摘要:

Near-Ultra-violet Spectrum of Propenal BY J. C. D. BRAND AND D. G. WILLIAMSON Dept. of Chemistry, The University, Glasgow, W.2 Received 1st January, 1963 The first, very weak band system has its origin near 4122A and is probably the triplet a*-n transition of trans-propenal. The origin of the corresponding transition between singlet states is at about 3865 A, and its associated vibrational structure has been analyzed in some detail. The results indicate that the singlet excited state is planar or nearly planar ; excitation increases the CO distance by ca. 0.12%1 and effects a considerable redistribution of a-electrons outside the CHO group, so that the greater CC bond order is associated with the central rather than the terminal CC link. The first electronic system of propenal (acrolein) commences near 4122A and is extremely weak, requiring about 1 m atm of vapour for development of the stronger bands.A more intense system (f = 6 x 10-4) occurs between 3865 and about 2800 A, and consists of somewhat diffuse bands superimposed on a continuum.ls2 The spectrum is of interest in that, though related to the long-wavelength system of formaldehyde,4 the presence of a double bond might be expected to offset any ten- dency towards non-planarity at the carbonyl group. In terms of ordinary valence formulae one might express this in terms of resonance between the excited state structures (1) and (2) with an important contribution from the structure (2). The spectrum of propynal, HC-C . CH=O, is considered from a similar point of view in an accompanying paper.5 H H H H H The planar trans-conformation of propenal is stable in the electronic ground state.6 The structure of the 3865A system is consistent with the assignment of all important bands to transitions emanating from the trans ground state, though it is difficult to exclude the possibility that some weak bands might originate in the cis form which is present to the extent of a few per cent at ordinary temperature.There is some evidence that the very weak 4122A system marks the transition to the lowest triplet state of the trans configuration. EXPERIMENTAL Propenal-d, CH2 : CH . CDO, was obtained by oxidation of CH2 : CH . CD20H 7 with oxygen on a copper catalyst at 300'. Spectra of the normal and d-isotope were photo- graphed on a Hilger E495 glass-prism spectrograph resolving about 1 cm-1 at 3800& though the intensities are taken from recordings made with a Perkin-Elmer 137 U instru- ment.Infra-red absorption was measured with a Unicam SP 100 prism-grating spectro- meter resolving 1-4 cm-1 in the range 3600-400 cm-1. Frequencies of the principal bands in the 3865 A system are listed in table 1. Detailed analysis was not attempted at frequencies greater than about 1700 cm-1 above the origin. 184POI. C C C a C C c C C a a a C C C a a a a a a a a a a C C c C C U a a a a a a C C a C J . C . D. BRAND AND D. G . WILLIAMSON 185 TABLE 1 . h I N C I P A L ASSIGNMENTS IN THE 3860 A SYSTEM CH2 : CH . CHO CH2 : CH . CDO cm- 1 25296.5 25549-8 25644.2 25705.2 25729.3 25746.5 25861 *4 26953.9 26055.6 261 13-2 261 94.5 26278.6 26349.7 26368.4 2643 1 -5 26443.6 26505.4 26512-9 26576.7 26592.3 26669-8 26747.0 26770.5 26994.1 27127.0 271944 2727 1 -8 27372.9 27407.2 27449.7 27484.1 - 564.9 - 31 1.6 -217.2 - 156.2 - 132.1 - 114.9 0 + 92.5 194.2 251-8 333.1 417.2 488.3 507.2 570.1 582.2 644.0 651.5 715.3 730.9 808.4 885.9 909.1 1 132.7 1410.4 1622.7 int .a - - - - - 8.0 4.0 0.9 1.1 1.4 0.6 5.4 2.2 2.2 I 0.6 0.5 0.9 0.9 1.1 1.1 2.5 9.0 10 2.2 2.2 cm-' 25327-1 25738.6 25890-1 25982.1 26135.1 262 19-7 26306.2 26372.5 263857 26455.3 26465.2 26444.2 26537.6 26601.8 26626.2 26688.3 26768.1 26789.8 26880.8 26923.0 271 65.5 26997.6 27292.2 27404.5 27488.9 27534.0 27574.0 27618-7 27643.2 27662-7 27732.0 27785.1 - 563.0 -151.5 0 + 92-0 245.0 329-6 416.1 482.4 495.6 565-2 575.1 554.1 647.5 71 1.7 736.1 798.2 878.0 899.7 990.7 1032.9 1275.4 1107.5 1402.1 1514.4 159848 1643.9 1683.9 1728.6 1753.1 1772.6 1841.9 1895.0 int.a - - 7-8 3.8 1.0 1.1 1.0 40 1.8 1.9 I - 0.5 0.7 0.4 0.7 1.1 1.1 I - 10 10 2.6 2.2 0.7 2.6 1.0 3.1 0-6 2.3 assignment' 12(0,1) 18(0,2) 18(1 ,2) 18(0,1) 12(0,1).18(0,4) 1 W,4) origin 18(1,1) 18(2,2) 18(1,0) 14( 1 ?O) 14(1,0). 18(1,1) 12(1 90) 1 W , O ) 12(1,0). 18(1,1) 16(1,0) 1 5( 1 ,o> 14(2,0) 12(1,0). 18(1,1) - - 12(1,0). 14(1,0) 12(1,0). 14(1,0) . 18(1,1) 17(1 $0) 8(1,0) 5 0 $0) 16(2,0) 6(1,0) 5(1,0). 18(1,0) 15(2,0) . 18( 1,O) 5(1,0). 14(1,0) 15(2,0). 14(1,0) 6(1,0) . 18(1,0) 5(1,0). 14(1,0) 6(1,0). 14(1,0) 5(1,0). 12(1,0) 5(1,0). 18(2,0) 5(1,0). 16(1,0) 17(1,0). 18(1,1) 12(1,0). 15(1,0) . 18(1,1) 6(1,0) 12(1,0) a maximum intensity (1.mole-1 cm-1) of discrete bands relative to the continuum. b A u'-v" transition in vk is indexed as k(v', u") : thus, the 1-0 transition in vg becomes 5(1 ,O), etc. Upper-state modes are identified by the same numbers used in the ground state (see table 2). Table 2 contains the analysis of the infra-red spectrum, portions of which are essential to an understanding of the electronic " hot " bands. A partial infra-red analysis has been given by Inumka,* some of whose assignments differ from ours. The list of infra-red frequencies in table 2 is incomplete to the extent that some fundamentals are undetectably weak in the spectrum. For the normal isotope the missing a" fundamental v15 can be186 SPECTRUM OF PROPENAL calculated from the product rule and appears to be buried under the strong absorption associated with ~ 1 4 and v16 in the region 960-1000 cm-1. Fundamentals below 400 cm-i are taken from the Raman spectrum 899 or the electronic analysis.In table 1 the vibrations of excited propenal are identified by the same numbers that apply in the electronic ground state : e.g., the vinyl torsion is labelledvl4 in both states, irrespective of whether this correctly expresses its position in a systematic table of upper-state vibrations. TABLE 2.-FUNDAMENTAL FREQUENCIES OF THE GROUND STATE approximate cm-1 (intensity)a description CH2 : CH. CHO CH2 : CH . CDO mode vinyl CH stretch formyl CH stretch CO stretch C=C stretch CH2 bend formyl CH rock vinyl CH rock vinyl CH2 rock LCCO bend LCCC bend Y ? Y Y C - 4 stretch vinyl C=C torsion formyl CH wag vinyl CH2 wag vinyl CH wag skeletal torsion 3 102 (4.7) 3000 (4.5) 2800 (21) 1723 (214) 1625 (13) 1422 (12) 1361 (4.0) 1276 (1.7) 1159 (38) 913 (36) 564 (6.6) 340 993 (25) 980 959 (48) 589 (8.3) 158 3101 (4.4) 2988 (2.3) 2060 (29) 1709 (210) 1621 (3.0) 1403 (14) 1275 (3.0) 1153 (30) 877 (14) 562 (4.1) 993 (23) 846 (4-0) 556 (7-1) 151 e - - 959 (43) apparent &max (1.moles-1 cm-1) ; b principal component of a Fermi polyad ; C from the Raman spectrum (ref. 8) ; d calc. from the product rule ; e from the electronic spectrum. RESULTS ROTATIONAL STRUCTURE IN THE 3865 A SYSTEM As first reported by Eastwood and Snow 2 the principal bands are perpendicular bands of a near-symmetric top, with well-marked K-structure running from the band centre towards higher frequencies : these perpendicular bands are, in fact, very similar in construction to the corresponding (type C ) bands of glyoxal.10 The K-structure degrades to the blue whereas the J-structure degrades to the red, con- sequently, 3’ <B” and (A’ - B’) > (A” - 2’).Quantitatively, it turns out that ( A ” - 3 ) - ( A ’ - z ’ ’ ) is about the same magnitude as B’ or B”, and thus A’>”’. Since any permanent twist about the C-C (single) bond tends to decrease A it follows that the excited state must be planar or nearly planar. Otherwise, it is difficult to draw any firm conclusion from the magnitude of the constant A’, for the CO and CC bonds have only a small projection on the a-axis, and their distances may well change in opposite senses and so tend to cancel.Besides the perpendicular bands the system contains a considerable number of comparatively weak bands whose envelope is consistent only with parallel structure (AK = 0). In C8 symmetry no sharp distinction exists between type A (parallel) and type B polarization, so that the description as parallel merely signifies a hybrid band largely type A in character. We have not identified any hybrid bands of predominantly type B character but the general lack of contrast in the spectrum, the increasing diffuseness towards shorter waves, and the large frequency rangeJ . C. D . BRAND AND G . D . WILLIAMSON 187 (ca. 150 cm-1) covered by each perpendicular band make it difficult to identify the contour of some weak transitions in the spectrum.Hence, the analysis which follows is qne we judge to be consistent with the appearance of the bands at the resolution achieved. THE TOTALLY-SYMMETRICAL FUNDAMENTALS Vibrational structure in the 3865A system is compatible with the assumption that the strong band at 25861 cm-1 is the electronic origin. The origin is then a perpendicular band separated from a number of parallel bands by intervals which must be assigned to non-totally symmetrical vibrations : this evidence requires that the system origin shall be polarized perpendicular to the molecular plane (type C band), and the same conclusion follows from its very close resemblance to the known type C bands of glyoxal. Granted this, the electronic transition is A” - A’, as expected for a one-electron n* - n transition.Prominent type C bands also occur near 0+490, O+ 1266 and 0 + 1410 cm-1 in the spectrum of the normal isotope. These intervals are assigned by Inuzuka 8 to CCO bending (vJ, CO stretching (v;) and CC stretching (v;), respectively, and we agree in principle with these assignments though we think that in two in- stances (0+490 and O+ 1266) the bands are doubled by Fermi resonance. All the strong bands above 0-1- 1410 cm-1 appear to be either pure overtones of 1266 cm-1 or their combinations with single quanta of 490 cm-1 or 1410 cm-1: thus the pro- gression forming vibration is v; (CO stretching), although vi (CC stretching) and vi2 (CCO bending) are strongly active for changes of one quantum. The propenal-d spectrum has the same construction, with prominent intervals of 482, 1275 and 1402 cm-1.The only other a’ fundamental active in the spectrum is the formyl CH rocking mode v;, marked in the normal isotope by a weak type C band at 0 + 1133 cm-1. The corresponding transition is not observed with propenal-d, probably because its intensity in the normal isotope owes more to the element of CO stretching in the normal co-ordinate than to any Franck-Condon effect associated with the HCO bond angle. Otherwise a small number of medium-to-weak type C bands do appear, but can be assigned either as hot bands or overtones of non-totally symmetrical vibra- tions. Of the former, only the band at 0-565 cm-1 (the 0-1 transition in vlz) seems to mark a single-quantum change. Among the parallel bands, however, one notices a negative sequence of about 32 cm-1(29 cm-1 for propenal-d) apparently associated with a low-frequency fundamental.This sequence does not involve v18 (see below) and so is probably connected with the next-lowest ground state funda- mental ~ 1 3 . As we have not been able to find the corresponding 1 4 transition, the sequence frequencies are omitted from table 1. THE NON-TOTALLY SYMMETRICAL FUNDAMENTALS Near the origin there are parallel bands at 0- 156 and 0+252 cm-1 which we assign to 0-1 and 1-0 transition in the skeletal torsion v18. This assignment immediately accounts for a moderately strong type C band at 0+92 cm-1 as the 1-1 sequence in v18, and for weak bands of the same polarization at 0-312 and 0-217 cm-1 as the 0-2 and 1-3 transitions respectively.Combining bands of the same polarization, in order to minimize errors resulting from the use of band centres rather than origins, we find vii = 157.7 and v,; = 250.2 cm-1 as the fundamental frequencies of the normal isotope. The magnitude of vlh suggests that the 2 4 transition should lie near Of500 cm-1 and, in fact, the type C absorp- tion in this region does appear to encompass two bands with centres at 0+488 and188 SPECTRUM OF PROPENAL 03-507 cm-1. A similar organization of bands occurs in the spectrum of glyoxal10 except that the 1 4 and 0-1 torsional transitions are absent, no doubt because in C2h symmetry they are forbidden by the rigorous g++g prohibition, Further non-totally symmetrical fundamentals occur at 0 + 333, 0 + 582, 0 + 644 and 0+909 cm-1.Of them, only 644 cm-1 changes substantially (to 554 cm-1) on deuteration and so can be attributed to the CHO out-of-plane fundamental vl; The overtone 2v15 (harmonic value 1288 cm-1) appears in Fermi resonance with the CO stretching fundamental which has two components (1266 and 1295 cm-1). With roughly equal intensities observed, one expects the unperturbed CO funda- mental would occur near 1281 cm-1 and this helps to remove an anomaly, namely, that the fundamental frequency is seemingly smaller for the normal molecule (1266 cm-1) than for the d-isotope (1275 cm-1). The remaining non-totally symmetrical intervals of 333,582 and 909 cm-1 must belong to modes of the CH : CH2 group. Within this group, judging by the results for propyna1,s one expects the CH wagging mode will maintain or increase its frequency relative to the ground state, and accordingly we assign vl; = 909 cm-1.The two unassigned a" fundamentals, the CH2 wagging mode (vld and the vinyl CC torsion (vl& then have to be allotted to the intervals 333 and 582 cm-1; that is to say, either vl; = 333 and v l i = 582 cm-1, or vice versa. For reasons given later, we prefer the second assignment, namely, v,; = 333 cm-1. The overtone 2vli appears as a weak type C transition at 0 + 651 cm-1. The product rule ratio for a'' fundamentals is 1.24, compared with an harmonic value of about 1.28. Other parallel bands can be explained as combinations of 250 or 333 cm-1 with the 488, 1266 and 1410 cm-1 totally symmetrical fundamentals. Two rather weak parallel bands, at 0+715 and Of916 cm-1, are left unexplained by this scheme.0+961 cm-1 might be attributed to the 3 4 transition in vli (the anhar- monicity is about right) but this assignment is difficult to reconcile with the evidence favouring a planar excited state. The assigned fundamentals are listed in table 3 where corresponding frequencies of glyoxal and propynal are included for comparison. TABLE 3 .-FUNDAMENTAL FREQUENCIES OF EXCITED (IA") PROPENAL mode description CHI : CH . CHO CH2 : CH . CDO CHO . CHO a CDO . CDO a C2H. CHO C2H. CDO v5 (a') CO stretch v6 CC stretch VS formyl CH v12 CCO bend ~ 1 4 (a") vinyl CC v15 formyl CH v16 vinyl CH2 wag ~ 1 7 vinyl CH wag v18 skeletal torsion rock torsion wag 1266 1410 1133 488 333 644 582 909 250 1275 1391 1394 1300 1402 - 1115 - - - - 482 C 509 500 507 330 - - 554 734 646 462 575 - - 390 900 244 232 216 - - - - 1268 824 - 500 - 41 1 375 - a ref.(1Ou) ; ref. (5) ; c principal component of a Fermi diad. DISCUSSION TORSIONAL MODES The excited state torsions must contain information about the distribution of n-electrons in the A" electronic state. The stiffening of the nominally single CCJ . C. D. BRAND AND D. G. WILLIAMSON 189 bond implies an increase of n-bond character in this region, evidently at the expense of the vinyl CC bond where the torsional frequency falls considerably. Consider first the ground-state skeletal torsion vii. The potential function assumed for small amplitudes is, in which Q is a mass-weighted normal co-ordinate, and a = h/4n2cao. According to first-order perturbation theory the energy levels in the potential (1) are given by (E, - EO)/hc = tloo - *(u + l)t~01.(2) With four levels of v i i observed (157.7, 311.6, 467.4 and 621 cm-I), one obtains from (2), 00 = 159.7 and 01 = 2.0 cm-1 for the normal isotope. The potential opposing a complete internal rotation can be expressed in the form, v = ~V,(l-COS 4)+*V,(l-cos 24), (3) where 4 is the angle subtended by the CHO and CH : CH2 planes. The functions (1) and (3) are consistent when and 01 = (h/8z2cI,)(Vl + 16V2)/(4V, + 16V2), in which I, is a reduced moment of inertia for internal rotation. A determination by ultrasonic methods of the cis +trans isomerization of liquid propenal gives V1 = 720 and V2 = 2080cm-1.11 Using a value of Ir calculated by Pitzer’s method 12 one then obtains o0 = 166 and o1 = 2.9 cm-l.Agreement with the spectral values leaves no doubt that the assignment of v[i is correct. Granted this, the vl& = 250 cm-1 assignment is also established, since the ground and excited state fundamentals are connected by n-n sequences. A number of quantitative difficulties enter in the excited state. First, the geometry is not accurately known and thus only approximate values of Ir can be calculated; and, secondly, the two torsions may well couple appreciably since their frequencies are not widely separated. In order to get rough values we adopt ground state 1,’s and neglect whatever coupling may actually be present. For the nominally-single CC bond torsion, on the further assumption that the cis-trans energy difference is unchanged relative to the ground state, we calculate from the zero-order frequency (253.2 cm-1) that the barrier opposing internal rotation is approximately 5300 cm-1.The theoretical 01 for this mode is 3.0 cm-1 : the calculated frequency, 200-301, of the 2-0 transition is then 497 cm-1 which is close to the mean frequency of com- ponents of the Fermi diad, v,;, 2vl& at 0+488 and O+ 507 cm-1. For the vinyl torsion ~ 1 4 the expected potential function is v = +V2( 1 - cos 24’), where # is the angle between the C-CH and CH2 planes. In this mode = h/8n2cIr, thus from the theoretical Ir 12 one calculates 61 = 11-3 cm-1. The frequencies of the 1-0 and 2-0 transitions give 00 = 347.7 and 01 = 14.6 cm-1 so that the large value of 01 observed is explained by the small I, for this mode.(This is also the principal reason for believing that 333 and 651 cm-1 should be assigned to the190 SPECTRUM OF PROPENAL fundamental level and overtone of v l ; ; no other vibration is expected to give so large an anharmonic defect.) The barrier opposing rotation about the vinylic CC bond in the excited state is calculated to be about 2700cm-1. The potential (4) generates equivalent minima by an internal rotation through 180" so that the question arises whether any splitting of the energy levels might be observed in consequence of tunnelling through the barrier. The theoretical splitting, however, is insignificantly small, of the order 10-6cm-1 for the higher (0+651 cm-1) of the two observed levels. The barrier heights are compared in table 4 with values for glyoxalloa where the twisting mode is analogous to the skeletal torsion of propenal.TABLE 4.-BARRIERS TO INTERNAL ROTATION barrier height (cm-1) torsion ground (A') state excited (A") state pro penal vinylic (C=C) 8700 a 2700 skeletal (C-C) 2270 (2420 b, 5300 glyoxal 1120 3800 a value for C2H4 (ref. 13)) ; b ref. (11) ; c ref. (1Oa). EXCITED STATE STRUCTURE The torsional barriers indicate that the bond order is greater in the central rather than the terminal CC bond, corresponding to an important contribution from the structure (2). A rough correlation between the barrier height and bond length implies that the CC bond distances, which are about 1.45 and 1-36 A in the ground state,6 alter in opposite senses by about 0.06-0-08 A on excitation.As the 1410 cm-1 vibration is then interpreted as the antisymmetrical stretching motion of the chain of three carbon atoms, a Franck-Condon calculation indicates that intensity in this mode should fall off rapidly along the series 1 4 , 2-0, . . ., which accords with the fact. A further consequence of the z-electron reorganization is that the CHO out-of-plane mode vI; is nearly harmonic in the A" state. The corresponding vibration of excited formaldehyde is grossly anharmonic owing to the non-planar equilibrium configuration of nuclei,4 whereas in propynal the vibration is noticeably anharmonic even though the nuclear configuration is probably ylanar.5 From the intensity distribution in the main progression, assumed to represent a pure CO group vibration, Inuzuka estimates that the CO distance increases by about 0.11 A (to 1-33 A) on excitation.8 With the same assumption, Badger's rule and Douglas Clark's rule give excited CO distances of 1-34-1.35 A, and all these values compare quite closely with the CO distances in excited CH2O (1.33&4 and propynal (1-33 &.5 The relative weakness of the bending modes suggests that bond angle changes are small ( < 5") everywhere.THE 4122A SYSTEM The system contains three prominent bands (24247, 24619 and 24834 cm-1: the first has a different contour from the other two) though a number of weaker transi- tions appear at high pressure. The three strong bands are conceivably 0 4 , 1-0 and 2-1 transitions in Ylg, an assignment which gives v,'! = 157 and vli = 372 cm-1. This implies that the 4122A system marks a transition of trans-propenal, presumably the triplet z*--n transition. The 4122 A band is reported in the magnetic rotation spectrum.14 We thank D.S.I.R. for a research studentship (to D. G. W.).J . C . D . BRAND AND D . G . WILLIAMSON 191 1 Luthy, 2. physik. Chem., 1923, 107,285. 2 Eastwood and Snow, Proc. Roy. SOC. A, 1935,149,446. 3 Blacet, Young and Roof, J. Amer. Chem. Soc., 1937, 59, 608. 4Brand, J. Chem. SOC., 1956, 858. Robinson, Can. J. Physics, 1956, 34, 699. 5 Brand, Calloinon and Watson, this Discussion. 6 Wagner, Fine, Simmons and Goldstein, J. Chem. Physics, 1957, 26, 634. 7 Bartlett and Tate, J. Amer. Chem. Soc., 1953, 75, 91. 8 Inuzuka, i3ull. Chem. SOC. Japan, 1960, 33, 678 ; 1961, 34,6,729. 9 Bourguel and Piaux, Bull. SOC. Chiin., 1935, 1967. Harrand and Martin, Bull. SOC. Chim., 10 (a) Brand, Trans. Faraday SOC., 1954, 50, 431. 11 de Groot and Lamb, Proc. Roy. SOC. A, 1957,242, 36. 12 Pitzer, 1. Chem. Physics, 1946, 14, 239. 13 Herzberg, Infia-red nnd Raman Spectra (Van Nostrand, New York, 1945), p. 227. 14 Eberhardt and Renner, J. Mol. Spectr., 1961, 6, 483. 1956, 1383. (b) King, J. Chem. Soc., 1957, 5054.

ISSN:0366-9033    

DOI:10.1039/DF9633500184

出版商:RSC    

年代:1963

数据来源: RSC

摘要:

Polarization and Assignment of the 3700 A Absorption Spectrum of 1,4=Diazine Vapour BY K. K. AND L. E. GIDDINGS, JR.* Dept. of Chemistry, Vanderbilt University, U.S.A. Received 7th January, 1963 A 70-m absorbing path of 1,Cdiazine vapour has been studied with a resolving power of about 200,000. Except for slight differences of intensity distribution, the rotational fine structure of the 0-0 band at 3729A is like that of the 0-0 band at 3239A, studied earlier. The structure is predominantly of the parallel type and the apparent polarization is therefore parallel to the axis of largest inertia and perpendicular to the plane of the molecule. The inertial constants determined from rotational analysis are 2 = 0-2050 cm-1 and 2 = 02041 cm-1 and the system origin is at 2682025 cm-I.This analysis does not distinguish whether the excited state is singlet or triplet, nor whether it is g or u. However, the fine details of intensity distribution indicate the presence of rota- tionaI transitions with AJ = 0, AK = &2. These are assigned as electric quadrupole which in turn requires that the main, parallel structure be magnetic dipole in character, that is, the excited state is g.? It has been assumed since the discovery of the band system by Hirt1 and by Goodman and Kasha 2 that the 3700 A absorption of pyrazine (l,4-diazine) arises from a triplet-singlet transition. Goodman and Kasha 2 integrated the absorption and found anf-value of about 10-7. At lower temperatures and in a different solvent they found emission of similar vibrational contour (dominant difference, about 600 cm-1) in the same spectral region.El Sayed and Robinson 3 offered convincing, though not conclusive, evidence that the origin ( 0 4 ) bands of the absorption and emission systems arise from the same two energy levels. Krishna and Goodman 4 measured the polarization ratio for emission from pyrazine molecules (in a rigid medium at 77°K) after excitation with (a) primarily 2537 A radiation and (b) mono- chromatic radiation, probably 3 13 1 A. Assuming polarization along the axis of intermediate inertia b for 2500A absorption and along the inertial axis c (per- pendicular to the molecular plane) for 3100A absorption 5 they derived that the 3700 A emission was polarized in plane and parallel to the axis of least inertia, which contains the nitrogen atoms.Two points emerge from the preceding history. First, there is a lack of definitive experimental evidence about the triplet-singlet character of the transition. The weakness of the transition is, of course, consistent with forbidden character of any kind. Secondly, pyrazine offers an unusually good opportunity for a high-resolution study of a (supposed) triplet-singlet absorption of a vapour, free of overlapping by other band systems. We have studied the spectrum with these points in mind. It will be shown that the origin band is type C. * present address, U.S. Naval Research Laboratory, Washington, D.C. f Dr. A. E. Douglas’s observation (this Discussion) of Zeernan splitting for the band discussed here is the first experimental evidence of the triplet character of the excited state.Since he indicates the field-free splitting to be less than 0.1 cm-1, the triplet state may be concluded to exhibit coupling analogous to Hund’s case (b) of diatomic molecules. The above A J accordingly should be changed to AN. Compare also the Spiers Lecture, this Discussion, in which another explanation of the selection rule AK = 5 2 is suggested. 192FIG. 1.-The 0-0 band of pyrazine; absorbing path, 70 m ; pressure, the equilibrium vapour pressure for 25°C. To face page 1931.K. K. INNES AND L. E. GIDDINGS 193 Our photographs offer no direct evidence of the triplet character of the excited state. However, the communication of Dr. A. E. Douglas to this Discussion reports unambiguous identification of the excited state as triplet. In accord with his results we assume excited state coupling closely analogous to Hund’s case (b) of diatomic molecules.N is the quantum number for total angular momentum apart from spin and J = N + 1, N and N- 1. For simplicity, we shall denote lines for AN = + 1, 0, - 1 by R-, Q- and P-branch, respectively. In this paper we adopt species symbols A,, Ra, Rb, R,, Au, Ta, T b , Tc for the point group D2h, where Ru, Rb, R, transform like rotations about the axes a, b, c, respec- tively, and Tu, T b , Tc transform like translations along the axes. For example, the 3100 A transition is designated ITc - 1A, and the 2500 A transition 1Tb - 1A,. EXPERIMENTAL The spectrum was photographed in the third order of a 3.4 m spectrograph. The dis- persion was 0-6A/mm and the resolving power observed was more than 125,000.The spectra were measured against iron lines using a David G. Mann model 300 comparator. The absorption tube was a 4-m multiple reflection cell, allowing up to 16 traversals. It was filled with pyrazine at the equilibrium vapour pressure corresponding to 25°C. The pyrazine was obtained from the Aldrich Chemical Company and was used without further purification. The continuous background for the absorption spectrum was provided by Hanovia high-pressure xenon arcs and was recorded on Kodak SA1 photographic plates. RESULTS The three bands for which fine structure was brought up at the absorbing paths available showed the simple appearance of the 0 - 4 band which is displayed in fig. 1. The structure is like that of the parallel bands (including the 0-4 band) of the 3100 A system of pyrazine 5 except that the P- and R-branches are seen to higher N-values and show a slight intensity alternation. The fine structure of the 0-0 band has been measured and the wave numbers are given in table 1.Each band showed a sharp Q-branch similar to that of fig. 1. The short wavelength edge of the Q-branch of fig. 1 is within 0.2 cm-1 of the band centre and we assume that this is true of each one of the set of @branch edges of table 2. DISCUSSION Methods for rotational analysis were summarized in our earlier 5 discussion of pyrazine. The only difference here is that the rotational constant can be obtained more accurately for both electronic states since P- and R- branch composite lines extend to N-values twice as great as in the former case.Assignments are given in fig. 1 and table 1. The new value for the ground-state inertial constant is then 0.2050 cm-1, which may be compared to 0-2048 cm-1 determined from the 3100A band system.5 The change in 5 for the transition is found to be only -0-00086 cm-1 and DO”, the centrifugal distortion constant, is 2 x 10-9 cm-1. As for the 3100A system, the 0-0 band is type C. There are many ways in which this effective polarization could arise. It appears that any triplet-singlet transition of pyrazine could give a band of the general type shown in fig. 1, if one assumes no knowledge of the precise nature of the coupling case. This difficulty in assignment should be quite common in the - B = h/167~2~(1/10 + 1 / I b ) G194 3700A ABSORPTION BY PYRAZINE TABLE l.-wAVE-NUMBERS OF THE LINES IN THE PYRAZINE 0 4 BAND (Vvac-, Cm-1)* N 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 R(N) 26825.77 (3) 26.17 (2) 26.54 (3) 26-86 (3) 27.29 (2) 27.68 (2) 28.08 (2) 28-45 (2) 28.85 (2) 29-26 (2) 29.60 (2) 29.96 (2) 30-31 (2) 30-67 (2) 3 1-03 (2) 31-37 (2) 31-73 (2) 32.18 (2) 32.48 (2) 32.84 (2) 33-16 (2) 33.50 (2) 33.84 (3) 34-23 (2)? 34.54 (3) 34-91 (2) 35.25 (3) 35.56 (2) 35-92 (3) 36-26 (2) 36-58 (3) 36-92 (2) 37.24 (3) 37-57 (2) 37-91 (3) P W ) 26816.05 (2) 15.63 (1) 15-21 (1) 14-76 (2) 14-27 (2) 13.85 (2) 13-41 (2) 12-96 (2) 12-54 (2) 12.22 (2) 11-77 (2) 11-25 (2) 10.83 (2) 10.37 (2) 09-92 (3) 09.46 (3) 09-01 (2) 08-58 (2) 08-11 (3) 07.66 (2) 07-21 (2) 06.74 (2) 06.27 (3) 05-79 (2) 05.32 (3) 04-84 (2) 04.39 (3) 03-92 (2) 03.47 (3) 02.98 (2) 02.50 (3) 02.04 (2) 01.54 (3) 01.09 (2) 00.59 (3) 800.11 (2) 799-61 (3) 99.10 (2) N 45 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 RCN) 38-23 (3) 38-55 (3) 38.88 (3) 39-20 (3) 39.52 (2) 39-84 (3) 40.15 (2) 40.47 (3) 40.77 (2) 41.07 (3) 41.37 (2) 41-69 (3) 42-00 (2) 42-30 (3) 42-60 (2) 42-91 (3) 43.20 (2) 43.54 (2)? 43.79 (2) 44-07 (3) 44.37 (3) 44-66 (3) 44.96 (2) 45.24 (3) 45-52 (2) 45.79 (3) 46.08 (2) 46.36 (3) 46.62 (2) 46-89 (3) 47.19 (2) 47-46 (3) 47.74 (l)? 47-99 (1) 48-28 (1) 48.54 (1) 45-83 (1) 49.21 (l)? P W ) 98.61 (3) 98.11 (2) 97.64 (3) 97-12 (2) 96-64 (3) 96-15 (2) 95-64 (3) 95.14 (2) 94.64 (3) 94.14 (2) 93.60 (3) 93-11 (2) 92.61 (3) 92-09 (2) 91-58 (3) 91.07 (2) 90-53 (3) 90.04 (2) 89-51 (3) 88-98 (2) 88-43 (3) 87-92 (2) 87.39 (3) 86.84 (2) 86.33 (3) 85-79 (2) 85.28 (3) 84-71 (2) 84.17 (2) 83.64 (2) 83-09 (2) 82-55 (2) 82-02 (3) 81-50 (2) 80.92 (2) 80.35 (3) 79-80 (2) 79-26 (2) * ( ) visually estimated relative intensities.TABLE 2.-BAND CENTRES OF 'THE 3700 A ABSORPTION SYSTEM OF PYRAZINE *t (vVaC., cm-1) 26640.87 (4) 26820.25 (20) 26887.05 (3) 27109.94 (1) 27294.24 (2) 27440.77 (9) 27793.25 (1) 27969.41 (3) 28042.32 (2) * Only the strongest bands are listed. From these the most active upper-state frequencies, based on voo = 2682025 cm-1, are 474.1, 6206 and 1149.3 cm-1, which are similar to three of the most active differences in the 3lWw system.(S. G. Tilford, Ph.D. Thesis, Vanderbilt University (1962) to be published.) We defer a complete listing and detailed analysis until it is quite certain which of the weaker bands are " hot " bands of the 3100 A system. t ( ) estimated relative intensities.K. K . INNES AND L. E . GIDDINGS 195 spectra of polyatomic molecules for which structure is incompletely resolved. De- tailed studies of Zeeman effects may be more urgent than has been realized. A tentative narrowing of the choice of assignment of the band of fig. 1 may be achieved by considering two further experimental facts. They agree in supporting the assignment 3Rc - lAg or 391Rc - lAg which we therefore, provisionally, accept. First, we recall the Krishna and Goodman experiment described in the introduction.4 The electric vector of the 3700A emission was found to be parallel to the plane of the molecule.However, if the relevant vector is the magnetic one (assignment &-A,), the magnetic transition moment would be perpendicular to the plane of the molecule and the (irrelevant) electric vector would be measured to be in the plane. Since the alternation was not observed in the type C bands of the 3200A system, it probably arises from an enhancement of intensity of every second line, that is, at intervals of approxim- ately 4B or 8(C-3). The most obvious source of added intensity is from transitions of AN = +2 which one may expect for selection rules A J = 0, +1. However, the degrading of the band spoils the coincidence of AN = f I and AN = +2 transitions which one would otherwise expect. Systematic enhancement of alternate lines cannot occur in that simple way. The remaining possibility is reasonable but harder to evaluate in detail.By analogy with perpendicular-type bands already analyzed for pyrazine 7 one would expect branches with AN = 0 and AK = +2 to stand out above a general background. These would show twice the spacing that they do in perpendicular bands, that is 2 x 2(C-B). The features could coincide closely with the P- and R-branch " lines '' of fig. 1, since 2F=4(C-B). But even and odd K-values are weighted by 13 and 11, respectively,7 so that the enhancement of the type-C structure would be alternately strong and weak. Detailed energy level expressions show that the intensity alternation would be out of phase in the P- and R-branches as is observed in table 1.We therefore adopt this description of the intensity alternation. The most obvious origin for such intensity is from electric quadrupole transitions. If one assumes the rotational selection rules of the Kaman effect to hold for quadrupole transitions, then AK = +2 may occur only for changes of the polarizability component uUb. But Club transforms like R,. Again we are led to the assignment 3Rc-1Ag or 3,1Rc-1Ag. It should be noted that Herzberg 1 1 has quite recently found several transitions AK = f2 in polyatomic molecules and has indicated that they are to be expected for any triplet-singlet transition. The details of the derivation of the selection rule AK = f2 are not available to the authors, but if the rule is established it will pre- sent a more attractive explanation than the quadrupole one given here. One would have, from fig. 1 alone, evidence for the assignment of the band to a triplet-singlet transition. We acknowledge financial support of this work by the National Science Foun- dation through Grant GP-402. We are grateful to Dr. Douglas for private dis- cussion of the results of his Zeeman experiments. The second result is the faint intensity alternation of fig. 1 . 1 Hirt, Spectrochim. Acta, 1958, 12, 114. 3 El Sayed and Robinson, Mol. Physics, 1961, 4, 273. 4 Krishna and Goodman, J . Chem. Physics, 1962, 36, 2217. 5 Merritt and Innes, Spectrochim. Acta, 1960, 16, 945. 6 McClure, J. Chem. Physics, 1949, 17, 665. 7 Innes, Int. Symp. Molecular Structure and Spectroscopy (Tokyo, 1962), paper B3 16. 8 Callomon and Innes, J. Mol. Spectr., 1963, 10, 166. 9 Wilson, D e c k and Cross, Molecular Vibrarioizs (McGraw-Hill, New York, 1955), p. 366. 10 Goodman, J. Mol. Spectr., 1960, 6, 109. 2 Goodman and Kasha, J . Mol. Spectr., 1958, 2, 58. 1 1 G. Herzberg, Spiers Lecture, this Discussion.

ISSN:0366-9033    

DOI:10.1039/DF9633500192

出版商:RSC    

年代:1963

数据来源: RSC

摘要:

Spectrum of the Tropyl Radical BY B. A. THRUSH AND J. J. ZWOLENIK* Dept. of Physical Chemistry, University of Cambridge Received 28 th December, 1962 The spectrum of the tropyl radical (C7H7) has been observed in the flash photolysis of ditropyl and of ditropyl sulphide. This spectrum consists of a Rydberg series in the ordinary ultra-violet converging to an ionization potential of 6237 eV. It is concluded that the ground state of the tropyl radical and the excited states observed have planar symmetrical structures like that of the C7HT ion. A value of AHl(C7H:) = 209h3 kcal/mole is deduced, which is about 13 kcal/mole less than that for the isomeric benzyl ion. Hiickel’s simple molecular orbital theory 1 applied to the n: electron system of monocyclic polyenes of the type (CH), established the (4n+2) rule, which pre- dicts high relative stability for those monocyclic, coplanar systems of trigonally hybridized atoms containing (4n + 2) n: electrons, where n is an integer.The stability of C&6, C5Hy and C7HT have all been explained as the case where n = 1 and the six electrons completely fill the three available bonding orbitals leaving the anti- bonding orbitals vacant. The cyclopentadienyl and the tropyl radicals are not expected to possess the stability of the corresponding carbanion and the carbonium ion, but flash photolysis in the vapour phase should allow production and identifi- cation of these radicals. Theoretical calculations of the spectra of both the cyclopentadienyl and the tropyl radicals have been made by Longuet-Higgins and McEwen.2 The absorption spectrum attributed to the cyclopentadienyl radical, produced by flash photolysis of cyclopentadiene and ferrocene under isothermal conditions, has been reported 3 and found to be in good agreement with the theoretical predictions.Recent mass- spectrometric work 4 has yielded ionization potentials for both radicals. For the tropyl radical, produced in the pyrolysis of ditropyl, a low ionization potential of 6.60+0.1 eV was found. This low value agrees well with the ionization potential of 6.41 eV calculated by Streitwieser.5 Since the unpaired seventh n: electron in the tropyl radical occupies an antibonding orbital, transitions involving the ex- citation and the removal of that electron should occur without a large change in molecular configuration.The loss of this electron should occur readily to yield the more highly stabilized tropylium cation. In this work we have observed the absorption spectrum of the tropyl radical from two parent molecules, ditropyl (bis-7-cycloheptatrienyl) and ditropyl sulphide (bis-7-cycloheptatrienyl sulphide) which were selected because they both show strong absorption in the ultra-violet where the photolytic energy absorbed is more than the 30-40 kcal/mole needed to rupture the appropriate bond for tropyl radical formation.4~ 6 EXPERIMENTAL A conventional flash photolysis apparatus with a quartz reaction vessel 50 cm long and 1.8 cm diam. was employed with a 2000 J photolytic flash of 30 psec duration.7 The * National Science Foundation Postdoctoral Fellow 1960-62.196B . A . THRUSH AND J . J . ZWOLENIK 197 duration of the 100 J, horizontal, end-on spectroscopic flash 8 was 7 psec. Hilger E. 498, E. 517 and E. 484 quartz spectrographs were used depending on the wavelength region studied. The absorption spectra were recorded photographically using Ilford Selochrome plates, sensitized if required with Kodak U.V. sensitizer no. 8269. A standard Cu arc was used for wavelength calibration. Since an adequate vapour pressure of the parent substances could only be obtained at elevated temperatures, the cylindrical reflector surrounding the reaction vessel and photolytic flash lamp was heated electrically. This apparatus and the tube connecting the reaction vessel to the reservoir of the parent substance were maintained at about 10°C warmer than the reservoir.Ditropyl vapour from a bulb at 52°C was admitted to the re- action vessel where it was diluted with excess inert gas. The ditropyl sulphide (which has a lower vapour pressure and is less stable than ditropyl) was placed in a U-tube which was maintained at 70°C and the inert gas was passed slowly over it into the reaction vessel. Optical density measurements using Emax = 6-7 x 103 at 2600 A showed that ditropyl has a vapour pressure of 0.1 mm Hg at 52"C, in good agreement with the previously reported vapour pressure of 0.1 mm Hg at 50T.4 The pressure of ditropyl sulphide used in this work was estimated to be about 0.05 mm Hg, assuming a similar extinction coefficient. In all cases 300-400 mm Hg of " oxygen-free " nitrogen was used as inert gas. The samples of ditropyl and of ditropyl sulphide used were generously provided by Prof.Hyp J. Dauben, Jr., of the University of Washington, Seattle. RESULTS AND DISCUSSION When 0.1 mm Hg of ditropyl in excess nitrogen is flashed photolyzed, three transient absorption bands are observed. These bands appear and decay together and therefore belong to the same species. They are listed in table 1 and a low dispersion spectrum taken on a Hilger E.484 spectrograph shown in fig. 1. Their maximum intensity is reached in about 40 psec and they can still be detected faintly after 200 psec. TABLE 1 n obs. calc. difference approx. width of band (cm-1) (cm-1) (crn-1) (cm-1) 3 150 38,500 38,501 - 1 4 30 43,654 43,626 + 28 5 30 45,991 46,O 1 9 - 28 The two longer wavelength bands (2600 A and 2292 A) have also been identified in the flash-photolysis of ditropyl sulphide in excess nitrogen ; the 2173 A band could not be identified with certainty in this case owing to strong continuous ab- sorption in this region.The only other transient absorption observed in any of these experiments was due to CS, the (0,O) band of which at 2575.6 appeared very weakly in the experiments with ditropyl sulphide. No other bands of CS or of Sz could be detected. On the basis of experiments on the flash photolysis of com- parable amounts of CS;! in excess nitrogen, it was concluded that CS was a very minor product of the photolysis of ditropyl sulphide; it may arise from the decom- position of a C7H7S radical formed in the primary act.The positions of the bands listed in table 1 agree well with a Rydberg series of the form where iz = 3, 4, and 5. This series converges to the very low ionization potential of 6.237+_0-01 eV, and is believed to be the first molecular Rydberg series observed outside the vacuum v (cm-1) = 50329 - R/(n + 0-046)2,198 SPECTRUM OF THE TROPYL RADICAL ultra-violet. Further members of the series could not be observed due to greatly increased absorption by the parent molecule at shorter wavelengths and to the decrease in strength which occurs with the higher members of a Rydberg series. The only species with such a low ionization potential which would be expected to be formed from both these parent molecules is the tropyl (cycloheptatrienyl) radical, and there can be little doubt that this spectrum is that of the tropyl radical.Further, the ionization potential agrees well with that calculated by Streitwieser 5 (6.41 eV) and satisfactorily with the mass spectrometric value of 6.6fO-1 eV deter- mined by Harrison, Honnen, Dauben and Lossing4 when it is remembered that such determinations commonly give values significantly greater than do spectro- scopic 9 and photo-ionization 10 methods. No vibrational structure was detected in any of the electronic transitions. The band at 2600 A was diffuse with a width of about 150 cm-1. The other two bands were much sharper, being ca. 30cm-1 wide. Owing to strong absorption by the parent molecule in these experiments, it would not have been possible to detect vibrational structure of the n = 4 and n = 5 transitions which had an intensity of less than one-tenth or one-fifth of that of the main band respectively. Nevertheless, the absence of strong vibrational structure shows clearly that the tropyl radical and the Rydberg states observed have closely similar structures, confirming the view that the most weakly bound electron in the radical is largely antibonding in character and has little effect on the structure of the radical.The tropyl ion has been shown to have a planar symmetrical D7h strucfure,lI and since the configuration of the Rydberg states must approach that of the positive ion, the lack of vibrational structure and the sharpness of the spectra show that the ground state of the tropyl radical and the Rydberg states observed have structures which are very close to symmetrical and planar.Since the unpaired electron occupies a degenerate orbital in the ground state of the tropyl radical, this state and any degenerate upper states of this radical would be expected to show Jahn-Teller distortion.12.13 The narrowness of the spectra observed indicates that such dis- tortions are small in this case, as might be expected from the non-bonding character of the degenerate orbital. A recent calculation gives 0.86 kcal/mole for the dis- tortion energy of the ground state of this radical.30 It is also interesting to note the small quantum defect (0.046) of the Rydberg states observed. The very low ionization potential observed for the tropyl radical is important in comparing the stabilities of the isomeric tropyl and benzyl ions, since mass- spectrometric studies of tropyl compounds give a value AHf = 217f6 kcal/mole for the tropyl ion which is not significantly lower than that for the benzyl ion (220f3 kcal/mole).4 This is a surprising result since there is clear evidence that benzyl ions formed with small excess energies from toluene 149 15 and from benzyl radicals 16 rapidly isomerize to the tropyl ion in which all the hydrogen atoms are equivalent, and that this ion is formed at the threshold from benzyl halides, where the appear- ance potential of the C7HT ion is 6-10 kcal/mole lower than that calculated from bond energies and the ionization potential of the benzyl radical.17 In comparing the heats of formation of the benzyl and tropyl ions, both mass- spectrometric and spectroscopic ionization potentials have to be used.Unfor- tunately, little comparable data on free radicals have been obtained from the two sources. The only other spectroscopic values of free radical ionization potentials are I(CH3) = 9-843 eV and I(CH2) = 10.396 eV determined by Herzberg.18. 19 The former agrees well with mass-spectrometric values of 9-95,20 9-85,21 and 9-88 22 eV, but the latter does not agree well, the accepted mass-spectrometric value being 11.8 eV.21 This discrepancy may be due to the very reactive nature of the methylene[To facepage 198.B . A . THRUSH A N D J . J . ZWOLENIK 199 radical. While molecular ionization potentials based on Rydberg series are gener- ally lower than those determined mass-spectrometrically, there is no obvious regularity in the difference.There is, therefore, no reason to assume that the mass-spectrometric ionization potential of the benzyl radical is in error by the 0-3 eV discrepancy observed with the tropyl radical. It is perhaps significant that Streitwieser's calculations 5 using a modified molecular orbital method, in which some of the coefficients depend on mass-spectrometrically determined ionization potentials, give excellent agreement with most mass-spectrometric ionization poten- tials, but predict a value of 6-41 eV for the tropyl radical, which is 0-2 eV below the mass-spec tr ometric value. A value of I(C6H5CH2) = 7.76 eV is well authenticated and appears to be the true vertical ionization potential of the benzyl radical.16, 23 The recommended value of A&(C&CHz) = 43 kcal/mole 249 25 is supported by recent work on the pyrolysis of toluene.26 These data give AHf(c6H~cHt) = 222 kcal/mole. The heat of formation of the tropyl radical has been estimated by Dauben, Lossing and co-workers 4 from a calculated value for ditropyl vapour (94.6 kcal/ mole) and mass-spectrometric observation of the temperature at which it decom- poses to yield tropyl radicals. Their data indicate D(C7H7-C7H7) = 35+5 kcal/mole and hence AHf(C7H7) = 65 k 3 kcal/mole, which with Z(C7H7) = 6.237 eV gives AHf(C7Ht) = 209+3 kcal/mole. Thus, the rearrangement of the benzyl ion to the tropyl ion is about 13 kcal/mole exothermic while the corresponding radical rearrangement is about 22 kcal/mole endothermic.TABLE 2.-APPARENT AHj-(C,H+) FROM APPEARANCE POTENTIAL MEASUREMENTS compound toluene ethyl benzene propyl benzene di benzyl cycloheptatriene 7-methyl cycloheptatriene benzyl chloride benzyl bromide benzyl iodide A.P. CC,H', eV 11.8 11.25 11.23 10.53 < 10.1 < 9.5 10.35 9-67 9.23 AHf of compound AHf(C7Hf) kcal/mole kcal/mole 12.0 7.1 1.9 28.0 43.5 37.2 4.0 19 28 232 234 236 228 Q224 <224 214 21 5 21 5 ref. 27 27 15 27 4, 28 4 17, 29 17, 24 17, 24 Table 2 gives heats of formation of the ion of formula C7HT based on its appear- ance potential from various compounds containing benzyl or tropyl groups, and their heats of formation used in the calculation. The < sign indicates evidence that the C7H; ion was being formed with excess energy. For the cycloheptatriene derivatives, this threshold is surprisingly high compared with that for the benzyl halides, where the dissociative ionization which must yield tropyl ions occurs with only ca.6 kcal/mole excess energy. (This excess energy would be zero or negative if the mass-spectrometric value of I(C7H7) were used.) The observation that the apparent heat of formation of C7Ht for benzyl hydrocarbons is greater than that of benzyl ion does not prove that this ion is produced directly in the dissociative ionization. Investigation of the metastable peak corresponding to the process C7H$+C7Hf+H for deuterated toluenes shows that randomization of the hydrogen is almost com- plete before dissociation occurs.14200 SPECTRUM OF THE TROPYL RADICAL Our spectroscopic observations of the structure and ionization potential of the tropyl radical are clearly in good agreement both with theoretical predictions and with mass-spectrometric studies of compounds containing benzyl and tropyl groups.The authors thank the National Science Foundation for the award of a post- doctoral fellowship to J. J. 2. 1 Huckel, 2. Physik, 1931,70, 204; 2. Elektrochem., 1937,43,752. 2 Longuet-Higgins and McEwen, J. Chem. Physics, 1957, 26, 719. 3 Thrush, Nature, 1956,178, 155. 4 Harrison, Honnen, Dauben and Lossing, J. Amer. Chem. SOC., 1960,82,5593. 5 Streitweiser, J. Amer. Chem. SOC., 1960, 82, 4123. 6 Dauben, private communication. 7 Norrish and Thrush, Quart. Rev., 1956, 10, 149. 8 Thrush, J. Phot. Sci., 1960, 8, 232. 9 Field and Franklin, Electron Impact Phenomena (Academic Press, New York, 1957). 10 Watanabe, J. Chem. Physics, 1957, 26, 542. 11 Fateley, Curnutte and Lippincott, J. Chem. Physics, 1957, 26, 1471. 12 Jahn and Teller, Proc. Roy. SOC. A, 1937, 161, 220. 13 Longuet-Higgins, Opik, Pryce and Sack, Proc. Roy. SOC. A, 1958, 244, 1. 14 Rylander, Meyerson and Grubb, J. Amer. Chem. Soc., 1957,79, 842. 15 Meyerson and Rylander, J. Chem. Physics, 1957, 27,901. 16 Pottie and Lossing, J. Amer. Chem. SOC., 1961, 83, 2634. 17 Meyerson, Rylander, Eliel and McCollum, J. Amer. Chem. SOC., 1959, 81, 2606. 18 Herzberg and Shoosmith, Can. J. Physics, 1956, 34, 523. 19 Herzberg, Proc. Roy. SOC. A, 1961,262,291. 20 Lossing, Ingold and Henderson, J. Chem. Physics, 1954, 22, 621. 21 Langer, Hipple and Stevenson, J. Chem. Physics, 1954,22, 1836. 22 Osberghaus and Taubert, 2. physik. Chem., 1955,4,264. 23 Farmer, Henderson, McDowell and Lossing, J. Chem. Physics, 1954, 22, 48. 24 Benson and Buss, J. Physic. Chem., 1957, 61, 104. 25 Sehon and Szwarc, Ann. Rev. Physic. Chem., 1957, 8,439. 26 Price, Can. J. Chem., 1962, 40, 1310. 27 Schissler and Stevenson, J. Chem. Physics, 1954, 22, 151. 28 Finke, Scott, Gross, Messerly and Waddington, J. Amer. Chem. SOC., 1956,78, 5469. 29 Kirkbride, J. Appl. Chem., 1956, 6, 11. 30 Hobey and McLachlan, J. Chem. Physics, 1960,33, 1695.

ISSN:0366-9033    

DOI:10.1039/DF9633500196

出版商:RSC    

年代:1963

数据来源: RSC

摘要:

Ionization and Dissociation Energies of the Hydrides and Fluorides of the First Row Elements in Relation to their Electronic Structures BY W. C . PRICE, T. R. PASSMORE AND D. M. ROESSLER The Wheatstone Laboratory, King’s College, London, W.C.2 Received 28th January, 1963 Curves of the bond dissociation energies of the hydrides and fluorides of the first row elements are given together with the curves for the corresponding processes in the isoelectronic ions. These show the dependence of the energies on the closed shell configurations appropriate to the shape of the radical or ion. When the closed shells are exceeded, stabilization of triatomic species is achievea by bending while planar tetratomic ones become pyramidal. The curves are also used to obtain the proton affinities of H20 and NH3.The availability of reliable data on the bond dissociation energies of some radicals and molecules together with spectroscopic or photoionization values for their ionization potentials enables the bond dissociation energies of the ionic species to be derived. With these data it is possible to plot isoelectronic curves such as those shown in fig. 1 and 2 for the dissociation energies of H,X-H and H,X+-H. FIG. 1.-Plots of dissociation energies of H,X-H and H,X+-H for the first row elements. The value for the dissociation energy of the isoelectronic ion is plotted along the same ordinate as that of the corresponding neutral molecule or radical; thus D(Be-H) and D(B+--H) are plotted at the same value of x, there being unit dis- placement along the x axis for X = Li, Be, B, etc.The curve connecting the dis- sociation energies of the ions should resemble in its general shape that connecting the dissociation energies of the corresponding isoelectronic neutral species. To get from the latter to the former all that has been done is to increase the charge on X by unity. It can be seen for the hydrides that the dissociation energies of the ions are uniformly greater than those of the isoelectronic species. This is mainly attributable to the polarization energy of the hydrogen atoms in the ionic species. G* 201202 IONIZATION A N D DISSOCIATION ENERGIES OF RADICALS While many of the points on these curves have been determined with certainty, others such as N-H and N+-H are less accurately known. In these cases the adjacent points CH, C+H, OH, O+H and FH, whose values are well established, enable the values for NH and N+H to be determined by virtue of the similarity which exists between the curves.The detailed discussion of the dissociation energy curves which follow refers with one or two exceptions to the ground states of the radicals or their ions. ktrahedrol FIG. 2.--Plots of dissociation energies of F,X-F and F,,X+-F for the first row elements. THE X--H AND X+-H CURVES Most of the values are well established. Those of B+-H, N+-H and Ff-H are interpolated. That of F+-H does not correspond to the lowest state of the ion which is formed from F+H+ not F++H. The value used corresponds to the latter pair of dissociation products, i.e., to the removal of a nF electron (the same may be true for N+-H).The ions are formed by the removal of sa electrons for Li, pa electrons for Be and B and non-bonding n electrons for C, N, 0 and F. Dissociation is into neutral H and the lowest state of X or X+. The general form of the XH curve can be understood as the passage from sa binding in LiH topa in BeH reduced by the promotion energy required by Be(1S). This is followed by a similar bond in BH, which is stronger because less promotion energy is involved. The ionization of BH may, in an oversimplified way, be described as that of the unbonded spa electron produced in the promotion which results in two oppositely directed spa orbitals. From CH to FH the ionization is of a non-bondingpn electron, the bond energies of these radicals increasing with increasing polarity. There appears to be no irregularity arising from the different amounts of energy required to decouple the spins in the ground states of C, N and 0.As can be shown for C this energy seems only to be about 0.2 eV with similar amounts for N and 0. [The accurate spectroscopic data on the ionization energies of C(3P), CH,1 CH22 and CH3 can be explained assuming a polarization stabilization in the ions of 0.4 eV per hydrogen atom, a spin correlation energy in C(3P) of 0-225 eV and a spin correlation energy in CH3(3E;) of 0.16 eV according to the following equations : I(C3P) -Z(CH) = 1 1.265 - 10.64 = 0-625 = 0.40 + 0.225 eV, I(CH)-I(CH2) = 10.64-1040 = 0-24 = 0-40-0.16 eV, I(CH2)-I(C€€3) = 1040-9.84 = 0-56 = 0.40+0.16 eV.]W. C .PRICE, T. R . PASSMORE A N D D. M. ROESSLER 203 The low value for Li+H represents a common phenomenon found when the ionization can be regarded as that of an electron from a single X-H or X-F bond. When X has a low ionization potential, e.g., X = Li (5.39 eV), HBe (8.4 eV), H2B (8.2 eV), H3C (9.84 eV) the low values found for the dissociation energy of the ion are 0.8, 0.9, 0.9 and 1.26 eV respectively. This can be explained in terms of the poor exchange to be expected, the electron spending most of its time on the hydrogen. It also explains the diminishing proportion of parent RHf ions in fragmentation in mass spectrometry with increase in the size of the alkyl radical R, this being associ- ated with the diminishing ionization potential (i.p.) of R, resulting in a low bond energy R+ .H. The values are even lower for the fluoride ions as expected from the increase in the i.p. of F relative to H which leads to a further reluctance to exchange its electron with an attached radical of much lower ionization energy. The polarity of the bonds is increased in going from XH to X+H with correspond- ing increases in their strengths. For the fluorides it is diminished or reversed with in most cases reduction in bond dissociation energy. The bond energy of HeH is taken as zero. That of Ne+H could be evaluated by extrapolation to be about 9 eV but this would be an excited state which might rapidly dissociate into Ne+H+. There are, however, many cases of radicals containing heavier ionized inert gases where this dissociation does not occur and which are strongly bound.For example, the energies of A+-H and Mr+-H are greater than 4-5 eV since they are formed by the interaction of the inert gas ion with H2.3 These are to be compared with 4.43 eV for HCl and 3.75 eV for HBr their isoelectronic analogues. THE HX-H A N D HX+-H CURVES The values of HC-H, HCf-H, HN-H, HN+-H, HO-H and HO+-H which establish the major parts of the curve are fairly well known. They are given in table 1. HLi-H is assumed to be zero and HBe+-H determined from the curve mentioned in the discussion on Li+H. Its low value is to be correlated with the high value of D(Be+-H). The values of HBe-H and HB-H presented some difficulty. They are based on interpolations from the accepted values of H,C-H, H,N-H and their ions through the isoelectronic curves.That of HB-H depends, for example, on taking the value of the total bond energy of BH3 as 11.6 eV,4-6 subtracting B-H to give the sum of HB-H plus H2B-H and dividing this between them as indicated by the XH2 and XH3 curves. The high value of HB-H (4.7 eV) arises from the large overlap of its colinear sp bonding. The actual magnitude is important in that when compared with HC-H (5.45eV) it shows that the major factor in the binding of CH2 is also sp bonding and the additional stabilization arising from the spin correlation in the 3& ground state of CH2 is only a few tenths of a volt. The low value of the i.p. of BH2 (8.2 eV) which is confirmed by a value of 8.1 eV from an analysis of appearance potential data,31 is to be associated with the fact that a non-bonding p n electron is removed (8.3 eV in B, 2P+) and that the dissociation energy of the ion HB+-H acquires 0.4 eV polarization stability relative to the neutral radical.In this latter respect it differs from CH;! where the gain of 0.4 eV polar- ization energy is offset by loss of spin correlation energy (0-16 eV). Note that the upward kinks in the Be, B, C and B+, C+, N+ curves of the triatomic radicals are due to the effect of the spin correlation energy. The downward ones for the tetratomic species arise from departures from planarity. Maximum bond energy is achieved for the linear XH2 structures for CH2 and N+H2. It appears that in addition to the closed shell (ug)2(au)2 system the structure can accept one p electron in each plane on the central atom.These two electrons in mutually perpendicular orbitals are virtually non-bonding because of their204 IONIZATION AND DISSOCIATION ENERGIES OF RADICALS TABLE TA TABLE OF DISSOCIATION AND IONIZATION ENERGIES OF THE HYDRDES OF THE FIRSTROW ELEMENTS EXPRESSED AS D(HnX-H) : D(HnX+- H) 1WnX-H) : I(HnX) LiH BeH BeH2 2.43 (0.8) 2-3 3.2 (4.4) (0.9) (7-0) 5.39 8-4 9.32 (11.9) 8.4 BH BH2 BH3 3.39 (3.0) (4.7) (5.2) (3.2) (0.9) (8.7) 8.296 (8.2) 8-7 (10.5) (8.2) CH CH2 CH3 CH4 3-47 4.10 5-45 5-69 3.90 4.46 4-40 1.26 10.64 11-265 10.396 10.64 9.84 10-396 12.98 9.84 NH NH2 NH3 NH4 3.7 (4.3) 4.2 (6.8) 4.4 5.6 (0.) (5.5) (13*9)* 14.545 11-3 (13.9) 10.15 11.3 (4.7) 10.15 OH OH2 OH3 FH 440 (5.0) 5-20 (5-6) (09 (6.4) 5.87 (6.5) 13-0 13.615 12-61 13.0 (6.2) 12-61 (1 6*8)* 17.42 Brackets indicate values interpolated from isoelectronic similarity curves of radicaIs and ions.* Starred values do not correspond to lowest i.p.s. The interpolated values differ in many cases from those given previously (Price, Harris and Passmore, J . Quant. Spectrosc. Radiat. Transfer, 1962, 2, 327, due to the inadequacy of the data then available for plotting the isoelectronic curves. central position in the molecule though they do occupy the lowest nu bonding orbitals. (Only terminal p n electrons entering this orbital can be bonding.) The system does, however, derive some stabilization from the unpairing of their spins. This is lost when they are paired in the singlet system. For this system the most stable configur- ation is the bent one. In this respect the XH2 radicals differ from the XF2 radicals which cannot accept anypn: electrons on the central atom in the linear configuration because of the large repulsions which would arise with thepn electrons on the terminal F atoms.These repulsions are comparable with those in the diatomic species NO or CF and are of several eV in magnitude. This is equivalent to saying that in XF2 the pF electrons push the pX electrons out of the lowest bonding n orbital to the first antibonding one. By symmetry they cannot enter the second bonding n orbital. A radical change in structure occurs in going from CH2 and N+H2 to NH2 and O+H2. It is no longer energetically profitable to promote to a linear (sp) structure which could not be assisted by spin correlation due to the necessary pairing of the additional electron.The repulsion between the rc2 and the bond orbitals is probably also important. The change from linear to bent structure shows up strongly as a sudden drop in HN-H relative to HC-H (and HOf-H relative to HN+--H). The fall is about 1.5 eV. The changes that occur as bent NH2 is ionized to the linear triplet N+H2 are in striking contrast to what happens in ionizing CH2 or H20 where little change in shape or bond type accompanies the ionization. This is re- flected in the changes in i.p. for XH and XH2 set out below : I(CH) - I(CH2) = 10.64 - 10.40 = 0.24 eV, I(OH)-I(OHz) = 13-0- 12-61 = 0.39 eV, I(NH) - Z(NH2) = 13-9 - 11 -3 = 2.6 eV.W. C. PRICE, T . R. PASSMORE AND D. M. ROESSLER 205 The first five ionizations involve simply the removal of a non-bonding electron but with NH2 a reorganization of the remaining electrons accompanies this removal.The increase of HO-H relative to HN-H is presumably due to the greater electronegativity of 0 relative to N as in the diatomic case and also because no spin decoupling is required in its formation from OH(2II). In spite of its high predicted total bond energy HF+-H is not stable because of dissociation into HF+H+. It is expected, however, that €€2C1+ should be stable. THE H2X-H AND H2X+-H CURVES The known points on these curves are HzC-H, H2Cf-H, HzNf-H and H2Ot-H ; the last is derived from the proton affinity of H20 7 given as 7.35 Ifr 0.1 eV in a private communication, a value which leads to D(HzO+-H) = 6.4+0-1 eV. The dissociation energy of H2B+. H can be expected to be low because it dissociates into the strongly bound " closed shell " linear B'H2, the stability of which is shown by its high abundance in mass spectrometry apart from the dissociation energy derived from the previous set of curves and the low i.p.of BH2. The value taken for H2B+ . H is 1.6 eV. This derives from a Rydberg series in BH, going to 9.8 eV found by Herzberg (private communication) which accords with an a.p. of 11.9 eV for BH+ and a value of 1.8 eV for BzH642BH3. In addition it requires I(BH2) = 8-2 and D(H2B-H) = 3.2 eV which comes from the known values of B-H 8 and the total bond energy of BH3 by dividing the difference between HB-H and H2B-H in the only way that allows the curves of the energies of the neutral and isoelectronic radicals to retain their feature of similarity.The bonding in BH3 is not as high as in BH2 presumably because it corresponds to sp2 as against sp bonding. H2C-H (3-9 eV) is weaker than expected (mean bond energy CH3 = 4.27 eV) because of the extra binding due to spin correlation in CH2, i.e., one of its dissoci- ation products has extra stability. Here is an independent indication that spin correlation amounts to only a few tenths of a volt. If it were larger, then H2C-H should be much weaker than it is found to be since all stabilization from this cause is lost in going from CH:! to CH3. Although it might be expected from the change in shape in going from BH3 and CH3 to NH3 that there is appreciable change in bond character, this is accompanied by little change in mean bond energies.These are 3.87, 4.27 and 4.1 eV, respectively, for BH3, CH3 and NH3. This is not sur- prising since it only requires 0.25 eV, the barrier height in NH3, to convert this molecule to the planar structure similar to BH3 and CH3. The fact that H2N-H is greater than H2C-H although their mean bond energies are in the reverse order is mainly a result of HN-H (bent) being lower than HC-H (linear). The value of H20+-H could be fixed by the known points involving Cf, C, N and N+ and it is satisfying that a value thus predicted should agree so well with results from ion impact 7 and considerations of heats of formation of crystalline perchlorates and lattice energies.99 10 It enables an effective ionization energy of 64eV to be associated with H3O (see table 1) which explains how this group can act as a metal in combination with electronegative atoms.The soundness of this extrapolation from the neutral NH3 to the isoelectronic positive ion O+H3 can also be supported from considerations of bond lengths and force constants. The positive charge contracts the bonds and tightens the binding of the system. The similar extrapolation of the properties of NH$ from those of CH4 will be demonstrated in the next section. THE H3X-H AND H3X+-H CURVES The known values are H3C-H and H3C+-H. Assuming H3B-H to be zero and drawing parallel curves gives 5.5 eV for H3N+-H, a value which agrees with206 IONIZATION A N D DISSOCIATION ENERGIES OF RADICALS the heat of formation of NHi(g) of 150 kcal calculated by a cyclic process from the heat of formation of crystalline ammonium chloride and the electron affinity of chlorine 8 .9 and with the value obtained in ion impact studies.7 Assuming (H3N-H) = 0 and using the i.p. of NH3 = 10.15 eV a value of 4.7 eV is obtained for the ionization energy of NH4 which explains why it acts as an alkali metal. By a similar method the ionization energy of PH4 can be found to be 6.4 eV, the higher value correlating with its lower electropositive character. The bond distances of NHt, CH4 and BH, are 1-03, 1.093 and 1.26 A and their v3 frequencies 3146,11 3020 and 2300 cm-1 respectively. These values fall in line with the progressive decrease in dissociation energies. BOND DISSOCIATION ENERGIES OF THE FLUORIDES OF THE FIRST ROW ELEMENTS A N D THEIR IONS The isoelectronic curves of the fluorides XFn do not show such close quantitative similarity as those of the hydrides because the bonds are strongly polar and in- creasing the charge on X greatly diminishes and possibly reverses the bond polarity.As a result, the bond dissociation energies of the ions are mostly less than those of the corresponding isoelectronic neutral species. The positive charge is no longer located mainly on the X atom but spreads out to the fluorine atoms. Because of the nature of the closed shells available to R: electron systems, the shapes of the radicals differ in certain cases from those of the corresponding hydride radicals (cf. CF2 and CF3 as compared with CH2 and CH3). Also, whereas it can be assumed that He-H, HLi-H, H2Be-H and H3B-H have zero dissociation energy, this is not so for the fluorine analogues because of the high electron affinity of F.How- ever, it does appear that these attachment energies are small. In the later discus- sions it is assumed that in the structures of the radicals and their ions only the ground states of X, X+ and F are involved. TABLE 2.-TABLE OF DISSOCIATION AND IONIZATION ENERGIES OF THE FLUORIDES OF THE FIRST ROW ELEMENTS EXPRESSED AS D(F,X-F) : D(FnX+-F) I(F,X-F) : I(F,X) LiF BeF BeF2 6-0 0.3 5.4 (5.6) (7.5) (0.5) 1 1 - 1 5.39 (9-1) 9.32 (16.1) (9.1) BF BF2 BF3 8.5 (5-2) (4-2) (6.3) (7.0) 0.8 (11-6) 8-296 (9.5) (11.6) (15-7) (9-5) CF CF2 CF3 CF4 4-9 7.3 5.2 3.0 (4.8) (5-8) 5.3 0 8-91 11.265 11.1 8.91 10.1 11.1 15.4 10.1 NF NF2 NF3 NF4 (3-15) (4.8) 3.0 (3.8) 2-5 (1.4) 0 (4.4) OF OF2 OF3 F2 (12-9) 14.545 (121) (12-9) 13-2 (12.1) (8-8) 13.2 (2.25) (3.7) (1.7) (0-3) (0.7) (0.0) 1-63 (3.3) (12-2) 13.615 13-6 (12.2) (14.3) 13.6 (15-7) 17-42 Brackets indicate values interpolated from isoelectronic similarity curves of radicals and ions.W.C . PRICE, T. R . PASSMORE AND D. M. ROESSLER 207 THE X-F AND X+-F CURVES The points well-established on these curves are those for LiF,12 Li+F,13 BF, CF, C+F 14 and F2. With a lower accuracy we have BeF, Be+F, OF and O+F, some of the values of which will be derived. A value of 3-3 eV for D(F,C) given by Iczkows and Margrave 15 and based on an ionization potential of 15-7 eV for F2 from a tentative Rydberg series was at first thought to be too high to be acceptable.However, it fits in with the flatter nature of the ion curve as compared with that of the neutral molecules. We take 3.3 and 15.7 eV respectively for D* and I. The first pair of points on these curves have low values. The bond energy of Li+ . F is low as is apparent from its appearance potential and in agreement with its expected small exchange. That of HeF is not known but in view of the electron affinity of fluorine and the high ionization potential of helium an attachment value of a few tenths of a volt is expected, slightly more than Li+. (Note I(He) = 24.6 eV. I(Li+) = 75 eV.) The next pair Li-F (6.0 eV) and Be+-F (5.6 eV) about equal in magnitude though probably different in character. Whereas in LiF the bond is mainly ionic, in BefF it is mainly covalent since the ionization energy of Be+ is 18.21 eV as compared with 17-42 eV for F.The near equality of these differen- bonds is not unexpected since the bond energy in HF (5.87 eV) which is probably nearly 50 % ionic is also almost identical with that in LiF. As in the hydrides, the dissociation energies of BeF and BfF are lower than the preceding pair due to the promotion energy necessary to convert the 1S to the valence state. These molecules can also be regarded as having “ closed shell minus one” configurations and the values are consistent with this approach also since they are slightly higher than those of the corresponding “ closed shell plus one ” structures. This agrees with the expectation that the ratio of the antibonding to the bonding power of an electron in these orbitals is as (1 + S)/(l- S ) , where S is the overlap integral. The bond energies reach their peak in BF and C+F which have closed shell structures of bonding electrons.They are isoelectronic with N2, COi, NO+ and have comparable binding energies. The next pair, CF and N+F, have structures involving a closed shell plus one antibonding n electron. The antibonding power of this electron is large whereas the effect of additional antibonding electrons be- comes progressively smaller. A simplified theory based on the assumption that in CF, for example, “ antibonding ” involves resonance to structures C=F+, predicts relative total antibonding powers of approximately 1 : 1.5 : 1.75 and 1-88 for 1 , 2, 3 and 4 antibonding electrons respectively (when two antibonding electrons are present as in NF, two p electrons cannot simultaneously leave the fluorine, thus restricting the operative time for repulsion).This is close to the ratio found for a number of sets of such molecules for which dissociation energies are known. The radicals NF and OF+ are isoelectronic with 0 2 , the 3Zg ground state of which is stabilized by interaction with triplet states of the same symmetry arising from 3P+1D atomic terms. The stabilization in 0 2 is about 1 eV as can be determined by its departure from a plot of the dissociation energies of BeO, BO, CO, NO, (00), OF as ordinate with unitary abscissa displacement for successive molecules. A curve through B2, CZ, N2, (02), and F2 also gives this value for the extra stabilization.However, an examination of the term values of the low states of N,O and F indicates that it is unlikely that stabilization of this magnitude is present in N F and OF+. The former is estimated as 3-15 eV both from the BF, CF, (NF), OF, FF curve and also from a BN, CN, NN, NO, (NF) curve. The dissociation energy of OF+ is Its large value is partly ionic in origin, cf Ne’F. -208 IONIZATION AND DISSOCIATION ENERGIES OF RADICALS taken as 3.7 eV from the CfF, N+F, (O+F), F+F curve and also a C+O, N+O, OfO, (O+F) curve. Both these values fit in well with the energy schemes and ionization potential data for NF3,2,1 and OF1,2 radicals.16917 The value 2.3 eV has been taken for OF partly because it fits in with the division expected for the total bond energy of OF2 (3.9 ev) between the fist and second bond dissociations when compared with the other halogen oxides.In these, steric factors make the first dissociation less than the second by amounts which are greater in C120 and Br2O than F20. The value taken for OF fits the present curve. It can hardly exceed the well-established value for F2 by more than 0.6 eV in view of the progressively diminishing effect of added antibonding electrons. THE FX-F AND FX+-F CURVES These curves show the dependence of binding on closed shell structures. They build up to a high maximum in BeF2 which is a linear closed shell structure analogous to C02. The additional p electron in BF2 causes the radical to bend as in the iso- electronic radical N02, thep orbital lying along the bisector in the BF2 plane.This structure can take another p electron which is paired with its predecessor giving a fairly stable CF2(1A) which is a subsidiary closed shell structure. The source of this stabilization is the high inductive effect of the F atoms on paC electrons. Their orbitals are oriented in a suitable way to be taken in by the fluorine atom to complete its closed shell and so to satisfy its electron affinity. This is not the case for sub- sequently added p n electrons which have to stand perpendicular to the plane and become involved in strong resonance repulsion with the pn electrons on the fluorine atoms. This leads to rapid reductions of bond energy for FN-F and FO-F relative to FC-F similar in magnitude to the repulsions found in the analogous diatomic cases.The first neutral radical and isoelectronic ion pair on these curves is FLi-F and FBe+-F. Both are assumed linear 2E. Although no work on FBef-F has yet been reported the ions of MgF2 and MgC12 are stable enough to be observed.18 The appearance potentials of these ions combined with bond dissociation data in- dicate values of up to 1 eV for XMg+ . X. A value of 0.5 eV is taken for FBef . X and 1.4 eV for FLi-F. This may seem arbitrary but they are both small and yet they cannot be zero as they correspond to structures which are only one electron less than linear triatomic closed shells. For comparison with FBef . F, we have Lif . F = 0.3 eV, F2B+ . F = 0.5 eV 19 and F3Cf. F = 0.0 eV. Their exact magnitudes do not greatly affect the discussions which follow.The next pair FBe-F and FB+-F are the linear closed shell species. The energy of the first of these is found from the total energy of BeF2 and the spectroscopic value for BeF. There is no experimental value for the former but a value can be obtained by extrapolation from experimental values for BeC12, BeBr2, Be12 using as a comparison the series MgF2, MgC12, MgBr2 and MgI2.18 This leads to a value of 12.9 eV from which we obtain 7-5 eV for FBe-F by subtracting 5-4 eV for Be-F. This high value is to be expected in view of the linear closed shell structure of BeF2. For the dissociation energy of FB+-F it is necessary to choose a value of 6-3 eV to fit in with several experimental facts. The derivation consists in fitting in values of the dissociation and ionization energies of the BF3 section of table 2.In this section all the values for BF are established and in addition the i.p. of BF3 is known (15-7 eV) and the dissociation energy of F2B+. F (0.5 eV) 19 can be derived from its a.p. The total bond energy of BF3 using the latest value for the heat of sub- limation of boron 20 is 19.7 eV from which subtraction of 8-5 eV for B-F givesW. C. PRICE, T. R . PASSMORE AND D. M. ROESSLER 209 11.2 eV for the sum of FB-F and F2B-F. The latter of these is expected to be much greater than the former because of its closed shell configuration, Also the former is weak because it dissociates into the strongly bound closed shell BF. It is also expected to be bent (cf. N02). The dissociation energy of FBf-F is expected to be high because it is a linear closed shell species, a fact supported by the high abundance of BF; in mass spectrometry.19 Fairly well-established values for the dissociation energies of the adjacent radicals CF2, C+F2 and N+F2 help to fix F2B-F at 7.0 eV and FB-F at 4.2 eV.These lead to a value of FB+-F of 6.3 eV, an ex- pected high value, and an i.p. of 9.5 eV for BF2, a low value which can be compared with the i.p. of 8-2 eV derived for BH2, also a low value (compare the i.p.s CH2, 10.40 eV and CF2, 11.1 eV, also OH2, 12.61 eV and OF2, 13.6 eV). It is clear that the variation in ionization and dissociation energies of the boron fluorides exemplify four types of closed shell, namely, BF, B+F2, BF3 and BF:. The values of the heat of formation 21922 of CF2 lead to values of FC-F which vary from 4-85 to 5.65 eV.We take a value of 5.2 eV as most probable and fitting our curve best. Combined with an i.p. of CF2 of 11-1 eV (originally given by Mar- grave 23) but supported by spectroscopic and photoionization work on C2F4 2425 and appearance potential data23 also comparison with the i.p.s of BH2, BF2 and OH2, OF2 given above), this leads to a value of 3.0 eV for FC+-F. For the nitrogen fluorides the values are established by taking the diatomic values from the curve, accepting a value of 2.5 eV for F2N-F 2 6 2 7 and thus obtaining FN-F by sub- traction from the total bond energy. The ionization energies of NF2 (12-1 eV) 28 and NF3 16 (13.2) permit the calculation of FN+-F and F2N+-F. The former of these together with FC-F and FC+-F enables FB-F to be located at 4.2eV by assuming similar behaviour of the isoelectronic curves.This then gives the line between FBe-F and FB-F and a roughly parallel line through FC+--F locates the value of FB+-F at 6.3 eV. THE F2X-F AND FzX+-F CURVES The curves of F2X-F and F2X+-F rise to maxima for the planar closed shell n:nz structures BF3 and C+F3. (n1 has no transverse nodal plane, n2 has a central nodal plane perpendicular to the molecular plane.) Similar closed shell structures are responsible, for example, for the stability of the carbonate and the guanadinium ions and govern chemical activity in a wide range of substances of similar structure. We would even include benzene with these structures, particularly since none of these structures can have a pn electron located on the central carbon atom.This can only happen in the hydrides XH, where a non-bondingp electron is the sole occupant of the n1 orbital. Because of the central nodal plane in 712, a centralpn electron can- not contribute to a n2 molecular orbital. Thus, when there are terminalpn electrons also contributing to the n1 molecular orbital, the central pn electron is pushed largely into the antibonding n3 molecular orbital the nodal surface of which passes through all three bonds. (The bonding of the n1 orbital formed from a centrally located pn atomic orbital in XH, can only be between the terminal H atoms. It is doubtful whether this can be different from zero.) Apn electron introduced on to the central atom of XF3, as in CF3 becomes involved in strong resonance repulsion with the p n orbitals of the terminal atoms.The interaction is between the ynC electron and electrons in a n(F + F + F) orbital. (It is convenient to consider sub-molecular orbitals made up only of F orbitals. Only those of the right symmetry can interact with the central C orbital which is the two electron orbital given above.) The inter- action is thus similar to that in CF where the antibonding relative to BF is indicated by the difference in dissociation energy, viz., 8-5-5.0 = 3.5 eV. Without exception,210 IONIZATION AND DISSOCIATION ENERGIES OF RADICALS this repulsion is relieved by bending out of linearity in the triatomic case or out of the plane in the tetra-atomic case. In this way the resonance causing the repulsion is reduced by the consequent departure from parallelism of the axes of the p orbitals.The following comparison giving the differences in the bond dissociation energies for " closed shell " and " closed shell plus one " structures illustrates this : BF-CF = 8.5-5.0 = 3.5 eV, (FBe-F)-(FB-F) = 7.5-4-2 = 3.3 eV, (F2B-F)- (F2C-F) = 7.0-4.8 = 22eV. Several factors affect these differences but the antibonding power of the additional electron is undoubtedly the major one. The antibonding power of a second antibonding electron does not fall off relative to that of the first in CF3 and NF3 as in CF and NF, the antibonding being almost equal for the tetra-atomics instead of reduced by half as it is in going from CF to NF. This is thought to be connected with the possibility of resonance structures occurring in the tetra-atomic case in which several pF electrons can interact simultaneously with the p X electrons without invoking structures involving doubly or triply charged fluorine atoms as would be necessary in the diatomic cases. The only points on these curves which have not been considered are those of Be, C and Cf.The first of these FZBe-F would be expected to be low but not necessary zero since the structure corresponds to one electron less than the closed shell BF3. A value of about 2 eV obtained by drawing the BBe line parallel to the C+B+ line seems to be about right though it might be less than this. The value of FZC-F depends on the heat of formation of CF4,29 that of F3C-F 30 and the values of FC-F and C-F already discussed.The i.p. of CF; 30 together with the above data gives FzC+--F. THE F3X-F AND THE F3Xf-F CURVES The two points known on these curves are F3C-F 30 and F3C+-F, the latter being zero. A low value of a few tenths of a volt is assumed for F3B-F (it might be zero). Joining points B to C and drawing the C+F+ line parallel indicates a value of F3N+-F only slightly less than that of F3C-F. Little is known of the NfF4 ion, in contrast to the N+H4 ion. Its ionization potential can be derived as ca. 8 eV from NF3+F = 0, N+F3+F = 5 eV and i.p. NF3 = 13-2 eV. It is clear that its ionization potential is too high and its dimensions too large for it to give enough lattice energy to form crystalline salts. We should like to acknowledge support from the Institute of Petroleum, Imperial Chemical Industries, the Department of Scientific and Industrial Research and the U.S.Dept. of the Army through its European Research Office under Contract DA-9 1-59 1 EUC-1683. Note added in proof: If a Rydberg series converging to 9.8 eV found for BH (Herzberg and Johns, private communication) corresponds to the lowest state of the ion then the following changes have to be made in table I : D(B+-H) = 1-9 eV, I(BH) = 9.8 eV, Z(BH2)=9.3 eV and I(BH3) = 11.6 eV. 1 Herzberg, J . Quunt. Spectrosc. Radiat. Transfer, 1962, 2, 319, and private communication. 2 Herzberg, Can. J. Physics, 1961, 39, 155. 3 Stevenson and Schlissler, J. Chem. Physics, 1955, 23, 1353. 4 Shepp and Bauer, J. Amer. Chem. SOC., 1954,76,265. 5 McCoy and Bauer, J. Amer. Chem. Soc., 1956,78,2061. 6 Gunn and Green, J. Chem. Physics, 1962, 36, 1 1 18. 7 Tal'rose and Frankevitch, Dokf. Akad. Nuuk. S.S.S.R., 1956,111,376, and later communication. 8 Hurley, Proc. Roy. Soc. A , 1961, 261. 237. 9 Grimm, Z. Elekrrochem., 1925,31,474. 10 Sherman, Chem. Rev., 1932, 11, 94.W. C. PRICE, T . R . PASSMORE AND D. M. ROESSLER 21 1 1 1 Morgan, J. Chem. Physics, 1959, 30, 1212. 12 Brewer and Brackett, Chem. Rev., 1961, 40, 425. 1 3 Berkowitz, Tasman and Chupka, J. Chem. Physics, 1962, 36, 2170. 14 Johns and Barrow, Proc. Physic. SOC. A, 1958, 71, 476. 15 Iczkowski and Margrave, J. Chem. Physics, 1959 30,403. 16 Reese and Dibeler, J. Chern. Physics, 1956, 24, 1175. 17 Dibeler, Reese and Franklin, J. Chem. Physics, 1957, 27, 1296. 18 Berkowitz and Marquart, J. Chem. Physics, 1962, 37, 1853. 19 Marriott and Craggs, J. Electronics Control, 1957, 3, 194. 20 Verhaegen and Drowart, J. Chem. Physics, 1962, 37, 1367. 21 Brewer, Margrave, Porter and Wieland, J. Physic. Chem., 1961, 65, 1913. 22 Majer and Patrick, Nature, 1961 , 192, 866. 23 Margrave, J. Chem. Physics, 1957, 61, 38. 24 Bralsford, Harris and Price, Proc. Roy. SOC. A , 1960, 258, 459. 25 Price, Bralsford and Roessler, Spectroscopy (Institute of Petroleum and Pergamon Press, 26 Kennedy and Colburn, J. Chem. Physics, 1961, 35, 1892. 27 Herron and Dibeler, J. Chem. Physics, 1961, 35, 747. 28 Colborn and Johnson, J. Chem. Physics, 1960, 33, 1869. 29 Cotterill, The Strengths of ChemicaZ Bonds, 2nd ed. (Butterworths Sci. Publ., London, 1958), 30 Farmer, Henderson, Lossing and Marsden, J. Chem. Physics, 1956, 24, 348. 31 Koski, Kaufman, Pachucki and Shipko, J. Amer. Chem. Sac., 1959, 81, 1326. London, 1961), p. 279. p. 245.

ISSN:0366-9033    

DOI:10.1039/DF9633500201

出版商:RSC    

年代:1963

数据来源: RSC

摘要:

GENERAL DISCUSSION Dr. J. H. van der Waals (KoninklijkelShell Lab. Amsterdam) said With reference to Kuppermann and Raff’s paper it is interesting that one can conclude spin-forbiddenness of a transition such as that at 4.6 eV in ethylene from the de-crease of peak height with increase of the energy of the incident electrons. Do the authors have any further information on how unambiguous this criterion is ? Their observation that the ratio of the intensities of the two lowest transitions in the He spectrum remains constant when raising the incident beam energy from 25 to 50 eV seems somewhat disturbing in this context. Prof. A. Kuppermann (University of Illinois) (communicated) In answer to Dr. van der Waals the dependence of the cross section of spin-forbidden transitions on the energy of the exciting electron was established in the past on the basis of theoretical and experimental work done mainly on rare gases.However the conclusions should also be applicable to molecular systems. Not enough work has yet been done with molecules to permit a quantitative assessment of this con-clusion. The 21 eV transition we observed in helium could contain an appreciable contribution from the 23P state (20.96 eV) which could account in part for the approximate constancy of the relative intensities of the two lowest transitions we observe as the incident beam energy goes from 25 to 50 eV. Dr. E. L. Mackor (KoninklijkelShell Lab. Amsterdam) said Kuppermann and Raff state that the lowest maximum in the excitation spectrum at 4-6 eV for ethylene obtained by electron-impact is in good agreement with the 4.6 eV derived from optical absorption spectra for the vertical T t N transition of the n-electron system.I would like to ask why one does not observe in the excitation spectrum intensity at energies lower than 46eV. In the u.-v. absorption spectrum of D. F. Evans, obtained by the oxygen intensification technique absorption sets in at 28,700 cm-1 or - 3.4 eV with a not very pronounced maximum at 4.6 eV. Is there a possibility that the twisting mode manifests itself differently in light absorption and electron impact excitation ? Prof. A. Kuppermann (University of Illinois) (communicated) In answer to Dr. E. L. Mackor our fig. 8 and 9 indicate that absorption corresponding to the maxi-mum at 4.6 eV sets in at about 33.5 eV.The relative cross sections for excitation of states by low energy electrons and electromagnetic radiation can be and usually is quite different. Prof. A. D. Walsh (Dundee) said Any technique that facilitates the study of optically forbidden transitions is of obvious importance; and I should like to con-gratulate Dr. Kuppermann and Raff on the technique they have developed; but there is one point that puzzles me. In the range 7.0-9-0 eV (1771-1377 A) quite a lot of electronic transitions of the C2H4 molecule are either known or expected to occur. The first Rydberg transition is well known to occur at 1744A; 1 this forms the first member of an s Rydberg series the second member lying at 1393 A. Other Rydberg transitions occur at 1501 and 1438.5 A 2 ; these begin p Rydberg series and a further optically forbidden series is expected to begin at a closely adjacent wavelength.In addition there should occur spin-forbidden transitions corresponding to transitions to singlet Rydberg states occurring at shorter wave-lengths. Finally there is the well-known intra-valency shell transition whose 1 Price and Tutte Proc. Roy. Soc. A 1940 174 207. 2 Wilkinson Can. J. Physics 1956 34 643. 21 GENERAL D I S CUSS ION 213 maximum intensity occurs around 1620A. This makes a total of more than five electronic transitions. Fig. 8 9 and 10 however show only two peaks in the range 7.0 to 9.0eV. I realize that lack of resolution will cause adjacent peaks to fuse together but can one say convincingly that the two peaks observed are " in good agreement with the optical values '' ? Can anything be said as to why some expected peaks apparently do not appear? Prof.A. Kuppermam (University of Illinois) (communicated) In answer to Prof. A. D. Walsh Prof. Walsh's point is well taken and points very clearly to the superiority of optical techniques insofar as resolution and accuracy is concerned. Within our reported resolution of 3 eV and the error in the intensities given by the vertical bars in fig. 8 and 9 it seems permissible to make the quoted statement about the agreement between the electron impact and optical spectra. It is hoped that improve-ments in the technique will permit a more meaningful comparison of the two types of spectra. Dr. S. F. Mason (Exeter University) said The answer to the question raised by Prof.Oosterhoff who asked why the vibrational structure is so much sharper in the circular dichroism than in the unpolarized absorption given by disymmetric ketones is probably as follows. In dissymmetric ketones with C2 symmetry such as trans-3,4-dimethylcyclopentanone (I) the electric moment of the 3000 8 ab-sorption has components orientated in the x- y - and z-directions. Only the z-component couples with the magnetic moment arising from the 2pz+2p transition FIG. I. at the oxygen atom to give a non-zero rotational strength which is orientated in the 2-direction along the carbonyl bond. The rotational strength is symmetry-allowed and only the totally-symmetric vibrations belonging to the A representation in C2 can appear in the circular dichroism spectrum.The most important of these vibrations is the carbonyl-stretching mode and it is the upper-state frequency of this mode which appears as a well-defined progression in the 3000A circular di-chroism absorption of dissymmetric ketones. The non-totally-symmetric vibra-tions are known to appear in the unpolarized absorption of ketones at 3000A, and these vibrations having frequencies different from that of the carbonyl stretching mode blur the vibrational structure obtained in unpolarized absorption measurements. Dr. E. Charney (Oxford University) said In view of Dr. Mason's reinter-pretation of the data in terms of the existence of different conformers is there any change in the interpretation of the origin of the optical activity of the 3000 A band? That is is it still required that the magnetic moment of the n+n* transition of the carbonyl groups couple or mix with a parallel component of the electric moment of the 71-71" transition of the skewed a /3 unsaturated system? In the skewed buta-diene systems containing no non-bonding valence electrons activities of very large magnitude are observed from the coupling of the electric and magnetic dipol 214 GENERAL DISCUSSION transition moments from the lowest n+n* singlet transition itself.I believe that in the saturated asymmetric ketones the activity is an order of magnitude smaller, but in this case no mixing with a low-lying n-m* transition having a component in the right direction is possible. The circular dichroism of the 3000 A band of (-) carvone appears from the data to be just about an order of magnitude less than that of the 2350A band arising from the conjugated a /? unsaturated system.Prof. C . A. Coulson (Oxford University) said Linnett and Sovers have shown how, by a suitable choice of non-pairing wave functions it is possible to get a remarkably close approximation to the best available 1.c.a.o.c.i. wave function for allyl. But a difficulty arises on account of the fact that different excited states are best described by different non-pairing functions. This prompts the questions (i) how do you know which non-pairing function to choose as a representation for any particular transition? (ii) in a new molecule for which no 1.c.a.o.c.i. functions are available with which to check any n.p.energy how will you know which n.p. function to choose, and what can be said about its accuracy? Except for the lowest level for which the variation method assures us that any n.p. function must lead to an energy a little above the best c.i. energy and for the top level when the opposite holds (see fig. 1 of Linnett and Sovers' paper) we cannot apparently say whether the n.p. energy will lie above or below the c.i. energy. As for the choice of n.p. function it seems impossible to avoid the introduction of at least one variable parameter such as k (in a localized orbital a+kb etc.). Does Dr. Linnett have any reasons for supposing that any particular value of k may be adopted and then used without change in other molecules as well as the original one? Dr. J. W.Linnett (Oxford University) said In reply to Prof. Coulson I would say that I feel that for a final answer one will have to wait till more experience has been acquired. However I would draw his attention to the paragraph in our paper in which we do suggest a procedure for choosing non-pairing functions for successive excited states. With these we would use in a step-wise manner those functions which were most orthogonal to those lower energy states which had already been formulated. I think there is some hope that this would- be adequate and we are proposing to examine it soon in a slightly more complicated situation than that of allyl. One of our objects in studying a wide variety of related species has been to dis-cover whether there is any regularity in the values of k that are best suited to the different problems.It is worth pointing out that the values of k that have been used for the allyl cation radical and anion are surprisingly constant and it is probable that good results could be obtained if the same value of k were used throughout. I certainly agree with Prof. Coulson that if the method is to be ad-vantageous it is essential that rules governing the values of k must be discovered and formulated. Dr. D. W. Davies (Groningen University) said Dr. Linnett has suggested the criterion of minimum overlap for choosing excited state functions in his non-pairing method. Let $1 be the ground state function and let 41 #z 43 be possible functions for the first excited state $2. Then if (#I $I)<(& $I) (#3 +I) Dr.Linnett would choose #I as a basis for $2. In this notation (a b) f Ja*bdz. Let us consider two sets of functions $ $f with energies E; Ef such that E,'<Ei <E;. Let $! $ span rl r 2 two irreducible representations of the symmetry group to which the Hamil-tonian of the system belongs. The $ are not eigenfunctions and although ($f $:) = 0, ($;,$;) # 0. Let H' be a perturbation applied to the system such that the symmetry group of the Hamiltonian is changed and rl r2 become I? in the perturbed system. I think there is a possible objection to this criterion. GENERAL DISCUSSION 215 Let x be the perturbed functions with energies Wi such that xi-+${ and W{+E as H’-+O. For small H’ (x xi) <(xi xi) but W < Wi < W:. Thus with Dr. Linnett’s criterion xf would incorrectly be chosen as the basis for the first excited state function.An example of a perturbation of this type is the replacement of >CH by >N. McWeeny and Peacock 1 have shown how the excited states of pyridine can be re-lated to those of benzene. I think that if Dr. Linnett applied his criterion to systems of this type he might find it unsatisfactory. Dr. J. W. Linnett (Oxford University) (communicated); In reply to the point raised by Dr. D. W. Davies having obtained a good non-pairing approximation for the ground state a function of this type suitable for an excited state would be constructed the detailed form being chosen to minimize overlap. Then the excited state function would be made orthogonal to the ground state in the manner described.Then this would be repeated for a second excited state and the procedure repeated except that now overlap with the two functions that have been decided would have to be considered. In more complicated systems it might be found that the energy of the second was lower than the first. If this occurred it would probably be worth altering the order of the procedure so that the states would be considered in the correct order of energy. However this might not matter. To avoid omitting any excited state functions the form of the m.0. functions would be considered together with the relation of the n.p. functions to them. Mr. D. M. Hirst (Oxford University) said The wave functions for the excited states of the ally1 cation can be treated in a way which is rather different from that of Linnett and Sovers.For the 1Al states Linnett and Sovers have selected from molecular orbital considerations BIB A/B AB/ and AA/ as a basis for constructing trial wave functions. For the first excited state one electron is excited from a semi-localized bonding orbital to the corresponding semi-localized anti-bonding orbital. In the second excited state, both electrons are in semi-localized anti-bonding orbitals but are in different parts of the molecule. In the third excited state both electrons are in the same semi-localized anti-bonding orbital. Wave functions each containing one adjustable parameter are constructed using the scheme B/B(st) A/B(sym) A/A(sym) AA/(valence bond A-see ref. (1)). The functions are : A An alternative basis is B/B A/B A/A and AA/.11) 12> 13) AIA (a-kb kb-c)+(kb-c a-kb) 14) AA/ (a-kb a-kb)+(kb-c kb-c) BIB (a+kb b+kc)+(b+kc a+kb)+(ka+b kb+c)+(kb+c ka+b) A/B (a-kb kb+c)+(kb+c a-kb)-(a+kb kb-c)-(kb-c a+kb) = (k2 + 1)$i i- 2k$2 +4k$4, = 2$2-4k2$4, = k$l-$2- 2k2+4, = - klC/1+~3 + 2k2+4. The notation is that used in ref. (1) with the superscripts (+) omitted. The value of k for the ground state is obtained by minimizing the energy of BIB. The k for the second state is then chosen to make A/B orthogonal to BIB. Similarly A/A is made xthogonal to A/B and AA/ to A/A. Energies calculated using these 1 McWeeny and Peacock Proc. Physic. Sac. A 1957,70,41 216 GENERAL DISCUSSION functions are compared with those obtained in other treatments in the table. How-ever the four functions do not form a mutually orthogonal set.The deviations from orthogonality which are small are (1 I 3) = 0.013 (1 1 4) = 0.050 and (2 I 4) = 0.066. state energy m.0. c.i. Linnett and Sovers - 30.20 1 - 29.434 - 30.396 - 30.200 - 22-41 - 18.850 - 22.785 - 22.839 - 16.059 - 18.190 - 16.247 - 16.313 - 11.216 - 13.420 - 10.460 - 10.541 11) 12) 13) 14) Starting from the symmetrical form for BIB is less successful. The energies are : I 1) -30.038 12) -21.749 I 3) -16.587 14) -11.198 and the deviations from orthogonality are (1 I 3) = 0-176 (1 14) = 0.065 and (2 14) = 0.074. For the 3Bz and 1B2 states it is possible to get the same results as in the c.i. treatment. 3B2 functions can be constructed from B/B and A/A. In the c.i. treat-ment the functions for these states are of the form cS$8 + c9$9-Apart from the normalization factor there is only one adjustable parameter.The non-pairing functions are BIB (a+kb kb+c)-(kb+c a+kb) = k$g+11/9, A/A (a - k'b k'b - c ) - (k'b - c a - k'b) = k'$8 -11/9. These functions are automatically orthogonal to the 1A1 functions because they are of different symmetry and spin multiplicity. The variation principle can, therefore be applied and k for the lower 3Bz state obtained by minimizing the energy with respect to k. This gives the same energy as the c.i. treatment. k' for the func-tion derived from A/A is obtained by making this function orthogonal to BIB. The energy calculated using this function is identical with the c.i. energy. 1B2 functions can be constructed in a similar way from BB/ and AA/.BBJ (a + kb a 4- kb) - (kb -k c kb 4- c> = k$g +$y, A A/ (a - k'b a - k'b) - (k'b - c k'b - c) = - k'$s +11/7. Dr. J. W. Linnett (Oxford University) said I should like to make one brief remark in relation to the comments of Mr. Hirst. The following table is taken from the paper we wrote and gives the coefficients of the component functions in the full c.i. function for the cation. From this it is clear because of the low importance of TABLE 1 State (0 b) etc. (a c) etc. (b 6) (a a) etc. 1st (gas.) 0.313 0.250 0-233 0-026 2nd -0.161 0.661 - 0.399 - 0.036 3rd - 0.323 0.229 1.957 - 0-261 4th - 0.362 0.158 0.438 1.712 ((a a)+(c c)) in the ground state function that a function of the type BIB is best, and that simple valence bond or molecular orbital functions cannot reproduce this feature unless configuration interaction is used.For the second state ((a a) + (c c)) is again unimportant and this combined with the pattern of signs of the other coefficients makes it clear that the most satisfactory type of function is A/B. For the highest state similarly the magnitudes and signs of the coefficients clearly diretc 1 Hint and Linnett J. Chem. Soc. 1962 1035 3844 GENERAL DISCUSSION 217 one towards employing AA/-. However for the third state the coefficients do not make it clear whether A/A (which Hirst used) or AB/ (which Sovers and I used) should be selected. The first would make ((a a)+(c c)) unimportant and the second (a c)+(c a)) unimportant. The table shows that these have the smallest coefficients but they are not by any means negligible.Prof. C. A. Coulson (Oxford University) and Dr. A. H. Neilson (Glasgow Uni-cersity) said Since communicating our paper we have completed entirely analogous calculations for the tetratomic molecule H@+ using the molecular wave functions calculated by Grahn.1 The results which will be published in detail elsewhere, are almost wholly similar to those for H20 reported in the present paper. In particular curves of ei(a) and ~ ( a ) against bond angle differ just as for H20. Their overall shapes are governed by their gradients. These differ in sign for any orbital i only when AEi and A(CJij-+Kij) have opposite sign and when also i where AEi is the difference in the core energies at 90" and 180" for HzO or 100" and 120" for H30+ and A[CJfj-3Kij] the difference of the coulomb and exchange energies.As with H20 this difference is greatest with the ( 3 4 molecular orbital, which is largely composed of the oxygen 2pz function. Similarly the nuclear repulsion energy varies strongly with bond angle a and it seems once more to be the combination of the (3al) m.0. and Vlv(a) which is responsible for increasing the bond angle above 90". It will be interesting when better wave functions are available to make similar calculations for other molecules of types AH2 and AH3 to see to what extent our conclusions depend on the nature of the central atom A. Until then the conclusion to which the study both of H20 and H30f lead us can be summarized as follows. It is clear from the uses to which Walsh's correlation diagrams are being put, that he is plotting something of physical significance in terms of its variation with valence angle.But this something does not appear to be either (a) the ionization potential (b) the core energy (c) the hydrogen-like one-electron energy or (d) the additive-partitioned energy ei defined in eqn. (6). At the present stage it is by no means certain just what the something really is. Dr. J. W. Linnett (Oxford University) (communicated) The paper by Coulson and Nielson shows how difficult it is to carry out a theoretical treatment including inter-electronic effects which leads to the very successful semi-empirical diagrams of Walsh. It seems to me worth remarking that the way of treating directed valency and the shapes of molecules commonly used by inorganic and organic chemists, and presented in the 1940 Bakerian Lecture of Sidgwick and Powell lays stress on the disposition of electron pairs in particular assemblies of valence shell electrons (e.g.trigonal bi-pyramidal for ten electrons as five pairs). This treats the mutual orientation of chemical bonds as resulting primarily and in fact almost entirely, from inter-electron effects. On the other hand the procedure of Walsh is at first sight entirely different and apparently fundamentally so in that it considers each electron individually examining separately the way in which the energy of each occupied molecular orbital changes when the inter-bond angle is changed from, say 90 to 180" in a triatomic molecule.I am inclined to think because of the very widespread success of the Sidgwick-Powell approach and also for more general reasons that inter-electronic effects are in fact of primary importance in fixing i 1 Grahn Arkiv. Fysik 1961 19 147 21 8 GENERAL DISCUSSION inter-bond angles. I believe that the semi-empirical construction of Walsh’s curves means that these inter-electronic effects have been included when constructing the curves in his diagrams. Also I think that they are the decisive factor in fixing inter-bond angles even though deductions based on the use of these diagrams which are presented as if such effects are of minor importance are extremely successful. The reason for their success is that the empirical construction of the curves means that such effects have been included with the correct weight.Dr. H. H. Greenwood (I.C.I. Bizzingham) said The paper of Coulson and Neilson provides a conspicuous example of the problems encountered in relating quantities arising in the quantum theory of molecules to those measured physically. An ambiguity in the theoretical interpretation which seems to be characteristically associated with these problems can be demonstrated by a small modification in the form of the equations. Thus substitution of Et from eqn. (3) in (7) gives ei = EL- E (Jij-SK) 1J ’ all j or ei = Ei-gij (A) where gii is the total interelectron repulsion energy. The quantity ei defined by Coulson and Neilson may therefore be expressed in terms of the ionization energy Q and either the core energy Ei or the interelectron repulsions gij in the two equivalent expressions (7) and (A) which provide therefore two equivalent physical inter-pretations of the particular theoretical form employed in the correlation.It is interesting to note further that if self-consistent-field functions are used the term separation of (A) is the same as that occurring in Roothaan’s form 3’ = H+G of the Hartree-Fock equations since E( is then an eigenvalue of F and el and gij the corresponding values of the effective electronic Hamiltonian H and interelectron operator G in the same representation. Prof. Coulson (Oxford University) and Dr. A. H. Neilson (GZasgo w University) said Dr. Greenwood is right in saying that there is ambiguity in the way in which we may decide to split up the total energy into parts.This is particularly true where one or more of the parts (e.g. exchange energy) has no genuine physical independence. No choice of wave functions can remedy this defect which is common to all attempts to represent many-electron properties in terms of a one-electron model. One can, however ask that the one-electron quantities used shall have as “ chemical ” a sig-nificance as possible. As our paper shows even this choice is not easy to make. To some extent the same situation applies to Dr. Linnett’s interesting reminder of the importance of lone-pair interactions. For even lone-pairs themselves have no independent objective existence. But they are nonetheless exceedingly important chemical concepts and we are inclined to agree with Dr. Linnett that in some at present unspecified way Walsh’s empirical curves do incorporate their influence in determining bond angles.Prof. A. D. Walsh (Dundee) (communicated) The original account 1 of the derivation of my “ correlation diagrams ” has been supplemented 2 by a simple and perhaps clearer description of the essential ideas involved. Construction of a correlation diagram starts with the writing down of a catalogue of the components from which the m.0. are to be built. For AH2 molecules these components are taken as the valency shell a.0. of atom A and the Is+ 1s and ls- I s orbitals of the H2 group. After each component the appropriate moZecuZar sym-1 Walsh J . Chem. SOC. 1953 2260 2266 2288 2296 2301 2306 2318 2321 2325,2330. 2 Walsh Photoelectric Spectrornetry Group bull.no. 13 1961 348 GENERAL DISCUSSION 219 metry label is then written for (a) a linear (b) a bent AH2 molecule. Components with the same molecular symmetry label are then combined yielding the qualitative forms of the sets of intra-valency shell 111.0. appropriate to (a) linear (b) bent AH2 molecules. Thus far the A atom components considered have been limited to s and p a.0. There is no doubt (as Dr. Linnett pointed out) that for A atoms not in the first row of the periodic table d a.0. should also be included. Their inclusion will have two effects. First the number of intra-valency shell m.0. and transitions available for the assignment of molecular spectra will be increased. There seems no obvious application of this consequence at present.Secondly the magnitudes of the rises and falls of curves on the correlation diagrams representing m.0. primarily built from s and/or p a.0. (as far as atom A is concerned) will be affected. For example, in constructing my diagrams I have laid stress on the fact that the s a.0. component of A because of its symmetry plays no part in the nu 111.0. of the linear molecule, but does combine with the relevant p a.0. in the bent molecule. Similarly the d a.0. components are all g with respect to centrosymmetric linear symmetry and therefore play no part in the nu m.0. of the linear molecule; but do play a part in the a1 111.0. of the bent molecule that correlates with nu. If we ascribe the fall from right to left of the a1 -nu m.0. on the correlation diagram (as conventionally plotted) to the mixing in of an extra component as the molecule bends (irrespective of whether that extra component is more or less tightly bound than a p a.0.in the free atom A) then the a1 -nu m.0. might be expected to fall more steeply in the diagram for H2S than in that for H20. Intuitively the extra effect of the d a.0. seems likely to be responsible for the fact that the ground state angle falls markedly from H20 to H2S and then remains almost constant to H2Se and H2Te. Similarly, the d a.0. can play no part in the a; m.0. of the planar AH3 molecule but will play a part in the a1 m.0. of the pyramidal molecule that correlates with a;. The HAH ground state angle falls markedly from NH3 to PH3 and then remains almost constant to AsH3 and SbH3. I agree with Dr.Linnett’s written comment that the well-known success of the Sidgwick-Powell (SP) approach to understanding and predicting the ground state shape of molecules means that inter-electronic effects are probably of primary im-portance in fixing inter-bond angles. The stress in the SP approach is on pairs of electrons (whether bonding or lone pair) getting as far away from each other as possible to minimize inter-pair repulsions. Such repulsions must be hidden in or correspond to the curves on my diagrams. One can in fact see a good deal of resemblance between ideas implicit in my diagrams and the SP idea. SP speak in terms of two localized pairs of bonding electrons in e.g. AH2 molecules; I speak of non-localized og and ou m.0. which is a description more suitable for discussion of spectroscopic transitions but which I take to be equivalent to the description in terms of two localized pairs.If the net H-H effect of two electrons in each of the al-o and b2-ou m.0. of AH2 molecules is taken to be anti-bonding the reason originally given for the b2-ou curve on my diagram falling from left to right (viz., that H-H anti-bonding implies tightest binding when the H atoms are as far apart as possible i.e. in the linear molecule) would be expected to apply to the net effect of all four electrons; and the language has an obvious correspondence to the SP description in terms of two interacting electron pairs whose net effect is repulsion. Consider also the effect of two electrons in the al-nu m.0. As the molecule is bent this m.0.must change from being pure p of atom A to part s ; the m.0. becomes a hybrid lone pair orbital whose axis lies in the plane of the molecule and is directed away from the H atoms (see ref. (2)) i.e. the centre of gravity of the electro 220 GENERAL DISCUSSION distribution changes from being at the nucleus of atom A to some point lying on the side of atom A remote from the H atoms. At the same time the a1 -0 m.0. changes, as far as the A atom component is concerned from being pure s to part p ; the m.0. becomes a hybrid orbital of atom A directed towards the H atoms and overlapping in phase with the H 1s a.0. (see ref. (2)) i.e. as far as the central atom component is concerned the centre of gravity of the electron distribution changes from being at the nucleus of atom A to some point lying nearer to the H atoms.Thus if we add two electrons to the a1 -nu m.0. the molecule bends these two electrons move away from the H atoms the two electrons in the a1-0 m.0. move towards the H atoms; and the effect parallels to some extent what would be expected from the idea of repulsion between electrons in the lone pair and bond orbitals. Incidentally, these remarks bring out the point that the rises and falls of orbital curves on my diagrams are not really independent but linked. Dr. H. H. Greenwood (I.C.1. Ltd. Billingham) said:- The written comments of Dr. Linnett and Prof. Walsh indicate alternative ways of describing the physical phenomena which closely resemble the alternative theoretical descriptions mentioned in my previous remarks.These suggest that a description given in terms of interelectronic effects which are represented by gij in equation (A) is associated with an equally valid alternative description in terms of orbital properties as represented by Ei in the equivalent eqn. (7) derived by Coulson and Neilson. It seems therefore that the two physical descriptions may be co-ordinated in the theory as equivalent interdependent but alternative ways of describing the same system though the physical connotations remain apparently distinct. Dr. A. E. Douglas (N.R.C. Ottawa) said The interpretation of the predissoci-ation of the A; state of ammonia given here differs from that in my paper. Here it is assumed that the A; state of NH3 arising from NH2 (2B1) and H (2s) crosses (a narrowly avoided crossing) the observed A; state near its potential minimum.The elementary method of characterizing states which I have used in my paper, seem to indicate that for normal bond lengths there is only one electronic state -46,000 cm-1 above the ground state and the next A; state should be at least 20,000 cm-1 higher. It appears that only by a very large drop in the upper potential curve for a small change in bond distances can two A; states be brought together near the minimum of the observed state. Such a sudden drop in the potential curve near the normal bond distances would imply that simple one-electron orbitals cannot be used to characterize states even in the Rydberg states of a hydride where they normally are considered most meaningful. Dr.G. Herzberg (N.R.C. Ottawa) said It must be admitted that a difficulty arises from the point raised by Dr Douglas which was not considered in the paper by Longuet-Higgins and myself. As Dr. Douglas rightly points out there is only room for one A; electronic state (derived from a 3s orbital of the united atom) at the position of the potential minimum. The other A; state derived from H(2S)+ NH2(2B1) must intersect the first fairly high above the minimum if one uses quasi-diatomic potential functions and it is then very difficult to see why a strong pre-dissociation should start already near the minimum of the 3s A; state (from u’’ = 0 on). The only way out seems to be to assume that in some other co-ordinate not represented in the usual potential diagram there is a strong interaction between the two potential functions bringing the peak between the two states down to almost the level of the minimum of the 3s A; state.Possibly this may happen only in a fairly narrow region in space but nevertheless it would be unwise on this basis to consider the predissociation of NH3 as a good example of a ‘‘ near intersection ” as we did in our paper GENERAL DISCUSSION 22 I Prof. C. A. Coulson (Oxford University) said In $4 of their paper Herzberg and Longuet-Higgins show that there is no difficulty in understanding the way in which H and NO can join together directly to form HNO provided that the H atom does not approach the NO molecule along the axis. This suggests that we should try to follow the formation of the combined molecule as the two parts approach along the reaction path.In the usual potential energy surface description this means that the representative point must go fairly near but not too close to the conical inter-section. If it has to " side-step '' this conical centre it might be expected that the molecule when formed would possess rotational energy perhaps in excess of that to be expected on a temperature basis. I believe that molecules formed in this sort of way do often possess relatively large rotational energies. Do the authors feel that this may be the explanation of this experimental fact? Perhaps a careful study of the direction around which the excess rotation is obtained would throw light on this matter. Dr. G. Herzberg (N.R.C. Ottawa) said We agree with the conclusion of Prof.Coulson that on account of the presence of the conical intersection there may be excess rotational energy produced in molecule formation. However this con-clusion assumes molecule formation in a two-body recombination which experi-mentally is very difficult to observe. The observations to which Prof. Coulson refers deal I believe largely with molecule formation in triple collisions. It may well be that in such collisions also the presence of the conical intersection makes itself felt by excess rotation but it would be difficult to extract this information in a clear-cut way from the observations. Dr. J. W. Linnett (Oxford University) said I should like to ask Dr. Lorquet whether it is certain from appearance potential or other data that CH2D+ and CD2H+ are derived from CH3CD and not from CH3CDz and CD3CHZ.If they are derived from the parent ion the bridge structure necessary to explain the ex-change must involve two hydrogen atoms. But if ethyl positive ions were the source a bridge involving only one hydrogen atom would be adequate to explain the observ-ations and this would be formed more easily I should think. Dr. J. C. Lorquet (Ligge University) said In our paper we implicitly assume that the methyl ions of ethane are formed directly from the parent ion, C,H$CH; + c H ~ . Prof. Skell and Dr. Linnett suggest the following mechanism : c,H,:c,H; + H C,Hf-+CH; +CH,. For non-deuterated ethane it is easy to show that the CH3 ion is formed directly from the parent ion (l) and not from the C2HZ ion (2).From thermochemical data one calculates an appearance potential of 13.6 eV for process (I) and of 17-1 eV for process (2). The experimental value is 13.95 eV which indicates that the CHS; ions are formed by process (1) with about 0.35 eV vibrational energy. A similar calculation shows that the CHZ and C2H3 ions are formed directly from the parent ion with propane and not via the C3H3 ion. If we now consider the deuterated species it might be argued that the rearrange-ment ions CHzDf and CHDZ are formed from CH3CD3 by process (2). The appearance potential of the rearrangement ions is not known in this particular case and it would be difficult to determine them (because of peak over-lap with other fragmentary ions of CH3CD3 and of the residual water). However it is 222 GENERAL DISCUSSION well-known fact that the rearrangement processes occur in the mass spectrometer with a very small activation energy.No case is known where this rearrangement energy is greater than 0.5 eV. It is thus extremely unlikely that the appearance potential of the rearrangement ions would be 3-5 eV greater than that of the normal frag-mentary ion as required by process (2). One might also add that the situation is expected to be quite different in the liquid state and in the gas phase. Apart from the role played by the solvent in the isomerization mechanism a fundamental difference arises from the fact that the molecule receives an important excitation energy from the impinging electron (several eV above the ionization potential). One therefore expects that rearrange-ments which are forbidden in the liquid phase will occur in the mass spectrometer.Dr. I. M. Mills (Reading University) said It seems to be worth drawing attention to the contrast between the CH stretching vibration frequencies and H . . . CO dissociation energies found by Johns Priddle and Ramsay in the two electronic states of the HCO radical. In the bent 2A' ground state of HCO they find VCH - 2700 cm-1 and in the linear 2A" excited state they find VCH- 3300 cm-1 (table 5) ; these results contrast rather surprisingly to their approximate dissociation energy diagram (fig. 6) which shows that-whatever the correct relative positions of the scales of energy on the two sides of their diagram-the dissociation energy in the ground state is 33 kcal higher than in the excited state although the vibration frequency is lower.Perhaps this emphasizes once again the danger of correlating vibration frequencies to dissociation energies. I t is also very striking that in the bent ground state the VCH frequency is closely similar to the exceptionally low frequency observed in formaldehyde and in the linear excited state it is similar to the high frequency observed in HCN. Dr. B. A. Thrush (Cambridge University) said With reference to Dixon's paper, in our experiments on carbon monoxide flame band emission from O+CO in flow systems we have found clear evidence that a third body is involved in forming the excited molecule.1 The intensity I of the emission obeys the relation I = Io[O][CO] where 10 depends on the nature but not on the pressure of the carrier gas used as in the chemiluminescent H+NO and O+NO reactions.The absence of a specific effect of 0 2 on I0 and the magnitude of 10 indicate that spin-reversal does not occur in a rate-determining process and we conclude that the mechanism is O(3P) + CO( 'Z') + M = C02*(3B2) + M (1) C 0 * ( 3 ~ 2 ) + ~ ~ ; ( 1 ~ ) CO,'( 93,) = C 0 2 ( 'C,') + hv COi('B,)+M = C02(1Cl)+M, where k4[M] % k3 and radiationless crossing between the singlet and triplet states is much more rapid than the other processes. The study which Dr. Clyne and I made of the kinetics of the carbon monoxide flame band emission 1 indicates that the barrier to combination of O(3P)+CQ(lC+) is between 3.0 and 5.5 kcal/mole. The top of this barrier presumably corresponds to the breaking-off in emission between levels K = 5 and K = 6 in the upper state of the bands which Dr.Dixon has analyzed. Dr. R. N. Dixon (Shefield University) said I should like to comment on the bending vibrational frequencies for C3 discussed by Gausset Merzberg Lagerqvist and Rosen. The three 0 1 0 levels of the upper I l l u state of C3 are given as 135 cm-1 1 CIyne and Thrush Proc. Roy. SUC. A 1962 269,401 GENERAL DISCUSSION 223 (IC;) 258 cm-1 (lAg) and 480 cm-1 (1X:) above the lowest vibrational level. These three energies can be represented to within 1 cm-1 by the use of Renner’s equations 1 for the Renner-Teller splitting of a 1II state with the parameters 0; = 307 cm-1 and E’ = +0-56. Thus there are three known cases of the Renner-Teller effect involving AB;! (or ABC) type molecules with first row elements (C3 NC0,2 B 0 2 3) in all of which there is an odd number of electrons in the (n,) orbital.The bending force constant in any state of a molecule depends on the combined effects of all the electrons and nuclei. However the difference in force constant between two electronic states of a molecule may be correlated in a simple m.0. theory with the difference between one-electron orbital contributions to the force constant. Table 1 gives the data for the two potential curves of the doubly-degenerate ll states TABLE BENDING FORCE CONSTANTS FOR THE II STATES OF NCO B02 ANDC~ molecule configuration 104 k s / P (dynes cm-1) difference NCO c3 . . . (x)4(a)2(7r a”)*(x a’) X2H{ . . . (Tc)4(a)2(7Tn a”)(x a’)2 4.25 3*08} 1.17 ”’”} 1.24 1 -73 of the above molecules.It is seen that the difference between the two force constants, i.e. between the one-electron contributions of (xg b2) and (ng7 a2) is remarkably constant even though the configuration for C3 includes (ng),2 whereas that for NCO and B02 includes ( 7 4 3 . This near equality may be associated with the fact that the (ng) orbital is defined by symmetry as a non-bonding orbital located on the end atoms. It may be shown from Renner’s equations and those for the bending of a linear triatornic molecule that where p2 is the reduced mass for w2. Since p2 does not vary greatly for ABC mole-cules with first row elements and the difference in bending force constants has been shown to be fairly constant it is therefore to be expected that any II state of such a molecule which is degenerate because of an odd number of electrons in the (n,) orbital will have a value of m& of the same order of magnitude as in the above molecules i.e.- 40,000-50,000 cm-2. Thus if w2 is small the Renner-Teller splitting 0.12~ will be large. Prof. A. D. Walsh (Dundee) said It may be worth pointing out that the state-ment that “ the rigidity of linear molecules like C02 is mainly determined by the four electrons in the n orbital” agrees (at any rate as far as comparison of the x and lower nu orbital goes) with the orbital correlation diagram that I constructed for AB2 molecules.4 On that diagram of the two curves leading to the lower xa orbital in the linear molecule one rises and one falls from left to right; whereas both curves leading to the xg orbital fall from left to right.Four electrons in the xg orbital should therefore have a much bigger effect in tending to increase the rigidity of the linear molecule than four electrons in the lower nu orbital. 1 Reniner 2. Plzysik 1934 92 172. 2 Dixon Phil. Trans. A 1960 252 165. 4 Walsh J . Chem. SOC. 1953 2265. Jokis Can. J. Piiysics 1961 39 1738 224 GENERAL DISCUSSION Prof. C. A. Coulson (Oxford University) said I must confess that I feel very troubled about one matter suggested by Coon Cesani and Loyd. This concerns the ease of tunnelling involved in passing from the excited configuration of C102 in which the two bond lengths are 1.685 and 1.555 A to the mirror configuration where the bond lengths are interchanged.The barrier height is claimed to be 2520 cm-1 or perhaps as much as 3000 cm-1. Now if the central atom moves its mass (35 units) is very much larger than that of a proton and the same is true if the outer atoms move (mass 2 x 16 = 32 for this motion) or if we use the reduced mass as given in (2.6) of the paper. Since the penetration of a barrier varies inversely with an exponential factor proportional to m3 and since the barrier height is of the same order as for the inversion of NH3 where the inversion splitting is known to be small it is not easy to see without further details of their calculations than are at present available how the authors can achieve anything like the relatively large splitting that they claim. Would we not expect a small splitting corresponding to the molecule flipping from one shape to the other relatively infrequently? And would not this mean that the selection rules for absorption would be more appropriate to an excitation from an initial wave function concentrated in one of the two wells, rather than from the " time average " wave function where the $ will be of equal magnitude in both wells? Dr.J. C. D. Brand (Glasgow University) said The interesting idea developed in the paper of Coon Cesani and Loyd that the antisymmetric stretching vibration y; of excited C102 may be associated with a double-rninimuni potential seems to depend on rather indirect evidence. A non-linear state with Cl-0 bonds of different length can be regarded as a CzV structure vibrating in a potential so anharmonic that the oxygen nuclei spend most of their time in positions not equidistant from the chlorine thus the CzV vibronic selection rules apply even though successive levels of the y j vibration are by no means equally spaced.Accordingly transitions from the ground state (000) terminate on alternate levels (O+ 1+ . . .) of y; the intermediate levels (0- I- . . .) combining only with odd-numbered vibrational states of 7;. Some transitions of the second type especially those originating from the fundamental level (0; = l) should be observable as hot bands in the spectrum. If they can be picked out it should then be possible to show that the levels (000) and (001) of the lower electronic state do in fact combine with alternate unevenly-spaced levels of 7; and so to prove the existence of a double minimum in the potential.Prof. A. D. Walsh (Dundee) said As regards the possible occurrence of vj in electronic spectra of BAB molecules I should like to make the following points. (i) It is difficult 1 to interpret the vibrational structure of the 2491 A transition of NO2 without introducing vj. (ii) The 2491 transition of NO2 is believed 1-3 to be the analogue of the C102 transition studied so much by Coon and his co-workers. Both transitions apparently transfer an electron from the a2 - ng orbital to the bl -nu orbital. (iii) Claims have been made for the occurrence of vj in certain transitions of SO2 ; each of these transitions almost certainly transfers an electron from a lower energy orbital to the b 1 - 5 ~ orbital.(iv) It is tempting therefore to associate the appearance of v; with the placing of an electron in the bl - nnu orbital. (v) It is im-portant to stress that the Za orbital of the linear molecule is the lowest energy orbital that is A-B anti-bonding. Bending of the linear molecule may be looked upon as a response to the placing of an electron in the Ea orbital; by bending the anti-bonding nature of the a1 component is largely lost. The anti-bonding nature of the bl component of ZU however cannot be lost by the molecule bending. It is -1 Ritchie Walsh and Warsop Spectroscopy (Institute of Petroleum London 1962) p. 289. 2 Walsh J. Chem. SOC. 1953 2266. 3 Mulliken Can. J. Chem. 1958 36 10 GENERAL DISCUSSION 225 precisely the bl -nu orbital whose occupation might be expected to produce a “ new ” change of shape.If any loss of symmetry - (not bending) can reduce the anti-bonding effect of placing an electron in the 61 -nu orbital the molecule may be expected to respond by adopting that less symmetrical form. According to Mulliken 3 the anti-bonding effect of an electron in the bl - nu orbital would be lessened by the molecule adopting unequal bond lengths. Unequal bond lengths in an excited state would supply a reason for the appearance of vi in an absorption spectrum. (vi) In AH2 molecules the bl -nu orbital is not anti-bonding but strictly non-bonding ; and so there is no suggestion that the ground state of e.g. the €320 molecule has other than Czv symmetry. (vii) Rotational analysis 1 of the NO2 2491 A transition appears decisively to lead to the conclusion that the upper state has CzV symmetry-in con-trast to the tentative conclusion from the vibrational analysis.However it is possible in principle for the opposite conclusions both to be correct. Slightly unequal bond lengths imply two minima in the potential energy surface with a small energy barrier between the minima. If this barrier is of a suitable size tunnelling through it may not occur in the time for one vibration but might take place many times in the much longer time for one rotation. In that case rotational analysis must necessarily indicate equal bond lengths ; while vibrational analysis will indicate unequal bond lengths. (viii) Since the ground state of the C1O2 molecule has one electron in the bl -nu orbital the question arises “ Does the ground state of C102 have unequal bond lengths? ” Microwave work appears to answer the question decisively in the negative; but (vii) makes it possible in principle that a study of the vibrational frequencies might indicate the opposite.However there is no evidence from the reported C1O2 values 2 of v; 2v,” and 317; for a double minimum in the ClOz ground state potential surface. (ix) There is evidence that the I; ion contain-ing in its ground state two electrons in the bl-Zu orbital and two in the al-iT, orbital (which is also A-B anti-bonding) has unequal bond lengths. However, the evidence applies to I; in solid CsI33 and solid NI3[&4 and may not apply to gaseous I;. A study of the gaseous XeF2 molecule might be interesting here.Prof. P. S. Skell (The Pennsylvania State University) said In 1955 we proposed a kinetic procedure for distinguishing triplet and singlet states of CH2 and thereby demonstrated that the primary CH2 of diazomethane photolysis was singlet. Sub-sequently others have employed this procedure to detect the presence of triplet CH2 obtained under modified reaction conditions. The procedure depends upon examination of the steric course of CH2 additions to olefins. A triplet CH2 and olefin react with spin conservation to yield a triplet 1,3-diradical incapable of cyclizing to the cyclopropane until the quasi-forbidden triplet-singlet transition occurs. A singlet CH2 can interact simultaneously with both carbon atoms of an olefinic linkage thus leading directly to the cyclopropane.Only in additions of triplet states to olefin are there intermediate long-lived open-chain species capable of undergoing rotations about single bonds (frequencies 10*-1010 sec-1) : ---\ / \ I/ slow / \ / * \ CH2 ( t ) + C=C + C-C (t)-((s)-+cyclspropane. -CH2 1 Ritchie Walsh and Warsop Proc. Roy. SOC. A 1962 266 257. 2 Nielsen and Woltz J. Chem. Physics 1952 20 1878. 3 Tasman and Boswijk Acta Cryst. 1955 8 59. Mooney 2. Krist. 1935 90 143. 226 GENERAL DISCUSSION This concept has been applied by using cis- and trans-Zbutenes in separate experi-ments. With CH2(s) and cis-2-butene the only cyclopropane should be cis-1,2-dimethyl cyclopropane and with trans-2-butane only trans- 1,2-dimethyl cyclopropane. With CHz(t) either olefin should yield a mixture of these two cyclopropanes because the open-chain intermediates live long enough for isomer equilibration by rotations about bond axes.More recently we have studied the chemical reactions of C3. With olefins C3 reacts at both ends of the molecule as though these carbons were bivalent. These experiments prove that C3 exists as a major coniponent of carbon vapour and that its chemical properties are consistent with a valence bond designation { C=C=C :). Studies with cis- and trans-2-butene demonstrate that C3 is singlet, in complete accord with the conclusions from spectroscopic studies. The chemistry of C3 being investigated may have some astrochemical significance in the comet “laboratory” for C3 reactions with the condensed phases of the comet nucleus may produce allenic derivatives which remain condensed during the journey away from the sun.On its return the exposure of these substances to solar radiation at distances less than 2 a.u. may be the source of some of the observed gaseous constituents of comet atmospheres. Dr. J. W. Linnett (Oxford University) said The treatments of the electronic structures of sulphur and chlorine dioxides that have been employed so far at this meeting have ignored the possible importance of d-orbitals of the sulphur and chlorine atoms in the description of the electronic structures of these oxides. Their electronic structures have been discussed in the same terms and with the same dia-grams as those of ozone nitrogen dioxide etc. in which d-orbitals cannot be of importance because only elements of the first short period are involved.Yet there are very considerable differences in properties between say 0 3 and SO2 (and FOa does not exist). For instance the 00 bond length in Q3 (1-28 A) is greater than that in 0 2 (1.21) whereas the SO bond length in SO2 (1.43) is less than that in the ground state (3E) of SQ (1.49). Also the 00 force constant in 0 3 is less than that in 02 whereas the SO force constant in SO2 is greater than that in SO. The most straightforward explanation of this appears to me to be that in constructing the molecular orbitals of SO2 the contribution of the sulphur d-orbitals is important. Or in other words the sulphur atom can accommodate in the valence shell more than eight electrons whereas with oxygen or any other first row element the Pauli principle excludes this.As an example of the effect of including 3d orbitals I will use the molecular orbitals of the symmetry class a2 for the bent molecule. For 0 3 there will only be a non-bonding orbital of this type constructed in the 1.c.a.o. approximation from 2pz atomic orbitals on the end oxygen atoms. With sulphur one of the d-orbitals will have this symmetry and will combine with the above combination of oxygen orbitals to give one bonding and one anti-bonding molecular orbital. Many chemists would ascribe the strength of the SO bond (see above) to the presence of two electrons in this a2 bonding orbital. The five d-orbitals have the following symmetries linear molecule cg ng ng 6, a, and for the bent molecule al al bl b2 and a2.Prof. C. A. Coulson (Oxford University) said Dr. Linnett has drawn attention to the way in which the 7cg molecular orbital of linear SO2 may be expected to be stabilized by the inclusion of some sulphur d-orbital. Thus if we consider a hypoth-etical linear 0-S-0 system and (see figure) if no d-orbitals are allowed the m.0 GENERAL DISCUSSION 227 si siniply 4(0,) -4(Ob). But if we may include some sulphur d-orbital an m.0. of type +(O,) -$(Ob) + k 4 ( S d ) might possibly convert the slightly anti-bonding m.0. into a bonding one for the d-orbital overlaps favourably with both terms of (b(0,) -4(Ob). 0 S O b I should like to point out that if we adopt Clementi's rules (or any other of the various alternatives) to estimate the size of these d-orbitals we find a radial factor of the type r2e-cr where c is of the order 0.8 (in a i l ) .The maximum of this orbital occurs when Y = 2/c = 2 . 5 ~ 0 . Now the S - 0 distance is about 3ao. It follows that the maxima of the four lobes of the sulphur d-orbital lie not very far from the maxima of the 7r-orbitals on the oxygen atoms. In fact +(0,)-4(Ob) without any modifica-tion or addition bears a close resemblance to S(3d). To some extent therefore, it is a matter of taste whether we specifically include the 3d orbital or merely say that the close resemblance of 4(Oa) -4(Ob) to a sulphur 3d-orbital implies additional stabilization from the electrostatic field around the S atom. Whatever language we do use however the fact remains that the presence of the sulphur atom confers additional stability on the otherwise slightly anti-bonding zg orbital.It also follows that this extra stability will be a maximum when the molecule is linear. This is because if the O,-S-Ob angle is 180" -2a then (see figure) we must choose for the S mid-plane of sulphur d-orbital _________ ;T ___________ a / \a / \ Ob 0, appropriate sulphur d-orbital one whose mid-plane makes an angle a with each S-0 bond. This reduces the overlap by a factor cos a. Fortunately however, unless a exceeds 30" this reduction is not large. Dr. J. H. van der Waals (KoninklijkelShell Lab. Amsterdam) said In his dis-cussion of the triplet-singlet transition of the 3800 A band system of S02 Dr. Merer notes that it contains four bright 3B1 bands but nearly as strong also some bands of vibronic symmetry 3A2 which involve excitation of the antisymmetric stretching frequency v3.By analyzing the manner in which spin-orbit interaction should admix singlet character into the lowest triplet state of SO2 this behaviour may qualitatively be understood. According to the table at the end of Dr. Merer's paper the lowest triplet state of SO2 has electron orbital symmetry B1 in the usual nomenclature of the group Czv.l In the absence of an external field the degeneracy of the triplet state will already be removed by spin-spin and spin-orbit coupling and the appropriate spin functions of the individual components may be labelled T, T, Tz.2 These three spin functions transform like rotations around the molecular axes x y and z and each of them corresponds to a situation in which the spin angular-momentum vector lies in one of the co-ordinate planes of the moIecule.3~ 4 1 Eyring Walter and Kimball Quantum Chemistry (John Wiley 1960) appendix VII.2 Hameka and Oosterhoff Mol. Physics 1958 1 358. 3 Weissman J. Chem. Physics 1950 18 232. 4 Van der Waals and De Groot Mol. Physics 1959 2 333. H 228 GENERAL DISCUSSION By taking the direct product of the representations of the orbital part of the electronic wave function and of the spin functions one gets for the symmetries of the lowest triplet manifold of SO2 : orbitaI spin symmetry x symmetry a sy ZkLy polarization TW2) A2 xv3 B1 TY(B1) A1 z Tz(A2) B2 Y The Hamiltonian for spin-orbit interaction is represented by 5 - + + Zso = I=Gi .si; i the summation is over all electrons in the system. Since Zs0 must be totally sym-metric in spin and orbital co-ordinates together the si as well as the quantities Gi, which depend on the orbital co-ordinates of all electrons transform like an angular momentum. The terms G i p S i x involving the x components are those responsible for mixing the Tz component with a singlet spin function etc. The terms in GI can be grouped as 1 -+ -+ -+ - + + -+ Gi - xFi(ri) A pi + two-electron terms. K (3) - + + Here Ft (Q) is the electric field due to the molecular skeleton acting on electron i when averaged over the positions of all other electrons ; pt = - ivi is the momentum operator. The product Ft Apt . Si corresponds to the interaction of the spin of electron i with the angular momentum of its orbital motion.As zs0 belongs to A1 the total symmetry of the triplet components given in (1) should be identical with that of the singlets which are admixed through spin-orbit coupling. Hence the TU component may participate in a parallel transition from the ground state 1A’; and the Tz component in a perpendicular one ; transitions to T, however cannot steal intensity from allowed singlet-singlet transitions without the intervention of the asymmetric vibration v3. These conclusions merely are a slightly different formulation of the selection rules for spin-orbit coupling given by McClure; 2 they do not involve any simplification of Xso. The strength of the intercombination transition 1Ay - 3B1 due to spin-orbit interaction which mixes the 3B1 state with a given singlet state say 1A1 will be proportional to + + -+ 4 - + The following qualitative conclusions can be drawn about the magnitude of (4) and its analogues for the other components of 3B1 when restricting oneself to the lower electronic states given in Dr.Merer’s table. Starting with the T component the absence of pronounced parallel bands in the 3800A band system may be related to the fact that the only excited state of 1Al symmetry in the table is doubly excited relative to 1AF and thus in (4) the transition 1 Bethe and Salpeter Handbirch der Physik (Springer 1957 vol. XXXV) p. 267. 2 McClure J. Chem. Physics 1949 17 665 GENERAL DISCUSSION 229 Next consider transitions to T, which by virtue of its spin symmetry can only mix with singlets through the terms CGizsie in (2).The spin-orbit matrix element in the analogue of (4) then reduces to the integral over the spatial co-ordinates i From detailed calculations on spin-orbit coupling in small organic molecules,lp 2 the results of which are strongly supported by experimental phosphorescence life-times and polarizations it follows that when a spin-orbit matrix element like ( 5 ) is expanded in integrals over atomic orbitals the result is completely dominated by the one-centre integrals in the expansion. If the latter vanish for the electron configuration concerned the matrix element is only very small. The major one-centre integral occurring in the expansion of (5) is that corresponding to the first terms in (3) e.g., where ZfK (riK) is the effective nuclear charge at a distance riK from the nucleus 3 and liz the z component of the angular-momentum operator acting on the electron.In more physical language one might say that the coupling between spin and orbital motion predominantly occurs whenever the electron moves very close to an atomic nucleus where the nucleus is largely unshielded and the field very nearly spherical. Accordingly in SO2 the one-centre integrals on the S atom with its greater nuclear charge must be several times greater than those on the lighter 0 atoms; this view is substantiated by the fine structure of the atomic spectra of 0 and S. We now notice that the electron configurations in the 1B2 and 3B1 states of ( 5 ) differ by the transfer of one electron from the (laa) to the (4~21) 111.0.in Dr. Merer’s table. Of these the (la2) m.0. is antisymmetric in the planes x = 0 y = 0 and SO it does not involve s- or p-type orbitals on the S atom. Thus the matrix element (5) for transitions to T, which give rise to the 3B1 bands in the spectrum must mainly arise from the relatively weak spin-orbit coupling on the 0 atoms in the molecule. Finally consider transitions to the component Tz. In the first instance these are orbitally forbidden but in combination with the vibration v3 they may appear as the perpendicular bands of 3A2 vibronic symmetry found by Merer. On excita-tion of v3 the T component can mix with a singlet state having the same total sym-metry i.e. with 1B1. The mixing of (3B1 T,) with 1B1 by spin-orbit coupling and the intervention of v3 should be quite strong comparatively because this is the only combination in which the most effective spin-orbit coupling near the sulphur nucleus participates.Moreover the 3B1 and 1B1 states arise from the same electron configuration and so have a maximum orbital overlap and a small energy denominator in (4). Prof. A. D. Walsh (Dundee) said Has not an error crept into the correlations given at the end of the paper by Merer? The states of the linear molecule derived from the nfn2 configuration are written in an order of energy that increases upwards, but the states of the bent molecule are written in an order of energy that increases downwards. The lAs linear state correlates with the 1A1 ground state (not with the lA1 state reached by an electron jump from the ground state) and the lB1 state.1 Hameka and Oosterhoff Mol. Physics 1958 1 358. 2 This Discussion. 3 Slater Quantum Theory of Atomic Structure (McGraw-Hill 1960) vol. 1 pp. 227-28 ; vol. 11, pp. 189-200 230 GENERAL DISCUSSION Nevertheless the fact that the 1B1 and 3B1 states correlate with different linear states (lAs and 3E.s respectively) is correct and possibly important. However I would have thought that the fact was more likely to result in an " abnormal " separation of the 1B1 and 3B1 states rather than in a gross difference of geometry. Dr. J. K. G. Watson (University College London) said With regard to the perturbations which Dr. Merer observes in his 110-000 band we have observed some Coriolis perturbations in the 3800A system of propynal (see paper in this Discussion).They fall into two categories : (i) AJ = 0 AK = 0. This type is due to coupling by rotation about the near-symmetric top axis and includes the degenerate vibrations of a true symmetric top. The J-structure is little affected by the coupling but the sub-bands are dis-placed bodily by an amount depending upon K. If the vibronic levels belong to the same electronic state so that A - B is nearly the same for both the gyro-vibronic levels will cross obliquely if they cross at all and the perturbation will extend over several successive sub-bands. (ii) AJ = 0 AK = f 1. The coupling here is produced by rotation about an axis perpendicular to the near-symmetric top axis ; for an asymmetric top the associ-ated vibronic selection rule depends upon whether the b or the c axis is involved.The perturbation is small for J = K and increases with increasing J for given K ; in consequence the primary effect of the coupling is to give an anomalous value of B for the sub-bands involved with little displacement of the sub-band origins. If A-k is nearly the same for both vibronic levels the cross-over is quite highly localized at one value of K and one finds an abrupt change in the effective value of 5 as one crosses the centre of the perturbation. Mr. A. J. Merer (Oxford University) said The Coriolis perturbations analyzed by Dr. Watson in the spectrum of propynal permit an explanation of the perturbations in the 110-000 band of the 3B1- 'A1 system of SOz. The sub-level K' = 12 of the 110 level suffers a perturbation of type AK = k 1 while the perturbation centred on k" = 16 is of type AK = 0.The third perturbation in K' = 3 seems to be a genuine perturbation and is probably of type AK = & 1. Recent investigations of the 2900A system of SO2 show that the polarization of the only band with sharp K-structure (32850 cm-1) is parallel. This means that the upper state of the 2900A is 1232 and implies that a 3B2 state should lie at longer wavelength. Since the selection rules for Coriolis perturbations indicate that a B1 vibronic level can only suffer Coriolis perturbations of type AK = 0 as a result of an A1 vibronic level it is possible that an antisymmetric level of the 352 state lies under the 110 level of the 3B1 state. The presence of this 3B2 state may account for some of the unexplained perturbations in higher levels of the 3231 state.A vibronic B1 state can undergo Coriolis perturbations of the type AK = & l by interaction with vibronic A2 or B2 states it is therefore not possible to determine the perturbing levels in the other two perturbations. However if it is assumed that these two AK = f l perturbations are caused by the same vibronic state then this state must have A-B = 1-66 cm-1 and an origin 1272 cm-1 above that of the 000 level of the 3B1 state. Prof. A. D. Walsh (Dundee) said Since writing our paper we have photographed the absorption spectrum of C102 at 90°C. We find that with increasing temper-ature absorption on the long wavelength slope of band H increases in intensity relative to the neighbouring bands.Evidently a hot band is present incompletely resolved from band H. Because of the incomplete resolution we cannot state the GENERAL DiSCUSSiON 23 1 position of the hot band with precision but it lies close to 445 cm-1 to the red of band I (the origin of the 1568 A transition). 445 cm-1 is the magnitude of vi for Cl02. The position of the hot band its intensity relative to band I and its increase of rela-tive intensity with temperature all suggest that it represents the (000)+(010) transi-tion of the 1568 A system and that the latter system really is due to C102. Accepting our assignments the upper states of the 1829 A and 1628 A systems (especially the latter since it lies closer to the ionization potential) should approxim-ate to the ground state of the ClOl ion.We thus deduce that for the ground state of ClO v1 is - 1050-1090 cm-1 and v2 is - 521 cm-1. In the ground state of SO2, v1 is 1152 cm-1 and v2 is 518 cm-1. The similarity of ClOz and SO2 was to be expected since the two entities are isoelectronic and contain much the same masses. Dr. R. N. Dixon (Shefield University) said Humphries Walsh and Warsop have suggested that Fermi resonance between v; and 2 ~ ; may be responsible for the observation of a staggering in the vibrational structure of the 1568 A system of C102. The observed staggering in the ground state of CO2 (with which a comparison is made) arises from two causes (i) as 2ul+ v2 increases from 0 the number of levels in resonance is 0 0 1 1 2 2 3 etc. ; (ii) levels with 201 + 212 even have even I and with 2vl +v2 odd have odd I and the matrix elements of the perturbation depend on the value of 1 in addition to those of 2'1 and u2.In a bent molecule such as C102 only the first of these two causes is operative and calculation for u)1 = 2~02 shows that in this extreme case the staggering of the levels is rapidly damped out as 2vl+u2 increases to more than 3. Thus it is suggested that Fermi resonance is not likely to be the main cause of the observed staggering which must therefore be due to the superposition of short progressions in both v; and vi with v; slightly less than 2v'. Calculation of the normal co-ordinates of C102 for a bond angle of 28 = 120", (01 = 1000 cm-1 0.102 = 500 cm-1 and using a simple valence force field leads to : 41 ; dr = +0*0364A, q 2 ; dr = +Om0228 A, rd8 = -0.0016A7 rd8 = + 0.062 A, where r is the length of either bond.Thus although 41 is almost a pure bond stretching motion q 2 involves both bond stretching and angle bending. A transition from the ground state of C102 to an excited state in which the only geometry change is one of bond-length will therefore show a progression in v;. However a transition to an excited state with a negligible bond-length change but a change of angle -5" will show short progressions in both v; and v; as does the 1568 A system of C102 (although a change in both angle and bond-length could give the same intensity distribution for dr > rd8). Dr. T. E. Peacock (King's College London) said There is a further transition of bl symmetry which should be included in table 4 of our paper.This is Y3-+@4 which has an energy of 26,200V and oscillator strength zero. This transition should clearly be grouped with the degenerate pair v3-+@1 and Y1-411. The interpreta-tion of the observed spectrum given in the discussion is unaffected by this as this band will be masked by the intense bands of 6 2 symmetry. Dr. J. L. Duncan (Reading University) said Results from the force constant calculations for ammonia phosphine arsine and stibine may throw some light upon the observed values for v2 in the excited states of these molecules. Force fields for the A1 species required to reproduce accurately the observed ground electronic state vibration frequencies Coriolis vibration-rotation interaction constants and centri-fugal stretching constants show that for NH3 (HNH angle -106") there is a large interaction between the NH stretching co-ordinate S1 and the deformation ( 6 c um-brella ") co-ordinate &.In the other molecules in the series where the HXH angl 232 GENERAL DISCUSSION in each case is nearly 90° this interaction is essentially zero. It appears that the interaction force constant is a sensitive function of the interbond angle and it may well be the case that opening up the pyramidal angle in going from the ground to the excited state would not significantly change the vibration frequency. This could occur if due to the transition both F12 and F22 in the force field are altered, as would be anticipated. If in the excited state the vibration levels are far above the inversion barrier then v; 4 7 2 as experimentally found.The excited state of ammonia however is known to be planar. In this case, therefore the point group changes from C3v in the ground state to D3h in the excited state. This has the effect of isolating the two A1 vibrations from each other since ~ A I ( C ~ ) + A ; + A;(&). Thus the large interaction between the two co-ordinates, present in the pyramidal molecule must be completely absent in the planar molecule. This state of affairs has the effect of raising the v2 frequency if the same bending force constant is retained (the change in the kinetic effect between the pyramidal and planar states being small). A simple calculation shows that in order to reproduce the same frequency in the excited as in the ground state it is in fact easier to bend the planar molecule than the pyramidal.This is in contrast to the authors’ proposi-tion that removal of one lone-pair electron from a planar molecule might be expected to make it more difficult to bend the molecule. Prof. W. C. Price and Dr. T. R. Passmore (King’s College London) said We have recently obtained by photon impact the following ionization potentials of the molecules discussed in the above paper PH3 9.98 ; AsH3 10-03 ; SbH3 9-58 ; PF3 9-71 and PC13 9.91 eV. These are in general compatible with the spectra and indicate that the ionization is that of an electron from the central atom, whereas in the hydrides of group 6 and 7 the corresponding spectra show little vibrational structure and their ionization potentials (i-p.) approach closely to the ionization potentials of the central atom with increasing atomic weight.This is not true for the hydrides of group 5 for which the atomic i.p. are P 10.48 ; As 9.81 ; Sb 8.64 eV. The most plausible explanation seems to be that in NH3 and PH3 where I (mole) -= Z (atom) the electrons involved are appreciably antibonding due possibly to orbital repulsion. For arsine and stibine where the molecular i.p. are greater than the atomic ones it does not follow that the electrons are bonding but it appears most likely that the equilibrium ionic configurations are so far removed from those of the neutral species that the photoionization values are not ‘‘ adiabatic ” in agreement with the suggested interpretation of the spectra.The electrons removed from the phosphorus halides are p P electrons which would antibond strongly with the p halogen electrons in planar configurations. This antibonding is removed by bending which is greater in the neutral than the ionized species. The fact that only bending vibrations occur in the spectra indicates the high degree of the compensation of bond strain by bending. The antibonding is expected to be greater for the fluoride than the chloride which is in accord with the relative magnitudes of the observed i.p. Prof. C. A. Coulson (Oxford University) said Brand and Williamson have shown very elegantly that in the 3650A band of propenal a non-bonding electron (n) of the oxygen atom is put into an antibonding (n*) orbital of the conjugated chain.In this process the C=C bond increases in length and the C-C bond decreases; at the same time the barrier heights (or torsional constants z) change in such a way that z(C=C) drops from 990 to about 330 cm-1 and z(C-C) increases from 160 to 250 cm-1. Some simple Huckel-type calculations for the somewhat similar conjugated system = - = of butadiene provide strong arguments to support these conclusions. The excitation from n to .n* in propenal could be simulated b GENERAL DISCUSSION 233 the addition of a fifth n electron to butadiene. This would go into the third mole-cular orbital (counting from the lowest). Standard Huckel calculations give for the 1.c.a.o. representation of this particular orbital the expression where 4I . . . 4 4 are 2pn atomic orbitals for the four carbon atoms in order along the chain and C1 = 0.60 C2 = 0.35.Adding an electron in $m therefore reduces the bond order of the 1-2 (i.e. C=C) bond by C1C2 = 0-22 and increases that of the 2-3 (i.e. C-C) bond by Cz = 0.14. Very roughly these correspond to an in-crease of 0-04 A in C=C and a decrease of 0.03 A in C-C. These are in the same directions as the changes inferred by Brand and Williamson and confirm that the proportional change in the “double” bond on excitation is greater than that in the “ single ” bond. Precise agreement (i.e. better than about 0.02 A) would not be expected in this rough treatment since we have used the n* molecular orbital for butadiene and not propenal and have assumed all resonance integrals /? to be equal. But since a more refined treatment would not change the signs of the individual terms in 1,,9m but would merely make small changes in the magnitudes of the coefficients C1 C2 etc.we can see that theory here provides strong support for all the main conclusions to which Brand and Williamson were led experimentally. A similar conclusion would apply to propynal discussed in the paper by Brand, Callomon and Watson. Dr. J. M. Hollas (N.P.L. Teddington) said In support of the conclusions reached by Brand and Williamson on the excited state structure of propenal (acrolein) re-sulting from the 3860A absorption I should like to report on work I have been doing on this system in absorption using higher resolution. Unfortunately the system suffers from diffuseness even in the 0-0 band so the chief gain in using a larger spectrograph is in the increased dispersion and accuracy of measurement.The limitation on the observation of the rotational structure is somewhat counter-acted by the very close similarity from the spectroscopic point of view of the acrolein molecule to glyoxal (CHO . CHO). Electronically they both have lone-pair elec-trons on the oxygen atoms which may be excited into a conjugated n-electron system. Their moments of inertia are also similar =O in glyoxal being replaced by =CH2 in acrolein. Fig. 1 shows the rotational contours of the 0-0 bands of the first singlet-singlet transition in glyoxall and in acrolein2. The similarities in the structures are obvious. The peaks diverging to high frequencies have been identified in the glyoxal band by King 3 as R-branch heads of the AK = + 1 part of a perpendicular band of a near-symmetric top.Fig. 2 shows the 0-0 band of acrolein at higher resolution. The J-structure is diffuse and only the AK = + 1 R-branch heads can be observed in detail. The K-numbering was obtained by a detailed comparison with the corresponding glyoxal band especially with the R-branch heads of low K-value where the effects of the molecule being a slight asymmetric top become apparent. The head positions for K>2 are given by the equation, v = v,+(A’-B’)+2(A’-B’)K+[(A’-B’)-(A”-B”)]KZ. Values of ( A ’ - 2 ) and (A”-?’) were obtained by fitting the head positions to this equation. The K-numbering was confirmed by comparison of the value obtained 1 Brand Trans.Faraday SOC. 1954 50 43 1. 2 Eastwood and Snow Proc. Roy. SOC. A 1935,149,434. 3 King J. Chem. SOC. 1957 5054 234 GENERAL DISCUSSION -for (A“ - W ) with that available from microwave data.1 After the K-numbering was confirmed a value for A’ was obtained from [(A’-B’)-(A”-B”)] by using the microwave value of A” and the value of @‘-”‘) obtained in glyoxal 3 (the latter quantity is very small compared with A’). The resulting value for A’ is 1-666fO.004 cm-1 and I, = 10.12+0-03 a.m.u. &. Now IA,-IAP = -0.55 a.m.u. & in acrolein and -0.47 a.m.u. & in glyoxal so it seems likely that a similar size change occurs in both molecules. GLYOXAL ACROLEIN frequency FIG. 1.-Microphotometer traces of the 0-0 bands of the first singlet-singlet transition in glyoxal and acrolein obtained by Brand 1 and Eastwood and Snow 2 respectively.King 3 derived three facts about the excited-state structure of glyoxal : (i) A LC-C=O = + 3” (ii) A ( ~ c o +UCC) = 0 (iii) Arc0 and A r c c ~ O but the errors are large (k0.1 A). Assuming that in acrolein similarly increases in LC-C=O and LC-C=C of 3” are the only changes the calculated value of IAt is 10.1 a.m.u. A2 which is in agree-ment with the observed value. But there is strong evidence that the angle increase is not the only change in size of acrolein. As Dr. Brand has pointed out there is a long progression in the C=O stretching vibration which has been interpreted by Inuzuka 3 as representing an increase in rco of 0.1 A. A long progression in 1 Costain private communication.3 Inuzuka B~dl. Chem. Soc. Japan 1960 33 678. 2 King J. Chem. Soc. 1957 5054 FIG. 2.- 0-0 band of acrolein GENERAL DISCUSSION 235 the C=O stretching vibration is common in aldehydes and also occurs in glyoxal. There is therefore good reason for believing that the rco increase of 0.1 A in acrolein is general in aldehydes including glyoxal. This is just possible in King’s results due to the large inaccuracy of rco but the fact that A(rco+vcc) = 0 requires that YCC decreases by 0.1 A. If we assume similar changes in acrolein i.e. that Arc0 = +0.1 A A r e c = +0.1 A (the C=C stretching vibration is strongly active) and Arcc = -0.1 A as well as ALC-C=O and ALC-C-C = +3” the calculated value of IAf is 10.0 a.m.u. A2 which is in reasonable agreement with the experimental value.These conclusions are only intended to be advanced as semiquantitative but they are in good agreement with those reached by Brand and Williamson using quite different arguments. Dr. A. E. Douglas (N.R.C. Ottawa) said Dr. Milton and I have recently measured the Zeeman effect of the 3728A band of pyrazine. It is found that a field as low as 1500 oersteds broadens the rotational structure to such an extent that it is no longer observable. This effect implies that the excited state is a triplet state and that the triplet splitting is small compared to the ordinary Zeeman splitting of the free spin at this field. Dr. G. Herzberg (N.R.C. Ottawa) said I should like to amplify the comments with regard to Dr. Innes’ paper contained in my lecture. It is well known that forbidden electronic transitions in polyatomic molecules can always be made slightly allowed by vibronic interaction certain vibrational transitions (even though not the 0 - 4 band) can occur as dipole radiation.It would appear therefore that long before quadrupole radiation becomes of importance for a given electronic transition vibronically induced dipole transitions will be observable. For this reason alone Innes and Giddings’ suggestion that the 3700A system of pyrazine is a quadrupole transition seemed unlikely to be correct. Their observation of rotational transitions with AK = +2 is similar to our observation of such transi-tions in HSiCl and HSiBr for which no symmetry-forbidden electronic transitions are possible. After a good deal of searching we believe that we have found the reason for the occurrence of AK = +2 transitions in HSiCl and HSiBr namely, spin-orbit interaction ; that is the observed electronic transition is a singlet-triplet transition.Since as is easily seen AK = +2 transitions are also possible in symmetry-allowed singlet-triplet transitions of D2h molecules it appeared to me that one must conclude that also in pyrazine the occurrence of AK = +2 is due to the singlet-triplet nature of the electronic transition and that it is not necessary to involve quadrupole radiation. This conclusion was reached and presented in my lecture before I knew of the result of A. E. Douglas’s beautiful Zeeman experi-ment. I feel however that even if this experiment had not been carried out the arguments presented would have left no way out but to accept the conclusion that the electronic transition is triplet-singlet.Dr. J. H. van der Waals (KoninklijkelShell Lab. Amsterdam) said It is fortunate that Prof. Innes agrees to the 3T,-1Ag assignment for the 3700A transition in pyrazine. This is consistent with an abundance of experimental material on the nature of the lowest (triplet) excited state which is responsible for phosphorescence, of conjugated hydrocarbons and their N and 0 derivatives.1 Moreover the phosphorescence lifetimes z (or inversely the f values of the long-wavelength transi-tion in absorption) of these compounds fit into a coherent theoretical picture without taking recourse to magnetic dipole radiation. 1 for a review cf. Kasha Radiation Res.Suppl. 1960 2 243 236 GENERAL DISCUSSION If the lowest triplet state is n-n* the observed z is of the order of milliseconds ; if predominantly n-n* as in aromatic hydrocarbons it is of the order of seconds. The reason for this contrast can be understood for instance from the detailed analysis of spin-orbit coupling (s.o.c.) in acetone (n-n*) and benzene (n-n*) of Hameka and Oosterhoff.1 When for a n - n* transition the relevant matrix elements for s.o.c. derived from selection rules analogous to those discussed earlier for SO2,2 are expanded in integrals over a.0. the result is dominated by one-centre, one-electron integrals on the hetero-atom(s) e.g., for pyrazine. From the atomic spectrum of nitrogen this integral is known to be about 50 cm-1. In this way the short lifetime and in-plane polarization for the n-n* case are understood (cf.also ref. (4) of Innes and Gidding’s paper). On the other hand when restricting oneself to a pure n-n* triplet state in aromatic hydrocarbons as Hameka and Oosterhoff did all one-centre integrals vanish and one arrives at matrix elements for S.O.C. of about 0.2 cm-1; phosphor-escence should then be in-plane polarized with a z of nearly 103 sec which is con-siderably too long. However from the isotropic hyperfine coupling in free radicals it is known that interaction between the n-electrons and the a-electrons of the C-C and C-H bonds may be important. Prof. Oosterhoff and I have made some cal-culations for naphthalene in which this interaction was considered. With only a few per cent of 0-n* character added into the n-n* state the matrix element for S.O.C.is again dominated by one-centre one-electron integrals and z is brought down to about 5 see in fair agreement with the observed value (20 sec). Moreover the phosphorescence should be polarized perpendicular to the plane as is indeed ob-served experimentally. Dr H. H Greenwood (I.C.I. Billingham) said I should like to describe briefly some results obtained from n electron calculations for pyrazine which may be relevant to the paper of Innes and Giddings. The calculations employ the configuration interaction method of Parr and Pariser but with electron configurations defined on a basis of self-consistent-field molecular orbitals which were calculated according to Pople’s modification of Roothaan’s equations.The nitrogen atoms were repre-sented by a coulomb integral UX = Uc+6U where UC applies to a carbon atom, and SU = kp where p is the s.c.f. carbon-carbon resonance integral. The s.c.f. parameter k is different from its counterpart in Huckel theory. The calculated spectroscopic bands of pyrazine can be conveniently referred to the parent hydro-carbon benzene and for this purpose Pople’s assignment in terms of Clar’s band notation will be used. For example the singlet transitions given in order of increas-ing energy follow the sequence la(l&,) lp(lBla) and 1p lp’(lEla). For singlet-triplet transitions now under consideration the sequence is 3p(3Blu) 3p3#?’(3ElU) and When k is increased from zero the 3p and 3 p bands change according to the figure.The 3p band results are in accord with the value obtained by Peacock and McWeeny with k = 0-35. Results for the 3 p band in the derivative do not appear to have been published previously but the position at which the bands cross falls well within the range of k values normally considered to be physically realistic. The 3p transition may therefore be relevant in itself or in its upper state to the discussion of Innes and Giddings especially if the 3p state has in the absence of other evidence been adjudged the lowest n electron transition. 3 a( 3B2u). 1 Hameka and Oosterhoff Mol. Physics 1958 1 358. 2 this Discussion GENERAL DISCUSSION 237 The above comments assume that the theory provides an adequate description of these spectroscopic properties. In fact the experimental evidence is as much a matter of verification of the present theory which has yet to be adequately tested for heterocyclic molecules.I F t \ 2 5,COOC I I I I I 1 0 - 1 .2 -3 -4 -5 k FIG. 1. Prof. K. K. Innes (Vanderbilt University) (communicated) Since the preparation of our discussion of pyrazine Dr. J. T. Hougen has kindly made available in advance of publication the results of his calculations of branch intensities for triplet-singlet bands. If we apply his expressions to pyrazine the important conclusion is that the band of fig. 1 of our paper is electric dipole in character rather than magnetic dipole as tentatively concluded there. Assuming case (b) coupling of nuclear and electronic angular momenta Hougen has found that the selection rules AK = 0 +2 should apply to two of the four possible electric dipole transitions namely 3Tc - 1A and 3A - lAg.If we assume that similar results apply for magnetic dipole transitions,s the possibilities are 3Rc-1Ag and 3Ag-1Ag but not 391R,-1Ag (which would arise from spin-orbit mixing of a 3R,- or 3Rb-state with an appropriate singlet state 6). Molecular orbital one-electron promotion systems for pyrazine 10 indicate that the 3700 A transition should be assigned n* - n. Likely excited configurations of these schemes are . . . (tu)(tc) . . . (t,)(a,) . . . (ag)(tc) and . . . (ag)(au) and the resulting singlet and triplet excited states are Rb R, Tc and A, respectively. From these the preceding discussion offers the choice 3Tc or 3A for the symmetry of the excited state of the 3700A system.The first of these would be the analogue of the excited singlet state of the 3200A system. The spin-orbit coupling scheme would for this case be that described by Krishna and Goodman.4 Hougen’s results emphasize that there is no conflict between a transition moment in the plane and the type C band structure of fig. 1 of our paper for a triplet-singlet transition 238 GENERAL DISCUSSION Unfortunately because the band structure is only partly resolved it will not be easy to use Hougen's expressions to decide unambiguously between the 3TTc - 1A, and 3A - 'A assignments. The greatest expected difference is that transitions with AK = 0 and AN = +2 should be missing in the first case and only relatively weak in the second. As we mentioned in our paper these transitions are certainly weak but it is not known whether they are missing altogether.Prof. A. D. Walsh (Dundee) said Having regard to the usual interpretation of quantum defects it would seem preferable to write the observed Rydberg series in the spectrum of the tropyl radical as R (n - 6)2' v = 50329 - ~ where 6 = 0.954 and n = 4 5 6 for the observed bands. This however leads to the expectation of an n = 3 member. If it is definite that no earlier member occurs, the difficulty can be avoided if one realizes that 6 cannot for a 3-membered series, be stated with great accuracy. The early members of molecular Rydberg series commonly deviate appreciably from the positions expected from the formula that fits higher members.Within the accuracy of the data and retaining n = 4 5 6 for the observed bands 6 could be increased from 0.954 to 1-0 or a little more. In other words the data are compatible with 6 = 0 or a little more and n = 3,4 5 for the observed bands. No earlier member would then be expected. Small changes in 6 make no appreciable difference to the stated value of the ionization potential and its range of accuracy since the extrapolation from the highest energy observed band to the limit is only over 0.5 eV. From the possible values of the quantum defect if no earlier member exists the series is presumably a d series ; if an earlier member exists the series is presumably an s series. Dr. B. A. Thrush (Cambridge University) said We have chosen (n+O-O46) rather than (d-0.95) to represent the observed Rydberg series of the tropyl radical because the first observed series member corresponds to n = 3 or n' = 4.We have been unable to detect any further transitions at longer wavelengths. This series appears to correspond to the R primed series in benzene.1 We have not detected any non-Rydberg transitions of the tropyl radical; the first member of the Rydberg series is a 4.77 eV which is between these strong n-n transitions at 4-26 eV and 5.06 eV predicted by Longuet-Higgins and McEwen.2 Prof. W. C. Price DP. T. R. Passmore and Dr. D. M. Roessler (King's College, London) said We have recently determined the ionization potential of cyclohepta-triene obtaining a value of 8.28 eV i.e. close to that of hexatriene -8.27 eV from Rydberg series.3 The difference between the ionization potential of the parent molecule and the tropyl radical-2-04 eV or 47 kcallmole-is equal to the differ-ence in the bond strengths (C~H~-H)-(C~H7f-H).The latter is clearly different from the former because of the greater stability of the n6 closed shell stricture of C7H3 relative to C7H7 which has one 7t electron in an antibonding 7t orbital. The antibonding power of this electron is equal to p the resonance integral for this radical which is thus evaluated as 47 kcal/mole a value close to that derived for hexatriene (45 kcal/mole) in the paper quoted above. Dr. I. M. Mills (Reading University) said Prof. Price has shown in his slides (although not in the preprint) an interesting way of plotting the total dissociation I Wilkinson Can.J. Physics 1956,34 596. 2 Longuet-Higgins and McEwen J. Chem. Physics 1957 26 719. 3 Price and Walsh Proc. Roy. SOC. A 1946 185 182. Liehr and Moffitt J. Clzem. Physics 1957. 26 1074 GENERAL DISCUSSION 239 energies of these molecules so as to draw attention to deviations from the ap-proximate additivity of bond energies. He has further suggested that the deviations from additivity observed in the fluorides which are not observed in the hydrides, can be completely understood in terms of the presence of Urey-Bradley repulsive forces between non-bonded fluorine atoms in the fluorides which are assumed to be unimportant in the hydrides (comparing for example the molecules CH4 and CF4). I find it hard to believe that the problem is quite as simple as this.The evidence for or against the Urey-Bradley model force field obtained from normal co-ordinate calculations is quite ambiguous and in any case the correlation of evidence from vibration frequencies with evidence from dissociation energies must be treated with great caution. I find that evidence obtained from equilibrium geometry-which is much more precise and reliable-is very hard to understand in terms of the Urey-Bradley model. The table shows the CF bond length and the interbond angles in CHlF and CF4 the changes are precisely the opposite of those that one would CH3F CF4 C-F/A 1 -39 10.0 1 1.320 f0-005 LHCH or LFCF 110" 20' 130' 109' 28' expect from the Urey-Bradley model which postulates a large F . . . F non-bonded repulsion a smaller H . . . F repulsion and an almost negligible H . . . H re-pulsion. It is of course easy enough to find an explanation of these results in terms of polarity and orbital hybridization effects the point that I wish to make is that these other eEects appear to be at least as important as repulsive forces between non-bonded atoms so that the apparent simplicity of the explanation offered by the Urey-Bradley model is probably rather misleading. Prof. W. C. Price (King's CoZZege London) said In reply to Dr. Mills it was not intended to minimize other effects but to bring out the magnitude of the Urey-Bradley repulsion in reducing the energies of the second and third bonds in the halides of nitrogen and oxygen relative to that of the first. This effect is very con-siderable in the radical ions where because of the reduced radius of the central atom the FXf-F bonds become very weak due to increased repulsion between terminal atoms. The molecules to which Dr. Mills refers are clearly affected by the difference in polarity of the H3C-F and F3C-F bonds which must have a large effect on these bond distances. Since CF4 has the tetrahedral angle whatever the force field it cannot be used in any argument concerning the nature of this field. Prof. J. B. Coon (Texas A and M University) (communicated) :-Dr. Brand pointed out that observation of the hot band (O-,l) would furnish direct proof of the double-minimum potential in ClOz. To determine the feasability of observing this band its intensity is calculated for the potential function of the present paper. The calculation yields an intensity of 0.0086 for the (O-,l) band relative to the (O+,O) band. If such a band occurs it is overlapped by a hot band at 20070 cm-1 which involves v;'. This hot band has an intensity of 0.13 relative to the (O+,O) band. Actually 15 hot bands have been observed to the red of the origin of this band system and all of them have been explained in terms of the totally symmetrical frequencies v; and vI;.

ISSN:0366-9033    

DOI:10.1039/DF9633500212

出版商:RSC    

年代:1963

数据来源: RSC

摘要:

Ballard, R. E., 43. Brand, J. C. D., 175,184,224. Buckingham, A. D., 48. Callomon, J. H., 175. Cesani, F. A., 118. Charney, E., 213. Coon, J. B., 118. Coulson, C. A., 71, 214, 217, 218, 221, 224, 226, Davies, D. W., 214. Dixon, R. N., 105,222, 23 1. Douglas, A. E., 158, 220, 235. Dows, D. A., 48. Dubois, I., 124. Duncan, J. L., 231. Gausset, L., 113. Giddings, L. E., Jr., 192. Greenwood, H. H., 218, 220, 236. Herzberg, G., 7, 77,113, 220, 221, 235. Hirst, D. M., 215. Hollas, J. M., 233. Humphries, C. M., 137,148. Innes, K. K., 192, 237. Johns, J. W. C., 90. Kuppennann, A., 30,212,213. Lagerqvist, A., 113. Linnett, J. W., 58,214, 215,216,217, 221, 226. Longuet-Higgins, H. C., 77. Lorquet, J. C., 83, 221. 232. Loyd, C. M., 118. Mackor, E. L., 212. Mason, S. F., 43,213. Merer, A. J., 127,230. Mills, I. M., 222, 238. Neilson, A. H., 71, 217, 218. Passmore, T. R., 201, 232. Peacock, T. E., 144,231. Price, W. C., 201,232,238, 239. Priddle, S. H., 90. Ramsay, D. A., 90. Rias-ur-Rahman, 144. Roessler, D. M., 201, 238. Roff, L. M., 30. Rosen, B., 113, 124. Skell, P. S., 225. Sleeman, D. H., 144. Sovers, O., 58. Thrush, B. A., 196, 222, 238. Tuckley, E. S. G., 144. Vane, G. W., 43. Waals, J. H. van der, 212, 227, 235. Walsh, A. D., 137, 148, 212, 218, 223, 224, 229, Warsop, P. A., 137, 148. Watson, J. K. G., 175, 230. Williamson, D. G., 184. Zwolenik, J. J., 196. 230, 238. AUTHOR INDEX * * The references in heavy type indicate papers submitted for discussion.

ISSN:0366-9033    

DOI:10.1039/DF9633500240

出版商:RSC    

年代:1963

数据来源: RSC