J . Chem. SOC., Faraday Trans. I , 1982, 78, 2239-2249 Acetonitrile on Silica-Magnesia Mixed Oxides Temperature-programmed Desorption and Infrared Study BY GABRIELE RITTER.* HEINRICH NOLLER AND JOHANNES A. LERCHER Institut fur Physikalische Chemie, Technische Universitat Wien, Getreidemarkt 9, A- 1060 Wien, Austria Received 7th October, 193 1 The adsorption of acetonitrile on silica, silica-magnesia mixed oxides and magnesia was studied by means of temperature-programmed desorption (t.p.d.) and infrared spectroscopy. With t.p.d. three types of desorption behaviour were distinguished. On silica, only one comparatively narrow peak was detected (type 1). the mixed oxides containing up to 50mol % magnesia showed one broad peak with pronounced high-temperature tailing (type 2) and the mixed oxides containing 60-90 mol % magnesia showed up to 4 peaks (type 3).The i.r. results gave evidence for at least 3 adsorption structures: acetonitrile bound to surface hydroxyl groups, acetonitrile bound to cations and a surface carboxamide. Assignments of desorption peaks to the different adsorption structures of acetonitrile are suggested. Silica-magnesia mixed oxides have been studied by several authors, with special attention being paid to acid-base properties and related catalytic behaviour. The acidity was attributed to mixed-oxide phases, while the basic properties were related to the magnesia ~ 0 n t e n t . l . ~ Our aim was to investigate the change in the acid-base behaviour over the whole range of composition from silica to magnesia, whereas most of the former studies have dealt with only a few mixed oxides.Acetonitrile was used as a test molecule since it showed a number of well resolved peaks in the thermal desorption spectra (magne~ia/acetonitrile).~ Both e.p.a. (electron- pair acceptor) and e.p.d. (electron-pair donor) sites were involved and the adsorption structures seemed to be very sensitive to pretreatment conditions and hydration states on the surface. Although acetonitrile is a small molecule and should therefore be suitable for studying interactions with the surface, there are not many papers dealing with the adsorption of acetonitrile on oxide surface^.^-^ Most authors found that acetonitrile cannot be used for the characterization of surface properties because of its high reactivity.However, even if acetonitrile is not suitable for the characterization of surface properties, the interactions of nitriles with oxide surfaces of various acidities and basicities should be of potential interest with respect to possible ring-closure reactions and the formation of heterocyclic compounds.lo9 l1 EXPERIMENTAL MATERIALS To obtain the silica-magnesia mixed oxides, Aerosil (Fluka) and magnesium hydroxide (Fluka purum p.a.) were stirred in water at 90 O C for 17 h, centrifuged and dried to constant weight at 120 "C. The samples covered the range 10-90 mol % magnesia. The B.E.T. surface areas are listed in table 1. 22392240 ACETONITRILE O N SILICA-MAGNESIA MIXED OXIDES TABLE 1 .-ACIDITY ( H , d + 3.3) AND B.E.T. SURFACE AREAS OF SILICA-MAGNESIA MIXED OXIDES acidity mol % surface area ( H , d +3.3) sample magnesia /m2 g-I /mmol m-I silica I I1 I11 IV V VI VII VIII IX magnesia 0 10 20 30 40 50 60 70 80 90 100 330 190 240 320 340 190 150 130 143 143 81 0.0085 0.01 15 0.0131 0.01 53 0.0027 0.01 17 0.0 148 0.0075 0.00 16" a Colour change not very clear SiO, 20 40 60 80 MgO mol 70 FIG.1 .-Acidity of silica-magnesia mixed oxides (H, < + 3.3) determined by titration method with n-butylamine. Debye-Scherrer diagrams taken from samples calcined at 650 OC (3 h) showed forsterite besides magnesia in the range 60-90 mol % magnesia, whereas another solid phase, probably talc, was detected in addition to silica in the range 10-50 mol%. The acidity of the samples pretreated at 650 "C for 3 h was measured by titration with n-butylamine according to Benesi using 4-dimethylaminoazobenzene (pK, = + 3.3).The acidity values obtained this way are listed in table 1 and shown in fig. 1. Acetonitrile (Merck Uvasol) was used without any further purification.G. RITTER, H. NOLLER A N D J. A. LERCHER 224 1 APPARATUS TE M PER AT U RE-P ROGR A M ME D DESOR PT ION T.p.d. was carried out in U ~ C U O (ca. 1 Pa) with a heating rate of 10 "C min-' in the range 30-750 "C. The reactor was a quartz glass tube (10 mm in diameter), which was connected to a vacuum pump and to a quadrupole mass spectrometer for detecting the desorbed species (m/e = 1-100). The mass spectrometer and the t.p.d. furnace were coupled to a process com- puter IBM S/7. For further details see ref. (5) and (12). INFRARED MEASUREMENTS The i.r.cell was constructed according to the description of Knozinger et PROCEDURE TEMPER AT U RE-P RO G R A M M E D D E S OR P T I ON The sample (100 mg) was calcined in the t.p.d. reactor at 650 OC, cooled to room temperature and contacted with 10 mm3 acetonitrile for 15 min. The sample was then evacuated at room temperature for 30-60 min, before t.p.d. was started. INFRARED MEASUREMENTS The hydroxides were pressed into thin self-supporting wafers (p = lo8 Pa), which were examined using a transmission technique. Thermal pretreatment of the samples (500 "C for 2 h) was carried out in the reactor before adsorption of acetonitrile. The spectra were taken at room temperature and recorded with a Perkin-Elmer (type 325) grating spectrograph in the range 4000-1000 crn-l.The scanning speed was 0.5 cm-l s-' using slit programme 7, which corres- ponded to a resolution of 3 cm-' at 3600 cm-l. RESULTS AND DISCUSSION TEMPERATURE-PROGRAMMED DESORPTION The temperatures of the desorption maxima of acetonitrile are listed in table 2 and the t.p.d. spectra are shown in fig. 2. The maximum intensities of the t.p.d. curves were adjusted to 1000 mV (by multiplying the intensity values of the curves by a suitable TABLE 2.-DESORPTION OF ACETONITRILE FROM SILICA-MAGNESIA IN THE RANGE 30-750 O C (1 0 OC min-l). TEMPERATURES OF T.P.D. MAXIMA AND MULTIPLICATION FACTOR APPLIED TO INTENSITY VALUES OF T.P.D. CURVES IN FIG. 2. desorption maxima/OC mult. sample peak I peak 2 peak 3 peak 4 factor silica I I1 I11 IV V VI VII VIII IX magnesia 67 76 81 90 135 107 86 78 90 78 70 2.40 1.73 0.8 1 - 1.32 4.35 - 2.49 1.82 1.26 164 260 0.74 I56 247 - 150 244 345 1.38 144 240 326 0.52 - - - - - - - - - - - - ._ - - - - - 150-260 -2242 n 4, ACETONITRILE ON SI LI C A-M A GNE SI A MIXED OX IDES - E 0 - 9 h W P L 0 NG.RITTER, H. NOLLER AND J. A. LERCHER 2243 factor given in table 2) in order to demonstrate how the shape of the curves varied with the composition of the sample. Three types of desorption behaviour may be distinguished : A comparatively narrow peak with its maximum below 100 O C appeared on silica (type 1). The mixed oxides I-V, on the other hand, showed broad peaks with tailing on the high-temperature side (h.t. tailing) (type 2). The temperature of the maximum increased in the order silica, I, 11,111, V, with IV being considered separately because of its exceptional behaviour.Its maximum appeared at the highest temperature and its intensity was the smallest in spite of its high surface area (340 m2 g-l), and the shape of its t.p.d. curve is more symmetrical than those of I, 11, I11 and V. T/"C FIG. 3.-T.p.d. spectrum of magnesia/acetonitrile. (- - -) T.p.d. spectrum without adsorption of water before adsorption of acetonitrile (m/e 41 ; 100% = 2000 mV); (-) t.p.d. spectrum with adsorption of 1 mm3 water before adsorption of acetonitrile (m/e 41 ; 100% = 500 mV). VII-IX and magnesia showed up to 4 peaks (type 3), which are subsequently referred to as peaks 1-4 (table 2). The Debye-Scherrer pattern of VI was similar to those of VII, VIII and IX (magnesia and forsterite), therefore VI was expected to show desorption behaviour of type 3.However, only one well developed peak was observed, although there were slight indications of others. In all cases, only acetonitrile and water were found to be desorbed. With magnesia, t.p.d. was carried out twice with the same sample without further activation between the runs. The t.p.d. spectra were practically identical. So it seems unlikely that any product had been formed during the first run and had remained on the surface. When water (1 mm3) was adsorbed on magnesia before adsorption of acetonitrile, peak 1 was more pronounced than in the case without preadsorption of water (fig. 3). Note that the intensities of peaks 2 and 4 were greatly reduced whereas that of peak 1 was visibly enhanced.The largest relative increase was observed for peak 3.- - I 1 1 I 1 0 , l I l l l l l l l l l r l l l l l r l l . ~ ~ ~ ~ _ _ ~ I N F R A R E D MEASUREMENTS 1.r. spectra are shown in fig. 4-6. S U R F A C E H Y D R O X Y L G R O U P S O F T H E O X I D E S Pure silica and magnesia showed one band (at 3740 cm-l) in the OH stretching region after evacuation (10-l Pa) at 500 O C for 1 h. This band was attributed to the valence vibration of free hydroxyl groups. l4G. RITTER, H. NOLLER AND J. A. LERCHER 2245 100 80 h $ 60 u E * .- E $ LO * 20 0 I 1 , \ 1 1 I \ I \ I 1 I I I I I LOO0 3500 3000 2500 2000 1800 1600 1400 1200 wavenum ber/cm -' FIG. 6.--I.r. spectra of acetonitrile on magnesia. (-) Adsorption of acetonitrile at 4 x lo3 Pa at room temperature; (- - -) after evacuation (lo-' Pa) at room temperature; ( .. . .) after evacuation (10-l Pa) at 200 OC. TABLE 3 .-C-H VIBRATIONS (cm-l) OF ACETONITRILE ADSORBED ON SILICA-MAGNESIA MIXED OXIDES pressure of acetonitrile, temperature sample 4 x lo3 Pa, 25 "C 10-l Pa, 25 "C 10-l Pa, 200 "C silica 3000, 2950 - - I 3000,2950 (2970, 2950, 2920) (2690, 2920) I11 3000,2960,2940 2940 2950,2920 VII 2995,2960,2940 2960,2920 2960,2920 VIII 2990,2930 2960,2920 - magnesia 2980,2940,2920 2980,2960, 2920 2960,2920 The mixed oxides exhibited two hydroxyl bands (3740 and 3670cm-l) after evacuation (lo-l Pa) at 500 O C (fig. 5). The two bands showed almost equal intensity in samples I, I11 and VII, while the band at 3670 cm-l was much weaker in sample VIII.Since the free hydroxyl groups of silica and magnesia have approximately the same i.r. frequency, we concluded that the band at 3670 cm-l was due to hydroxyl groups of talc or forsterite (mixed-oxide phases). This conclusion is in accordance with the results of Koubowetz et aL9 and Kermarec et al.,15 who found the band at 3670 cm-l to increase with magnesia content. The position of the 3760cm-l band was independent of the composition of the sample. We conclude from this that the environment of that (free) hydroxyl group is similar in both solid phases (talc and forsterite).2246 A C ETON I T R I LE ON S I L I C A-M AG NES I A MI XED OX IDES ADSORPTION OF ACETONITRILE The wavenumbers of CH vibrations and CN vibrations are listed in table 3 and 4.Silica. After adsorption of acetonitrile at 40 mbar* the free hydroxyl band of silica (3740cm-l) was weakened and a perturbed, broad hydroxyl band at 3435cm-l appeared (fig. 4), indicating that hydroxyl groups were involved in the adsorption process . TABLE 4.-C-N VIBRATIONS (cm-l) OF ACETONITRILE ADSORBED ON SILICA-MAGNESIA MIXED OXIDES pressure of acetonitrile, temperature sampie 4 x lo3 Pa, 25 O C 10-1 Pa, 25 O C 10-1 Pa, 200 OC silica 2295, 2261 I 2318, 2290, 2260 23 12,2290,2250 I11 1700, 1445, 1375 2310,2282,2245, (2200) VII 1665, 1445, 1375 VIII 2305,2280,2245 2190,2150 - - 2315, 2290, (2260) - 2310,2282 2315,2285, 1700, 1610 1490, 1447, 1375 2195, 1670, 1590, 1450 1700, 1445, 1370 2200, 2190, 1667, 1445 2190,1640, 1570 - (2305), 2295,2275,2250 magnesia 2150, 1610, 1530, 1390 1375, 1368, 1325, 1190 2180, 1610, 1600, 1570 1530, 1457, 1412, 1390 1378, 1192 2180,2160, 1600, 1530 1414, 1390, 1378, 1358 The CH valence bands of the adsorbed acetonitrile were found at 3000 and 2950 cm-l, similar to the bands in the liquid phase (3000 and 2942 cm-l).Two bands (2295 and 2261 cm-l) were observed in the CN stretching region. The band at 2295 cm-l was attributed to a combination band of the symmetrical CH, deformation vibration and the CC stretching vibration, and that at 2261 cm-l was attributed to the CN stretching vibration.16 The corresponding bands in the liquid phase were found at 2293 and 2254cm-l. These bands are known to shift to higher wavenumbers when acetonitrile interacts with an electron-pair acceptor. The stronger the interaction, the larger the shifts.Therefore the interaction of acetonitrile with the surface hydroxyl groups of silica must be assumed to be only slightly stronger than the interaction of acetonitrile molecules with each other (in the liquid phase). After evacuation at room temperture (10-l Pa) the bands due to acetonitrile disappeared and the original intensity of the free hydroxyl band was restored. The low desorption temperature as well as the slight upward shift of the bands in the CN region indicated that acetonitrile interacted rather weakly with silica. Mixed-oxides Iand III. After evacuation at 500 "C for 1 h, acetonitrile was adsorbed at 40 mbar. Only the band at 3740 cm-l was reduced while that at 3670 cm-l was not affected. This might be explained by the asumption that the hydroxyl groups, which gave rise to the band at 3670 cm-l, were not accessible for acetonitrile molecules.The perturbed hydroxyl bands were found at 3450 cm-l with sample I and at 3440 cm-l * 1 bar = lo5 Pa.G. RITTER, H. NOLLER AND J . A. LERCHER 2247 with sample 111. The smaller shift of these hydroxyl bands (295 and 300crn-', respectively) in comparison with that of silica (305 cm-l) indicated that the acid strength of the mixed oxides is lower than that of silica. On sample I the CH valence bands of acetonitrile were at 3000 and 2950 cm-l, which suggested surface properties similar to those of silica. On sample 111, bands occurred at 3000,2960 and 2940 cm-l, indicating that there must be more than one adsorption structure.The larger shift to lower frequencies (2940 cm-l compared with 2950 cm-l) must be caused by a stronger interaction of the methyl group with the surface oxygen.17 Three bands situated at 2318, 2290 and 2260 cm-l with sample I and at 2312,2290 and 2250cm-l with sample I11 were observed in the region of the CN stretching vibrations. According to Krietenbrink16 vibrations of weakly adsorbed acetonitrile and vibrations of acetonitrile coordinated to cations contribute to the band at 2290 cm-I. The bands at 2318 cm-l (and 2290 cm-l) (I) and the bands at 2312 cm-l (and 2290 cm-l) (111) were attributed to acetonitrile molecules coordinated to cations, while those at 2260cm-l (and 2290cm-l) (I) and 2250cm-l (and 2290cm-l) (111) should be due to acetonitrile adsorbed on hydroxyl groups.The band at 2260 cm-I (2250 cm-l) disappeared after evacuation at room temperature, which again indicates the weak interaction of acetonitrile with surface hydroxyl groups. The downward shift in the frequency of the CN stretching vibrations from sample I to sample I11 indicates weaker interaction with cations and hydroxyl groups, which is interpreted in terms of decreasing acid strength. The bands at ca. 1445 and 1370 cm-l were also due to adsorbed acetonitrile molecules. Mixed-oxides VII and VIII and magnesia. For samples VII and VIII the band at 3740 cm-l (due to magnesia or silica hydroxyl groups) was perturbed after adsorption, while the band at 3670 cm-l remained unperturbed. The i.r. spectra of acetonitrile adsorbed on these samples and especially on magnesia showed several bands in the region of perturbed hydroxyl bands, for which Koubowetz et aZ.9 provided a detailed discussion.The bands of the CH stretching vibrations were shifted to lower wavenumbers with increasing magnesia content (table 3). Since that shift, found on all mixed oxides, seems to be too large for a long-range effect, a direct interaction of the methyl group with surface oxygen was assumed. An analogous conclusion was drawn for the adsorption of acetone on magnesialS and was supported by results of Takezawa and K0baja~hi.l~ The CN vibrations at 2310, 2282 and 2245 cm-l on sample VII and at 2305, 2280 and 2245 cm-l on sample VIII were found only during adsorption of acetonitrile at 4 x lo3 Pa. After evacuation at room temperature these bands disappeared.Compared with samples I and 111, acetonitrile bound via cations has less thermal stability. The reason for the disappearance of the CN vibrations could be due either to a decrease in the acid strength (e.p.a. strength) of the cations or to a transformation of the initial surface structures. The decrease in the e.p.a. strength of the cations was shown by the decrease of the frequency for cationically bound acetonitrile (2310 and 2305 cm-l). Further bands situated between 2150 and 2200 cm-l were observed on these samples. The bands were thermally more stable than the CN bands at 2310 and 2305 cm--l. Their intensity increased with increasing magnesia content. This led to the assumption that they were caused by acetonitrile molecules bound to the surface via oxygen and a cation; such adsorption might be a precursor of an acetimidato anion on the surface.In this case the CN stretching frequency was expected to be lowered. The suggestion of Krietenbrink and Knozinger7t l6 that the band at ca. 2200 cm-l is due to condensation products seems unlikely, since acetonitrile was the only desorption product.2248 ACETONITRILE O N SILICA-MAGNESIA MIXED OXIDES After evacuation the band shifted to 2190 cm-I on samples VII and VIII and to 21 80 cm-I on magnesia. Raising the temperature caused further shifts (table 4) of these bands which persisted up to 400 "C. The formation of precursors of acetamide leads us to expect a CN double bond as well as a CO bond. The band situated at 1667 cm-l (sample VII), 1640 cm-l (sample VIII) and 1610 cm-l (magnesia) could be due to a CN double bond.The stronger the bond between the carbon atom and the surface oxygen is, the weaker the CN bond and the lower its vibration frequency. A similar intermediate structure was found in acetonitrile hydrolysis in aqueous medium.19 Some of the acetonitrile molecules seemed to form carboxamide structures on the surface after evacuation at higher temperatures. The antisymmetrical vibrations of such carboxamide structures were found between 1600 and 1580 cm-l, while there was some uncertainty regarding the symmetrical mode, which should be located between 1510 and 1500 cm-l. Nevertheless, note that a water molecule would be needed to form acetamide and it cannot be provided by the hydroxyl group of the surface.Similarly acetone forms carboxylate structures on oxide surfaces, while the desorption product is acetone.'* CORRELATION OF T.P.D. PEAKS WITH ADSORPTION STRUCTURES SILICA The only adsorption structure which was found in the i.r. spectra was acetonitrile bound to surface hydroxyl groups. Therefore the single t.p.d. peak at 67OC is attributed to the desorption of acetonitrile attached to hydroxyl groups. However, there is some discrepancy. The i.r. bands corresponding to adsorption structures via hydroxyl groups disappeared after evacuation at room temperature while the t.p.d. peak attributed to the same surface species appeared at higher temperatures. This behaviour, however, may be due to the difference in the conditions of the t.p.d. and i.r.experiments, in particular to the vacuum for the i.r. experiment being approximately one order of magnitude better than that of the t.p.d. apparatus. It is possible that this difference between t.p.d. and i.r. results is due to most of the weakly bound species being pumped off in the i.r. experiments. MIXED-OXIDES I, 11, 111, I v AND v With samples I, 11, I11 and V a strongly asymmetric t.p.d. peak was found, the maximum temperature of wh-ich went from 76 to 107 OC. The i.r. data (I and 111) give evidence of acetonitrile being bound via both hydroxyl groups and cations. Instead of two peaks (one for the hydroxyl groups, one for the cation), only one peak, with broad h.t. tailing, appeared. Whatever the reason for such tailing may be, e.g. lateral interactions or distribution of the adsorption strength, centres with higher adsorption strength than that of hydroxyl groups must have been formed.Note that the tailing extends up to > 400 O C , whereas on silica desorption is practically complete by ca. 150 O C . As hydroxyl groups can be ruled out as adsorption sites (shown by the h.t. tailing of the desorption peak) it is very likely that cations are involved in this strong adsorption. On silica the number of these cationic sites is practically zero, because of the high shielding of the small Si4+ cation by four 02-, as discussed in former papers. If in the mixtures, especially those with low magnesia content, Mg2+ is substituted for Si4+, the stoichiometry of the compound is changed, in such a way that some oxygen can be omitted, which would mean that the cations are less shielded and hence more accessible to donor molecules.G.RITTER, H. NOLLER A N D J. A. LERCHER 2249 The desorption maximum attributed to desorption from hydroxyl groups increases in the order I < I1 < I11 < V, although the acidic strength of the surface e.p.a. centres decreases in this order. In our opinion a further interaction must be assumed. We believe that interaction of (basic) surface oxygen with CH, groups of acetonitrile contributes to the adsorption of acetonitrile. This type of interaction will become more important as the magnesia content of the sample is raised. Indeed, the i.r. spectrum reveals interaction with the CH, group, shown by the shift of wavenumber, which increased with increasing content of magnesia. Thus we have to take into account two interactions, one decreasing and the other increasing with increasing magnesia content.MIXED-OXIDES VII A N D VIII A N D MAGNESIA Of the four t.p.d. peaks, that with the lowest maximum temperature was attributed to acetonitrile interacting with hydroxyl groups. Beside the low temperature of desorption, the increase of the peak when water was adsorbed before acetonitrile supports our interpretation. The surface species which caused the i.r. absorption between 2200 and 21 50 cm-l was thermally less stable than the surface carboxamide species. Therefore peak 2, which appeared at a lower temperature than peak 3, was attributed to the adsorption form absorbing between 2200 and 2150 cm-l, and peak 3 was assigned to the carboxamide species.For peak 4, we have not yet found a valid interpretation. We thank the ‘ Fonds zur Foerderung der wissenschaftlichen Forschung ’ for providing the i.r. spectrometer. H. Bremer and K-H. Steinberg, 4th Int. Congr. Catalysis, Moscow, 1968 (Akademiai Kiado, Budapest, 1971), p. 1377. H. Niiyama, S. Morii and E. Echigoya, Bull. Chem. Soc. Jpn, 1972, 45, 655. Y. Okamoto, T. Imanaka and S. Teranishi, Bull. Chem. Soc. Jpn, 1973, 46, 4. H. Vinek, H. Noller, J. Latzel and M. Ebel, Z . Phys. Chem. N.F., 1977, 105, 319. G. Hatzl, Diplomarbeit (TU Wien, 1978). H. Knozinger, Forschungsberichte Wehrtechnik, Luft- und Raumfahrt, 1976, 2, 53. H. Krietenbrink and H. Knozinger, 2. Phys. Chem., 1976. 102, 43. G. Della Gatta, J. Calorim. Anal. Therm., 1976, 7, 1/5/1. F. Koubowetz, J. Latzel, H. Noller, J . Colloid Interface Sci., 1980, 74, 322. lo T. Tokuyama, J. Pharm. Sac. Jpn, 1955, 75, 957. l 1 S. Pasynkiewicz, Pure Appl. Chem., 1972, 30, 509. l 2 J . Latzel and G . Kaes, React. Catal. Lett., 1978, 9, 2, 183. l 3 H. Knozinger, H. Stolz, H. Buhl, G. Clement and W. Meye, Chem. Ing. Tech., 1970, 42, 548. l 4 F. H. Van Canwelaert, P. A. Jacobs and J. B. Uytterhoeven, J. Phys. Chem., 1972, 76, 1434. l6 H. Krietenbrink, Diplomarbeit (Ludwig-Maximilian Universitat, Miinchen, 1974). l8 J. A. Lercher, H. Noller and G. Ritter, J. Chem. Soc., Faraday Trans. I , 1981, 77, 621. l9 H. R. Christen, Grundlagen der organischen Chemie (Sauerlander Aarau, Diesterwag Salle Frankfurt- 2o V. Gutmann, The Donor-Acceptor Approach to Molecular Interactions (Plenum Press, New York, M. Kermarec, M. Briend-Faure and D. Delafosse, J. Chem. Soc., Faraday Trans. I , 1974, 70, 2180. N. Takezawa and H. Kobayashi, J. Catal., 1973, 28, 335. am-Main, 1975), p. 619. 1978). R. T. Sanderson, Chemical Bonds and Bond Energy (Academic Press, New York, 1976). (PAPER 1 / 1558) 7 3