Phys. Chem., 1980,84, 3079. A. Fakishima and K. Honda, Nature (London), 1972,238, 37; Bull. Chem. SOC. Jpn, 1971,44, 1148. M. S. Wrighton, Chem. Eng. News, 1979, 57, 30. F. T. Wayner and G. A. Somorjai, J. Am. Chem. SOC., 1980, 102, 5494. J. C. Hemminger, R. Carr and G. A. Somorjai, Chem. Phys. Left., 1978, 57, 100. * H. Courbon, M. Formenti and P. Pichat, J. Phys. Chem., 1977, 81, 550. J. Cunningham, E. L. Goold and E. M. Leahy, J. Chem. SOC., Faraday Trans. 1, 1979,75, 305. lo E. R. S. Winter, J. Chem. SOC. A , 1968, 2889; 1954, 1522. l1 G. K. Boreskov, Adv. Catal., 1964, 15, 285; Discuss. Faraday SOC., 1966, 41, 263. l2 J. Novakova, Catal. Rev., 1970, 4, 77. l3 Y. Ben Taarit, C. Maccache, M. Che and A. J. Tench, Chem. Phys. Lett., 1974, 24, 41. l4 M. Che, B. N. Shelimov, J. F. J. Kibblewhite and A. J. Tench, Chem. Phys. Lett., 1974, 28, 387. l5 V. S. Zakharenko, A. E. Cherkashin, N. P. Kleier and S. V. Korcheev, Kinet. Katal., 1975, 16, 182. l6 (a) J. Cunningham and N. Samman, Dynamic Mass Spectrometry, ed. D. Price and J. F. J. Todd (Heyden, London, 1975 and 1978), vol. 4, chap. 17; (b) J. Cunningham, B. Doyle and D. J. Morrissey, Dynamic Mass Spectrometry, ed. D. Price and J. F. J. Todd (Heyden, London, 1975 and 1978), vol. 5, chap. 18. J. Cunningham and E. L. Goold, J. Chem. SOC., Faraday Trans. 1, 1981, 77, 837. J. Cunningham, J. J. Kelly and A. L. Penny, J. Phys. Chem., 1970, 74, 1992. and E. Thull, Faraday Discuss. Chem. SOC., 1974, 58, 176. l7 R. D. Iyenyar and M. Codell, Adv. Colloid Interface Sci., 1972, 3, 365. 2o (a) F. Steinbach and R. Harborth, Faraday Discuss. Chem. SOC., 1974, 58, 143; (b) W. Hirschwald 21 E. Arijs, F. Cardon and W. M. Van der Vorst, 2. Phys. Chem., 1975, 94, 255. 22 J. Cunningham, B. Doyle, D. J. Morrissey and N. Samman, Proc. 6th Int. Cong. Catalysis, London, 1976, ed. G. G. Bond, P. B. Wells and F. C. Tompkins (The Chemical Society, London, 1977), p. 1093. 23 J. M. Hermann, J. Disdier and P. Pichat, Proc. 7th Int. Vac. Cong. and 3rd Int. Conf. Solid Surfaces, Vienna, 1977, p. 951. 24 J. Cunningham and H. Zainal, J. Phys. Chem., 1972, 76, 2362. 25 W. Maenhaut-Van der Vorst and F. Van Craynest, Phys. Stat. Solidi, 1964, 5, 357. 26 A. Bielanski and J. Haber, Catal. Rev., 1979, 19, 1. 27 P. Meriaudeau and J. C. Vedrine, J , Chem. SOC., Faraday Trans. 1, 1976, 72, 472. 28 J. Herrmann, J. Disdier and P. Pichat, J. Phys. Chem., in press. 2* A. H. Boonstra and C. A. H. A. Mutsaers, J. Phys. Chem., 1975, 79, 1694. (PAPER 1/514)
ISSN:0300-9599
DOI:10.1039/F19827800785
出版商:RSC
年代:1982
数据来源: RSC
|
14. |
Analysis of the factors affecting selectivity in the partial oxidation of benzene to maleic anhydride. Part 3.—Mechanism of benzene surface oxidation on a vanadium pentoxide–molybdenum trioxide catalyst |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 803-815
Raymond W. Petts,
Preview
|
PDF (909KB)
|
|
摘要:
J . Chem. SOC., Faraday Trans. 1, 1982, 78, 803-815 Analysis of the Factors Affecting Selectivity in the Partial Oxidation of Benzene to Maleic Anhydride Part 3.-Mechanism of Benzene Surface Oxidation on a Vanadium Pentoxide-Molybdenum Trioxide Catalyst BY RAYMOND W. PETTS AND KENNETH C . WAUCH* The Corporate Laboratory, I.C.I., P.O. Box 11, The Heath, Runcorn, Cheshire WA7 4QE Received 6th April, 1981 Rate measurements are described in which benzene, hydroquinone and p-benzoquinone were oxidised over a vanadium pentoxide-molybdenum trioxide catalyst. These experiments showed : (i) that the catalytic oxidation of benzene did not involve any homogeneous component as has been previously reported and (ii) that identical selectivities (ca. 60%) to maleic anhydride could be obtained from benzene or hydroquinone, p-benzoquinone being oxidised mainly to carbon oxides with negligibly small selectivities (< 6%) to maleic anhydride. The selectbe reaction pathway from benzene to maleic anhydride is therefore taken to involve hydroquinone, the oxidation of which can result in maleic anhydride or p-benzoquinone, the latter being the main intermediate in the non-selective pathway.An explanation of these results is found in orbital symmetry conservation arguments which show that the concerted addition of a chemisorbed oxygen molecule (i.e. having an extra electron in an antibonding n orbital) para, across the ring of a chemisorbed benzene molecule (i.e. a benzene molecule deficient of electrons in the highest occupied orbital), is allowed.This adsorbed adduct is presumed to rearrange to form hydroquinone which, by another identical molecular addition of oxygen across the ring, followed by the elimination of a C,H, fragment and of water, forms maleic anhydride. The low surface oxidation activation energy, the necessity for the involvement of molecular chemisorbed oxygen and the identity of the selectivities of benzene and hydroquinone are accounted for in this mechanism. The near inability of p-benzoquinone to form maleic anhydride is also accounted for, since its structure and probable mode of chemisorption (donation of electrons to the catalyst from the carbon+arbon double bond) are unlikely to produce the 1 $-oxygen adduct necessary for maleic anhydride formation. Previous paperslv have examined the detailed kinetics of the vanadium pentoxide- molybdenum trioxide (3 : 1) catalysed oxidation of benzene to maleic anhydride.The combination of the transient techniques of temperature programmed desorption (t.p.d.), temperature programmed reaction spectroscopy (t.p.r.s.) and gas adsorption chromatography have shown the reaction to be rate limited in the desorption of the product, the surface oxidation of an adsorbed benzene molecule to an adsorbed maleic anhydride molecule being particularly facile, having an activation energy of only 36 kJ mol-l. Furthermore no evidence was obtained for the existence of the benzene- oxygen adduct which is proposed by Trimm and coworkers3 to be formed on the surface, to desorb and to oxidise further in the gas phase to maleic anhydride.The experiments described in this paper were undertaken with a two-fold purpose. First, they would provide a rigorous test of the possibility of there being a homogeneous component in the reaction and, secondly, by oxidising two of the more likely intermediates (hydroquinone and p-benzoquinone) in the oxidation of benzene to maleic anhydride they would provide mechanistic information on the selective reaction pathway. 803804 SELECTIVITY I N OXIDATION OF BENZENE EXPERIMENTAL THE CATALYST The method of preparation of the catalyst, a supported 3: 1 molar V,O,: MOO, catalyst, has been described previously.' It had a particle size of 0.5 cm diameter and a surface area of 2.2 m2 g-l. MATERIALS The benzene was AnalaR grade supplied by Hopkin and Williams (Essex, England).Its specification was that not less than 95% of it boiled in the range 352.5-353.5 K and its im- purities were: sulphur-containingcompounds 3 x lo-, %, thiophene 2 x %, water 5 x lo-, %. The hydroquinone and the p-benzoquinone were both supplied by B.D.H. (Poole, England); the former was AnalaR grade having a melting point in the region of 444-448 K, the major impurity being catechol(O.O2%). The latter was technical grade having a melting point in the region of 386-388 K. Its impurities (ca. 1 %) are hydrobenzoquinone and benzoquinol. Both were used without further purification. The gases, oxygen and helium, were supplied by British Oxygen Company. Their specified purities were oxygen: oxygen 99.7%, carbon dioxide < 2 x lo-,%, carbon monoxide < 1 x hydrocarbons as CH, < 2 x lo-,%; helium: helium 99.5%, oxygen (1-2) x lo-,% [the oxygen content was measured at 1.5 x lo-,% on a Hersch oxygen meter (Englehard Industries Limited)], carbon monoxide nil, carbon dioxide nil, nitrogen < %.BENZENE OXIDATION: TUBULAR REACTOR RATE MEASUREMENTS The temperature dependence of benzene conversions and selectivities to maleic anhydride listed in table 1 were obtained in a glass tubular reactor shown in diagram in fig. 1. The reactor, which was mounted vertically, was 91 cm long, 1.9 cm internal diameter and was heated over N FIG. I.-Diagram of the apparatus used for the air oxidation of benzene over the vanadium pentoxide- molybdenum trioxide catalyst. its entire length by a fluidised bed sand bath. To test the possibility of there being a homogeneous-heterogeneous interaction in the benzene oxidation,, i.e.that a component of the overall reaction could include the gas-phase oxidation of a desorbed intermediate, provision was made for the sampling of products (a) immediately after the catalyst bed, (b) after 210 cm3 free space and (c) after 356 x 0.635 cm diameter glass balls packed on top of the catalyst. The benzene + air mixtures were prepared by injecting a continuous stream of benzene from a burette under pressure into an air flow. Minor fluctuations in the reactant stream wereR. W. PETTS AND K. C. WAUGH 805 removed by passing through a large mixing vessel (250 cm3, see fig. 1). The benzene + air mixture was pre-heated by passage down the length of the sand bath and then over a s h r t column (5 cm long) of glass beads.Calibration was made by titrating the amount of benzene injected during the course of the experiment, knowing the total gas flow. The system gave a very stable gas mixture. The reactant and product streams were analysed by gas chromatography using two columns: a poly-2,2-dimethyl propane- 1,3-succinate column [ 10% on Embacel (60-80 mesh)] separated benzene, maleic anhydride, acrylic acid and acetic acid, while a Porapak Q column separated CO and CO,. OX1 D AT10 N OF p-BEN ZOQ U I NON E The line diagram for the apparatus used in the oxidation of p-benzoquinone is shown in fig. 1. However, the apparatus itself differed from that used to oxidise benzene in two major respects: (i) the p-benzoquinone+air feed was achieved simply by passing air over a column ofp-benzoquinone held at 333 K and (ii) the tubular reactor was replaced by a stirred gas solid rea~tor.~ OXIDATION OF HYDROQUINONE Because of the highly reactive nature of hydroquinone neither the tubular reactor described previously nor the stirred gas solid reactor could be used for its oxidation; considerable blank conversions of the hydroquinone to p-benzoquinone were observed in the lengths of heated metal tubing leading to the tubular reactor and also on the walls of the stirred gas solid reactor.A line diagram of the apparatus used in hydroquinone oxidation is shown in fig. 2. M S PR I O2 N2 1 l25OC L - - - J r - - -6cTi , SAV3 , FIG. 2.-Diagram of the apparatus used for the oxidation of hydroquinone over the vanadium pentoxide- molybdenum trioxide catalyst. C1 = Apiezon column; C2 = poly-2,2-dimethyl propane- 1,3-succinate column; C3 = Porapak columns; C4 = CO to CO, converter; CT = cold trap; FID = flame ionization detector; HC = heating coils; HT = heating tape; HS = hydroquinone saturator; MS = molecular sieve; OM = oxygen meter; PR = pressure regulator; R = rotameter; SAV = sample valve; SW = switching valve; TR = tubular reactor; WGM = wet gas meter.The hydroquinone + air feed was produced by bubbling nitrogen through a saturator containing liquid hydroquinone at 453 K; the saturator was housed in a Pye series 104 gas chromatography oven. Oxygen and nitrogen to make a final inlet composition of 20% oxygen were introduced to the nitrogen + hydroquinone mixture and this feed was diverted by switching valve 1 either to the glass tubular reactor (8 cm long, 1 cm internal diameter) or to the feed analytical column [Apiezon L (2 ft, 10% on Embacel 80-100 mesh)] all of which were housed in the Pye oven.The reactor itself was maintained at the temperature of interest (580-650 K) by heating tape, the majority of the piping being glass. In this way blank conversion of the hydroquinone was minimised to a negligible value. The products were analysed on two columns identical to those used in benzene oxidation.806 SELECTIVITY IN OXIDATION OF BENZENE CALIBRATION Calibration of the gas chromatographic analysis system for the composition of hydroquinone feeds was made by correlating peak areas with weighed amounts of hydroquinone collected in the cold trap after a given length of time at a measured flow rate.Calibration of the flame ionization detector response to p-benzoquinone was made using standard solutions in p-xylene. RESULTS AND DISCUSSION RATE MEASUREMENTS BENZENE OXIDATION It has been reported3 that the oxidation of benzene to maleic anhydride over a vanadium-molybdenum catalyst proceeds mainly by the gas-phase oxidation of a desorbed, partially oxidised, benzene adduct. The tubular reactor used in this work offers a more rigorous test of any homogeneous involvement than was available previ~usly.~ The free space provided after the reactor (2 10 cm3) for the homogeneous reaction to proceed was much greater than before3 where 16.32 cm3 was the maximum volume available. Furthermore it was possible tb sample the gas at the end of the catalyst bed and after the 210 cm3 free space for the same catalyst on the same experiment (in previous experiments it was necessary to change reactors to do this). Inspection of table 1 shows that at 643 K benzene conversions and selectivities to maleic anhydride are identical at the end of catalyst bed and after 210 cm3 free space.As a check that the homogeneous reaction was not being propagated in the 8.8 cm3 volume of the sample tube to the end of the catalyst bed (70 cm long, 0.4 cm internal diameter) the free volume in this tube was reduced by packing it with glass wool (it had also been claimed3 that any inert substance would quench the homogeneous reaction). Results with the packed sample tube were identical to those after 210 cm3 free space.At 618 K the conversions and selectivities show a time dependence (due possibly to some phase change in the catalyst at the outset or to the deposit of a small amount of carbon) but after 180 min the results obtained were again identical at the end of the catalyst bed and after 210 cm3 free space. Equally, in contrast to T ~ i m m , ~ packing the free space above the catalyst with glass balls (356 x 0.635 cm diameter, i.e. 451 cm2 surface area) produced no effect, the results at the end of the catalyst bed and after the glass balls again being identical. We conclude therefore that oxidation of benzene to maleic anhydride occurs totally on the surface of the catalyst and has no homogeneous, i.e. purely gas phase, component. An explanation of the previous author^'^ observations of the effects of free space after the packed bed might be found in the suggestion that their reactor design would produce eddies in the free space after the catalyst and that these would cause a re-mixing of the reactants and catalyst, leading to further reaction.The longer reactor and the packed length of inerts after the catalyst required for an analysis of the effect of inert packing would reduce this re-mixing effect considerably and so would apparently reverse the reaction. Indeed, the suggestion that a catalyst by the reaction of two adsorbed species could produce a species which could desorb and be decomposed by any inert substance is a definition of a system for the creation of energy.The adsorbed species would contain the heats of adsorption of benzene, oxygen and of the adduct itself, plus its heat of formation. The gaseous benzene and oxygen molecules produced by its desorption and collisional decomposition with an inert substance would contain,TABLE TUBULAR REACTOR RATE MEASUREMENTS FOR THE AIR OXIDATION OF BENZENE OVER THE VANADIUM MOLYBDENUM CATALYST gas-phase composition/mole fraction x 1 O2 ~~ ~ exit (a) exit (b) exit (a) exit (b) time to temp flow rate sample inlet maleic maleic conversion selectivity conversion selectivity /K /cm3 s-' /min benzene benzene anhydride benzene anhydride (%) (%I (%I (%I 618 12.7 12.7 12.7 12.7 12.7 12.7 12.7 12.7 643 643 6.6 6.6 6.6 6.6 30 60 90 120 150 180 210 240 30 60 90 120 1.079 1.079 1.079 1.079 1.079 1.079 1.079 1.079 0.763 0.763 0.763 0.763 0.986 0.0455 - - - - 1.022 0.034 1.025 0.033 - - 1.029 0.033 0.152 0.328 0.155 0.328 - - exit (a) 6.6 6.6 6.6 6.6 6.6 6.6 30 60 90 120 150 180 0.816 0.816 0.816 0.8 16 0.816 0.816 0.149 0.324 - - - - 0.133 0.355 0.126 0.344 - - 0.994 0.048 1.011 0.037 - - 1.029 0.035 1.028 0.035 0.158 0.323 - - 0.154 0.328 exit (c) - - 0.123 0.340 0.127 0.353 0.129 0.340 - - 8.6 48.9 - - - - 5.3 59.7 5.0 61.1 - - - - 4.6 66.0 - - 80.1 53.7 79.7 54.0 exit (a) - - 7.9 48.0 6.3 54.4 - - 4.6 69.6 4.7 68.6 79.3 53.4 - - 79.8 53.9 exit (c) 81.7 48.6 - - - - 83.7 52.0 84.6 49.9 - - - - 84.9 49.1 84.4 51.2 84.2 49.5 - - - - Exits: (a) gas stream sampled at end of catalyst bed, (b) gas stream sampled after 210 cm3 free space, (c) gas stream sampled after 356 x 0.635 cm Weight of catalyst, 15 g ; surface area of catalyst, 2.2 m2 g-l.benzene (inlet) -benzene (exit). Conversion = , selectivity = benzene (inlet) diameter glass balls (ie. 451 em2 glass surface) packed on top of the catalyst bed. maleic anhydride (exit) benzene (inlet) - benzene (exit) *808 SELECTIVITY I N OXIDATION OF BENZENE distributed between them in vibrational or translational degrees of freedom, the heats of adsorption of benzene and oxygen on the catalyst, i.e. these identical reactants would somehow have gained energy. Additionally, were the collisional decomposition energies for the adduct quoted by Brown and Trimm3 correct (a value of 152 kJ mol-l was estimated for the decompo- sition of the adduct to oxygen and benzene by collision with an inert substance), at the temperature studied (ca.670 K) the mean translational energy is only ca. 5 kJ mol-l and so virtually none of the molecules colliding with an inert substance would be expected to decompose. In total then the arguments presented previously3 about the involvement of a homogeneous component in the oxidation of benzene to maleic anhydride over a vanadium pentoxide-molybdenum trioxide catalyst are both infeasible and logically inconsistent. p-BEN Z OQU I N ONE 0 X ID A TI 0 N The results of the rate measurements of the oxidation of p-benzoquinone over the vanadium pentoxide-molybdenum trioxide catalyst in a stirred gas solid reactor4 in the temperature range 543-563 K are listed in table 2. (Blank experiments, i.e.catalyst removed, showed negligible oxidation of the p-benzoquinone and of the maleic anhydride product at these temperatures.) TABLE 2.-sTIRRED GAS SOLID REACTOR RATE MEASUREMENTS FOR THE AIR OXIDATION OF P-BENZOQUINONE OVER THE VANADIUM PENTOXIDE-MOLYBDENUM TRIOXIDE CATALYST gas-phase composition/mole fraction x 1 O2 exit time to inlet maleic carbon con- selec- temp flow rate sample p-benzo-p-benzo- an- carbon mon- version tivity /K /cm3 s-l /min quinone quinone hydride dioxide oxide (%) (%) 563 9.5 563 9.5 563 9.5 563 9.5 543 9.83 543 9.83 543 9.83 553 9.34 17 42 72 90 30 51 80 93 0.60 0.28 0.60 0.35 0.60 0.37 0.60 0.48 0.62 0.33 0.62 0.42 0.62 0.49 0.60 0.40 0.015 0.013 0.010 0.008 0.0096 0.0058 0.0034 0.062 0.79 0.70 0.51 0.49 0.54 0.37 0.32 0.566 0.25 53.3 4.7 0.24 41.7 5.2 0.14 38.3 4.3 0.12 20.0 6.7 0.10 46.8 3.3 0.039 32.3 2.9 0.026 21.7 2.6 0.102 31.0 3.1 p-benzoquinone (inlet) -p-benzoquinone (exit).p- benzoquinone (inlet) Weight of catalyst, 13.27 g; conversion = 9 maleic anhydride (exit) p-benzoquinone (inlet) -p-benzoquinone (exit) * selectivity = Two points are immediately apparent : (i) significant conversions ofp-benzoquinone are observed at similar contact times (l/flow rate x weight of catalyst) but at lower reaction temperatures than required for comparable benzene conversions (at only 50 % greater contact time the amount of p-benzoquinone converted is 3 times that of benzene, the latter however having been oxidised at a 50 K higher temperature) and (ii) the selectivities to maleic anhydride from p-benzoquinone are much lower than observed in benzene oxidation. Taken in combination these results indicate that theR.W. PETTS AND K. C. WAUGH 809 mechanism and kinetics of p-benzoquinone oxidation over the vanadium pentoxide- molybdenum trioxide catalyst are totally different from those of benzene, the former having an overall activation energy to its main product, carbon dioxide, lower than that of benzene to its main product, maleic anhydride. Germain's mechanism5 which depictsp-benzoquinone as the main intermediate in the oxidation of benzene to maleic anhydride can therefore no longer be taken as valid. [The time dependence of the p-benzoquinone oxidation (table 2) is probably due to carbon deposit.] HYDROQUINONE OXIDATION The results of the rate measurements of the oxidation of hydroquinone in the temperature range 580-660 K are listed in table 3.The difficulty of maintaining a constant hydroquinone feed was overcome by saturating nitrogen with hydroquinone using a low flow rate of the carrier gas (0.83 cm3 s-l). An oxygen+nitrogen mixture was then added to this saturated nitrogen stream to bring the oxygen content to 20% and the total flow rate to 2.5 cm3 s-l. The resulting long contact times (ca. 10 times longer than for benzene) produced complete conversion of the hydroquinone. TABLE 3.-TUBULAR REACTOR RATE MEASUREMENTS FOR THE AIR OXIDATION OF HYDROQUINONE OVER THE VANADIUM PENTOXIDE-MOLYBDENUM TRIOXIDE CATALYST gas-phase composition/mole fraction x 1 O2 exit inlet temp flow rate hydro- hydro- p-benzo- maleic carbon carbon selectivity /K /cm3 s-' quinone quinone quinone anhydride dioxide monoxide (%) 588 2.5 1.45 0 0.41 0.49 2.16 1.08 33.8 597 2.5 1.45 0 0.25 0.62 1.80 1.14 42.8 620 2.5 1.45 0 0.04 0.81 2.52 1.26 55.9 628 2.5 1.45 0 0 0.91 2.10 1.26 62.8 634 2.5 1.45 0 0 0.84 2.40 1.38 57.9 652 2.5 1.45 0 0 0.93 1.80 1.38 64.0 maleic anhydride (exit) hydroquinone (inlet) - hydroquinone (exit) Weight of catalyst, 8.034 g; selectivity = In the temperature range 620-650 K the selectivities to maleic anhydride from benzene and from hydroquinone are virtually identical, suggesting that hydroquinone is an intermediate in the selective reaction pathway from benzene to maleic anhydride.However, at lower temperatures the selectivities to maleic anhydride are significantly lower and considerable quantities of p-benzoquinone are formed.(The selectivities to p-benzoquinone are 28.3% at 588 K, 17.2% at 597 K and 2.8% at 620 K.) From the inverse relationship between the maleic anhydride selectivities, which increase over the temperature range 588-628 K, and those to p-benzoquinone, which decrease over the same temperature range, it is inferred that the activation energy for the adsorption of hydroquinone into that state which producesp-benzoquinone has a small advantage over the energy barrier for adsorption into the species which eventually forms maleic anhydride. However, no hydroquinone was ever detected in the products of benzene oxidation over the vanadium pentoxide-molybdenum trioxidecatalyst, while only trace quantities 21 FAR 1810 SELECTIVITY I N OXIDATION OF BENZENE ofp-benzoquinone have been noted, which, when considered with our previous finding that the maleic anhydride is formed at the site at which the benzene was first adsorbed,2 suggests that the difficulty in adsorbing hydroquinone propitiously does not exist in benzene oxidation itself.The reaction is confined entirely to the surface of the catalyst. MECHANISM A rationale for the observation that benzene and hydroquinone have identical selectivities, while that of p-benzoquinone is extremely low, and for our previous finding that the selective oxidant is molecularly held oxygen2 is to be found by consideration of the electronic perturbations of the reactants on chemisorption and the reactions allowed, on the basis of orbital symmetry arguments,g to these electronically perturbed adsorbed species.[While acknowledging that the Woodward- Hoffmann rules are strictly qualitative, dealing mainly with the angular contribution to the total wavefunction, in effect neglecting the radial contribution, they are introduced as a form of justification of the feasibility of reactions which are postulated to occur between chemisorbed oxygen and benzene and chemisorbed oxygen and hydroquinone (in the absence of this justification these reactions might otherwise be viewed as speculative). The arguments are considered solely in respect of the adsorptions and reaction on the V5+, V4+ ions at the surface but apply, with minor modifications for ionic size, to Mo6+, Mo5+ ions, should they also occur at the surface.] The 30% molar solution of MOO, in V205 coincides with its maximum solubility in V205.7 The role of the former, by the substitution of the larger Mas+ ion for the V5+ ion, has been judged to produce inclusion strain in the vanadium pentoxide lattice7 and therefore to maximise the number of V4+ ions.Shvets et aZ.,8r9 using e.s.r. spectroscopy, have shown that the mechanism of oxygen adsorption on V,05 is O,(gas) -+ 02(ads) + O;(ads) -+ 20-(ads) + 202-(1at) (1) [where 02- (lat) is oxygen of the lattice] and that 0; (ads) and 0- (ads) co-exist on the surface. The 0; species is formed by interaction of the precursor 0, (ads) with a V4+ ion on the surface, donation of the electron being made from either the d,, or d,, orbitals of the vanadium ion to one of the antibonding II orbitals of the oxygen Q D Q D FIG.3.-Symmetry allowed combination of the oxygen K* orbitals and of the dyL and d,, orbitals of the vanadium ion V4+.R. W. PETTS AND K. C. WAUGH 81 1 atom. These are symmetry allowed reactions (see fig. 3). The electronic configuration of the 0, (ads) will therefore be [+)I2 [o*(s)l2 [a(p)I2 [zI4 [z*I3, the extra electron being located in either the x or y, n* orbitals. [The oxygen molecule in fig. 3 is depicted as adsorbing ' end-on ', although a ' sideways-on ' configuration is also symmetry allowed. The spatial arrangement of the n* orbitals of oxygen in 0; and of the dvz or dzz orbitals of the V4+ ion suggest that the end-on configuration will lead to the maximum orbital overlap and hence the stronger bond.This conclusion derives from charge density considerations for 0; lo which show that the 0.002 contour of the total charge density (95% of the total charge) in 0; has the dimensions of 4.45 A along the intermolecular axis and 3.39A at right angles to it, while the maximum charge density of the d,, FIG. 4.-(a) Symmetries of the constituent atomic orbitals of the allowed combinations which comprise the six molecular orbitals of benzene. (b) Bonding orbitals involved in the chemisorption of benzene on the V6+ ion of vanadium pentoxide; the yl(E,a) orbital of benzene overlaps with the d,, or the d,, of the vanadium ion. 21-2812 SELECTIVITY I N OXIDATION OF BENZENE or dz5 orbitals is 0.8 A apart, 95% of the total charge lying within a 3 A distance.llU The end-on adsorption therefore more closely coincides with the spatial distribution of charge in the d,, or dzz orbitals.] The weakly bound oxygen which has been shown to be the selective oxidant in benzene oxidation2 is likely to be 0;.The signs of the atomic wavefunctions of the allowed combinations which constitute the six molecular orbitals of the electron system of benzene are shown in fig. 4(a).12 The energies of these orbitals are y/(A) < y/(Ela) = y/(E,b) < y/(E2a) = y/(E,b) < y/(B) so that the ground-state electronic distribution in benzene is [y/(A)I2, Chemisorption of benzene will probably form a compound shown in fig. 4(b), which is strictly analogous to the chromium-benzene complexes well known in transition- metal chemistry.llb The bond will be formed, on distance and overlap considerations, by electron donation from the y/(Ela) orbital to either the duz or the dz5 orbitals of a surface V5+ ion [fig.4(b)]. The electronic configuration of chemisorbed benzene will The correlation diagram for the concerted reaction of chemisorbed oxygen with [y/(El 412, [y/(El b)I2, ME2 41°, [y/(E2 b)I09 "3)1°. therefore be [W)l2, [y/(El 4 3 ' 9 [y/Y(El 412, [y/(E2 41°, [ W 2 b)I0, [y/(B)I0. chemisorbed benzene [reaction (2)] is shown in fig. 5. CJc FIG. 5.--Correlation diagram for the 1,4- addition of a chemisorbed 0; species across a chemisorbed benzene molecule.R. W. PETTS AND K. C . WAUGH 81 3 When orbital electronic occupancy is considered the combination of an electron deficient y/(E,a) orbital and a singly occupied 7t* oxygen orbital is both symmetry allowed and makes a net bonding contribution of two electrons in the o o orbital. Surprisingly, reaction (2) is also perfectly feasible on bond-distance considerations, since in spite of the oxygen molecule bond length being only 1.2 A as described earlier, the charge density of the 0; extends to 4.45 A, while the 1,4- distance across the benzene ring is only 2.79 A.(Indeed, 1,3- addition of 0; to chemisorbed benzene, which is also bonding and symmetry allowed, is less favoured on bond-distance considerations, it being only 2.42 A; the resultant overlap with the oxygen antibonding orbitals will be smaller and therefore weaker.) The mode of transport of the end-on chemisorbed oxygen molecule anion to the chemisorbed benzene molecule (a previous paper2 has shown that it is the oxygen which migrates to the immobile, adsorbed benzene) is envisaged as involving a pendular vibration of the oxygen molecule as a whole with respect to the vanadium cation to which it is attached.Adsorbed vicinal to the chemisorbed benzene molecule, the rocking vibrational mode of the oxygen molecule [indicated by the bent arrow, n, in reaction (2)] could easily result in 1,4- addition to the benzene molecule, which itself is held planar to the catalyst at the $6- positions. The product of reaction (2), the benzene-oxygen adduct (identical to that proposed by Dmuchovsky et all3), will because of the heat released in forming two carbon-oxygen bonds and because of the extra electron in the antibonding oxygen orbital, probably rearrange to form hydroquinone.Although the addition of gas-phase oxygen to the electronically deficient, chemisorbed benzene molecule is symmetry allowed, the net bonding contribution is only one electron in the 0-0 orbital; this would result in a lower heat release on formation of the benzene-oxygen adduct, which, together with the absence of an extra electron in the oxygen antibonding orbital, would result in the adduct having a higher activation energy (lower probability) of rearranging to hydroquinone. As always, the apparently lower energy reaction pathway (in this case the one involving the chemisorbed 0; species) is favoured, but this nevertheless accords with our previous observation2 that the overall activation energy for the oxidation of an adsorbed benzene molecule to an adsorbed maleic anhydride molecule is low, 31.4 kJ mol-l.Adsorption of oxygen, now to a position vicinal to a chemisorbed hydroquinone molecule, followed by 1,4- addition of the adsorbed oxygen molecule anion to the chemisorbed (and therefore electronically perturbed) hydroquinone molecule produces a hydroquinone-oxygen adduct, strictly analogous to that formed between chemisorbed benzene and chemisorbed oxygen [reaction (3)]. The correlation diagram for this reaction is identical to that employed for the formation of the benzene oxygen adduct (fig. 5) and the reaction is therefore symmetry allowed and bonding. Like the benzene-oxygen adduct, the hydroquinone-oxygen adduct will rearrange but in this case to adsorbed maleic anhydride, an adsorbed acetylide residue and water.The complete reaction sequence is shown in reaction (3), all the reactions of which are confined to the surface of the catalyst [the reaction of gas-phase molecular oxygen with gas-phase hydroquinone, or with the benzene-oxygen adduct as required by Trimm and c o ~ o r k e r s , ~ is symmetry forbidden, while the formation of the comparable anthracene-oxygen adducts, common in organic chemistry, requires the involvement of the excited singlet state of oxygen (a state not far removed from the 0; species) for the reaction to occur].14 The near identity of the selectivities of benzene and hydroquinone is explained by this mechanism as is the necessary involvement of a weakly bound, molecularly held oxygen.The stability of the strongly chemisorbed2 maleic anhydride molecule to814 SELECTIVITY I N OXIDATION OF BENZENE \ I 3 \ I + v5+ 5 +(+el 1 \ / 5 + t + e ) + V 4 + O-(ads) OzPi[at I I o=y+ + C,H,( ads) - \ I + H20 \ / ,, 5 t (+el COz+ CO+ HZO t v4+ 1 further oxidation can also be understood since 1,2- addition of 0; to the carbon-carbon double bond of the chemisorbed maleic anhydride molecule is forbidden on bond- distance considerations; the reaction therefore must involve the 0- (ads) or 02- (lat) species, the latter of which, according to Bond,15 is difficult to remove. The non-involvement ofp-benzoquinone in forming maleic anhydride (the low selectivities to maleic anhydride obtained from it could almost be due to hydroquinone impurities) follows from its structure. It will not be involved in 1,4- molecular oxygen addition, 2,3- reaction with the carbon-carbon double bond being forbidden on distance considerations, and so, like maleic anhydride, it can only be involved with the non- selective reactions of 0- (ads) or 02- (lat). J. Lucas, D. Vandervell and K. C. Waugh, J. Chem. SOC., Faraday Trans. I , 1981, 77, 15. J. Lucas, D. Vandervell and K. C. Waugh, J. Chem. SOC., Faraday Trans. I , 1981, 77, 31. D. M. Brown and D. L. Trimm, Proc. R. SOC. London, Ser. A, 1972,326, 215. M. L. Brisk, R. L. Day, M. Jones and J. B. Warren, Trans. Inst. Chem. Eng., 1968, 46, No. I , T3. J. E. Germain, Catalytic Conversion of Hydrocarbons (Academic Press, New York, 1969), chap. 5, p. 259. F. D. Mango, Aduances in Catalysis (Academic Press, New York, 1969), vol. 20, p. 291. I. I. Ioffe, Z. E. Ezhkova and A. G. Lyubarskii, Kinet. Katal., 1962, 3, 194.R. W. PETTS AND K. C. WAUGH 815 V. A. Shvets, V. M. Vorotyntser and V. B. Kazanskii, J. Catal., 1968, 10, 287. * V. A. Shvets, M. E. Sarichev and V. B. Kazanskii, J. Catal., 1968, 11, 378. lo P. E. Cade, R. F. W. Bader and J. Pelletier, J . Chem. Phys., 1971, 54, 3517. L. E. Orgel, An Introduction to Transition Metal Chemistry: Ligand Field Theory (Methuen, London, 1960), (a) p. 21, (b) p. 160. l2 F. A. Cotton, Chemical Applications of Group Theory (Wiley-Interscience, New York, 1971), p. 136. l3 B. Dmuchovsky, M. C. Freerks, E. D. Pierron, R. H. Munch and F. B. Zienty, J. Catal., 1965,4,291. l4 H. Heaney, Comprehensive Organic Chemistry, ed. J. F. Stoddart (Pergamon Press, Oxford, 1979), I6 G. C . Bond, Catalysis by Metals (Academic Press, New York 1962), chap. 1, p. 2. vol. I, p. 330. (PAPER 1 /545)
ISSN:0300-9599
DOI:10.1039/F19827800803
出版商:RSC
年代:1982
数据来源: RSC
|
15. |
Influence of the bulk and surface properties on the performance of iron–antimony catalysts |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 817-833
Nicola Burriesci,
Preview
|
PDF (1057KB)
|
|
摘要:
J . Chem. SOC., Faraday Trans. I , 1982, 78, 817-833 Influence of the Bulk and Surface Properties on the Performance of Iron-Antimony Catalysts BY NICOLA BURRIESCI, FABIO GARBASSI,* MICHELE PETRERA AND GUIDO PETRINI Istituto Guido Donegani SPA, Centro Ricerche Novara, Via G. Fauser 4, 28100 Novara, Italy Received 6th April, 1981 The modifications induced in Fe-Sb catalysts by the introduction of an excess of antimony oxide, which is needed in order to obtain highly selective catalysts in the ammoxidation of propylene to acrylonitrile, were investigated by means of X-ray diffraction (X.r.d.), X-ray photoelectron spectroscopy (X.P.S.) and Mossbauer spectroscopy. More imDortant than increasing the surface Sb: Fe ratio, a promoting effect by an excess of Sb was found to develop during activation through the formation of structurally distorted and defective FeSbO,, which appears to be the active phase.Fez+ ions are thus introduced into the iron antimonate rutile structure near oxygen vacancies. These vacancies are possibly connected with the adsorption sites for the more strongly bound oxygen species that is responsible for allylic oxidation. Iron-antimony catalysts are known to be active in the allylic oxidation of olefinsl and have been patented, inter alia, for the ammoxidation of propylene to acrylonitrile, a reaction of relevant industrial interest2? which can be sketched as follows:4 O,,NH, H,C=CH--CH, - H,C=CH-CN + 3H,O. H,C=CH-CHO Apart from acrolein, HCN and acetonitrile are the other products of the reaction. The predominant evidence suggests, however, a direct path to acrylonitrile not involving the acrolein intermediatee5 The Fe-Sb oxide system exhibits catalytic behaviour that strongly depends on the Sb: Fe ratio.,? Various interpretations of this phenomenon have been proposed6-8 but not established so far.Concerning the nature of the sites where partial oxidation of propylene can occur, the conclusions drawn by various authors are contradictory. It is generally accepted that the main role in such reaction is played by Sb5+ ions, but there is no agreement on the surface antimony compound present in the catalyst. Boreskov et d5 propose that the formation of FeSbO, is the main cause of the increased selectivity of the catalyst, while Sala and Trifirb6 consider this compound as a deep oxidation catalyst and propose as the selective phase Fe,Sb,O, or FeSb,O,.Yamazoe et aL8 support the role of this latter phase by giving evidence for the presence of Fe2+ ions through X.P.S. analysis. To gain better understanding of the role of the catalyst composition in selectivity, we investigated two series of samples with very different catalytic behaviour by Mossbauer spectroscopy and X.P.S. These two techniques have been widely applied 817818 I R 0 N-AN T I MON Y C A T A L Y STS in the chemical characterization of a variety of catalytic systems and their combination often allows a detailed insight into the bulk and surface properties of the material under inve~tigation.~ The aim of this work is to attempt a correlation between chemico-physical features and catalytic behaviour.EXPERIMENTAL SAMPLE PREPARATION Catalysts were obtained by preparing a slurry of Sb,03 (Carlo Erba RPE) in aqueous iron nitrate (Carlo Erba RPE) solution. After drying overnight at 383 K, the samples were calcined in air at various temperatures (from 573 to 1173 K) in subsequent heating steps of 100 K for 24 h. In this investigation two series of samples, having molar Sb/Fe ratios of 1 or 2, were prepared and examined. In the following we will refer to these samples as C1 and C2, respectively. X-ray diffraction analyses were made by means of a Philips powder diffractometer (Cu Ka radiation), equipped with a scintillation counter and a pulse-height analyser. MOSSBAUER ANALYSIS The Mossbauer spectra were recorded with a constant-acceleration driver using a Co-Rh source maintained at room temperature (r.t.).The calibration of the velocity scale was made using a metallic iron absorber. Measurements at liquid-nitrogen temperature (1.n.t.) were performed using a variable- temperature cryostat (Oxford DN704) equipped with a digital temperature controller (Oxford DCT-2). An RSP3 Reuter-Stockes proportional counter was used connected to an Elscint AME40C computerized spectrometer. Data points were fitted with lorentzian curves using an iterative least-squares fitting program. Centre shifts (6) are always given with respect to metallic iron. Standard deviations are in the range f0.05 mm s-' for the IS values, kO.1 mm s-' for quadrupole splitting (A) values and & 10 kOe for hyperfine field values.Larger deviations were found in the case of relaxation phenomena. X.P.S. MEASUREMENTS The experiments were carried out on a PHI 1.e.e.d.-A.e.s.-X.p.s. instrument equipped with a double-pass cylindrical analyser (model 15-2556). During the measurements the basic pressure into the analysis chamber was maintained in the lo-' Pa range. After insertion in In foils (Goodfellow 99.999%)1° the powdered samples were analysed using a Mg Ka 400 W source. X-ray photoelectron spectra were collected under high-resolution conditions (pass energy 50 eV), by means of a PDP 11/50 computer (Digital Equipment Corp.) connected to the spectrometer through a digital-analogue interface. Details of the system performance (signal averaging, storage, smoothing, background subtraction, integration, deconvolution and plotting of spectra) are described elsewhere." The 3d, and 3 4 peaks of Sb were measured, the latter being overlapped by the 0 1s transition.A flood electron gun was used to counterbalance sample charging and the C Is contamination peak at 285.0 eV was used as an independent check. Quantitative X.P.S. analyses were carried out using the same instrumental conditions in order to minimize the instrumental influence. From 10 to 30 runs were averaged for each peak. Even if the separation of Sb 3 4 and 0 1s contributions at ca. 530 eV was found possible in the case of pure antimony oxides, using a fitting model with gaussian components,ll this procedure was not considered necessary in the present study. The oxygen peak intensities, I,, were instead calculated as follows: OSb5 '0 = I(Sb5+O) - 'Sb3 - OSb3 where I(Sb5+0) represents the measured intensity (by digital integration) of the compound peak at 530 eV, ISb3 the measured intensity of the Sb 3 4 peak and Osb5/O?,3 the ratio of the photoelectron cross-sections for the two terms of the spin-orbit doublet.This ratio was obtainedN. BURRIESCI, F. GARBASSI, M. PETRERA AND G. PETRINI 819 from values published by Scofield12 and previously found to be in good agreement with the experimental intensity rati0.l' The concentration C, of the element x was calculated as c, = ~ IXlfX x 100 C (Iilf,) a where Ii is the measured peak intensity andf, the elemental sensitivity factor, which takes into account the variation of the electron escape depth and detection efficiency of the analyser with the kinetic energy of the electron~.'~ Values of 2.50 for Fe 2pr and 2.45 for Sb 3 4 were adopted.With the above procedure, composition comparisons among different samples are highly reliable, while some uncertainty can affect the absolute values of Sb: Fe ratios, which are of particular interest in this work. For this reason, some mechanical mixtures of Fe203 and Sb20, powders were prepared and analysed. CATALYTIC CHARACTERIZATION The catalytic activities were measured in a flow stainless-steel reactor, with an internal diameter of 10 mm, containing ca. 4.5 cm3 of cataly~t.~ Sample activations were carried out as follows: 16 h at 923 K, 16 h at 1023 K and 24 h at 1123 K. The gas-feed composition (molar ratios) was air: propylene = 1 1 ; ammonia: propylene = 1.35.Several temperatures (673-773 K) and contact times (2.5-7 s) were used. Product analysis was carried out by a gas chromatograph directly connected to the reactor. Selectivities to acrylonitrile, HCN, acetonitrile, acrolein, CO, and CO were taken into account. RESULTS X-RAY ANALYSIS The X-ray powder spectra revealed the presence of orthorhombic Sb,O, (ASTM 1 1 -689), orthorhombic Sb204 (ASTM 1 1 -694), a-Fe,O, (ASTM 13-534) and FeSb0,,l4? l5 depending on the sample calcination temperature. Phase analysis as a function of temperature, T, is sketched in fig. 1, which shows the complete disappearance of Sb204 at 1173 K in Cl and the anticipated disappearance of free ion oxide in C2. The first item suggests that a quantitative formation of FeSbO, was reached, while the second recalls that a solid-state reaction between the two oxides is easier in the presence of an excess of Sb,O,.An accurate quantitative analysis was hindered by peaks overlapping and broaden- ing. However, a rough estimate of the iron oxide (d= 2.69) to iron antimonate - Sb204 l ~ l ~ l ~ 1 1 ' 1 400 600 800 1000 1200 T/K FIG. 1 .-Occurrence of the different phases as a function of temperature in Cl (dashed bars) and C2 (empty bars) samples.820 IRON-ANT I M ON Y CAT A LY S TS 3- 0 .* +b 2 2 - c 2 .* c, .r( 1 - 0- T K FIG. 2.-Intensity ratio of X-ray diffraction peaks of a-Fe,O, and FeSbO, (0, C1 samples; @, C2 samples). (d = 1.72) peak intensity ratio as a function of firing temperature is reported in fig.2. The haematite phase progressively disappears with increasing T. As expected, the a-Fe,O, consumption is faster in C2 than in C1 samples. A discontinuity in the reaction rate was observed at 973 K for C1 catalysts, which is probably due to a change in the reaction mechanism (i.e. surface to bulk diffusion). MOSSBAUER ANALYSIS Some representative room temperature (r.t.) and liquid-nitrogen temperature (1.n.t.) spectra are reported in fig. 3 and 4. The experimental data were fitted by introducing an appropriate number of different spectral components (Dl, D2, S1, S2). All the spectra show a doublet D l whose isomer shift (6) and quadrupole splitting (A) values (0.38 and 0.74 to 0.91 mm s-l at r.t., respectively) are characteristic of the distorted octahedral high-spin Fe3+ in FeSbO,.ls9l7 A second doublet (D2) was observed in various amounts in the 1.n.t.spectra of samples fired at T 2 873 K and it was related to a small amount of divalent iron. Mean values of the Mossbauer parameters of the D2 component are 1.3 mm s-l for 6 and 3.0 mm s-l for A, which are typical of octahedral high-spin Fe2+. The S1 sextet is attributed to a-Fe,O,, whereas a second magnetically split component S2 has been introduced in some 1.n.t. spectra in order to account for the relaxation phenomena occurring in FeSbO,. For better accuracy, the constraint for the area values discussed in a previous work on pure FeSbO, l8 has been applied to the lines of the S2 component.N. BURRIESCI, F. GARBASSI, M. PETRERA AND G. PETRINI 82 1 1 .oo 0.98 0.96 e :: 0.94 E 2 E $ 2 1.00 2 Y .- Y 0.98 0.96 D I n s1 I I I I I 1 I I I I I I I I I 1 1 I I I I I 1 I -8 -6 -4 -2 0 2 4 6 velocity/mm s-' FIG.3.-Mossbauer r.t. spectra of C1 (above) and C2 (below) samples fired at 1123 K. Curves reported in fig. 5 clearly show that the relative amount of a-Fe,O, increases for calcination temperatures up to 573 K as a consequence of the haematite particle size growth, which determines an increase of the magnetic component spectral area. When 573 < T/K < 973, this ratio remains practically constant in C1, indicating a certain stability of both phases in the given temperature range. A sharp decrease of the haematite phase is observed at T > 973 K, indicating a faster rate of the solid-state reaction between antimony and iron oxides leading to FeSbO,.For the C2 samples, fig. 5 shows different behaviour for the Sl/(Dl + S2) ratio, with a remarkable reaction rate even at T > 773 K. This is probably connected to the higher content of free antimony oxides in C2 and to the smaller particle size in these samples, at least for the iron-containing phases. The comparison between relative spectral areas measured at r.t. and 1.n.t. indicates that a-Fe,O, is partially present as small particles exhibiting superparamagnetic behaviour at r.t. This is confirmed by the relatively low hyperfine field values of the S1 magnetic component with respect to bulk haematite (Ifeff = 518 and 542 kOe at r.t. and l.n.t., respectively), as can be seen from fig. 6. In fact, a decrease in Heff is typical of superparamagnetic particles subjected to the appearance of collective spin excitations.19 The Mossbauer analysis confirms for this system the possibility that Fe3+ is reduced in air to Fez+ at temperatures higher than ca.800 K. The fraction of reduced iron apparently increases when a larger amount of free antimony oxides is present, i.e. in the case of C2 catalysts. As the firing822 I R ON-AN T I MO N Y CAT A LY S TS D? n D2 s1 I I 1 I I 1 s 2 r I I I I 1 1 .oo 0 98 0.96 E 2 094 .I .g c e 4- 0 5 100 - E 0.98 0.96 0 94 I t 1 I I 1 I I I -10 -8 -6 -4 -2 0 2 4 6 8 1 0 velocity/mm s-' FIG. 4.-Mossbauer 1.n.t. spectra of C1 (above) and C2 (below) samples fired at 1123 K. temperature is increased, the amount of divalent Fe reaches a maximum at 1 123- 1 173 K and then decreases again (fig.5). The dependence on firing temperature of the iron antimonate quadrupole splittings, i.e. of the iron site symmetry, is shown in fig. 7. The A values for both series of catalysts are noticeably larger than the corresponding values for pure FeSbO,, especially for those firing temperatures where the concentration of haematite is relatively high. The influence of Sb,O, can be ruled out, since samples with very different amounts of this oxide show the same A value above 1100 K. SURFACE ANALYSIS Table 1 reports the surface composition (in at. %) of the two series of samples, calculated from the X.p. spectral intensities. The same results are shown in fig. 8, where the surface Sb:Fe ratios are reported as a function of temperature. In C1 a continuous increase in the measured Sb: Fe ratio was found, with the increase in the temperature from 573 to 1073 K.A higher value was shown by samples dried at 383 K. Quantitative analyses of several mechanical mixtures of iron and antimony oxides, reported in fig. 9, show that errors inherent in the method of analysis or variations in the elemental sensitivity factors due to matrix effects are limited to 30% in favour of Fe. Thus we can conclude that strong Sb enrichment occurs at the surface for high activation temperatures.N. BURRIESCI, F. GARBASSI, M. PETRERA A N D G. PETRINI 823 FIG. 5.-Normalized area values of the single components in 1.n.t. Mossbauer spectra. Empty symbols refer to C1 samples, full symbols to C2 samples. 0, 0, D1 +S2; 0, B, S l ; A, A, D2.For an explanation of abbreviations, see text. 540- 450 I I I I I I 1 I I 400 600 800 wxx) lz00 TIK FIG. 6.-Hype&ne field values of the S1 magnetic component as a function of firing temperature (empty bars: C1 samples; full bars: C2 samples). Dashed lines refer to 1.n.t. spectra, unbroken lines to r.t. spectra. Bar lengths represent the standard deviation.824 I R 0 N-A N T I M 0 N Y CAT A L Y S T S 1 .o 0.9 . I v) E E 2 0.8 0.7 I I I I I I I I I 400 600 800 lo00 1200 TIK FIG. 7.-Quadrupole splitting values of FeSbO, in C1 (empty bars) and C2 (full bars) samples as a function of temperature. Empty squares represent A values for pure FeSb0,.18 Bar lengths indicate the standard deviation. TABLE SU SURFACE COMPOSITION FROM X.P.S. PEAK INTENSITIES (at.%) C1 samples C2 samples heating temp./K 0 Sb Fe 0 Sb Fe 383 573 673 773 873 973 1073 1123 1173 81.3 79.9 79.9 80.8 79.9 79.8 75.6 78.5 79.9 12.2 4.7 6.2 7.5 10.7 11.9 19.6 19.2 16.1 6.5 15.4 13.9 11.7 9.4 8.3 4.8 2.3 4.0 82.1 71.3 74.5 73.1 74.6 73.9 73.3 71.4 72.5 16.6 23.6 18.5 19.2 19.8 19.5 23.6 23.4 23.5 1.3 5.1 7.0 7.7 5.6 6.6 3.1 5.2 3.9 C2 samples show similar behaviour, with some differences. The Sb: Fe variation with temperature is less pronounced and the ratio values are always above the bulk Sb: Fe ratio. Also this series exhibits a decrease in the metal ratio (i.e. an increase in surface iron) after the first activation step, at 573 K. In both cases the scattering of Sb: Fe values at T > 1100 K can be attributed to the very low intensity of the Fe photoemission peaks and to the larger errors in the background subtraction.The result is that in this temperature range the Sb:Fe ratio is definitely higher than that in the bulk. Examining table 1, the oxygen content appears largely constant between 573 and 973 K in C1 samples, and between 673 and 1073 K in C2. On the low-temperatureN. BURRIESCI, F. GARBASSI, M. PETRERA A N D G . PETRINI 825 10- 5- h .1' a 2- x, 4 =: '- 0.5- 0.2- I\, \ \ \ \ \ \ i \ \ \ \ L I I I I I I I I I 400 600 800 1000 1200 TIK FIG. 8.-Sb: Fe surface ratio as a function of temperature for C1 (0) and C2 (0) samples. Arrows indicate the expected values on the basis of bulk composition. Sb : Fe (bulk) FIG. 9.--Comparison between Sb: Fe ratios drawn from X.P.S. data and expected values for mechanical mixtures of a-Fe,O, and Sb,O, powders.826 383 K 973 K i I I I I I I I 1 I IR ON-AN T I M O N Y CAT A L Y S T S side a decrease occurs, probably connected with the loss of H20 and OH- species, as is suggested by the asymmetry of the peaks near 530 eV before calcination (fig. 10).In contrast, at high temperature the variations in the amount of oxygen are probably caused by the solid-state reaction or by some mobility of the lattice oxygen atoms. Apart from the behaviour described above, the oxygen concentration in C 1 samples is always higher than in C2. A mean difference of ca. 6% was observed between 573 and 1173 K without significant variations, apart from at 1073 K where the rate of formation of FeSbO, reaches a maximum (fig. 5).Binding-energy values of Sb 3 4 and Fe 2pi photoemission peaks are shown in fig. 1 1 . Positions and profiles of iron photoemission peaks are consistent with the presence of Fe3+ on the surface. Up to 1073 K, on both series of samples, binding energies of 710.8-711.1 eV are in good agreement with published data for a- Fe203.13919920 Above this temperature a shift of ca. 1 eV towards higher binding energies was observed. In table 2 binding energies previously foundll for antimony oxides are reported, together with some literature 22 The separation between the binding-energy levels of Sb3+ and Sb5+ is ca. 1 eV in pure oxides, and was measured as 0.6 eV in mixed-valence Sb204.23 In such conditions, the determination of the Sb valence is not easy, and is complicated by the presence of electrostatic charge shifts.However, it seems clear from the data shown in fig. 1 1 that at low temperature in both series of samples Sb is present at the surface as Sb204. At higher temperatures a trend in Sb binding energy towards values typical of Sb5+ can be reasonably assumed. At T > 1100 K, Sb binding energies further increase up to a value around 541 eV,N. BURRIESCI, F. GARBASSI, M. PETRERA A N D G. PETRINI 827 4 712 540 4 0 0 600 800 1200 TIK FIG. 1 1 .-Binding-energy values of Fe 2pa (above) and Sb 3 4 (below) peaks as a function of temperature for C1 (empty bars) and C2 (full bars) samples. Bar lengths represent the estimated errors. TABLE 2.-BINDING-ENERGY VALUES OF ANTIMONY OXIDES Sb 3d2/eV oxide ref. (1 1) ref. (21) ref. (22) 539.6 540.0 540.6 539.6 540.0 540.5 540.6 539sa - 540.4a a Binding-energy value of Sb 3 4 plus 9.4 eV.accompanied by a corresponding shift of the Fe peak to 712.2 eV. As most of the FeSbO, is formed in such a temperature range, it could be argued that the new binding- energy values are typical of the two cations in such a compound. Consequently, the Me-0 bonds ought to be less covalent than in free Although the FeSbO, structure is not known so accurately as to rule out this statement, it is however scarcely convincing, since comparable binding-energy values of Sb or Fe in similar chemical compounds have not been found so far, apart from a recent result for FeS0,.25 Bearing in mind that as the temperature here is increased a Sb-rich surface layer is formed, it can be suggested that the formation of such a layer, possibly a stronger insulator than the bulk, causes an electrostatic charge shift of the electrons photoemitted by the sub-surface layer.The larger shift of Fe with respect to Sb is consistent with this interpretation, as a large contribution by the surface layer is present in the Sb signal. CATALYTIC BEHAVIOUR The influence of the activation temperature on activity and selectivity was studied for representative samples from both series of catalysts. As an example, the results828 IRON-AN T I MON Y CAT A LY S T S ’ O 0 1 ? 000 1100 1200 T/K FIG. 12.-Selectivity of C2 catalysts as a function of firing temperature (0, propylene conversion; x , acrylonitrile selectivity; 0, CO + CO, selectivity). obtained on a C2 sample are reported in fig.12. An enhancement of activity in the conversion of propylene and a strong improvement in selectivity were found on increasing the calcination temperature from 1023 to 1123 K. Catalytic data as a function of the Sb: Fe ratio are reported in fig. 13 for samples activated at 1123 K. Very different selectivities were observed on varying Sb: Fe from 1 to 2 (corresponding to C 1 or C2 samples, respectively), while a substantial constancy in the catalytic properties was observed for higher Sb: Fe contents. In particular, the selectivity to acrylonitrile increases from 20 to over 70% on passing from C1 to C2, reducing the CO + CO, production, which falls from 75 to 18 %. The contribution to activity and selectivity of the small amounts of free a-Fe,O, present in the C1 samples appears negligible.Such results agree with previous literature DISCUSSION The formation of a compound of Fe and Sb is much slower in C1 than in C2 catalysts, as is shown both by X.r.d. (fig. 2) and Mossbauer spectroscopy (fig. 5). Furthermore, a-Fe,O, completely disappears at 1073 K in C2, while some small amounts remain unreacted in C1. About the nature of the compound containing the two cations, Tarasova et a1.26 (on the basis of X.r.d. and i.r. spectroscopic measure- ments), Malakhov and Abdik~va,~ (by chemical and diffractometric analysis) and Anufrienko et concluded that the formation of FeSbO, occurs, when starting from a coprecipitated mixture of oxides. Kriegsmann et al.” (on the basis of X.r.d. data) did not take sides between FeSbO, and Fe,Sb,O,. Sala and Trifiro6 (mainly from i.r.spectra) suggested the formation of FeSb,O, or Fe,Sb,O,. On the basis of X.r.d. and Mossbauer results, we conclude that the main compound formed in our samples is FeSbO,.N. BURRIESCI, F. GARBASSI, M. PETRERA A N D G. PETRINI 829 100 - 9 0- 80- 70 - 60 - n E ;L, Y 's 50- ?A .- Y u w 40- 30- 20- 10- 0- \ 'a FIG. 13.-Catalytic properties of iron-antimony catalysts as a function of the bulk composition (0, propylene conversion; x , acrylonitrile selectivity; 0, CO + CO, selectivity; A, HCN selectivity). The correlation suggested by Kriegsmann et a1.l' between acrylonitrile yield and FeSbO, quadrupole splitting is not confirmed by our findings, as very different acrylonitrile selectivities were found for catalysts showing about the same value of quadrupole splitting (see fig.7 and 13). A remarkable result shown by Mossbauer spectroscopy was the detection (at 1.n.t.) of divalent iron in C2 samples heated at T 2 873 K, while in C1 catalysts its presence was confirmed only after heating at 1073 K, and in a lesser quantity. The presence of Fe2+ in similar samples was similarly observed by Yamazoe et al., who were able830 I R ON-A N T I M ON Y CAT A L Y S T S to detect this chemical species by X.P.S. In our case, the Fe photoemission peaks were too broad and weak to show the same evidence. An important point is to determine whether divalent iron participates in a phase other than FeSbO,, or is rather present as a foreign ion in the lattice of this compound.On the basis of the Sb: Fe value of ca. 2 found by X.P.S., it was suggested that Fe2+ exists as FeSb,O, in an ultrathin layer (6 A) at the surface of the catalyst.* Such an interpretation is not satisfactory for our results. First, divalent iron was detected by Mossbauer spectroscopy (in this case, a technique more sensitive than X.P.S.) both in C2 and in C1 samples, indicating that the improved selectivity of the C2 catalysts can hardly be related to the formation of a surface Fez+-Sb5+ compound, which should be present in both samples. Secondly, Fez+ was found in amounts much greater than a few layers at the surface of the sample (fig. 5). With respect to the Mossbauer parameters no comparison is possible because values for the hypothetical compound FeSb,O,, envisaged by analogy with other compounds of Sb with Ni, Co or Zn,6 are lacking.The’formation of Fe,Sb,O, was also excluded, taking into account the work of Gakiel and Malam~d,,~ who concluded that Fe,Sb,O, was erroneously assumed as the chemical formula of the mineral tripuhjite. However, the parameters found by us for divalent iron are rather close to those reported for FeSb,O, (6 = 1.20 mm s-I and A = 3.01 mm s-l at 8 K).30 Taking into account the reported values for the crystal-field splittings A1 and A2 (83 and 500 cm-l, re~pectively),~~ which confirm the nearly regular octahedral Fe2+ coordination in FeSb,O,, the slightly larger A value found by us at l.n.t.31932 could be easily explained by a higher degree of distortion possibly induced by surface effects.Therefore, the formation of such an iron-antimony compound would account for both the Mossbauer results and the role played by the excess Sb, facilitating the bulk reduction of ferric to ferrous iron. Owing to the considerable amount of Fe2+ in some samples, we expect that an additional phase would be detected by X.r.d. Another possibility is that no new phase is formed and Fez+ ions are incorporated in the rutile structure of FeSbO,. It has previously been shown that divalent iron forms complexes with the normal crystalline defects of r ~ t i l e . ~ ~ In order to clarify this point, we further calcined the CI and C2 catalysts at TABLE 3 .-UNIT-CELL PARAMETERS OF FeSbO, sample temp./K %/A %/A c1 c1 c 2 1173 4.633 & 0.001 3.072 & 0.001 1373 4.603 & 0.007 3.069 f 0.009 1373 4.615&0.007 3.071 & 0.009 1373 K for 20 h and looked at the X-ray diffraction patterns.Again no new phase was detected, and no indication of the presence of amorphous compounds was observed. We also measured by the Debye technique the cell parameters of the FeSbO, phase in some samples heated at high temperature. Results are reported in table 3. Rather than making a comparison with literature data,15 attention must be paid to differences with respect to the Cl sample heated at 1373 K, where FeSbO, is the only phase present and Fez+ is practically absent: both C1 heated at 1173 K and C2 heated at 1373 K show an a, parameter larger than that sample, and both contain a greater amount of Fe2+ (fig. 5). It can be reasonably suggested that the cell distortion is caused by the insertion of Fe2+ ions into the FeSbO, lattice.Unfortunately, it was not possible to carry out the same measurement on C2 heated at I173 K with anN. BURRIESCI, F. GARBASSI, M. PETRERA AND G. PETRINI 83 1 acceptable standard deviation ; however, the results reported seem sufficient to support the above conclusion. The mechanism of formation of divalent iron in FeSbO, is likely to be associated with the formation of oxygen vacancies during heating, in analogy with other oxide Quantitative X.P.S. data (fig. 8) show also that the surface suffers a remarkable chemical evolution during heating, resulting in a final surface composition showing strong antimony enrichment. An Sb surface segregation was found in other mixed-oxide catalysts, such as the Sb-Sn-0 ~ y s t e m .~ ~ - ~ ' It is likely that this element segregates at the surface as Sb3+, while its oxidation is the second step of the process.36 In the Sb-Sn-0 system antimony was finally found at the surface as Sb,O,. In contrast, in the Sb-Fe-0 system, our observations and those of Yamazoe et ala8 suggest that Sb is present in the surface layers in its highest oxidation state. From a quantitative point of view, our data and those of Yamazoe et a1.8 agree only partially, even taking into account the difficulty in comparing X.P.S. data obtained in different laboratorie~.~~ We found a strong Sb enrichment for both series of catalysts, not only for Sb:Fe = 2 but with surface Sb:Fe ratios much higher than 2. Thus, our data do not support as a general rule the formation of a thin layer of FeSb,06 on the surface of the catalyst particles.8 We cannot, however, exclude that this compound can form in favourable conditions.that catalyst oxygen is involved in acrolein synthesis, while the formation of carbon oxides by deep oxidation involves weakly bound oxygen. In temperature-programmed desorption experiments carried out by Yamazoe et a1.,8 two types of adsorbed oxygen were found, a and /?, the former dominating in non-selective catalysts and corresponding to weakly bound oxygen. From this point of view, the catalyst selectivity towards acrylonitrile is a function of the type of oxygen accom- modated on the surface, and possibly of the oxygen mobility in the catalyst lattice, which influences the degree of reduction of the surface during the reaction.A correlation has been tentatively proposed between the prevalence of either a or oxygen forms and the surface compound possibly formed on the Sb-Fe catalysts.8 In our case very similar Sb:Fe surface ratios were found for both C1 and C2 catalysts with a strong Sb enrichment. The slightly larger value for C2 could be related to a higher mobility of excess Sb. Therefore, the different catalytic behaviour can hardly be associated directly with the Sb enrichment. On the other hand, the larger amount of a-Fe203 persisting in C1 samples at a given firing temperature cannot explain the different selectivity with respect to C2, as shown by the comparison between C1 at 1123 K and C2 at 1073 K, where both catalysts have similar iron oxide concentrations (fig.5). The activity and selectivity of the second sample (fig. 12) are much better than those of the first (fig. 13), suggesting that a-Fe203 cannot play more than a secondary role in the catalytic behaviour of these samples. We feel that an important parameter differentiating the two catalysts is the partial concentration of oxygen on the surface, which is lower in the case of C2, where a larger amount of divalent iron is formed. It can thus be inferred that under reaction conditions the C2 catalyst can accommodate a larger amount of B-oxygen as a consequence of its 'more reduced' surface. The role of the bulk Sb excess therefore seems to be evidenced mainly during catalyst activation (i.e. promoting the formation of Fe2+ and oxygen vacancies in the FeSbO, structure) and not in the catalytic conditions.The catalysts of the C2 series show little variation with temperature of the surface Sb:Fe ratio, but become selective only after the formation of the FeSb0,-defective structure. The mechanism of the promoting effect of excess antimony oxide should be further investigated. The role of the lattice distortions in determining FeSbO, selectivity was suggested It was pointed832 IR ON-A N T I MON Y CAT A L Y S TS by Kriegsmann et all7 They demonstrated that a medium lattice distortion, induced by the presence in the lattice of a foreign ion having an ionic radius near 0.75 A, increases the production of acrylonitrile towards a maximum value. Our statement on the presence of Fe2+, which has an ionic radius of 0.74 in the FeSbO, structure is perfectly in agreement with the results of Kriegsmann et al.CONCLUSIONS In the ammoxidation of propylene, a high selectivity towards acrylonitrile rather than deep oxidation is obtained using a catalyst with a bulk Sb:Fe ratio 2 2, fired at a high temperature (ca. 1173 K). Under these conditions, the quantitative formation of FeSbO, is achieved, together with a strong Sb surface enrichment. However, neither the presence of FeSbO, nor a high Sb:Fe surface ratio is a sufficient condition for obtaining a high selectivity, since both characteristics were also found in poorly selective catalysts. The larger amount of Fe2+ in C2 samples (Sb: Fe = 2) with respect to C1 (Sb: Fe = 1) could be connected with the better selectivity towards acrylonitrile through the formation of oxygen vacancies, which are necessary to maintain electrical neutrality in the FeSbO, lattice.Such vacancies are possibly associated with the adsorption sites for #?-oxygen, which were indicated to be responsible for the allylic oxidation. As no evidence was found for the formation of any other Fe-Sb compound, the role of bulk excess antimony in the more selective catalysts during activation is in promoting the formation of structurally distorted and defective FeSbO,. We thank Mr A. Gennaro and Mr L. Pozzi for their valuable technical support. I. Aso, S. Furukawa, N. Yamazoe and T. Seiyama, J. Catal., 1980, 64, 29. T. Yoshino, S. Saito and M. Sobukawa, Japanese Patent 71 03.438, 1971.G. Petrini and P. Marinozzi, 6th Znt. Congr. Chem. Eng. (CHZSA), Praha, 1978 (unpublished). J. L. Callaghan, R. K. Grasselli, E. C. Milberger and H. A. Strecker, Ind. Eng. Chem., Prod. Res. Deu., 1970, 6, 134. G. K. Boreskov, S. A. Ven’yaminov,V. A. Dzis’ko, D. V. Tarasova,V. M. Dindoin,N. N. Sazonova, I. P. Olen’kova and L. M. Kefeli, Kinet. Katal., 1969, 10, 1350. F. Sala and F. Trifiro, J. Catal., 1976, 41, 1. N. Yamazoe, I. Aso, T. Anamoto and T. Seiyama, Proc. 7th Znt. Congr. Catal., Tokio, 1980 (Elsevier, Amsterdam, 1981), part B, p. 1239. C. A. Clausen I11 and M. L. Good, in Characterization ofMetal and Polymer Surfaces, ed. L-H. Lee (Academic Press, New York, 1977), vol. I, p. 65. F. Garbassi, Surface and Interface Analysis, 1980, 2, 165. ‘I V. Fattore, Z.A. Fuhrmann, G. Manara and B. Notari, J. Catal., 1975, 37, 223. lo G. E. Theriault, T. L. Barry and M. J. B. Thomas, Anal. Chem., 1975, 47, 1492. I2 J. H. Scofield, J. Electron Spectrosc. Relat. Phenom., 1976, 8, 129. l3 Handbook of X-ray Photoelectron Spectroscopy, ed. C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder and G. E. Muilenberg (Perkin-Elmer Corp., Physical Electronics Div., Eden Prairie, 1979). l4 K. Brandt, Ark. Kemi, Mineral. Geol., 1943, A17, 15. l5 J. Korinth and P. Royen, 2. Anorg. Allg. Chem., 1965, 340, 146. l8 G. M. Bartenev, R. R. Zakirov and A. D. Tsyganov, Sou. Phys. Solid State, 1975, 16, 2409. l7 H. Kriegsmann, G. Ohlmann, J. Scheve and F. J. Ulrich, Proc. 6th Int. Congr. Catal. (The Chemical Society, London, 1976), paper B23. M. Petrera, A. Gennaro, N. Burriesci and J. C. J. Bart, 2. Anorg. Allg. Chem., 1981, 472, 179. lB N. S. McIntyre and D. G. Zetaruk, Anal. Chem., 1977, 49, 1521. 2o G. T. Allen, M. T. Curtis, A. J. Hooper and P. M. Tucker, J. Chem. SOC., Dalton Trans., 1973, 1675. 21 T. Birchall, J. A. Connor and I. H. Hillier, J. Chem. Soc., Dalton Trans., 1975, 2003. 22 T. A. Carlson, Photoelectron and Auger Spectroscopy (Plenum Press, New York, 1975). 23 A. F. Orchard and G. Thornton, J. Chem. SOC., Dalton Trans., 1976, 13, 1238. 24 G. C. Allen and P. H. Tucker, Znorg. Chim. Acta, 1976, 16, 41.N . BURRIESCI, F. GARBASSI, M. PETRERA A N D G. PETRINI 833 25 K. Yabe, K. Arata and I. Toyoshima, J. Catal., 1979, 57, 231. 26 D. V. Tarasova, V. A. Dzis’ko, I. P. Olen’kova, L. T. Tsikoza, L. G. Karakchiev and V. V. Molodt- 27 V. V. Malakhov and F. G. Abdikova, Kinet. Katal., 1972, 13, 196. 28 V. F. Anufrienko, N. G. Yarmukhametov, D. V. Tarasova and I. P. Olen’kova, Kinet. Katal., 1973, 29 U.’Gakiel and M. Malamud, Am. Mineral., 1969, 54, 299. 30 G. A. Petitt and M. R. Meder, Hyperfine Interactions, 1978, 5, 323. 31 M. Eibschutz and U. Ganiel, Solid State Commun., 1968, 6, 775. 32 F. Varret, P. Imbert, A. Gerard and F. Hartmann-Boutron, Solid State Commun., 1968, 6, 889. 33 T. R. Sandin, D. Schroeer and C. D. Spencer, Phys. Reu. B, 1976, 13, 4784. 34 W. A. Weyl and T. Forland, Ind. Eng. Chem., 1950, 57, 257. 35 Y. Boudeville, F. Figueras, M. Forissier, J-L. Portefaix and J. C. Vedrine, J . Catal., 1979, 58, 52. 36 Y. M. Cross and D. R. Pyke, J. Catal., 1979, 58, 61. 37 H. J. Herniman, D. R. Pyke and R. Reid, J. Catal., 1979, 58, 63. 38 T. E. Madey, C. D. Wagner and A. Joshi, J. Electron Spectrosc. Relat. Phenom., 1977, 10, 359. 39 A. I. Gel’bshtein, S. S. Stroeva, A. N. Parfenov and T. V. Zinov’eva, Kinet. Katal., 1976, 17, 412. sova, Kinet. Katal., 1973, 14, 481. 14, 716. Handbook of Chemistry and Physics, ed. C. R. Weast (C.R.C. Press, Cleveland, 51st edn, 1970), p. F-152. (PAPER 1 /546)
ISSN:0300-9599
DOI:10.1039/F19827800817
出版商:RSC
年代:1982
数据来源: RSC
|
16. |
Effect of Ce3+ions exchanged in NiX zeolites on the location and reducibility of Ni2+ions and on the stabilization of a highly dispersed metallic nickel |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 835-843
Samir Djemel,
Preview
|
PDF (731KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I, 1982, 78, 835-843 Effect of Ce3+ Ions Exchanged in Nix Zeolites on the Location and Reducibility of Ni2+ Ions and on the Stabilization of a Highly Dispersed Metallic Nickel BY SAMIR DJEMEL, MARIE-FRANCE GUILLEUX,* JANINE JEANJEAN, JEAN F. TEMPERE AND DENISE DELAFOSSE E.R. 133 Reactivite de Surface et Structure, Laboratoire de Chimie des Solides, UniversitC P. et M. Curie, 4 Place Jussieu, 75230 Paris Cedex 05, France Received 13th April, 1981 Highly dispersed nickel particles, with diameters ranging from 0.7 to 3 nm, can be obtained by hydrogen reduction of NiZ+ ions exchanged in X zeolites, in the presence of Ce3+ ions. Samples with various Ce3+/Ni2+ ratios have been studied. The structure and surface properties of the solids were investigated by X-ray diffraction and e.p.r.The particle size was determined by magnetic methods. It is shown that the Ce3+/Ni2+ ratio influences the initial location of the Ni2+ ions in the different cationic sites and the redox properties of the zeolite. The effect of Ce3+ ions on the obtention of highly dispersed metallic nickel is discussed in terms of strong electrostatic field and electron-donor properties associated with Ce3+ ions. During the last ten years, there have been many investigations of the properties of very small particles of various supported Unusual catalytic behaviour of highly dispersed metals has been reported and ascribed to metal-support interactions which can greatly modify both the activity and the selectivity of metal catalysts. In order to obtain metallic particles of homogeneous size, zeolite carriers were chosen because of their crystallographic structure and cation-exchange properties, the metal ions being uniformly dispersed within the framework and strongly held to the solid.Further reduction can lead to the growth of metallic particles inside the rigid cavities of the zeolite. Owing to their importance in catalysis, Pt and Pd in Y zeolite have been studied extensively.' Furthermore, complete reduction of these ions is easy (at relatively low temperatures) and generally produces a highly dispersed metallic state inside the zeolite structure. More recently, nickel has also received a lot of a t t e n t i ~ n . ~ - ~ Nevertheless, this element is difficult to reduce and it was observed that metallic particles formed by molecular hydrogen reduction migrate towards the external surface of the zeolite and agglomerate into large crystallites.Attempts have been made to prevent the sintering of nickel particles towards the external surface and to obtain a stabilized state of nickel particles inside the framework. Numerous studies have shown that one of the main factors determining the metallic dispersion is the strength of interaction between the metal and the support.'O These interactions may arise from electron transfer of metallic particles towards the electron- acceptor sites of the support.' In other respects it has been observed5 that the presence of a 'modifying' element which can interact strongly with the support and also with 835836 EFFECT OF Ce3+ IONS I N NIX ZEOLITES the metal atomic species or particles decreases their mobility and as a result favours a high degree of dispersion.In studies involving Pt-silica systems, Yermakov and coworkersll have shown that low-valence ions of transition elements (Mo, W, Re) tend to prevent the sintering of the supported metal when chemically bonded to the support surface. On zeolitic carriers, it has been observed that introduction of Cr3+ or Ce3+ in various nickel zeolites could inhibit the diffusion of small nickel particles to the outside surface and thereby favours a high degree of dispersion.8y12 Thus, in the case of NiCeX zeolites, it has been shown that it is possible to obtain a highly dispersed state of metallic nickel with particle size < 3 nm.Nevertheless, the nature of the Ce3+ stabilizing effect has not been elucidated. The present study reports on the results of the influence of the Ce3+ ions exchanged in Nix zeolites on the cation location, the redox properties of the system, the reducibility of Ni2+ ions and the metallic dispersion obtained. EXPERIMENTAL Samples were prepared by conventional exchange of a sodium X zeolite (samples I1 and 111) or a partially decationated X zeolite (sample I) with 0.1 mol dm-3 solutions of Ni2+ and Ce3+ nitrates, successively. Sample composition and nomenclature are given in table 1. TABLE 1 .-SAMPLE COMPOSITION AND NOMENCLATURE chemical composition" unit cell sample composition Ni2+ (wt. %) Ce3+ (wt. %) Na+ (wt. %) I Ni,Ce,Na,,H,,X 3.65 5.9 3 I1 Ni,Ce5Na,,H4X 4.11 5.78 8.75 I11 Ni21Ce6.5Na20H4X 9.23 5.75 3.8 a Calculated on dry weight basis.Before reduction, structural and surface characterizations were performed on samples dehydrated at a pressure of Torr for 16 h at 623 or 773 K. Reduction, whatever the pretreatment of the samples, was carried out at 623 K in a flow (5 dm3 h-l) of high purity hydrogen for 16 h. For the sake of simplification, samples will be referred to as D623-R623 and D773-R623, according to the temperatures at which desorption (623 or 773 K) and reduction (623 K) took place. When X-ray determinations were required, the samples were treated in a Pyrex vessel connected to a Lindemann glass capillary (0.2 mm diameter); after thermal treatment, the zeolite was transferred under vacuum into the Lindemann capillary which was then sealed off for the X-ray investigations.X-ray powder diffraction patterns were taken at room temperature using a large diameter Debye-Scherrer camera and a narrow collimated X-ray beam (CuK, radiation) to assure optimum resolution of lines. A multiple film technique was used to collect line intensities, which were photometered and then planimetered. The redox surface properties of the different samples were investigated by e.p.r., by means of charge-transfer complexes, using perylene as an electron-donating molecule and tetracyano- ethylene (TCNE) as an electron-accepting molecule. Solutions of perylene and TCNE were prepared in benzene, free of water, at concentrations of 5 x lop3 mol dm-3. The samples (36 mg), after the different pretreatments adopted, were contacted, in situ, with the appropriate solution.Experiments showed that spin equilibrium values were attained in 48 h. The e.p.r. spectra were recorded on a Varian spectrometer (model C.S.E.-109, X band). This spectrometer is connectedDJEMEL, GUILLEUX, JEANJEAN, TEMPERE A N D DELAFOSSE 837 to an E. 900 e.p.r. data acquisition system fitted for double integration of the experimental curves. Spin concentration calculations have been evaluated in comparison with Varian’s standard spectra registered under the same operating conditions. The error in the determination of radical ion concentration is less than & 10%. Magnetic methods were used to determine both the degree of reduction of ferromagnetic samples and the particle size distribution.l3 The magnetization measurements were performed by the Weiss extraction method,14 using either an electromagnet, in the temperature range from 77 to 300 K and fields up to 21 kOe, or a superconductive coil reaching 70 kOe (measurements at 4.2 K). The metallic nickel content was calculated from the saturation magnetization value, us. For the samples studied, the experimental determination of us is difficult to obtain, these samples exhibiting superparamagnetic or paramagnetic behaviour. After the magnetic measurements, the samples were sintered at 1073 K for 16 h under flowing helium. The larger ferromagnetic crystallites thus formed are easily saturated and the us value was measured precisely. The crystallite size distribution was determined from the magnetization field curves 0 =AH).The experimental 0 =AH) curves were corrected for the diamagnetism of the zeolitic support as well as for the paramagnetism of the Ce3+ ions and of the residual Ni2+ ions. When the reduced samples had paramagnetic behaviour at 77 K, the magnetization measurements were performed in the superconductive coil at 4.2 K. RESULTS STRUCTURAL CHARACTERIZATION The structural characterization of sample I, I1 and I11 dehydrated at 773 or 623 K have been determined from powder data as previously described.15 Populations and coordinates of extraframework atoms are given in table 2. The complete list of crystallographic parameters including framework atoms and the listing of observed and calculated structure factors can be supplied by the authors if requested.Note that structure analysis gives only the electron density and so identification of extraframework atoms is often very difficult, especially when many ions share the same type of site. In samples I and 111, site attribution is relatively easy because the coordinates of Ce3+ and Ni2+ ions at SI, sites are different. As we have previously shown,lG the Ce3+ ions not located at SI/ sites are found at SI sites. On the other hand, in sample 11, because of the large occupancy factor of Na+ ions at SI/ sites, populations of Ce3+ and Ni2+ cations located at SI/ sites cannot be detected. However, taking into account the rule according to which whenever a SI site is occupied, the two adjacent SI. sites are unoccupied [NI/ < 2(16-N,)] we can assume that Ce3+ and Ni2+ cations share the SI sites in the proportions given in table 2; Ce3+ not located at SI sites occupy SI/ sites.We must specify that, for almost all the samples, we have indicated in the table: (1) the extraframework atom population which corresponds to the electron density given by structure analysis and (2) the population of cations which effectively occupy a given type of site when many ions share the same type of site [e.g. for sample I11 dehydrated at 623 K we have reported the cation population at SI sites (14.7 Ni) and we have also indicated the concentration of cations which actually occupy this site (1 Ce+ 12.6 Ni)]. INFLUENCE OF PRETREATMENT A N D PROTON CONCENTRATION ON CATION LOCATION The temperature of pretreatment determines the amount of residual water molecules and therefore the cation positions.In sample I11 dehydrated at 623 K, we have located oxygen Ow in an extraframework position; this species is attributed to water moleculesTABLE 2.4ATION DISTRIBUTION sample 111, Ni,,Ce,.,Na,,H,X: dehydrated sample I, sample 11, Ni,Ce,Na,,H,,X : Ni,Ce,Na,,H,X : dehydrated dehydrated 773 K 623 K 773 K 623 K cation site 623 K 773 K S I 7.4(2)" Ce i.e. 5.0(2) Ce+5.3(4) Ni (0.0)b 13.4(4) Ni i.e. 2.7(5) Ce+7(1) Ni 2( 1)Ni (0.0) 12.6(5) Ni i.e. 3( 1) Ce+ 6(2) Ni 24(2) Na i.e. Ni 2(1) ce+{ + Na (0 .O 5 6) (0.0) 15.1(5) Ni 14.7(5) Ni i.e. i.e. 3.8(5) Ce+7(1) Ni 1.0(3) Ce+ 12.6(9) Ni (0.0) (0.0) 8(1) Ni - 10.5(5) Ni i.e. 2( 1) Ce + 6(2) Ni (0.0) 32(2) Na i.e. Ni (0.056) (0.043) 2.3(5) Ce (0.045) - 2.7(5) Ce 5 3 3 ) Ce (0.063) (0.068) - (0.066) lO(2) o w (0.157) 4.8(7) Ni - (0.202) 31(1) Na 28(1) Na i.e.i.e. (0.232) (0.23 3) 24.67 24.74 0.102 0.1 10 19(1) Na+4.8(8) Ni 19(1) Na+3.6(8) Ni - 23(1) Na i.e. 17( 1) Na + 2.4(9) Ni (0.230) 24.69 0.1 16 - 22(1) Na i.e. 17(1) Na+2(1) Ni (0.238) 24.75 0.096 S I I 29(2) Na or Ni + Na (0.232) 24.79 0.119 33(2) Na or Ni + Na (0.232) 24.82 0.106 ac Rd a Standard error of the last significant figure. Atomic coordinates. Cell parameter in 8, (kO.01 A). R = E~~l$~-~l$ll/E~l$~.DJEMEL, GUILLEUX, JEANJEAN, TEMPERE AND DELAFOSSE 839 or hydroxyl groups. In this sample, Ce3+ located at SI. sites are coordinated with three framework 0, oxygensl' and one to three water molecules occupying SII, sites. In the other samples, we have not located residual water molecules.However, according to Dempsey and multivalent cations (such as Ce3+ and Ni2+) located at S,! sites are probably bonded to extraframework oxygen which form with three framework oxygen 0, an octahedral or a tetrahedral coordination; these framework oxygens cannot be detected because their too low occupancy factor. In these conditions we can assume that only sample I at 773 K is completely dehydrated (in this sample, scattering matter was not located at SI~ sites). In conclusion, dehydration increases with temperature but also with proton concentration (sample I). INFLUENCE OF Ce3+ IONS ON Ni2+ ION LOCATION The Ce3+ concentration at SI sites increases with the temperature of dehydration. As we have seen, for partial dehydration Ce3+ are essentially located at ST/ sites.In the absence of water, Ce3+ cations similar to the Ni2+ ions are found in an octahedral coordination at SI sites. Therefore during the dehydration process there is competition between Ce3+ and Ni2+ ions to occupy the SI sites. After dehydration, S, sites are preferentially occupied by Ce3+ ions leaving Ni2+ ions essentially out of the hexagonal prisms. The parameters on which we have successively commented, temperature of dehydration and proton concentration on the one hand and Ce3+ ion location on the other, are closely related. Thus the presence or absence of water determines the Ce3+ ion location and therefore that of the Ni2+ ions. OXIDIZING AND REDUCING PROPERTIES The oxidizing and reducing properties of the samples are measured by the number of perylene radical cations and TCNE radical anions formed per 1 g of zeolite dehydrated at 773 K in the case of perylene and at 623 or 773 K in the case of TCNE, respectively.The results are summarized in table 3. In this table are also reported the Ce3+/Ni2+ ratio of the various samples and the calculated number of Lewis centres generated during thermal pretreatments. This number was estimated from the dehydroxylation scheme proposed by i.r. ~ t u d i e s . ~ ~ ~ ~ ~ In column a, we have recorded: first, the number (L,) of Lewis sites formed by dehydroxylation of OH groups present initially at the sample surface according to the scheme: H (1) O\ /O\-/O 2 - + ,Si, ,A( I 0 00 0 0 0 and, secondly, the number (L,) of Lewis sites obtained by dehydroxylation of acidic OH groups created by hydrolysis of water molecules by nickel cations, according to : H Ni2+ (OH)- I Ni2+ ( H 2 0 ) I (11) \ P-./O 0 0 0 0 + ,SiyAlb The L, values are to be found in column b. The contribution of Ce3+ cations to Lewis site formation has not been taken into account for the samples studied owing to their similar Ce3+ contents.840 EFFECT OF Ce3+ IONS IN NIX ZEOLITES Note that the concentration of perylene radical cations depends on the number of Lewis sites in the samples whereas the TCNE radical anion concentration is related to the Ce3+/Ni2+ ratio. Generally, it is assumed that the electron-donating and electron-accepting centres of the zeolite are generated by dehydroxylation of OH groups present or created at the zeolitic surface, according to scheme (I). In most cases,21y 22 the electron-accepting centres are ascribed to the three coordinated aluminiums and the electron-donating sites to the (A10,)- entities. Concerning the oxidizing properties of the samples studied, it appears (table 3) that the number of perylene radical cations formed is increased (by a factor 100) for the initially highly protonated sample I compared with TABLE 3 .-OXIDISING AND REDUCING PROPERTIES OF SAMPLES radical ion concentration number of Lewis spins g-' sites/lO-,O g-l TCNE- Ce3+/NiZ+ column a column b Per+ sample composition ratio (L,+L,) (L,) ,, 773 Ka 773 K a 623 Ka I Ni,Ce,Na,,H,,X 0.62 10.7 8.83 23 5.3 8.5 I1 Ni,Ce,Na,,H,X 0.55 2.79 0.93 0.3 5.2 7 I11 Ni,,Ce,.,Na,,H,X 0.31 5.81 0.93 0.6 1.4 4.4 Ni,,Xb Ni,,Na,,X 0 7.67 0 0.49 0.28 - a Dehydration temperature.Ni,,Na,,X sample without Ce3+ was studied for the purpose of comparison. the other samples. This variation has to be correlated to the variation of L, Lewis sites concentration. Thus the acidic strength of the L, sites will probably be stronger than that of the Lewis sites due to the nickel cations. On the other hand, it has been previously noted that the reducing properties of the samples containing Ce3+ depend on the Ce3+/Ni2+ ratio and do not vary directly with the number of Lewis sites created. These properties are the more important for higher Ce3+/Ni2+ ratios and, for a given sample, for lower dehydration temperatures (623 K). Note that sample Ni3,X without Ce3+ cations, used as a comparison, shows rather poor reducing activity.It is to be expected that the polarizing power of the different cations exchanged in zeolites and their electron affinity will modify the reducing power of the neighbouring electron-donating So, the presence of Ce3+ ions, strongly electron-donating and with a high polarizing power, would be able to facilitate electron transfer from the reducing centres of the zeolite towards the electron-accepting molecule. On the other hand, the presence of electron-accepting Ni2+ cations will lead to a decrease in the electron-donating power of these centres. Therefore, we can expect that both the Ce3+/Ni2+ ratio and the Ce3+ ion concentration determine the reducing properties of the samples studied. Moreover, it has been observed that, for a partial dehydration of the support (623 K), the reducing properties are more important.For this dehydration state, first, the number of electron-acceptor sites, that is to say the number of Lewis sites, is lower compared with a total dehydration (773 K) and secondly, the Ni2+ ions are to be found, for the most part, as Ni2+ (OH)- entities which would exhibit smaller electron-acceptor properties. As a result, the oxidizing properties of the support are decreased. Consequently, the role of the electron-dorlor Ce3+ ion in the formation of TCNE radical anions will be enhanced.DJEMEL, GUILLEUX, JEANJEAN, TEMPERE AND DELAFOSSE 84 1 DEGREE OF REDUCTION AND CRYSTALLITE SIZE DISTRIBUTION The degree of reduction and crystallite size distribution obtained for the various samples studied are summarized in table 4.We can see that, except for sample I, the reduced samples are heterodispersed, but the particle size never exceeds 3 nm in diameter. Sample I leads to a homodispersed metallic state with particles 0.7 nm in diameter. It appears that the degree of reduction depends on several factors, closely related to each other, such as: (a) the pretreatment conditions which determine the cation location; (b) the redox properties of the support; (c) the extent of Na+ exchange. TABLE 4.-MAGNETIC DETERMINATION OF THE DEGREE OF REDUCTION AND CRYSTALLITE SIZE DISTRIBUTION OF SAMPLES reduction conditionsa D623-R623 D773-R623 sample composition a D/nm a D/nm I Ni,Ce,Na,,H,,X 0.20 100% 0.7 0.17 100% 0.7 I1 I11 Ni,Ce,Na,,H,X 77% 2.5 0.38 79% 2.5 0*46 23% 1.5 21% 1.5 13% 3 60% 2.5 87% 0.7 0'45 40% 1.5 a D623-R623 and D773-R623 : Dynamical reduction at 623 K after desorption at 623 and 773 K, respectively; a: degree of reduction; D: crystallite diameter.We now discuss these different points. In table 4 it is observed that the samples are less reduced after a pretreatment at 773 K than after a pretreatment at 623 K. We could invoke a different location of Ni2+ ions before reduction, which, during pretreatments at increasing temperatures, generally migrate towards less accessible sites. Nevertheless, X-ray results show that, whatever the pretreatment temperature, a non-negligible percentage of Ni2+ cations are found in accessible sites. Moreover, in an X-ray concerning the variation of Ni2+ site occupation during reduction, a simultaneous depopulation of the different crystallographic sites involved was observed.On the other hand, e.p.r. measurements have shown that the reducing properties are decreased for a total dehydration (773 K) compared with a partial dehydration (623 K) when the oxidizing properties are increased. From this, it follows that the pretreatment conditions strongly control the redox properties of the system and, as a result, the redox equilibrium of Ni2+ ions according to: Ni2+ + 2e + Nio so the displacement of this equilibrium to the right and, consequently, the extent of reduction may be favoured by a partly dehydrating pretreatment. For a given pretreatment, it appears that sample I, which exhibits the strongest oxidizing properties, is the less reduced.As for the two other samples, sample 11, which has the more important reducing properties, is still less reduced than sample 111. Hence, the redox properties of the system are not the only factors which determine the degree of reduction. Another factor has to be considered, which is related to the 28 FAR 1842 EFFECT OF ce3+ IONS IN NIX ZEOLITES Na+ ion concentration. So, sample I1 exhibits the highest cation concentration, consisting mainly of Na+ ions. It follows from this that, whatever the pretreatment temperature (623 or 773 K), all the S,, sites near the hexagonal windows are occupied by Na+ ions (table 2). It may be assumed that the presence of these ions in such a location can reduce the ease of migration of Ni2+ ions through the hexagonal windows towards the supercage and, so, the extent of reduction.For this sample the metallic distribution obtained is inhomogeneous. This point will be discussed later. However, the particle size does not exceed 3 nm in diameter. This emphasizes the role played by Ce3+ cations upon the metallic dispersion state obtained. This influence may be discussed in terms of the electron-donor properties of these ions and of the strong electrostatic field associated with the trivalent cations. DISCUSSION As previously mentioned, it is well known that one of the main factors determining the metallic dispersion is the strength of interactions between the metallic particles and the support. In the case of zeolitic carriers, the electron-deficient character of small metallic particles was shown, by X.P.S., for several PtY sample^.^ This was explained by the withdrawal ofelectrons from the Pt particles (1 nm in diameter) by the electron-acceptor sites of the support.These interactions of metal atoms towards the support may hinder the migration of the metallic particles. The presence of trivalent cations such as Ce3+ in these samples increases the electrophilic character of the platinum particles and favours the formation of smaller particles (0.8 nm in diameter). first, Ni2+ ions are difficult to reduce and the reduction process requires high temperatures (2 573 K) where the diffusion rate of the atomic species becomes non-negligible, and secondly, in contrast to platinum, the metallic nickel, owing to the value of the redox potential of nickel (E, = -0.25), is easily oxidizable in the presence of oxidizing centres of the zeolitic support such as Brarnsted or Lewis sites.Thus, in general, the reduction of Ni2+ ions into Nio is difficult and leads to a very heterogeneous metallic state, the particle size distribution ranging from ca. 0.7 nm to > 25 nm. The presence of Ce3+ ions, due to their electron-donor power and the value of the redox potential (E, = - 1.7), could stabilize the nickel atomic species or particles in limiting the nickel oxidation by electron-acceptor sites of the support. This stabilizing effect is not observed with La3+ exchanged Ni2+X zeolite: La3+ does not possess electron-donor properties. With this cation, nickel reduction is very difficult, requires high temperatures and leads to a poorly dispersed metallic In other respects the presence of Ce3+ ions, highly charged and bonded to the support, can lead to some local perturbations, such as an energetic well, generating preferential sites for the growth of the metal and inhibiting the atomic species migration by an interaction of Ce3+ with NiO.The larger the Ce3+/Ni2+ ratio, the larger would be this effect. Thus, for sample I, with the highest Ce3+/.Ni2+ ratio, the particle size obtained is 0.7 nm in a homogenous distribution. For sample 11, the Ce3+/Ni2+ ratio is about the same, but, in this case, the very high Na+ ion number, located in the supercages, could act as a screen for the Ce3+-Nio interactions and so increase the atomic species migration.Finally, sample 111, for which the Ce3+/Ni2+ ratio is ca. 0.3, leads to a heterogeneous particle size distribution. This result may be correlated with the higher Ni2+ ion concentration with regard to the Ce3+ ion concentration. For nickel zeolites, experimental results have shownDJEMEL, GUILLEUX, JEANJEAN, TEMPERE A N D DELAFOSSE 843 In conclusion, this study has shown the effect of Ce3+ ions in the obtention of a highly dispersed metallic state of nickel in different NiCeX zeolites. The main factor involved in this stabilization phenomenon is related to the Ce3+/Ni2+ ratio which greatly influences the initial location of the Ni2+ ions in the different cation sites and the redox properties of the system. Moreover, the presence of Ce3+ ions with a high polarizing field can act as anchorage points, preventing the migration of nickel particles towards the external surface of the zeolite.P. Gallezot, Catal. Rev. Sci. Eng., 1979, 20, 121. K. H. Minachev and Ya. I. Isakov, in Zeolite Chemistry and Cata1;sis (American Chemical Society, Washington D.C., 1976). Tran Manh Tri, J. Massardier, P. Gallezot and B. Imelik, Preprints 7th Znt. Congr. Catal., A 16, Tokyo, 1980. V. N. Romannikov, K. G. Ione, L. A. Pederson, J. Catal., 1980, 66, 121. Yu. I. Yermakov, Preprints 7th Int. Congr. Catal. P5, Tokyo, 1980. W. Romanowski, Rocz. Chem., 1971, 45,427. M. F. Guilleux, D. Delafosse, G. A. Martin and J. A. Dalmon, J. Chem. SOC., Faraday Trans, 1,1979, 75, 165. M. Briend-Faure, M. F. Guilleux, J. Jeanjean, D. Delafosse, G. Djega-Mariadassou and M. Bureau-Tardy, Acta Phys. Chem., 1978, 24, 99. M. S. Ioffe, B. N. Kumetsov, Yu. A. Ryndin and Yu. I. Yermakov, Proc. 6th Znt. Congr. Catal. (The Chemical Society, London, 1976), vol. 1, p. 131. l 2 N. P. Davidova, N. V. Peshev, M. L. Valcheva and D. M. Shopov, Acta Phys. Chem., 1978,24, 113. l 3 P. W. Selwood, Chemisorption and Magnetization (Academic Press, New York, 1975). l4 P. Weiss and R. Forrer, Ann. Phys. (Paris), 1926, 5, 153. l5 P. Gallezot, Y. Ben Taarit and B. Imelik, J. Phys. Chem., 1973, 77, 652. l6 J. Jeanjean, S. Djemel, M. F. Guilleux and D. Delafosse, J. Phys. Chem., in press. l7 J. V. Smith, Adv. Chem. Ser., 1971, 101, 188. l9 J. B. Uytterhoeven, L. G. Christner and W. K. Hall, J. Phys. Chem., 1965, 69, 21 17. *O J. W. Ward, in Zeolite Chemistry and Catalysis (American Chemical Society, Washington, D.C., 1976). 21 I. Turkevich and Y. Ono, Adv. Catal., 1969, 20, 135. 22 M. I. Loktev and A. A. Slinkin, Russ. Chem. Rev., 1976, 45, 9. 23 J. T. Richardson, J. Catal., 1967, 9, 172. 24 M. Briend-Faure, J. Jeanjean, M. Kermarec and D. Delafosse, J. Chem. Soc., Faraday Trans. I, 1978, 25 M. Briend-Faure and G. Spector, unpublished results. 'I K. G. Ione, V. N. Romannikov, A. A. Davidov and L. B. Orlova, J. Catal., 1979, 57, 126. lo E. Ruckenstein and B. Pulvermacher, J. Catal., 1973, 29, 224. E. Dempsey and D. H. Olson, J. Phys. Chem., 1970,74, 305. 74, 1538. (PAPER 1/594) 28-2
ISSN:0300-9599
DOI:10.1039/F19827800835
出版商:RSC
年代:1982
数据来源: RSC
|
17. |
Mechanism of the catalytic oxidation of hydrogen on copper |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 845-854
Kazunari Domen,
Preview
|
PDF (691KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I, 1982, 78, 845-854 Mechanism of the Catalytic Oxidation of Hydrogen on Copper BY KAZUNARI DOMEN, SHUICHI NAITO, MITSUYUKI SOMA,? TAKAHARU ONISHI* AND KENZI TAMARU Department of Chemistry, Faculty of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 11 3, Japan Received 13th April, 198 1 The mechanism of water formation from H, and 0, on copper at 393-423 K was studied using a closed circulating system and X-ray photoelectron spectroscopy. Under the whole pressure range of 0, studied the surface was covered with a Cu,O layer during the reaction. Where the partial pressure of 0, was higher than ca. 9 x lo3 Pa, the reaction rate was expressed as follows: r = k[H,]'[O,]O and H,-D, exchange did not proceed during the reaction. On the other hand, where the partial pressure of 0, was lower than ca.9 x lo3 Pa, the rate of water formation showed a maximum value at 1.3 x lo3 Pa and the H,-D, exchange reaction proceeded gradually. From these results, a possible mechanism of the reaction is proposed. The catalytic oxidation of hydrogen on various metals, such as silver, gold, platinum and palladium,l-' has been extensively investigated because of its apparent simplicity and technical interest. In comparison with these metals, the reaction on copper has not been equally well clarified because of the complexity of the reaction. As was first demonstrated by Bone and Wheelar,* the formation of water from hydrogen and oxygen on reduced copper is always accompanied by the simultaneous oxidation of the copper surface. Pease and Taylor9 and Larson and Smithlo concluded that the oxide thus formed accompanying water formation underwent alternate reduction and oxidation. Under their experimental conditions, however, the reaction seems to be in a transient state and the concentration of oxygen in the gas phase was low.On the other hand, Wilkins and Bastowll attempted to correlate the velocity of oxidation and reduction with the rate of water formation, and suggested that some direct catalytic combination between hydrogen and oxygen was taking place in addition to the cyclic oxidation-reduction process. Cleave and RideaP also observed that the catalytic combination occurred on the oxide, but the detailed reaction mechanism and the nature of adsorbed' species have not yet been clarified. In this study a closed circulating system was used to measure the amounts as well as the reactivity of adsorbed hydrogen and oxygen during the reaction over a wide range of partial pressures of oxygen and hydrogen.This method makes it possible to estimate the extent of oxidation of the catalyst in the course of the water-formation reaction and the reactivity of surface oxygen under the working conditions. By studying the dependences of the rate of water formation upon the amounts of the adsorbed species, as well as the partial pressures of ambient gases, it was possible to elucidate the reaction mechanism of water formation on copper. The H,-D, exchange reaction in the course of water formation was investigated to obtain information about 7' Present address: National Institute for Environmental Studies, Yatabe, Tsukuba, Ibaraki 305, Japan. 845846 CATALYTIC OXIDATION OF H, ON Cu hydrogen activation.X-ray photoelectron spectroscopy (X.P.S.) was also applied to study the state of the catalyst surface during the reaction. On the basis of the results obtained, the mechanism of the catalytic H,-0, reaction over copper is examined. EXPERIMENTAL The copper catalyst employed was prepared by the reduction of Cu(OH), with hydrogen at 423-523 K in a conventional closed-circulation system (ca. 450 cm3). Cu(OH), was precipitated from an aqueous solution of Cu(NO,), (99.9% purity) by NaOH solution, washed thoroughly and dried at 373 K. The copper reduced at lower temperatures (393-403 K) exhibited very high activity for oxidation, probably owing to the higher surface area, and the reaction took place so rapidly that it was impossible to follow the reaction rate exactly in our system.Hence, only the samples reduced at higher temperatures (423-523 K) were studied, where the water-formation reaction occurred at a moderate rate. The amount of catalyst was ca. 5 g, and its surface area was determined as 7 m2 g-l by the B.E.T. method using N, adsorption. By X.P.S., small amounts of C and C1, but no Na, were detected on the surface of the reduced copper. Hydrogen was purified by passing through an Engelhard De-oxo unit and a dry-ice-methanol trap to remove any H,O formed. Oxygen was purified by passing through a dry-ice-methanol trap. The amounts of H, and 0, in the gas phase were analysed by gas chromatography. All experiments were carried out in a closed circulating system, and the H,O formed by the reaction was collected by a dry-ice-methanol trap located just after the catalyst and its weight measured gravimetrically at appropriate intervals in the course of the reaction.To investigate the surface state during the reaction, X.P.S. was applied, using a McPherson ESCA 36 spectrometer. The X.P.S. samples were prepared in a closed circulating apparatus and transferred without contact with air to a dry box, which was purged by nitrogen, attached to the sample chamber of the spectrometer and then mounted on the sample holder of the apparatus. RESULTS H,-0, REACTION ON REDUCED AND OXIDIZED COPPER Before the H,-O, reaction the catalyst was pretreated with the mixture of H, and 0, gas at 423 K until a constant reaction rate was obtained, followed by the reduction with H, at 423 K for 12 h, by which time the consumption of H, had finished completely, and evacuation at the same temperature for 2 h.Then the mixture of H, and 0, (each 1.6 x lo4 Pa) was introduced onto the catalyst at 423 K as shown in fig. 1, which indicates clearly that H,O formation and oxidation of copper occurred at the same time. The amounts of H,O, O,(a) and H,(a) are also represented in fig. 1. Even after the formation of more than ten layers of surface oxide (the surface area was estimated by the B.E.T. method), when metallic copper no longer seems to exist on the surface, the uptake of oxide layer still continued. In comparison with the amount of surface oxygen, that of adsorbed hydrogen was small and its surface coverage (H atom/Cu atom) was < 10%.The H,-O, reaction was also studied on the oxidized copper surface at 393 K, as shown in fig. 2. The oxidized copper was formed by the reaction of H, and 0, and its thickness was ca. 10 layers of C U , ~ . In carrying out the reaction, the system was evacuated for 1 h, then H, alone was admitted onto the catalyst at the same temperature. Immediately after the admittance of H,, the rate of H,O formation or of decrease of H, admitted was equal to that before the evacuation, and the rate of reduction increased gradually. This suggests that the rate of H,O formation was retarded by adsorbed oxygen, which is not in adsorption equilibrium with gaseous 0,.As will be shown later, during the H,-0, reaction, the surface is covered with Cu',K. DOMEN, S. NAITO, M. SOMA, T. ONISHI AND K. TAMARU 847 2.0- 2.0 1 . 5 a" 2 * 1 1 .( 4 0 .! - 8.0 evacuation - (H2 + 0 2 ) (H* 1 8.0 6.0 n c: a W m 4.0 --- - 8 $ 2.0 0 0 100 200 tlmin FIG. 1.-H24, reaction on reduced copper at 423 K: 0, H,(g); A, O,(g); 0 , H,O; 0, H,(a); A, O,(a). FIG. 2.-Effect of O,(g) on the rate of H,O formation. After H , 4 , ( 1 : 1) reaction on the oxidized copper surface ( C U , ~ ) at 393 K for 180 min, the gas phase was evacuated for 1 h, then only H,(g) was reintroduced at 393 K. 0, H,(g); 0, O,(g); 0 , amount of H,O formed.848 CATALYTIC OXIDATION OF H, ON Cu and the increase in the rate of H,O formation corresponds to the appearance of Cu metal.To study the effect of the coexistence of H, gas on the rate of oxidation of copper, 0, alone was first introduced onto the oxidized copper surface (ca. 10 layers of Cu*) which was formed by the H,-O, reaction at 423 K. As shown in fig. 3, the oxidation ? a x ""L I 0, only I I . I I I I n 1 0 30 60 90 tlmin FIG. 3.-Effect of H,(g) on the rate of oxidation of copper. (I) p ( 0 , ) = 1.3 x 104 Pa was introduced on the oxidized surface (ca. 10 Cu,O layers) at 423 K. (11) After evacuation for 1 h at 423 K, p(0,) = 1.3 x 1 O4 Pa and p(H,) = 1.0 x lo4 Pa were reintroduced. 0, Amount of 0, uptake in the catalyst. of copper proceeded at a constant rate, and after 1 h the gas phase was replaced by the mixture of H, and 0,. Then the rate of oxidation of copper was enhanced by the coexistence of H,, in accordance with the results of the conductivity measurements by Palmer.13 Note that, upon oxidation by O,, the surface is covered by Cu", which is not the case for the H,-02 reaction.DEPENDENCE OF THE RATE OF H20 FORMATION ON 0, OR H, PARTIAL PRESSURE The dependence of the rate of H,O formation upon the partial pressure of 0, at 403 K under the constant H, pressure (7.3 x lo3 Pa) was investigated on the oxidized surface (with > 10 oxide layers on average), where the H,O formation reaction proceeds steadily. The measurement began with the highest 0, pressure (2 x lo4 Pa), continuing to the lower partial pressures of 0,. As shown in fig. 4, in the 0, pressure range above ca. 9 x lo3 Pa the rate of H,O formation was independent of the partial pressure of 0,.Below 9 x lo3 Pa 0, the rate increased gradually, showing a maximum at ca. 1.3 x lo3 Pa, and then decreased to zero 0, pressure. However, note that at zero 0, pressure in fig. 4 the rate of formation of H,O by repeated reduction with 7.3 x lo3 Pa of H, increased gradually, as is indicated from the results shown in fig. 2. The effect is considered to be relevant to the results from X.P.S., i.e. Cuo could be detected at this stage for the first time, as will be shown later in detail. On the other hand, even at very low 0, pressure (ca. 6.7 x 10, Pa) oxidation of copper occurred during H,O formation.K. DOMEN, S . NAITO, M. SOMA, T. ONISHI AND K. TAMARU 0.25 h 0.00 0 ." U 2 2 0, 3: - 0 . 2 5 - cr a +-' E! W Bn -. 0 0 0.5 1 .o 1.5 P ( 0 2 ) l l o4 Pa - 849 FIG.4.-Dependence of the rate of H,O formation upon the partial pressure of 0, at 403 K. p(H,) = 7.3 x lo3 Pa: before the sequential measurements, the catalyst surface was oxidized by the H,-0, reaction and > 10 Cu,O layers were formed. Considering the results of the dependence of the rate on 0, pressure mentioned above, two different constant 0, pressures were chosen to study the dependence on H, pressure at 403 K, i.e. a high pressure [p(O,) = 1 .O x lo4 Pa], where the rate of H,O formation was independent of 0, pressure, and a low pressure [p(O,) = 2 x lo3 Pa], where the rate did depend on 0, pressure. Fig. 5 shows that the rate of H,O formation - 0 . 7 5 -O.I0t--- 0.5 0.75 log b ( H 2 ) l 1 o4 1 Pal .o 1.25 FIG.5.-Dependence of the rate of H,O formation upon the partial pressure of H, at 403 K; 0, p ( 0 , ) = 1.0 x lo4 Pa; 0, p ( 0 , ) = 2.0 x lo3 Pa.850 CATALYTIC OXIDATION OF H, ON CU was directly proportional to H, pressure in the higher 0, pressure range. On the other hand, in the lower 0, pressure range it showed a dependence of 0.6 order. The reaction order seems to depend upon the 0, pressure and/or reaction temperature. H2-D2 EXCHANGE REACTION DURING THE H,O FORMATION REACTION A gaseous mixture of H,, D, (1 : 1) and 0, was introduced at 403 K onto a surface having an average of > 10 oxide layers, and the rates of HD and H,O formation were followed by gas chromatography. The experiments were carried out under two different 0, pressures, 1.4 x lo4 and 2.0 x lo3 Pa.The results are shown in fig. 6 and 7, where the rate of decrease of hydrogen pressure (H,+D,) is regarded as the rate 0 ’ 5 ~ 0 0 5 10 15 20 t/min FIG. 6.-H,-D, exchange reaction on Cu,O (ca. 10 layers) at 403 K. Initial pressure of 0, was 1.4 x lo4 Pa. 0, Total pressure of H,, HD and D, (initial ratio of H,:D, was 1: 1); 0, O,(g); A, water; A, HD(g). 1 0.5 0 0 5 10 15 t/min FIG. 7.-H,-D, exchange reaction on Cu,O (ca. 10 layers) at 403 K. Initial 0, pressure was 2.0 x lo3 Pa. 0, Total pressure of H,, HD and D, (initial ratio of H,:D, was 1 : 1); 0, O,(g); A, water; A, HD(g).K. DOMEN, S. NAITO, M. SOMA, T. ONISHI AND K. TAMARU 85 1 of H,O formation. Fig. 6 shows that, in the higher 0, pressure range where H,O formation rate was proportional to H, pressure, the H,-D, exchange reaction proceeded very slowly.For example, after 20 min the decrease of hydrogen amounted to 25% but the amount of HD formed was only ca. 1 % of the total hydrogen. In contrast, as shown in fig. 7 where 0, pressure was low and the dependence of the rate of H,O formation upon H, pressure was less than first order, the H,-D, exchange reaction began to proceed gradually during water formation. Even at this stage, however, the rate of the H,-D, exchange was slow in comparison with the rate of water formation, and after the consumption of most of the gaseous 0, the rate of HD formation increased. Cu(L3M4,,M4,,) AUGER SPECTRA AND 0 (1s) X-RAY PHOTOELECTRON SPECTRA During the reaction the catalyst was investigated by X-ray excited electron spectroscopy to determine the mechanism of this reaction in more detail.Cu(L,M,,,M,,,) Auger spectra and 0 (1s) X-ray photoelectron spectra of the catalyst in the course of H,O formation at 413 K are shown in fig. 8(B) and (C). Fig. 8(B) shows the spectra in case of the lower 0, pressure (1.3 x lo3 Pa) and fig. 8(C) that in case of the higher 0, pressure (1.3 x lo4 Pa). In the Cu(L,M4,,M4,,) Auger spectra 530.4 4 11 cu cul f cuo j I t . # , I t 1 1 1 + ' l " . l # 904 924 536 524 k .e .lev b.e./eV FIG. 8.-Cu(LMM) Auger spectra (kinetic energy, k.e.) and 0 (1s) X.P.S. spectra (binding energy, b.e.) of the catalyst: (A) reduced copper at 423 K for 10 h; (B) after H, (1.3 x 104 Pa) and 0, (1.3 x 10" Pa) reaction for 20 min at 41 3 K; (C) after H, (1.3 x lo4 Pa) and 0, (1.3 x lo3 Pa) reaction for 20 min at 41 3 K; (D) after 0, (1.3 x lo4 Pa) oxidation for 20 min at 413 K; (E) after H, (1.3 x 104 Pa) reduction for 10 min at 413 K; (F) Cu(OH),.852 CATALYTIC OXIDATION OF H, ON Cu peaks at 915.1, 914.4 and 913.3 eV (kinetic energy) have been assigned to Cuo, Cull and Cul, respectively, by Larson.14 Consequently, it is thought that in the case of both higher and lower 0, pressures the catalyst surfaces were covered with Cul in the course of the reaction in the steady state.When the catalyst was reduced by H, only, the new peak, which was assigned to Cuo (915.4 eV), appeared as shown in fig. 8(E). On the other hand, when the surface was oxidized by 0, at ca. 1,3 x lo4 Pa and the same temperature (413 K), only the 914.4 eV peak was detectable, which means that the surface is covered mainly with Cu"; however, the peak is rather broad, which indicates that it might be mixed with Cul and/or Cuo.The 0 (1s) peaks at 530.4 eV (binding energy) during the reaction [fig. 8(B) and (C)] indicate the formation of Cu,0.15 0 (1s) X.P.S. of Cu(OH),, which was precipitated by the same procedure as mentioned in the Experimental section and dried at room temperature, were also examined as shown in fig. 8(F), and gave a broad peak at 53 1.5 eV. Perhaps there exists a small amount of -OH species during H,O formation which could not be detected clearly by our method. DISCUSSION H,O formation from H, and 0, on copper was studied at 403-423 K using a closed circulating system and X-ray excited photoelectron spectroscopy.From the results in fig. 1 it is clear that under the reaction conditions the oxidation of copper proceeds steadily, accompanied by the formation of water. Therefore, various measurements in the steady state were carried out on the oxide layer of known thickness which was formed during the reaction of H, and 0,. It was possible to neglect the effect of gaseous H,O on the reaction rate because H,O formed during the reaction was constantly removed by a dry-ice-methanol cold trap placed immediately after the reactor. As has been clarified in the Results section, to elucidate the reaction mechanism attention should mainly be paid to the partial pressure of 0,. WE define Cuo, Cul and CuI1 as the states of copper in metal, cuprous oxide and cupric oxide, respectively.HIGHER OXYGEN PARTIAL-PRESSURE REGION In the region as follows: of 0, partial pressure above 9 x lo3 Y = k[H,]l[O,]O. Pa, the reaction rate is expressed (1) As shown in fig. 6, the H,-D, exchange reaction does not occur when water formation proceeds, and the amount of adsorbed hydrogen is small. As is demonstrated by X.P.S. and A.e.s. (fig. 8), the surface was fully covered with a Cul layer during the catalytic formation of water, whereas in the absence of hydrogen, copper ions were easily oxidized to Cull by oxygen. The absence of Cull during the catalytic reaction and the rate law (1) suggest that the redox cycle Cul/CulI does not contribute to the catalytic formation of H,O significantly, because if that were the case the zero-order behaviour of oxygen could not be explained.Consequently it seems most probable that in this 0, pressure range the rate-determining step of H,O formation is the activation of the hydrogen molecule on the oxidized surface. The results in fig. 2 also show that the evacuation of gaseous oxygen does not affect the rate of H,O formation immediately, which means that H,O is formed from the adsorbed oxygen on the Cul surface, which does not desorb easily by evacuation at the reaction temperature, 423 K. Further support for the existence of adsorbed oxygen on Cu,O is found in the work of Ostrovsky et al.lS They investigated the heat of adsorption of oxygen on copper and its reactivity with hydrogen at 373-423 K, and found that the heat ofK. DOMEN, S. NAITO, M. SOMA, T.ONISHI A N D K. TAMARU 853 adsorption of 0, on the oxidized surface was lower than that of the formation of Cu,O (334 kJ mol-l). They also reported that oxygen having a heat of adsorption of ca. 293 kJ mol-1 was most reactive with hydrogen, which may be the same species discussed here. As for the activation of hydrogen, it is possible to suppose two active sites where the hydrogen molecule can attack, i.e. on copper and on oxygen. To interpret the data mentioned above, the latter seems to be more probable, and the hydrogen molecule will attack dissociatively adsorbed oxygen on the surface which is 'covered' by adsorbed oxygen. This adsorbed oxygen may be the oxygen atom in the uppermost layer of the Cu,O lattice. In other words, the surface of Cu,O can catalyse H,O formation from H, and 0,.LOWER OXYGEN PARTIAL-PRESSURE REGION In the region of the partial pressure of 0, below 9 x lo3 Pa the reaction kinetics are still complex, but the model of adsorbed oxygen on Cu,O seems able to interpret it. The rate of H,O formation shows a maximum value at a certain oxygen pressure (ca. 1.3 x lo3 Pa) and the reaction order with respect to H, becomes lower. The H,-D, exchange reaction also starts to proceed gradually, although the rate is considerably slower than that of water formation. The surface of the catalyst is still covered with Cul but the 0 (1s) peak [fig. 8(B)] seems to be narrower on the lower binding-energy side in comparison with that of fig. S(C). These phenomena can be understood by considering another reaction mechanism, i.e.the reaction between adsorbed oxygen and adsorbed activated hydrogen on the surface of Cu,O. The amount of adsorbed oxygen on the surface of Cu,O will decrease gradually with the decrease of 0, partial pressure, and if the adsorption of oxygen and hydrogen is competitive the number of sites available for the activation of hydrogen will increase, which will cause the maximum rate shown in fig. 4. The existence of such a maximum rate may also exclude the redox mechanism via the Cuo/Cul cycle in this region. If the reduction of lattice oxygen in Cu,O proceeds and Cuo is produced on the surface during the reaction of H, and O,, Cuo should be oxidized rapidly to CuI. The progress of the oxidation of the catalyst during H, and 0, reaction was confirmed even at the lowest 0, pressure (ca.6.7 x 10, Pa) in fig. 4, and X-ray excited Auger electron spectroscopy could not detect the existence of Cuo on the surface during H,O formation [fig. 8(B)]. However, since the rate of reduction of the lattice oxygen is accelerated when the surface is further reduced and Cuo is clearly detected by X.P.S. (as will be shown later), this redox mechanism (Cuo/Cul) could not predict the existence of the maximum rate in fig. 4. In this region the dependence of the rate of H,O formation on H, is less than first order, which suggests that the rate-determining step changes from the activation of molecular hydrogen to the reaction between H(a) and O(a) on the surface. In view of the fact that the amount of adsorbed hydrogen is not so great (0 < O.l), and the H,-D, exchange reaction-is rather slow in comparison with H,O formation, the probability that H(a) reacts with O(a) must be larger than that of the recombination. The number of active sites for the reaction should be small (probably < 10% of copper atoms on the surface), considering the above results.The H,-D, exchange reaction is supposed to proceed on Cul slowly or on Cuo which is occasionally present on the surface. Although hydroxyl species, which give the 0 (1s) peak at 53 1.4 eV as Cu(OH), [fig. 8 (F)], might be expected to be present on the surface during the reaction, we could not find clear evidence of such a peak.854 CATALYTIC OXIDATION OF H, ON CU REDUCTION OF THE LATTICE OXYGEN The initial rate of the reduction of Cu,O is shown at zero 0, pressure in fig.4, which may correspond to the rate where there is no adsorbed oxygen on the surface. The successive reduction rate is much faster than the initial rate, and at the apparent stationary state which gives constant reaction rate it is faster than the maximum rate in fig. 4. At this stage the H,-D, exchange reaction proceeds rapidly (fig. 7), and Cuo is clearly detected on the surface [fig. 8 (E)]. Accordingly the increase in the reduction rate is attributable to the increase in the number of active sites for the activation of molecular hydrogen, i.e. the number of Cuo atoms on the surface. On the other hand, these phenomena can clearly exclude the possibility that Cuo on the surface is relevant to the phenomena shown in fig. 4, i.e.the rate of water formation has a maximum value at low partial pressures of 0,. The reduction may occur at the boundary of copper metal and copper oxide, as indicated by Pease and Taylor.e CONCLUSION The essential points of the proposed reaction mechanism between hydrogen and oxygen on copper are as follows. Within the range of our experimental conditions for the catalytic formation of water, the reduction of adsorbed oxygen predominates over that of lattice oxygen, and the reaction proceeds on the surface of the Cu,O layer which grows during the reaction. G. C. Bond, Catalysis by Metals (Academic Press, London, 1962). A. T. Larson and P. H. Emett, J. Am. Chem. SOC., 1925, 47, 346. A. F. Benton and P. H. Emett, J. Am. Chem. SOC., 1926,48, 632. A. F. Benton and J. C. Elgin, J. Am. Chem. SOC., 1926, 48, 3027. A. F. Benton and J. C. Elgin, J. Am. Chem. SOC., 1927, 49, 2426. F. E. Smith, J. Phys. Chem., 1928, 32, 719. W. A. Bone and R. V. Wheeler, Philos. Trans. R. SOC. London, Ser. A, 1906, 1, 206. R. N. Pease and H. S. Taylor, J. Am. Chem. SOC., 1922,44, 1637. lo A. T. Larson and F. E. Smith, J. Am. Chem. SOC., 1925, 47, 346. l1 F. J. Wilkins and S. H. Bastow, J. Chem. SOC., 1931, 1525. l9 A. B. Van Cleave and E. K. Rideal, Trans, Faraday SOC., 1937, 33, 635. l3 W. G. Palmer, Proc. R. SOC. London, Ser. A., 1923, 103, 444. l4 P. E. Larson, J. Electron Spectrosc. Relat. Phenom., 1974, 4, 213. l5 G. Schon, Surf. Sci., 1973, 35, 96. l6 V. E. Ostrovsky and N. N. Dobrovolsky, Proc. 4th Int. Congr. Catal., Moscow, 1968 (Akademia Kaido, Budapest, 1970), p. 46. 'I M. Boudart, D. M. Collins, F. V. Hanson and W. E. Spicer, J. Vac. Sci. Technol., 1977, 14, 441. (PAPER 1/596)
ISSN:0300-9599
DOI:10.1039/F19827800845
出版商:RSC
年代:1982
数据来源: RSC
|
18. |
Alternating copolymerization of conjugated dienes with methyl acrylate. Part 1.—Butadiene |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 855-868
Clement H. Bamford,
Preview
|
PDF (848KB)
|
|
摘要:
J. Chem. SOC., Faraday Trans. I, 1982, 78, 855-868 Alternating Copolymerization of Conjugated Dienes with Methyl Acrylate Part 1 .-Butadiene BY CLEMENT H. BAMFORD* AND XIAO-zu H A N ~ Department of Inorganic, Physical and Industrial Chemistry, University of Liverpool, Liverpool L69 3BX Received 22nd April, 198 1 A study has been made of the alternating copolymerization of methyl acrylate (MA) and butadiene (Bd) in the presence of the Lewis acid ethylaluminium sesquichloride (Al,Et,Cl,), photoinitiated (A = 436 nm) by the system Mn,(CO),, + CCl,. Alternation occurs over a wide range of reactant composition, and the contents of trans-1,4 and 1,2 butadiene units in the alternating copolymers have been estimated as 90% and 10 %, respectively. Measurements at low conversion show that the rate of copolymerization is proportional to [MA- - -all [MAireel0 [Bd]"Yf, where MA- - -a1 represents the methyl-acrylate-Lewis-acid complex and 9 is the rate of initiation.This reaction is accompanied by the Diels-Alder addition yielding methyl cyclohex-3-ene-1-carboxylate (MCC), proceeding at a rate proportional to [MA- - -all [Bd]. At longer reaction times the overall kinetic features reflect the competition between these two processes. Kinetic data on the copolymerization are shown to be consistent with the simplest mechanism of alternation, namely that based on predominating cross-propagation. Propagation between M Bd ' and MA, which is relatively slow in the absence of Lewis acid, is greatly accelerated (ca. 33-fold) by Al,Et,Cl, under the conditions used, and so gives rise to alternation.Chain-transfer to CBr, has been observed and the appropriate kinetic parameters evaluated. Since Hirooka et a1.l first reported that, in the presence of ethylaluminium dichloride, propylene and acrylonitrile yield alternating copolymers, the alternating copolymerization of many other pairs of vinyl monomers has been studied,293 the alternating copolymerization of conjugated dienes and acrylic monomers being of considerable interest. Alternating butadiene-acrylonitrile rubber has been synthesized suc~essfully.~~ Furukawa et al. studied the copolymerization of butadiene and methyl methacrylate in the presence of a Lewis acid (such as Al,Et,Cl,) and a compound of a transition metal of Groups IV or V in the Periodic Table (such as VOC13).6-8 They considered that 1 : 1 and 1 : n complexes of the aluminium compound and methyl methacrylate and the 1 : 1 : 1 ternary complex of the aluminium compound, methyl methacrylate and butadiene participated in the reaction.s* Alternating copolymers of butadiene and methyl acrylate, prepared in a similar way, have been reported.l09 l1 Kuran and his coworkers found that, when benzoyl peroxide was used as initiator in the copolymerization of butadiene and methyl acrylate in the presence of some Lewis acids, the yield of polymer could be increased, while the yield of the Diels-Alder adduct of the monomers was reduced.', The mechanism of alternating copolymerization is still a matter of debate, much argument being centred around the possible involvement in the reaction of complexes containing both monomers.2* In mechanistic investigations it is highly desirable to f Present address : Changchun institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, China.855856 COPOLYMERIZATION OF CONJUGATED DIENES separate the initiation and the regulation processes;13 this was achieved in an earlier study of the alternating copolymerization of methyl acrylate and styrene in the presence of ethylaluminium sesquichloride by the use of free-radial initiators based on transition-metal derivative + organic halide The results of this work were consistent with the simplest proposal for alternation, in which cross-propagation plays a predominant part. This and the following paper describe similar investigations of the alternating copolymerizations of methyl acrylate with butadiene and with isoprene.In both cases photoinitiation (A = 436 nm) by Mn,(CO),, + CCl, was employed,15 with Al,Et,Cl, as the Lewis acid. EXPERIMENTAL MATERIALS Butadiene (Bd) was purified by passing through a column packed with molecular sieve 4A, followed by cooling in a dry-ice + methanol bath overnight, then filtering off frozen water and drying over calcium hydride at -78 "C. The butadiene was then distilled on the vacuum line into a graduated vessel, in which a 4 mol dm-, solution in toluene was made for use. Three grades of butadiene were used in our experiments, the initial purities being 99% (B.D.H. Laboratory gas), 99.5 % (Cambrian Gases) and 99.86% (Phillips Petroleum Co.). No significant differences between them were observed.Methyl acrylate (MA) having a purity of 99% was washed several times with an aqueous solution containing 5% sodium hydroxide and 20% sodium chloride. After drying over anhydrous calcium chloride, it was distilled and stored over calcium hydride in a refrigerator. Before using, the monomer was distilled on the vacuum line. Ethylaluminium sesquichloride (25 % toluene solution) was used without further purification. Manganese carbonyl [Mn,(CO),,] was sublimed in vacuum at room temperature. Toluene (A.R.) was dried over sodium wire. APPARATUS AND TECHNIQUE All the experiments were carried out in a laboratory illuminated by inactive (sodium) light. Reactions were initiated by light of wavelength 436 nm, the optical system being the same as described in an earlier paper.Is Reaction vessels consisted of Pyrex glass ampoules with internal diameter 16 mm, which were washed, dried, flamed and filled with nitrogen.Mn,(CO),,, toluene, CCl,,Al,Et,Cl,, MA and Bd were introduced by pipetting required volumes into the reaction vessel in the above order (CCl,,Al,Et,CI, and Bd being in toluene solution). Reaction mixtures, of which the total volume was 10 cm3, were thoroughly degassed by the conventional freeze-thaw technique; when the reaction vessel had been sealed it was illuminated in a thermostat at 25 f 0.1 OC for the required time by filtered radiation from a 250 W high-pressure mercury arc (AEI, type ME/D). The reaction mixture was then poured into 400cm3 methanol containing a small amount of phenyl-/3-naphthylamine as stabilizer for the copolymer.The latter was precipitated, filtered off, dried and weighed; it was a rubbery solid free from gel. Samples for analysis were reprecipitated from toluene solution into methanol. The random copolymer of MA and Bd required for spectral observations was prepared by copolymerizing the two monomers (Bd: MA = 1 : 2.4 mol dm-3) in toluene with benzoyl peroxide as initiator at 40 O C . The conversion was only ca. 2% after 40 h. Rates of initiation (f) were determined by calibration experiments in which the homopoly- merization of methyl methacrylate was photoinitiated by Mn,(CO),, + CCl, with similar light intensities and concentrations of Mn,(CO),, and CCl,.14 This procedure is justified by earlier inve~tigationsl~ which show that in homopolymerizations the rate of photoinitiation by this system is effectively independent of the nature of the monomer.In our experiments 9 was 9.18 x Number-average molecular weights of copolymers were measured osmometrically with a Hewlett Packard 503 high-speed membrane osmometer, with toluene as solvent. mol dm-, s-,, except when stated otherwise.C. H. BAMFORD A N D XIAO-ZU HAN 8 57 80 n 5 L3 60- x 0 a - 4 0 - .+ -€ E 20 Infrared absorption spectra of copolymers were recorded by a Perkin-Elmer 5 17 grating spectrophotometer with films cast from chloroform solution. Copolymer composition was determined by elemental analysis and the degree of alternation was estimated from the 60 MHz n.m.r. spectrum in deuterochloroform. The Diels-Alder adduct of Bd and MA in the reaction system was estimated on a Pye 104 gas-liquid chromatograph (g.1.c.) having a 3 m column packed with 3 % OV-17 on Supelcoport (100-120 mesh) at 110 OC.- 0 n W 0 - RESULTS AND DISCUSSION COMPOSITION A N D STRUCTURE OF THE COPOLYMERS Over a wide range of initial concentrations of methyl acrylate [MA], at constant [Bd],, the copolymer was found to maintain a 1 : 1 composition (fig. 1). When [MA], and [Bdl0 were both 0.8 mol dm-3 and polymerization had proceeded for 10 h, the content of MA in the copolymer was still 50%. '0° T 01 1 I 1 I I I 0 0.4 0.8 1.2 1.6 2.0 2.4 [MA], in reaction mixture/mol dm-3 FIG. 1 .-Dependence of copolymer composition on [MA],. Initial concentrations/mol dm-3 : Bd 0.8, Al,Et,Cl, 0.2, Mn,(CO),, 2 x lo-,, CC1, 0.1 ; 25 OC, reaction time 1 h, 1 = 436 nm.The infrared spectra of alternating and random copolymers of Bd and MA are shown in fig. 2. The absorption band at 1730 cm-l is characteristic of the C=O bond in MA units. The bands at 970, 915 and 720 cm-l are characteristic of trans-1,4, 1,2 and cis-1,4 isomers of Bd units, respectively. In the spectrum of the alternating copolymer there is no absorption band at 720 cm-l, implying effectively no cis-1,4 isomer. The contents of trans-1,4 and 1,2 isomer in the alternating copolymer were estimated as 90% and lo%, respectively, according to the method of Haslam et a1.18 The contents of trans-1,4, 1,2 and cis-1,4 isomers in the random copolymer were 74.7%, 9.0% and 16.3%, respectively. The n.m.r. spectra of alternating and random copolymers of Bd and MA are shown in fig.3. The peak near 26 (ppm relative to TMS) in the spectrum of the random copolymer [fig. 3 (a)] arises from the adjacent methylene groups in butadiene-butadiene diads. The absence of this peak from the spectrum of fig. 3 (b) confirms the alternating structure of the copolymer. The peaks (i) at 5.1-5.46 arise from -CH= and CH2=CH- protons in Bd units and the peak (ii) at 3.65 6 from H3CO- protons858 COPOLYMERIZATION OF CONJUGATED DIENES 2000 1800 1600 1400 1200 1000 aoo 600 wavenumberlcm-' FIG. 2.-Infrared spectra of MA-Bd copolymers. (a) Random copolymer prepared by free-radical copolymerization with initial concentrations/mol dm+: MA 6, Bd 2.5, benzoyl peroxide 40 OC, 40 h. (b) Alternating copolymer, prepared with initial concentrations/mol dm-3 : MA, Bd 0.8, A12Et3C13 0.2, Mn,(CO),, 2 x CCl, 0.1 ; 25 OC, 1 h, II = 436 nm.I I I I , 6 5 4 3 2 1 0 6 (PPm) FIG. 3.-N.m.r. spectra of MA+Bd copolymers in CDCl,: (a) random, (b) alternating, prepared as described for fig. 2.C. H. BAMFORD AND XIAO-ZU HAN 859 in MA units. The observed ratio of peaks (i)/(ii) determined from the integral line in fig. 3 (b) is ca. 2/3, while the predicted ratio from the 1 : 1 composition and the trans-l,4 and 1,2 isomer contents derived from the i.r. observations is 2.1/3. The spectra of copolymers of Bd and MA are similar to those of Bd and methyl methacrylate.s~19 KINETICS In general, the concentrations of MA (A) and Bd (B) were both 0.8 mol drn-,, and the concentrations of Al,Et,Cl,, Mn,(CO),, and CCl, were 0.2, 2 x lo-, and 0.1 mol drn-,, respectively. To determine the rate of copolymerization we used the gravimetric method, which involved weighing the copolymer formed at low conversion : d([MA]+[Bd]) - W 1000 dt 70.09 10t - X- rate of copolymerization co = where W is the copolymer weight (in g), t the reaction time (in s) and 70.09 is the average molecular weight of MA and Bd monomers.Previo~slyl~ both Mn,(CO),, + CCl, and Ni(CO), {P(OPh),}, + CCl, systems were found to be very effective initiators for the alternating copolymerization of MA and styrene. In the present work only the former was effective; it is likely that the nickel complex forms a n-ally1 derivative with Bd and so does not initiate.,O The concentration of CCl, (0.1 mol drn-,) is in the ‘plateau’ range’, over which the rate of initiation is independent of [CCl,].The rate of copolymerization was found to depend linearly on [Mn,(CO),,]~ up to [Mn,(CO),,] = 8 x lo-, mol dm-, (fig. 4), showing that the reaction proceeds by a free-radical mechanism. 3.2 - t m 2.4 E -0 - 1.6 t 2 3 Oa \ 0 0 1 2 3 4 5 [Mn,(CO),,l$l 0-2 moh dm -* FIG. 4.-Dependence of rate of copolymerization o of MA and Bd on [Mn,(CO),,]fo. Initial concentrations/mol dm-3: MA, Bd 0.8, Al,Et3C13 0.2, CCl, 0.1 ; 25 OC, 1 = 436 nm. Furukawa et aLs considered A1,Et,Cl3 to behave both as an initiator and a regulator in the copolymerization of Bd and MMA, the rate of copolymerization being proportional to [Al,Et,Cl,]% In our system the rate of reaction was negligible in the absence of Mn,(CO),, or light and Al,Et,Cl, was only a regulator; the effect of its concentration on o is shown in fig.5 . The presence of Al,Et,Cl, leads to greatly enhanced rates and without Al,Et,Cl, no polymer was obtained even at very long reaction times. The linear relation (fig. 5 ) is observed when the concentration of Al,Et,Cl, exceeds 0.05 ml drn-,, up to the highest value studied (0.4 mol dm-,).860 COPOLYMERIZATION OF CONJUGATED DIENES 2.4 * I m m E a 1.6 - z 4 0.8 --- 3 0 0 01 0.2 03 0.4 0.5 [Al,Et3C13]o/mol dm-3 FIG. %-Dependence of rate of copolymerization w of MA and Bd on [Al,Et3Cl,],. Initial concentrations/mol dm-3: MA, Bd 0.8, Mn,(CO),, 2 x lo-,, CCl, 0.1; 25 OC, 1 = 436 nm. 3 0 0.8 1.6 2.4 3.2 [MA],/mol dm- FIG. 6.-Dependence of rate of copolymerization w of MA and Bd on [MA],.Initial concentrations/mol drn-,: Bd 0.8, Al,Et,Cl, 0.2, Mn,(CO),, 2 x CCI, 0.1 ; 25 OC, 1 = 436 nm. 0 0.8 1.6 2.4 3.2 [ Bd],/mol dm-3 FIG. 7.-Dependence of rate of copolymerization w of MA and Bd on [Bd],. Initial concentrations/mol drn-,: MA 0.8, Al,Et,Cl, 0.2, Mn,(CO),, 2 x lop4, CC1, 0.1; 25 OC, 1 = 436 nm. When [MA], was lower than 0.4 mol dmm3, the reaction mixture became hetero- geneous after reaction for a few minutes. The dependence of cc) on the concentrations of monomers is shown in fig. 6 and 7, from which it can be seen that w is effectively independent of [MA], from 0.6 to 2.4 mol dm-3, and [Bd],'from 0.4 to 2.0 mol dm-3. Rates of copolymerisation were determined for short reaction times (15 min) in the above experiments.Data for longer reaction times are shown in table 1 and fig. 8 andC. H. BAMFORD A N D XIAO-ZU HAN 86 1 TABLE EFFECT OF REACTION TIME ON THE COPOLYMERIZATION OF MA AND Bd IN THE PRESENCE OF Al,Et,Cl, Initial concentrations/mol dm-3: Bd 0.8, Al,Et,Cl, 0.2, Mn,(CO),, 2 x lo-,, CCl, 0.1 ; 25 OC, A = 436 nm. reaction conversion of Bd [MA]/mol dm-, time/h into copolymer (%) lo-, M, 0.6 0.8 1.6 2.4 0.25 1 3 5 10 0.25 1 3 5 10 0.25 1 3 5 10 16 0.25 1 3 5 10 6.1 13.7 17.8 21.1 20.7 6.3 17.0 25.8 29.7 33.2 6.1 17.7 32.2 41 .O 53.2 53.4 6.9 18.2 35.1 45.0 57.9 70 81 112 152 65 91 110 119 64 102 132 - - - 0 2 4 6 8 10 reaction time/h FIG. 8.-Dependence of polymer yield on reaction time for various values of [MA],. Initial concentrations/mol dm-3: Bd 0.8, A1,Et3C1, 0.2, Mn,(CO),, 2 x CCl, 0.1; 25 OC, I = 436 nm.[MA],/m~ldrn-~: 0, 2.4; 0, 1.6; 0, 0.8; 0, 0.6.862 COPOLYMERIZATION OF CONJUGATED DIENES 0 2 4 6 8 10 reaction time/h FIG. 9.-Dependence of polymer yield on reaction time for various values of [Bd],. initial concentrations/mol drn-,: MA 0.8, A12Et,Cl, 0.2, Mn,(CO),, 2 x lo-,, CCl, 0.1; 25 OC, d = 436 nm. [Bd],/mol dm-a: 0, 1.6; 0, 0.8; 0, 0.4. 9. It is obvious that, when the reaction time is short, the yields of copolymer are almost independent of [MA], and [Bd],; when the reaction time is > 1 h, the yields increase with increasing [MA], but decrease with increasing [Bd],. It is well-known12* 21 that, in the presence of a Lewis acid, Diels-Alder addition can occur between methyl acrylate and butadiene to form methyl cyclohex-3-ene-l- carboxylate (MCC) : O\,,,M, Lewis acid (= + CH?=CH-COOMe -P M CC The yield of MCC is proportional to the concentrations of complex of MA with Lewis acid (MA- - -al, see below) and Bd, but is independent of the concentration of free MA : -- - k,[MA- - -all [Bd].d[MCC] dt 0 1 2 3 4 5 reaction time/h FIG. 10.-Dependence of yield of MCC on reaction time for various values of [Bd],. The conditions were the same as in fig. 9.C. H. BAMFORD AND XIAO-ZU HAN 863 In our system the formation of MCC occurs simultaneously and competes with copolymerization.8* l2 Concentrations of MCC in the reaction mixture for a series of reaction times and a range of [Bdl0 are shown in fig. 10. 1 3 Kb+A+A' Rb+B-+B' B' +A -+ A' kba A'+B+B' k,, TABLE 2.-EFFECTS OF SOME CONDITIONS ON THE YIELDS OF COPOLYMER AND ADDUCT MCC Initial concentrations/mol dm-3: MA, Bd 0.8, CCl, 0.1 ; 25 OC, 1 h.B' +S + P+S' kfb S'+A+A' kra S' +B -+ B' krb A'+A'-+P ktaa A'+B' -+P ktab 0.2 2 x 10-4 dark 0.2 2~ 10-4 light 0.2 2 x 10-3 light 0.1 2 x 10-3 light 0.29 0.25 0.22 0.12 0.02 0.13 0.30 0.15 From fig. 9 and 10 it is clear that the higher [Bdl0 the greater is the ratio adduct : polymer for long reactions times. The effects of other conditions on the yields of copolymer and adduct are shown in table 2. The concentration of Al,Et,Cl, affects the yields of both copolymer and adduct. Irradiation and [Mn,(CO),J have no signifiant effect in the Diels-Alder reaction but of course greatly influence the yield of copolymer.MECHANISM (3)864 COPOLYMERIZATION OF CONJUGATED DIENES Homopropagations are not included. Initiating radicals R, are formed from the initiator C and P represents the dead polymer. Termination by combination is assumed, this seeming most appropriate for the two monomers under consideration; our general conclusions are not affected by this assumption. Conventional stationary-state treatment leads to relations (4) and ( 5 ) for long chains : O = 2kba[A] (t)! (4) In deriving these equations the total rate of termination has been written as k,[B'I2, so that k, contains contributions from all three termination steps in reaction (3) and is given by Relations (4)-(6) are quite general and involve no assumptions about specific termination mechanisms. Two limiting cases are of interest.If [B] is sufficiently large, kt = ktbb, i.e. termination occurs predominantly between B' radicals, so that from eqn (4) the rate of copolymerization is given by eqn (7a) W = 2kba[A] (L)t. (7 a) ktbb On the other hand, for sufficiently small [B] the third term on the right-hand side of eqn (6) (termination between A' radicals) is large and k, approximates to ktaa(kba[A]/kab[B])2; thus the rate of copolymerization follows eqn (7b) In our system (A = MA, B = Bd) under present conditions the former alternative appears to hold and the rate of copolymerization given by eqn (7a) is almost independent of butadiene concentration for short reaction times (fig. 7 ) . HirookaZ2 showed that complexing between methyl acrylate and ethyl aluminium sesquichloride is very strong and that the complex may be represented by MA---AIEtl~,Cll~, or MA---al: Krn MA + a1 + MA- - -al.Thus in the general kinetic scheme (3) A must be understood to refer to this complex and [A] = [all = 2 [Al,Et,Cl,] so long as [all < [MA] (approximately). Under these conditions eqn (7a) predicts that w cc [all; according to fig. 5 this proportionality is found except at low values of [all. A similar observation was reported for the methyl acrylate + styrene system and evidence was adduced showing that at low [all primary radical termination occurs and interferes with the simple proportionality.'* It seems likely that such an explanation may hold in the present case. Eqn ( 5 ) does not allow explicitly for chain-transfer to species other than the added transfer agent S .Since under our conditions k, = ktb, eqn ( 5 ) predicts that, at constant [all and 9, Pn should be independent of the concentrations of methyl acrylate (> [all) and butadiene (since [A] = [all), provided no significant transfer to these monomers occurs. This is consistent with the data for short reaction times in table 1. The increase (8)C. H. BAMFORD A N D XIAO-ZU H A N 865 in molecular weight with increasing reaction time reflects the consumption of the photoinitiator Mn,(CO),, and consequent decline in the rate of initiation. For short reaction times, the yields of adduct increase with [Bd], as expected (fig. lo), although as explained above the yield of copolymer remains constant. By virtue of the Diels-Alder reaction, the rate of consumption of MA becomes a function of [Bd], increasing with the latter [cf. eqn (2)].For sufficiently long reaction times [MA] eventually falls below [all, and [MA- - -a11 (and hence the rate of copolymerization) becomes dependent on [MA]. Other things being equal, this situation will be reached more rapidly with higher initial butadiene concentrations, so that at long reaction times the mean rate of copolymerization will depend on the initial Bd concentration, [Bd],, decreasing with increase in the latter. These considerations are supported by the data in fig. 9, which show that, for constant [MA],, [all, and initiation conditions, the polymer yield decreases with increasing [Bd],, except for short reaction times. Similar reasoning enables us to understand why the polymer yield for constant [Bd],, [all, and initiation conditions increases with increasing [MA],, except for short reaction times (fig.8). As would be expected, the polymer yield is not much affected by [MA], when the latter is sufficiently high (cf. the curves for [MA], = 1.6 and 2.4 mol dmM3 in fig. 8). When equimolar concentrations of MA and Bd are used, the total conversion to polymer and MCC is very high at long reaction times. It must be stressed that not only methyl acrylate but also the copolymer and MCC can form complexes with the aluminium derivatives, so that, in addition to equilibrium (8), there exist the equilibria KP P + a1 + P-- -a1 KlL MCC + a1 e MCC- - -al. (9) Thus the aluminium is distributed between MA, P and MCC in proportions determined by the equilibrium constants Km, Kp and Ka.Quantitative treatment of the data in fig. 8-10 requires a knowledge of these constants. Unfortunately there is little information about the magnitudes of Kp and K, except that these are probably much less than Km. In principle, therefore, as the reaction proceeds a decreasing fraction of aluminium is available for complexation with MA, and a progressive (probably small) decrease in the overall rate would be expected for this reason. CHAIN TRANSFER TO CARBON TETRABROMIDE: REACTION PARAMETERS Chain transfer occurs in alternating copolymerizations, but usually to a relatively small extent. Thus in the alternating copolymerization of methyl acrylate and styrene in the presence of Al,Et3C13 the rate coefficient for the transfer reaction between the radical M MA-St and CBr, is ca.30 times smaller than that for the reaction without complexing agent. Possible reasons for this, based on complexing of the radical with Lewis acid, have been advanced.13 Corresponding transfer data for the MA + Bd system with short reaction times (10 min) are presented in table 3. The presence of CBr, leads to copolymers of lower molecular weight but has little effect on the rate of copol ymeriza tion. The plot of l / e against [CBr,] presented in fig. 11 is satisfactorily linear with slope 0.545 mob1 dm3. According to eqn (5) the slope is866 COPOLYMERIZATION OF CONJUGATED DIENES TABLE 3.-cHAIN-TRANSFER TO CBr, IN THE ALTERNATING COPOLYMERIZATION OF MA AND Bd AT 25 OC Initial concentrations/mol dm-3: MA, Bd 0.8, A1,Et3C13 0.2, Mn,(CO),, 2 x lo-*, CC1, 0.1 ; 9 = 3.2 x lo-' mol dmP3 s-l.[CBr,]/ 1 O-, mol dm-3 o/ lo-, mol dm-3 s-l 0 2 5 10 1.82 1.80 1.88 1.84 1244 1028 939 746 14 1 2 E '5 2 u 8 6 0 2 4 6 8 10 [CBr,]/104 mol dm-3 FIG. 11 .-Dependence of 8 on CBr,; plot according to eqn (5). Initial concentrations/mol drn-,: MA, Bd 0.8, AI,Et,CI, 0.2, Mn,(CO),, 2 x CCI, 0.1 ; 25 OC, L = 436 nm, 9 = 3.2 x lo-' mol dm-3 s-*. For systems without Lewis acids, the first term in expression (10) is equal to rA CA/2[B], rA and CA denoting reactivity ratio and transfer constant, respectively. If we assume rA = 0.0524 and CA = 0.325 we obtain kfa = 0.0094 mol-l dm3. 2kab[Bl This is only a small contribution towards the observed slope. In the presence of A12Et3Cl,, C , is probably greatly reduced so that eqn (1 1) is an overestimate; however, its use will not introduce significant error.From eqn (10) and (1 1) we find k,, = 0.43. (12) kba The rate data in table 3 permit the.evaluation of kbak& from eqn (7a); we find that kbakcib = 0.40 mOl4 dmi S b . kfb k;&, = 0.17 m o l t dm; S-1 (1 3) (14) Combination of eqn (12) and (13) shows thatC. H. BAMFORD AND XIAO-ZU HAN 867 a value very close to that of the same parameter in the methyl acrylate + styrene system (0.18 mol-g dmi s t ) (styrene = B).14 No experimental determinations of kt,, have been reported; however, it seems unlikely that this coefficient is less than ktbb for the MA+St system,13 uiz. 6 x lo6 mol-l dm3 s-l. With this value we find from eqn (13) that k,, = 980 mol-l dm3 s-l.According to Walling and D a ~ i s o n ~ ~ the reactivity ratios for the system in the absence of Lewis acids are rA = 0.05 and rB = 0.76 at 5 O C ; no determinations are recorded at other temperatures. Assuming the values at 25 OC are not very different, we find, with the aid of the propagation coefficient in the homopolymerization of butadiene (19 mol-1 dm3 s-l) determined by Morton et a1.,26 that in the simple free-radical copolymerization k,, x 25 mol-l dm3 s-l. Then the presence of Lewis acid brings about a 39-fold increase in this coefficient. The other cross-propagation coefficient, k,,, estimated similarly from the homopropagation coefficient of methyl a ~ r y l a t e ~ ~ (592 mol-1 dm3 s-l) turns out to be 11 840 mol-l dm3 s-l in the absence of Lewis acids.The influence of A12Et3C13 on k,, is not yet known but analogy with the MA+ St system2* suggests that presence of the Lewis acid would produce an increase in k,,. The high estimates of kba and k,,, compared with values of homopropagation rate coefficients in the system, are clearly consistent with the proposed mechanism of alternation [eqn (3)]. With the aid of the value of kt,, assumed above we find from eqn (14) that k,, z 4 16 mo1-l dm3 s-' ; unfortunately no data on this rate coefficient for systems free from Lewis acids are available for comparison. To summarise, we believe that the detailed kinetic data strongly support the cross-propagation mechanism of alternation. Further, the results presented indicate that the conditions favouring the copolymerization reaction over the Diels-Alder addition are high [MA],, low [Bd],, high light intensity and high initiator (Mn,(CO),,) concentration; the findings of Kuran et a1.12 are consistent with this conclusion.M. Hirooka, H. Yabuuchi, S. Morita, S. Kawasumi and K. Nakaguchi, J. Polym. Sci., Part B, 1967, 5, 47. M. Hirooka, Doctoral Thesis (Kyoto University, 1971). For reviews see C. H. Bamford in Molecular Behaviour and the Development of Polymeric Materials, ed. A. Ledwith and A. M. North (Chapman and Hall, London, 1975), chap. 2; H. Hirai, J. Polym. Sci., Macromol. Rev., 1976, 11, 47. J. Furukawa and Y. Iseda, J. Polym. Sci., Part B, 1969, 7, 47. N. G. Gaylord and A. Takahashi, J. Polym. Sci., Part B, 1969, 7, 443. J. Furukawa, E.Kobayashi, Y. Iseda and Y. Arai, Polym. J., 1970, 1, 442. J. Furukawa, Y. Arai and E. Kobayashi, J. Polym. Sci., Part A , 1974, 8, 417. J. Furukawa, Y. Iseda and E. Kobayashi, Polym. J., 1971, 2, 337. a J. Furukawa, E. Kobayashi, Y. Iseda and Y. Arai, J. Polym. Sci., Part B, 1971, 9, 179. lo British Patent, 1,186,461, 1968 (to Bridgestone Tyre Co). l1 Japanese Patent, 23,181, 1972 (to J. Furukawa, Y. Iseda, Y. Kazuo and K. Nobuyaki, Japan). l2 W. Kuran, S. Pasynkiewicz, R. Nadir and B. Kowaleweska, Macromol. Chem., 1976, 177, 1291. l 3 C. H. Bamford and P. J. Malley, J . Polym. Sci., Polym. Lett. Ed., 1981, 19, 239. l4 C. H. Bamford, S. N. Basahel and P. J. Malley, Pure Appl. Chem., 1980, 52, 1837. l5 C. H. Bamford in Reactivity, Mechanism and Structure in Polymer Chemistry, ed. A. D. Jenkins and A. Ledwith (John Wiley, London, 1974), chap. 3. C. H. Bamford and S. U. Mullik, Polymer, 1973, 14, 38. l7 C. H. Bamford, R. W. Dyson and G. C. Eastmond, Polymer, 1969, 10, 885. la J. Haslam, H. A. Willis and D. C. M. Squirrell, ZdentiJication and Analysis of Plastics (Iliffe Books, l9 J. R. Ebdon, J. Macromol. Sci., Chem., 1974, 18, 417. *O J. Ashworth and C. H. Bamford, J. Chem. Soc., Faraday Trans. I , 1973, 69, 302, 314. 21 T. Inukai and T. Kojima, J. Org. Chem., 1967, 32, 872. 22 M. Hirooka, J. Polym. Sci., Part B, 1972, 10, 171. London, 1972), p. 443.868 COPOLYMERIZATION OF CONJUGATED DIENES 23 B. Golubev, V. P. Zubov, G. S. Georgiev, I. L. Stoyachenko and V. A. Kabanov, J. Polym. Sci., 24 C . Walling and J. A. Davison, J. Am. Chem. SOC., 1951, 73, 5736. 26 G. C. Eastmond, in Comprehensive Chemical Kinetics, ed. C . H. Bamford and C. F. H. Tipper 26 M. Morton, P. P. Salatiello and H. Landfield, J. Polym. Sci., 1952, 8, 215. 27 M. S. Matheson, E. E. Auer, E. B. Bevilacqua and E. J. Hart, J. Am. Chem. Soc., 1951, 73, 5395. 28 C. H. Bamford and P. J. Malley, to be published. Polym. Chem. Ed., 1973, 11, 2463. (Elsevier, Amsterdam, 1976), vol. 14A, p. 226. (PAPER 1/644)
ISSN:0300-9599
DOI:10.1039/F19827800855
出版商:RSC
年代:1982
数据来源: RSC
|
19. |
Alternating copolymerization of conjugated dienes with methyl acrylate. Part 2.—Isoprene |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 869-879
Clement H. Bamford,
Preview
|
PDF (649KB)
|
|
摘要:
J. Chem. Soc., Faraday Trans. I, 1982, 78, 869-879 Alternating Copolymerization of Conjugated Dienes with Methyl Acrylate Part 2.-Isoprene BY CLEMENT H. BAMFORD* AND XIAO-zu H A N ~ Department of Inorganic, Physical and Industrial Chemistry, University of Liverpool, Liverpool L69 3BX Received 22nd April, 198 1 The alternating copolymerization of methyl acrylate (MA) and isoprene (Ip) in the presence of ethylaluminium sesquichloride (Al,Et,Cl,) has been studied. Photoinitiation (3, = 436 nm) was effected by the system Mn,(CO),, + CCl,. Alternation occurs over a wide range of reactant composition. Copolymerization competes with a rather rapid Diels-Alder addition yielding a mixture of 3-methyl- and 4-methyl- (methyl cyclohex-3-ene- 1 -carboxylate). Evidence is adduced indicating that the rate of copolymerization is proportional to [MA- --all [MAfree]' [Iplo9# where MA- --a1 represents the methyl- acrylate-Lewis-acid complex and 9 is the rate of initiation.All the kinetic data are consistent with the cross-propagation mechanism of alternation. The propagation process -1p' +MA is relatively slow in the absence of Lewis acid, but is markedly accelerated (ca. 88-fold) by Al,Et,Cl, under the conditions used, with resulting alternation. This enhancement in rate is attributed mainly to weakening of the C=C bond in MA which accompanies complexation of the monomer with the Lewis acid. Chain-transfer to CBr, has been observed and the appropriate kinetic parameters evaluated. In the previous paper1 we have reported an investigation of the alternating copolymerization of methyl acrylate and butadiene in the presence of ethylaluminium sesquichloride, with photoinitiation by a transition-metal derivative + organic halide system [Mn,(CO),, + CCl,].We concluded that the kinetic data are compatible with the simplest alternation mechanism, namely that arising from the predominance of cross-propagation reactions. A similar conclusion was reached earlier for the methyl acrylate/styrene alternating copolymerization.2 In both cases it appears that com- plexation of methyl acrylate with the Lewis acid greatly increases the rate of reaction of this monomer with propagating chains carrying terminal units derived from the hydrocarbon monomer. This paper describes an extension to our work to the methyl acrylate/isoprene copolymerization in the presence of ethylaluminium sesquichloride.Furukawa et aL3 have reported the synthesis of an alternating copolymer of methyl acrylate and isoprene by photoinitiated copolymerization in the presence of aluminium chloride and vanadyl butoxide, and Akimoto and Ohtsuru, have investigated the influence of water on the alternating copolymerization of the same monomers with Et,AlCl,-, + VOCl, as catalyst. However, there is not much mechanistic information in these works. EXPERIMENTAL MATERIALS Isoprene (Ip) of research grade (Phillips Petroleum Co) was distilled and dried over calcium hydride; before use, it was distilled on the vacuum line. t Present address : Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun, 869 China.870 COPOLYMERIZATION OF CONJUGATED DIENES Methyl acrylate (MA), manganese carbonyl [Mn,(CO),,] and carbon tetrabromide (CBr,) Ethylaluminium sesquichloride (A1,Et3Cl3, 25 % toluene solution) and carbon tetrachloride Toluene (A.R.) was dried over sodium wire.The apparatus and techniques used were similar to those described in the previous paper.' All kinetic experiments were carried out in a laboratory illuminated by inactive (sodium) light. Reactions were initiated by light of wavelength 436 nm, the optical system being the same as previously de~cribed.~ Mn,(CO),,, CCl,, A1,Et3C13 (in toluene solution as required), MA and Ip were introduced by pipetting the necessary volumes into the Pyrex glass reaction vessel in the order stated. The reaction mixture (total volume 10 cm3) was then thoroughly degassed by the conventional freeze-thaw technique. When the vessel had been sealed it was irradiated in a thermostat at 25 0.1 OC for the required reaction time.The reaction mixture was then poured into 400 cm3 methanol containing a small amount of phenyl-8-naphthylamine to prevent oxidation of the copolymer. Precipitated copolymers were filtered off, dried and weighed ; they were rubbery solids free from gel. Samples for analyses were reprecipitated from methanol. Random copolymers of MA and Ip required for spectral observations were prepared by copolymerizing the two monomers ([MA] = 6.5 mol dm-3, [Ip] = 2.5 mol dm-3) in toluene solution using benzoyl peroxide (0.008 mol dm-3) as initiator at 40 OC. The conversion was 1.7% after 26 h.Number-average molecular weights of copolymers were measured osmometrically with a Hewlett-Packard 503 high-speed membrane osmometer, with toluene as solvent. Infrared absorption spectra of copolymers were recorded by a Perkin-Elmer 5 17 grating spectrophotometer using films cast from chloroform solution. Compositions of the copolymers were determined by elemental analysis and the extent of alternation was estimated from the 60 MHz n.m.r. spectra in, deuterochloroform. The Diels-Alder adduct of MA and Ip in the reaction mixture was estimated on a Pye 104 gas-liquid chromatograph using a 3 m column at 160 OC, packed with 3% OV-22s on Supelcoport (100-120 mesh). The solution for g.1.c. analysis was made by pouring the reaction mixture into 90 cm3 methanol containing 5 cm3 water, to facilitate separation of copolymer and destroy AI,Et,Cl,.RESULTS AND DISCUSSION were purified as previously reported.' (A.R.) were used as supplied without further purification. APPARATUS AND TECHNIQUES COMPOSITION AND STRUCTURE OF THE COPOLYMER Elemental analyses showed that the composition of the copolymer does not depend on the initial concentration of MA in the range 0.4-3.2 mol dmb3; the 1 : 1 composition is maintained except when the concentration of MA is so low ( e g . 0.1 mol dm-3) that homopolymerization of isoprene to cyclic polymers occurs (see later) (fig. 1). Fig. 2. shows the n.m.r. spectra of an alternating and a random copolymer of MA and Ip. The peak near 2 6 (ppm relative to TMS) in the spectrum of the random copolymer arises from adjacent methylene groups in isoprene-isoprene diads and the absence of this peak from the spectrum of copolymer (b) confirms the alternating structure.s The peak at 5.1 6 arises from -CH= protons in isoprene units and the peak at 3.65 6 from -OCH3 protons in methyl acrylate units.The ratio of the above two peaks, estimated from the integral line of spectrum (b), is close to 1:3 and is therefore consistent with the 1 : 1 composition. The peak at 1.55 6 originates from -CH3 protons in isoprene units. Spectra of copolymers of MA and Ip, both alternating and random, are very similar to those of methyl methacrylate and isoprene. The infrared spectra of alternating and random copolymers of MA and Ip are shown in fig. 3. The characteristic absorption band^^?^ for cis-1,4 and trans-1,4 isomers ofC.H. BAMFORD A N D XIAO-ZU H A N loo I 00 h E - 4 6 0 - - 5 0 a O0 4 0 - .r( d E - / P S n 0 / I I I I -d 1 I 1 I 87 1 0 Q0 1.6 2.4 [MAlo in reaction mixture/mol dm-3 FIG. 1 .-Dependence of copolymer composition on [MA]. Initial concentrations/mol dm-, : Ip 0.8, Al,Et,Cl, 0.2, Mn,(CO),, 2 x CCI, 0.1 ; 25 OC, 1 = 436 nm, reaction time 10 min. FIG. 2.-N.m.r. spectra of MA/Ip copolymers in CDCI,. (a) Random copolymer. Initial concentrations/mol drn-,: MA 6.5, Ip 2.5, benzoyl peroxide 8 x lo-,; toluene solution, 40 OC, 26 h. (b) Alternating copolymer. Initial concentrations/mol dm-3 : MA 1.6, Ip 0.8, Al,Et,Cl, 0.2, Mn,(CO),, 2 x CCl, 0.1; toluene solution, 25 OC, 10 min, 1 = 436 nm.872 COPOLYMERIZATION OF CONJUGATED DIENES 2000 1000 1600 1400 1200 1000 000 600 wavenumber/sm-' FIG.3.-Infrared spectra of MA/Ip copolymers: (a) random, (b) alternating. Copolymers prepared as described for fig. 2. isoprene units at 840,1131 and 1 152 cm-l overlap absorptions which arise from methyl acrylate units, so that no significant difference between them is observed. KINETICS In general, the concentrations of MA and Ip were both 0.8 mol dm-,, and the concentrations of Al,Et,Cl,, Mn,(CO),, and CCl, were 0.2,2 x lo-, and 0.1 mol dm3, respectively. The rate of initiation, 9, is based on those derived from calibration experiments in which the homopolymerization of methyl methacrylate was photoinitiated by Mn,(CO),, + CCl, with similar light intensity and concentrations of Mn,(CO),, and CCl, as described in ref.(1). In our experiments Y = 2.07 x lo-' mol dm-3 s-' for short reaction times, except in the case of the chain-transfer experiments. To determine the rate of copolymerization we used the gravimetric method, which involved weighing the copolymer formed at low conversion : d w 1000 rate of copolymerization o = --([MA] + [Ip]) = -- dt 77.1 lot where Wis the copolymer weight (in g), t the reaction time (in s) and 77.1 is the average molecular weight of MA and Ip. The plot of w against [Mn,(CO),,]b was found to be effectively linear over a wide range of [Mn,(CO)l,]O (fig. 4), consistent with a conventional free-radical mechanism. The dependence of o on [Al,Et,Cl,], is shown in fig. 5. The presence of Al,Et,Cl, leads to greatly enhanced rates of copolymerization, as observed in the copolymeri- zation of methyl acrylate and butadiene.l An approximately linear relation is followed for [Al,Et,Cl,], between 0.025 and 0.2 mol dm-, when [MA], = 0.8 mol drn-,.C. H.BAMFORD AND XIAO-ZU HAN 873 0 1 2 3 4 5 [ Mn2 (CO), I i/ 1 (T2 rnol&drn-# FIG. 4.-Dependence of rate of copolymerization, w, on [Mn,(CO),,]~. Initial concentrations/mol dm-, : MA, Ip 0.8, Al,Et,Cl, 0.2, CC1, 0.1 ; 25 OC, L = 436 nm. 25 2 0 - 'Y) IE 15 rn a z 10 - \ 3 5 o 0.1 0.2 0.3 a4 [Al, Et, Cl,]/mol dm-, FIG. 5.-Dependence of o on [Al,Et,Cl,]. Initial concentrations/mol drn-,: MA, Ip 0.8, Mn,(CO),, 2 x CC1,O.l; 25 OC, 1 = 436 nm. In the presence of Al,Et,Cl,, MA and Ip undergo a Diels-Alder reaction (1) forming a mixture of methyl 4-methyl- and 3-methyl-3-cyclohexene- 1 -carboxylates (MMCC)O (95%) COOCH, (1) H3c'Q COOCH3 (5%) H3cx+ ( < H3 c -0- COOCH, 29 FAR 1874 COPOLYMERIZATION OF CONJUGATED DIENES m 'E 2 a - --..u u E E 41 rr .- x reaction time/min FIG. 6.-Variation of yield of MMCC with [Ip],. Initial concentrations/mol dm-3 : MA 0.4, Al,Et,Cl, 0.2, Mn,(CO),, 2 x lo-,, CCl, 0.1; 25 OC, 1 = 436 nm. [Ip],/mol dm-3: 0, 0.8; 0, 0.4; 0, 0.2. Curves calculated from eqn (24, (4) and (5). 0 60 120 180 reaction time/min FIG. 7.-Variation of copolymer yield with [Ip],. Conditions as for fig. 6. Ap represents a unit of methyl acrylate in the copolymer. This reaction has a relatively high rate so that it is difficult to disentangle the kinetics of the competing copolymerization.Fig. 6-8 show the progress in time of the two types of reaction. For a given reaction time, the yield of MMCC increases with [Ip], at constant [MA], (fig. 6) and increases with [MA], at constant [Ip], (fig. 8). These results are consistent with the reportlo that the rate of the adduct formation is proportional to [MA---all [Ip]. On the other hand, for a given reaction time, the yield of copolymer falls with increasing [Ip], at constant [MA], under the conditions of fig. 7. We believe this is a result of the rapid consumption of MA at the higher Ip concentrations by reaction (1). Thus, in the experiments of fig. 7, for a fixed reaction time [MA] falls below [all, to an extent which is greater for greater [Ip],; consequently, if the rate of copolymer formation is proportional to [MA---all [Ip], we should expect the co- polymer yield to decrease as [Ip], increases.Under the conditions of fig. 8, the copolymer yield at constant [Ip], increases rather sharply with [MA],. This situation arises because the lower [MA], (0.4 mol dm-3) is initially just equal to [all and falls below the latter as the reaction proceeds, while with [MA], = 0.8 mol dmb3 the methyl acrylate unreacted exceeds [all for most of the reaction. Consequently under the latter conditions [MA- - -all, and hence the rate of copolymerization, is always higher.C. H. BAMFORD A N D XIAO-ZU HAN 875 -.. 0.6 . 0.5 - 'E -0 0.1 1 /-* 60 120 1 80 reaction time/min 0.2 * I€ a 0 0.1 E - 1 c-.l Y 4 0 FIG. 8.-Variation of yields of MMCC and copolymer with [MA],.Initial concentrations/mol drn-,: Ip 0.8, AI,Et,CI, 0.2, Mn,(CO),, 2 x lo-,, CCl, 0.1; 25 OC, 1 = 436 nm. [MA],/mol dm-3: 0, 0.8; 0, 0.4. Fig. 6-8 demonstrate that the conversion of monomers into MMCC and copolymer can approach 100%. The dependence of the rate of copolymerization on the initiai monomer concen- trations for relatively short reaction times (10 min) is presented in fig. 9 and 10. According to fig. 9, w passes through a maximum for rather low [Ip],, then decreases monotonically with increasing [Ip],. An interpretation of the latter (decreasing) part of the curve has already been given, namely the progressive depletion of MA with increasing [Ip], attributable to the rapid Diels-Alder adduction (1). It seems clear from 2 ) 0 0.8 1.6 2 A [ Ip], /mol dm-' 3.2 FIG.9.-Dependence of o on [Ip],. Initial concentrations/mol drn-,: MA 0.8, AI,Et,CI, 0.2, Mn,(CO),, 2 x lo-,, CCI, 0.1 ; 25 OC, 1 = 436 nm. 29-2876 COPOLYMERIZATION OF CONJUGATED DIENES 10 0 0.8 1.6 2.4 3.2 4.0 4.8 [MA],/mol dmW3 FIG. 10.-Dependence of o on [MA],. Initial concentrations/mol drn-,: Ip 0.8, AI,Et,Cl, 0.2, Mn,(CO),, 2 x lo-,, CCl, 0.1; 25 OC, 1 = 436 nm. The dashed curve represents cyclopolyisoprene formation. this result that the rate of copolymerization shows a lower dependence on [Ip] than does reaction (1) (unity); over the range we are considering the order is probably zero. If the copolymerization is similar mechanistically to the methyl acrylate/styrene and methyl acrylate/butadiene reactions we should expect that at sufficiently low [Ip] the rate would be proportional to [Ip]: this is apparent in fig.9. The existence of the maximum in co is therefore understandable. Fig. 10 shows that at high [MA],, the rate of copolymerization is effectively independent of [MA],. Under these conditions the concentration of MA is sufficiently high to maintain complete complexation with A1,Et3C13 throughout reaction. This finding is therefore consistent with the view that co is proportional to [MA---all but is independent of the free methyl acrylate concentration. At lower [MA],, when complete complexation cannot be maintained throughout the reaction, mainly on account of the depletion of methyl acrylate by reaction (l), the rate falls off (fig. 10). The latter figure also demonstrates another complication, namely the existence of a fast reaction at very low (or zero) [MA].A white powdery solid product is formed which, from the n.m.r. spectrum, appears to be cyclized polyisoprene.ll9 l2 This reaction does not seem to be significant for [MA], > 0.1 mol dm-3; below this concentration some cyclized polyisoprene may be formed as well as the alternating copolymer, We have concluded above that the results presented are qualitatively consistent with the rate expressions below, = k,[MA- - -all [Ip] d[MMCC] dt which are essentially the same as those applicable to the methyl acrylate/butadiene system. In eqn (2b) Ap, Bp represent methyl acrylate and isoprene units, respectively, in the copolymer.C. H. BAMFORD AND XIAO-ZU HAN 877 We have attempted to quantify the kinetic treatment by applying the integrated forms of eqn (2a) and (2b) to the results in fig.6 and 7. In the previous paper it was pointed out that the aluminium derivative may complex with the copolymer and the Diels-Alder adduct as well with methyl acrylate; in the latter case complexation is strongest, but no reliable values for any of the equilibrium constants exist at present. It is therefore difficult to estimate [MA- - -all reliably. We thought that calculations could be carried out most usefully for the case [MA], < [all since in these circumstances no recycling of aluminium to MA from other species can occur, so that knowledge of the equilibrium constants is not required provided MA and the aluminium derivative are strongly complexed. According to the kinetic treatment set out in the previous paper [see eqn (7a)], k in eqn (2b) is equal to kbakrib 94 (MA = A, Ip = B), where 9, the rate of initiation, is given by13 kd being the first-order rate coefficient for photodecomposition of manganese carbonyl.Thus we obtain from eqn (2b) 9 = 2kd[Mn,(CO)10] = 2kd[Mn2(CO)1010 exp ( - kd t> (3) dm = 1/2kba k& k\[Mn2(CO>,,]i[fMA- - -all exp ( -ikd t ) . dt (4) We also have the stoichiometric relations [MA- - -all = [MA- - -allo - [Ap] - [MMCC] [Ip] = [Ip], - [MA- - -all0 + [MA- - -all. ( 5 ) Values of [Ap] and [MMCC] have been computed from eqn (2a), (4) and ( 5 ) for a series of starting conditions. The parameters used are shown below: k, = 2.86 x mol-l dm3 s-l kd = 5.17 X 10-4S-' (6) k,,k& = 0.366 m o l t dmt s t .Of these, kd was evaluated from eqn (3), the rate of initiation being determined by calibration with methyl mechacrylate as described in the previous paper. The other parameters were obtained from the initial slopes of the curves for [MA], = [Ip], = 0.4 mol dm+ in fig. 6 and 7. There is a reasonable degree of agreement between the calculated and observed data which strengthens our confidence in kinetic relations of the form shown in eqn (2a) and (2b) and the kinetic mechanism proposed for the alternating copolymerization. CHAIN-TRANSFER AND VALUES OF REACTION PARAMETERS Chain-transfer in alternating copolymerization is of mechanistic interest on account of the information it provides about the character of the propagating species. As a rule, transfer to active agents such as halides and mercaptans occurs much less readily in alternating copolymerizations in the presence of Lewis acids than would be expected from data on homopolymerizations.For example, in the methyl acrylate/styrene system,14 the rate coefficient for transfer to CBr, has a value 441 mol-1 dm3 s-l, ca. 30-fold less than that for the (uncomplexed) radical wv MA-St * . The difference has been attributed to complexation of the radical with the Lewis acid. Results for methyl acrylate/isoprene are presented in table 1. The presence of carbon tetrabromide leads to copolymers of lower molecular weight but has little effect on878 COPOLYMERIZATION OF CONJUGATED DIENES TABLE 1 .-CHAIN-TRANSFER TO CARBON TETRABROMIDE IN THE COPOLYMERIZATION OF METHYL ACRYLATE AND ISOPRENE AT 25 O C Concentrations/mol dm-3: MA 0.8, Ip 0.4, Al,Et,Cl, 0.2, Mn,(CO),, 2 x lo-,, CCl, 0.1 ; 10 min, A = 436 nm, 9 = 1.57 x lo-' mol dm-, s-l.0 2 5 10 20 1.37 1.34 1.33 1.31 1.25 1380 1205 960 758 566 I 1 1 I L 0 4 0 12 16 2 0 [CBr, 1 / 1 (r4 mol dm-3 FIG. 1 1 .-Plot of 1 /E against [CBr,]. Conditions as in table 1. the rate of copolymerization. The plot of l / E against [CBr,] is satisfactorily linear (fig. 1 l), with a slope 0.525 mol-1 dm3. Following the procedure outlined in the previous paper' [see eqn (lo)] we find kfb - 0.384. The rate data in table 1 allow kbak;&, to be evaluated as 0.43 mol3 dmi sf [see ref. (l)], hence we obtain from eqn (8) -- kba k,, k<tb = 0.17 molf dmi s-4. (9) Thus this parameter has closely similar values in the alternating copolymerizations of methyl acrylate with styrene, butadiene and isoprene.The value for kbak;jb quoted above is different from that in eqn (6); in theC . H. BAMFORD AND XIAO-ZU HAN 879 experiments of table 1 the ratio [MA],/[al] was greater than that in the runs of fig. 6 and 7, so possibly the difference implies incomplete complexing in the latter. No estimates of ktbb are available, but it seems unlikely that this is less than the corresponding coefficient in the methyl acrylate/styrene system, uiz. 6 x los mol-1 dm3 s-l. Assuming this figure, we find kb, x 1053 mol-l dm3 s-l. From the data of Morton et aZ.15 we may estimate that, in the homopolymerization of isoprene at 25 O C , k , (i.e. kbb) x 9.3 rno1-l dm3 s-l; hence, with the aid of the reactivity ratio data of Ida et alls (assuming that their values at 50 O C are not very different from those holding at 25 "C) we obtain kba x 12 mol-l dm3 s-l (without Lewis acids).As with the corresponding styrene and butadiene systems the great enhancement in k,, brought about by the Lewis acid is evident. In a similar fashion we find that kab x 4933 mol-1 dm3 s-l; analogy with the MA/St system suggests that when Lewis acid is present k,, may have a larger value than this. Both cross-propagation coefficients are therefore sufficiently large to ensure effective alternation at normal monomer concentrations in the presence of Lewis acid. In conclusion, the kinetic data on all three systems MA/St, MA/Bd and MA/Ip are consistent with the cross-propagation mechanism of alternation.If no Lewis acid is present the propagation rn B' +A (B = hydrocarbon monomer) is relatively slow and effective alternation does not occur. We believe the enhancement in the rate coefficient of this process, which is the crucial feature in achieving alternation, arises essentially from the weakening of the carbon-carbon double bond which accompanies complexation of monomer A with Lewis acid. We are indebted to the Government of the People's Republic of China for financial support for one of us (X-z.H.). C. H. Bamford and X-Z. Han, J . Chem. Soc., Faraday Trans. 1, 1982, 78, 855. C. H. Bamford, S. N. Basahel and P. J. Malley, Pure Appl. Chem., 1980, 52, 1837. U.S. Patent, 3,840,449, 1974 (to J. Furukawa, E. Kobayashi, Y. Iseda and T. Yukuta). A. Akimoto and M. Ohtsuru, J . Polym. Sci., Polym. Chem. Ed., 1975, 13, 549. C. H. Bamford and S. U. Mullik, Polymer, 1973, 14, 38. 0. Eizo, T. Akiji and S. Tadamasa, Kobunshi Kagaku, 1973, 30, 22. W. S. Richardson and A. Sacher, J. Polym. Sci., 1953, 10, 353. J. C. Binder and H. C. Ranshaw, Anal. Chem., 1957, 29, 503. T. Inukai and T. Kojima, J . Org. Chem., 1966, 31, 1121. lo T. Inukai and T. Kojima, J. Org. Chem., 1967, 32, 872. l1 M. A. Golub and J. Heller, Tetrahedron Lett., 1963, 2137. l 3 C. H. Bamford in Reactivity, Mechanism and Structure in Polymer Chemistry, ed. A. D. Jenkins and l4 C. H. Bamford and P. J. Malley, J . Polym. Sci., Polym. Lett. Ed., 1981, 19, 239. l5 M. Morton, P. P. Salatiello and H. Landfield, J. Polym. Sci., 1952, 8, 279. l6 F. Ida, K. Uemura and S. Abe, Kagaku To Kogyo (Osaka), 1965,39, 565; cJ C.A. 1966,64, 3695a. I. Kossler, M. Stolka and K. Mach, J. Polym. Sci., Part C, 1964, 4, 977. A. Ledwith (John Wiley, London, 1974), chap. 3. (PAPER 1/645)
ISSN:0300-9599
DOI:10.1039/F19827800869
出版商:RSC
年代:1982
数据来源: RSC
|
20. |
Effects of added acetonitrile on the heat capacities of activation for the solvolysis of simple organic esters in water |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 78,
Issue 3,
1982,
Page 881-886
Michael J. Blandamer,
Preview
|
PDF (399KB)
|
|
摘要:
J. Chem. Soc., Faraday Trans. I , 1982, 78, 881-886 Effects of Added Acetonitrile on the Heat Capacities of Activation for the Solvolysis of Simple Organic Esters in Water BY MICHAEL J. BLANDAMER,* JOHN BURGESS AND PHILIP P. DUCE Department of Chemistry, The University, Leicester LE1 7RH AND Ross E. ROBERTSON Department of Chemistry, University of Calgary, Calgary, Alberta, Canada AND JOHN W. M. SCOTT Department of Chemistry, Memorial University, St. John’s, Newfoundland, Canada Received 23rd April, 1981 Kinetic data for the solvolysis of a range of simple organic esters in water and in water + acetonitrile mixtures are examined in terms of the two-stage Albery-Robinson mechanism. The analysis is directed at understanding the dependence of previously reported heat capacities of activation on the composition of solvent.This dependence is shown to stem from a dependence on composition of the kinetic parameters describing the two-stage mechanism and on the variation of the temperature range over which the kinetic data were measured. For many years, Robertson1y2 has argued that the large negative heat capacities for the solvolysis of organic esters arise to a large extent from the enhanced water structure surrounding the hydrophobic initial state. It was also noted that addition of small amounts of monohydric alcohols or tetrahydrofuran produces a more negative value of AC$ for the solvolysis of t-butyl chloride, whereas addition of acetonitrile produces2 an increase in ACZ (i.e. less negative). The latter trend is consistent with the widely accepted hypothesis that added acetonitrile disrupts the hydrogen-bonded structure of waters3 Indeed this trend in AC$ with increase in mole fraction x, of acetonitrile has been used to identify the contribution to AC$ from reorganisation of water structure on activation from other solvolytic Although these generalisations rationalise the observed trends, there remain several problems which have often been overlooked.s Thus the rate constant and enthalpy of activation for t-butyl chloride decrease when all organic cosolvents are added. It is not therefore immediately apparent why the solvent dependence of AC$ is so discriminating between cosolvents.Various hypotheses may be advanced in this context. A more fundamental reassessment questions the significance of the heat capacity of activation for solvolysis in water, i.e.the starting point of the argument outlined above. More recent analyses of the kinetic data for this class of reactions7* * have prompted this reconsideration of the significance of the calculated ACZ. In particular, attention has been directed to the Albery-Robinson mechanism for the solvolysis of t-butyl chlorides [eqn (1)-(3)] : k, k, kp RX R+X- --+ products (1) 88 1882 SOLVOLYSIS OF ORGANIC ESTERS I N WATER where and k(obs) = kJ( 1 +a) a = k,/k3. If AH,Z and AH$ are independent of temperature, the dependence of k(obs) on temperature leads to an apparent heat capacity of activation AC,Z(app) which is relatedg to a and AAH# (= AH$ -AHt) by eqn (4): a (AAH#), (1+a), RT2 ' AC$(app) = -- (4) The ratio a, depending on the sign of AAHf, increases or decreases with increase in temperature, passing through unity at a temperature T,.However, AC$(app) is always negative and depends on temperature such that the final plot has an inverted bell shape with a minimum at a temperature near Ta. The foregoing mechanism and associated equations raise the question as to the possible re-interpretation of the effect of added acetonitrile on the kinetics of solvolysis. We have examined the data for six esters, and compared the outcome with a similar analysis for a range of different added solvents on the kinetics of solvolysis of t-butyl chloride.s ANALYSIS The kinetic data describing the dependence of rate constants on temperature have been fitteds to eqn (5) where a, is related to AHf and a, to AAHf : k = a,exp(-a,/T)/[l +a3exp(-a,/T)].( 5 ) The analysis used a modified Gauss-Newton technique based on explicit calculation of the relevant Hessian and Jacobian matrices.8 The computer program (FORTRAN) calculated the dependences of k,, a and ACg(app) on temperature. RESULTS The outcome of the analysis is summarised in table 1. In all cases eqn (5) satisfactorily fitted the data to within experimental error, a plot of the residuals { = 100 x [k(obs) - k(calc)]/k(calc)) against temperature showing a satisfactory scatter. Table 1 reports a number of quantities of interest. An important entry concerns the temperature range, TR, over which the kinetic data were measured. The temperature T, is an experimental temperature close to the mean temperature of the experimental range.Thus a(T,) is the calculated value of a at this temperature. AC,Z(app, T,) is the heat capacity of activation calculated using eqn (4) at T,, where the activation enthalpy AH# (app, T,) is the corresponding enthalpy of activation. Included in table 1 is the temperature at which a is unity [T(a = l)] and the temperature at which ACZ(app) is a minimum, together with this value of AC,Z(app). The dependence of AC$(app) on temperature is shown in fig. 1 for benzyl chloride in waterlo and in water + acetonitrile mixture where x, = 0.05. The plots for the other systems reported in table 1 are similar in their general appearance. DISCUSSION Before embarking on a detailed discussion of the results, we make two important points which clarify several features of the analysis.First, the values of AC,f reported by Robertson and coworkers were obtained1 by fitting the data to the three-constantTABLE EFFECT OF ADDED ACETONITRILE ON THE KINETIC PARAMETERS DESCRIBING THE SOLVOLYSIS OF VARIOUS SIMPLE ORGANIC ESTERS IN WATER A c t 290K) (app, Tm) A H # AC$ bPP9 AC$ number temp. AH? @PP, max) mole fraction of of data range T, k(obs, T , ) k,(T,) (298 K) AAH# T(a = 1) /J at T /J /J. (~PP, Tm) acetonitrile, x2 points /K /K s-' a(T,) /10-8 s-* /kJ mol-I /kJ mol-l /K mol-I K-l /K mol-1 K-' mol-l K-I /kJ mol-l t-butyl chloride8 0 20 274-293 0.05 40 273293 0.10 44 275-297 0.20 50 286-313 benzyl chloridelo- Is 0 51 288-338 0.048 44 315-335 0.177 40 324-346 0 39 324-344 0.084 45 316-345 0.084 34 321-337 pchlorobenzyl I4 p-methylbenzenesulphonyl c h l ~ r i d e ' ~ ~ l5 0 34 278-296 0.05 34 278-300 p-methoxybenzenesulphonyl I5 0 41 275-293 0.048 41 274-300 0 9 273291 0.2 (ethyl alcohol) 38 283-315 1 -adamantyl nitrate'" 284 283 287 300 41.45 19.98 12.96 9.542 0.13 02.27 0.10 8.33 4.68 6.55 1.43 8.90 104.6 109.4 89.1 117.1 -47.2 -28.9 -37.9 - 34.4 316.7 265.4 336.0 260.5 - 672 - 364 -391 - 531 315 210 330 255 -431 - 222 - 192 - 234 - 348 - 265 - 173 -150 99.2 89.3 85.7 86.4 365 220 340 280 - 56.7 -51.4 - 122 -691 - 120 -22.7 - 30.7 - 124.2 89.88 83.84 84.31 89.49 313 326 329 333 0.7663 0.8668 0.5334 0.1571 0.106 0.456 12.93 15.23 0.8413 1.197 0.0769 0.252 93.10 99.71 95.53 131.3 -33.71 - 17.09 - 35.86 -44.56 378.9 23 1.7 350.2 285.0 - 246 - 172 - 323 - 740 d fi m w 0 a m P 3.213 0.3269 0.089 0.220 0.350 0.398 91.12 95.64 - 51.25 -43.4 383.0 365.9 - 545 -431 375 360 -21.2 -61.4 -214 - 306 86.93 87.81 333 33 1 4 0 z m - 244 - 159 288 289 13.35 7.625 5.814 9.255 9.107 7.845 106.1 97.59 - 38.5 - 36.2 259.6 251.9 - 667 - 628 255 250 - 269 - 166 73.25 64.9 1 9.48 26.30 98.09 96.04 - 29.6 -31.6 250.0 227.0 -430 - 589 245 225 - 133 -357 - 171 -44.2 73.06 65.42 284 286 14.59 8.331 5.480 3 1.23 m 283 298 13.69 2.002 0.277 5.18 1.745 1.22 126.7 127.1 -41.9 -33.2 305 266 - 575 -477 300 260 - 524 - 270 -448 - 202 117.6 99.2 00 00 w884 SOLVOLYSIS OF ORGANIC ESTERS IN WATER 100 200 300 400 500 600 T/K I I I 1 I 100 200 300 400 SO0 600 TIK -20 - I -60 - 0 E a -100 h 1 h 4 I;p W 1L -140 FIG.1.-Calculated dependence of AC$(app) on temperature for solvolysis of benzyl chloride in (A) water and (B) water + acetonitrile, mole fraction = 0.048.Valentiner equation" which assumes at the outset that AC; is independent of temperature. Although this equation has, it is now realised,12 several unsatisfactory features, the dependence of AC; on composition forms the background of the problems discussed here. Consequently, we designate these previously reported values as AC$ (V). Secondly, it is convenient to examine the relationship between the range, TR, and the dependence of AC,Z on temperature required by eqn (5). This is illustrated in fig. 2 where, rather than indicating the range, we have indicated five possible values of T,. In a given investigation the range will extend to differing amounts about these mid-points.Part of our argument is that AC,f(V), as previously reported, is some averaged quantity over the sampled section of the dependence described by eqn (5). Suppose, therefore, that throughout a series of investigations TR remains the same. Although the temperature T* corresponding to the minimum in AC,f and the value of AC$ at the minimum depends on the solvent and substrate, we can consider what happens to AC$(V) as T* changes relative to Tm. Therefore, if by adding acetonitrile, T, moves from (a) to (b) or (c) (fig. 2), AC$(V) becomes more negative. Similarly, if Tm moves from (b) or ( d ) to (e), AC$ becomes less negative. Of course, a dramatic decrease in T* together with an increase in T, might lead to a small [i.e. ( d ) -+ (e)] or negligible [ie.(a) --* (e)] change in ACZ(V). With these points in mind, we turn to a consideration of the results in table 1. The pattern is set by the data for t-butyl chloride.* For solvolysis in water the experimental temperature range was just below T(a = 1) and T* [(b) in fig. 23 such that AC$(V) is large and negative. When acetonitrile is added x, = 0.05, T(a = 1) and T* move to below the experimental range [(d) in fig. 21 and so ACZ(V) increases (i.e. becomes less negative). The increase is marked because at the same time the value of AC #(app) at the maximum is almost halved, a consequence of a dramatic fall in AAH? As more acetonitrile is added, T(a = 1) and T* move above and then below the experimental range, but the minimum value of ACZ (app) increases. Consequently AC,Z(app) at Tm increases gradually.On the other hand, AC$(app) at a common temperature, 290 K, passes through a maximum. A similar trend is observed for benzyl chloride. For reaction in water, TR is slightly further away from T* and T(a = 1) than for t-butyl chloride. Since ACf(app, max) is not so negative, these two effects combine to yield a smaller value for lAC$(V)l.BLANDAMER, BURGESS, DUCE, ROBERTSON A N D SCOTT 885 When x, = 0.048, T, has increased but T* and T(a = 1) have dropped to below TR with the result that ACZ increases. With an increase in x,, T, increases but T* and T(a = 1) increase and then decrease although AC$(app, max) decreases. The result is a decrease and then increase in AC$(app) at T,. A similar complicated solvent dependence of T(a = l), T,, TR and AC$(app, max) for the other substrates accounts for the trends in AC,Z(app) at T, and hence in AC$(V).0 n a Ii- L!! Q \- FIG. 2.-Diagramatic representation of the relationship between the mean of the experimental temperature range and the dependence of AC,Z (app) on temperature. The data in table 1 also include the effect of added ethyl alcohol on the kinetic data for 1 -adamantyl nitrate. For reaction in water, ACZ(V) is strikingly more negative than that for t-butyl chloride. This is the result of the upper limit of TR being close to T(a = 1) and T*, even though IAC$(app)( at T* is not as large as that for t-butyl chloride. Ethyl alcohol at x, = 0.2 also exerts a structure-breaking influence on water.3 The dramatic decrease in ACf(app) at Tm [and hence AC$(V)] is attributed to a decrease in T(a = 1) and T*, which are now below the lower end of TR.The major conclusion which follows from this analysis is that the changes in AC$(V) with added acetonitrile are not as straightforward as hitherto discussed. The determining factors are the solvent sensitivity of the ratio a and the related enthalpy term AAHf:. There appears to be a general trend for a to increase with increase in mole fraction of acetonitrile, indicating that recombination becomes increasingly favoured as the kinetic fate of the intermediate. Unfortunately, a is the ratio of the two rate constants k , and k3 so we have no indication of their absolute changes. Similar problems emerge in understanding the solvent sensitivity of AAH # .It is possible to understand these trends by combining possible effects of the acetonitrile on the water structure and hence on the solvation characteristics together with the gradual change in dielectric properties of the solvent. However, these arguments turn out to be qualitative at best and sweeping rationalisations at worst. What is required is a more detailed examination of solvent effects along the lines indicated here for a wider range of substrates and some way of splitting a into the component rate constants, k , and k3. We thank the S.R.C. for a grant to P.P.D.886 SOLVOLYSIS OF ORGANIC ESTERS IN WATER R. E. Robertson, Prog. Phys. Org. Chem., 1967, 4, 213. M. J. Blandamer, Adv. Phys. Org. Chem., 1977, 14, 203. E. C. F. KO and R. E. Robertson, Can. J. Chem., 1972, 50,946. K. M. Koshy, R. K. Mohanty and R. E. Robertson, Can. J. Chem., 1977,55, 1314. M. J. Blandamer, R. E. Robertson, J. M. W. Scott and A. Vrielink, J. Am. Chem. SOC., 1980, 102, 2585. M. J. Blandamer, J. Burgess, R. E. Robertson and J. M. W. Scott, J. Chem. SOC., Faraday Trans. I , submitted for publication. * M. J. Blandamer, J. Burgess, P. P. Duce, R. E. Robertson and J. M. W. Scott, J. Chem. SOC., Faraday Trans. I , 1981, 77, 1999. W. H. Albery and B. H. Robinson, Trans. Faraday SOC., 1969, 65, 980. * R. E. Robertson and S. E. Sugamori, Can. J. Chem., 1972,50, 1353. lo R. E. Robertson and J. M. W. Scott, J. Chem. SOC., 1961, 1596. l1 S. Valentiner, Z. Phys. Chem., 1907, 44, 253. l2 M. J. Blandamer, R. E. Robertson and J. M. W. Scott, Can. J. Chem., 1980, 58, 772. l3 R. E. Robertson, unpublished data. l4 K. M. Koshy, R. E. Robertson and W. M. J. Strachan, Can. J. Chem., 1973, 51, 2958. R. E. Robertson and B. Rossall, Can. J. Chem., 1971,49, 1441. K. M. Koshy, R. K. Mohanty and R. E. Robertson, Can. J. Chem., 1977,55, 1314. (PAPER 1/654)
ISSN:0300-9599
DOI:10.1039/F19827800881
出版商:RSC
年代:1982
数据来源: RSC
|
|