The electronic structure of some heterocycles with bridgehead nitrogen: photoelectron spectra and ab initio molecular orbital calculations

Journal

QRsevier

of Molecular Structure, 42 (1977) Scientific Publishing Company,

85-101 Amsterdam

-

Printed

in The Netherlands

THE ELECTRONIC STRUCTURE OF SOME HFaTEROCYCLES BRIDGEHEAD NITROGEN: PHOTOELECTRON SPECTRA AND AB INI’I’IO MOLECULAR CALCULATIONS

MICHAEL

H. PALMER,

Department

of Chemistry,

DEREK

LEAVER,

University

JOHN

of Edinburgh.

D. MSBET,

ROSS

West Mains Road,

WITH ORBITAL

W. MILLAR

Edinburgh

EH9 355 (Gt. Britain) and RUSSELL Department (Gt. Britain) (Received

EGDELL of Inorganic

Chemistry,

University

of Oxford.

South

Parks

Road.

Oxford

3 June 1977)

ABSTRACT He(I) and He(U) photoelectron spectra are reported for thd cycl[3,3,3]azine (l), cyc1[3,2,2 ]azine (2), indolizine (6) and imidazo[ 1,2-a] pyridine (7), as well as He(I) spectra for related compounds (3-5)_ Ab initio molecular orbital calculations have been used to assign the spectra of 1, 2,3, 6 and 7, and to give information about the nature of the a-electron energy levels. The first IP for 1 is singularly low (5.86 eV), and this has vacant in been interpreted in terms of occupancy of the la,” orbital which is normally related compounds. In the cyclazines, the nitrogen lone pair seems to be split into two n-levels. INTRODUCTION

The well-known Hiickel rule for cyclic conjugated hydrocarbons predicts fundamentally different properties for molecules if the number of n-electrons is given by (4n. + 2) rather than 4n (n = integer) 111. The former are the stable singlet aromatic series, whereas the latter are predicted to be triplet states unless deviations from regularity occur. The rule, originally devised for monocyclic systems, has been widely applied to the periphery of polycyclic systems; in this way, cross-linking of the periphery is regarded as a minor perturbation [ 21. Owing to ring strain many of the annulenes (C, H, ), of interest to the study of these predicted differences, adopt non-planar conformations [3] I One way in which the molecules can be constrained to planarity is by incorporation of an internal nitrogen atom as in the cyclazines (1) [4a-4c] and (2) [5a, b]. These have 4n and (4n + 2) n-electron peripheries respectively, and are usually known as cycl]3,3,3] azine and cyc1[3,2,2] azines respectively.

b=Me

~,R=H;

5

4

This study is based upon our earlier independent synthetic studies [4a-c] of l-5, and photoelectron and molecular orbital studies of various heterocycles [Ga-c 3. We report ab initio calculations for l-3,6 and 7 zs well as He(I) and He(II) photoelectron (PES) spectra for all except 3 (He(I) only)* and partial He(I) spectra for 4 and 5a, b. Indolizine (6) and its aza derivative (7) are included as reference molecules with bridgehead nitrogen atoms. The principal objectives are to obtain and interpret the photoelectron spectra and investigate the extent of interaction of the nitrogen atoms with the hydrocarbon system. In addition we extend the analysis of the recent He(I) PES [‘7] of the related 1,5,9&ridehydro[12] annulene, which can be represented by 8 or 9, and is another 412system. A number of calculations have been reported for 1,2,6 and 7 by earlier workers; these can be summarised as empirical (HMO): 1 [Sa, 9],2 [8a, 8g, 9],6 [8b-8f, 91 and 7 [8b, 8e, 8f] ; and semiempirical: 1 [lOa], 2 [ll], 6 [lo, 12-151 and 7 [ 12c,1% 141. The present paper records the first all electron, ab initio studies of l-3,6 and 7.

6

7

IO *Spectra of 4, 5a and 5b are deposited in BLLD, together in mol. l-3, as Sup. Pub. No. SUP 26075 (9 pages).

with expansion

of fine

structure

EXPERIMENTAL Synthetic methods All the compounds were prepared by previously reported procedures

[4,5b,

16,171.

Instrumental

measurements

The photoelectron spectra were obtained with a Perkin Elmer PS18 modified by incorporation of an Helectros hollow cathode discharge lamp, which gives both He(I) (21.22 eV) and He(I1) (40.81 eV) radiation. In all cases the spectra were obtained in admixture with the calibrants Xe (2Pa,2 12.13, 2P,,2 13.44 eV), and Ar (‘Ps12 15.75 eV) or N2 (2C, 15.58, 2ZU 18.75 eV); further calibration was by means of He’ (2S 24.59 eV) and He(I)p satellites of Xe. Computational

methods

In order to compare directly with our previous work on the PES of N-heterocycles [6b, c] we used an identical gaussian basis set. More than one geometry was investigated for l-3,8 and 9 (see below); there are no experimental data for 6 and 7, and two attempts at optimisation of 6 within the CNDO/B framework [12,13] have led to varying results, although one was close to a MINDO/B result [ 131. In the present work we have adopted a superposition of pyrrole or imidazole on 2-pyridone (yielding 10 and ll), a method which gives satisfactory agreement with experiment for other N-heterocycles. In practice the geometry of 10 is close to the average of the semi-empirical ones cited above 112,131. RESULTS

AND CORRELATIONS

Total energy

The experimental geometry [ 181 of 1,4dibromocycl[ 3,2,2] azine is likely to be similar to that of the parent compound (2), notwithstanding an earlier calculated geometry [lOa], and thus was utilised (12) for the latter; it gave (Table 1) a total energy only 13 kJ mol-’ lower than another geometry based upon benzenoid lengths (rcc = 1.398 A, angles 120” and 105” at carbon in the 6- and 5-membered rings respectively). In fact the major difference between the two geometries lies in Cs-C6/C6-C, which are 1.441 A in the experimental geometry, and may possibly relieve some of the strain from the 6,5,5-fused tricyclic system. It is interesting to note that I

I

when =N was substituted for =CH at position 6 in 12, (leading to 3) the longer CN (1.441 A) bonds gave the lower energy. There is no experimental geometric information on cycl[3,3,3] azine (l), and in view of the computational size, it is impracticable to consider an optimisation by ab initio methods. However, the alternating geometry (13)

led to a slightly lower singlet energy (by 30 kJ mol-‘) than the singlet benzenoid geometry (rcc = 1.398 A, angles 120”), with the triplet benzenoid geometry a further 129 kJ mol-’ above this (Table 1). The significance of the singlet differences is uncertain, although it is in agreement with earlier semiempirical calculations [ 1 f&la] ; indeed the geometries used in the two studies for 1 are effectively identical. There is no experimental information for 112 3 annulene, but the recent observation 1191 that the low temperature 13C magnetic resonance sp ectrum of heptalene, a further bridged [ 123 annulene (as is l), shows nonequivalence of C-UC-5 and of C-2/C-4, has been interpreted in terms of fast n-bond exchange at higher temperatures, in agreement with calculations [20 J. This could well point to a similar situation with 1, but this must be regarded as unproven since other heptalene derivatives are known to be non-planar at normal temperatures [21]. The orbital energies and most atomic populations are little affected by these minor changes in bond length; we have normally adopted the data from the computation of lower total energy. For all geometries in l-3 we have calculated substantial resonance energies (RE), based upon earlier bond energies [6b, 6c, 221. The value for the alternating form being higher (30 kJ mol-‘) than the benzenoid form in 1,isan anti-aromatic characteristic, in agreement with the HMO 4n periphery predictions, and earlier calculations (ref. 1Oa gives a difference of 50 kJ mol-I). However, both values for 1 are higher than for 2 or 3 (Table 1) in contrast to that of the empirical [9] and semiempirical calculations [lOa], the latter of which show 2 to be more stabilised than classical 1 by 53 kJ mol-‘. In the latter work the u- and n-energies were calculated separately as an approximation; of course this does not occur in ab initio studies. We attribute the present result to a combination of lower u-bond strain in 1 than 2 (or 3), and to more efficient cross-linking of the ring in the symmetrical [3,3,3] system. The difference in reactivity of 1 and 2 is discussed below. The RE’s for indolizine (6) and its azaderivative (7) are much lower than in either indole (308 kJ mol-‘) or isoindole (250 kJ mol-‘) [SC] as expected_ Photoelectron

spectra and their assignment

Combined He(I) and He(I1) spectra for indolizine (6), its aza-derivative (7) and the cyclazines (1 and 2) are shown in Figs. 1-4; with the He(I) spectrum for the aza-cyclazine (3) in Fig. 5. The compounds 4 and 5a, b show little in the way of well defined IP’s beyond 10 eV; hence the low energy regions only are available* with these and other observed IP’s being *From BLLD as Sup. Pub.

No.

SUP

26075 (9 pages).

Aherneting geometry

1.12

Benzenoid geotne try -436.00727 2.14847 217

-512.66743 2.59335 171

-512.71688 2.64280 300

Eenzenoid geomelry

1.646 = 27.21 eV.

1.546

Geometry B Geometry A hi/N6 1.441 A) (k5/~6 1.399 a) -451.90458 -451.90120 1.87765 1.87427 186 196

0.861

“Experimental ” geometry -436.01228 2.15346 230

a1 a.u = 627 kcal mol-I, = 2626 kJmo1”

Total energy (a.u.) Binding energy (a.u.) Resonance energy (kJ mol-‘) Dipole moment (D)

6-Aza cyc1/3,2,2] azine

Total energy (a&) Binding energy (a.u.) Resonance energy (kJ mol”) Dipole moment (D)

Cycl/3,2,2]azrne

(a) Singlet Total energy (a-u ) -512.72828 Binding energy (a-u.) 2.65420 Resonance energy 330 (kJ mol-‘) (b) Triplet Total energy (a.u.) Binding energy (a.u.) Resonance energy (kJ mol”)

Cyc1[3,3,3Jazine

Indolizine (6)

3.5

-376.37263 1.63651 175

Imldazo[1,2-aj pyrldme (7)

(I2]Annulene (Regular duodecagon) (a) Smgle t -459.80494 Total energy (a.u.) (b) TrIplet Tutal energy (a.u.) -459.81046

1,5,9.Trrdehydrof12]onnufene (8 and 9) (a) Acetylenic structure (8); orbital lo ,I’ doubly occupied -456.18244 Total energy (a.u.) (b) Acetylenic structure (8); orbital 20,” doubly occupied Total energy (a.u.) -456.52422 (c) Cumulenic structure (9) -456.40797 Total energy (a.u.)

Total energy (a,u.) -360.48442 1 .a4624 Binding energy (a-u.) Resonance energy 220 (kJ mol”) Dipole moment (D) 1.47

Molecular total energies (a.~,)~ and dipole moments (Debye) from ab initio calculations

TABLE 1

IO

14

I8

22

26

eV

Fig. 1. Photoelectron spectrum (He(I) and He(IT)) for indolizine (calculated, scaled levels inset).

it

15

19

23

eV Fig. 2. Photoelectron spectrum (He(I) and He(n)) scaled levels inset).

for imidazo [1,2-alpyridine

(calculated,

91

II

I

a

16

12

18

24

eV

Fig. 3. Photoelectron speckurn (He(I) and He@)) levels inset).

5

9

17

I3

21

for cycl[3,2,2]azine

(calculated, scaled

25

eV

Fig. 4. Photoelectron spectrum (He(I) and He(II)) for cycl[3,3,3]azine levels inset).

(calculated, scaled

92

given in Table 2. Other than the first IP, very few bandsshow fine structure*, and all of these are thought to arise &om r-type ionisations [23]. Although a considerable number of IP’s have been identified (Table 2), multiple assignments have been necessary for certain portions of the envelope where no separate maxima are observed. AsGgnment has been assisted by (a) previous experience [6c] with indole and isoindole which are isomeric with the indoIjzines (6 and 7); (b) by variations in band intensity in the , immediate neighbourhood, where cross section differences should be small; (c) variations between IP’s both observed and calculated, and also spaces between groups of IP’s across the series l-3,6 and 7. Finally, the range of accessible_IP’s in the present work is 5-25 eV; the restrictian at the high energy end arises from the unfiltered nature of the He(I) plus He(II) radiation and thus overlaying of the kinetic energy ranges from the tFRoexcitation sources. The !2sNand 2sc levels in pyrrole are at 29.5, and 23.8 plus 22.3 eV respectively [24] ; some other 2sc levels relevant to the present compounds are: benzene 25.9,22.7 [24]; pyridine 24.2, 23.3 [25]; pyrimidine 24.4, 20.5 [26a]; pyrazine 24.1 and 21.0 eV [26b]; hence it seems probable that at least 3 levels (2sN + 2~2s~) will not be observed in each of 1,2,6 and 7. We have assigned the high energy end to account for variations across the series with this in mind. If we place all the low energy IP’s (5.8-11.5 eV) for the molecules l-3,6 and 7 in strict sequence and compare with the calculated sequence using Koopmans’theorem (23 levels), then only 5 calculated levels are out of position; of these, 3 arise from the aza cyclazine (3) whose geometry (and hence calculation) is less well defined. Over this short range

0

_ 12

16

20

eV

Fig. 5. Photoelectron *See footnote

p_ 88.

spectnun

(He(I))

for 6-azacycl[&Z.Z]azine

(calculated,

scaled levels inset).

93

TABLE

2

Experimental ionisation potentials and their assignments with calculated data IP(eV)

-q&V)

Cyc1[3,3,31azine 5.87 8.26 9.66 10.3 10 69 11.53 12.01 12.92 13.77 14.52 15.53 15.80 19.91 22.68 23.58

(I)= 5.86 10.22 11.84 13.45

Cyc1[3,2,2/azine

(2)=

7.63 8.41 9.26 10.77 11.11 12.85 (broad) 13.59 14.47 16_15(broad) 15.56 18.45 22.15

e'

14.25 15.25 16 87.16.67 17.08 17.70.18.32 19.15 20.05 23.52.23.71 28.02 29.46.30.80

a' e' 0: a' e' 'Ia' e a,s e' et a' e' a: e'

8.12 8.67 10.70 13.18 13.58 14.32.14.75 14.86.15.13 16.24 16.92.17.16 17.94.18.38 18.54.18.92 21.69. 22.11 22.69, 23.85 26.64, 27.58

16.60.17.15 17.66.17.67 18.94.18.96.19.29 22.22. 22.49 23.48 24.24, 24.48

19.28 Iontsation

(cxh) a" e" a- (LPN) err

13.51

28.33 6-Azacycl[3,2.2lazrne (3)= 8.59 7.66 9.07 8.51 11.33 9.30 &boulder) 11.13 9.50 13.71 11.00 13.97 11.90 14.78 13.07 1540 13.49 15.46 13.92 15.1 15.35 16.06 17.05 1802

Assignment

(C,") 46, % 3b,
IP(eV)

Indolnine

--E,(eV)

(C,)

(6)

7.24 8.60 10.27 10.96 12.10 12.76 13.44 14.42 14.70 14.90 15.68 18.45 19.70 21.76

7.53 9.14 1200 1299 13.62 14.31.16.03 16.36 16.68.17.05 17.63.17.96 18.35 19.65 21.66.22.08 23.52 26.24. 27.28

1.5.9-TridehydroKl2Jannulene (8) 7.69 8.53 10.76.11.05 9.5 11.67 10.4 13.48 11.67 14.49 12.4 13.2 15.33 13.9 16.01 15.1 17.45. 17.59 16.1 19.01 16.75 19.73 Imi&zo[l.2-alpyridine 8.19 9.08 10 09

11.07 11.41 1264 13.50 13.94

Assxgnment

a” (77) a”*. a..

a”

a’ (0)

a;.a’

a’ a: a’ a" cLPp~=),a' a' a’ a: a’

a’ a: a’ I,

a2 e”.

aI e”

;

e’

I,

a2

e’

I

(12 alI ’ e’ aI

e’

(7) a”

8.62 9.77

a”

a’ (LPN)

11.33 13.07 13.88 14.41 15.42 15.87

a”

a” a’

a’ a’

1500

17.21.

16.02 17.03 19.20 20.18 22.73

18.51. 18.55. 18.90 20.29 22.6. 23.09 24.10 27.98, 28.13

17.48

a; P’ a: a’:

a’

a: a’ a’ a: a'

potentials

3-MethylcycIopentaCcdlc~cl[3,3.3]azine (4) IP 6.37. 7.35. 8.41.9.15. 9.55.10.33.12.1. Cyclopenta[hlc~clC4.2,2]azine<6a)~ IP 7.06.7.63. 8.90. 10.05.14.47.12.58.

14.0

13.7. 15.07. 16.46. 18.4

6.8-DimethyIcyclopenta[hlcycl[4.2.2]azine(6b) IP 6.99. 7.57. 8.82.9.9

*The JUPAC systematic rules give the following

names: pyrido[ 2,1.6-delquinolizine (1); pyrrolo[2,1,5_cd]indolizine (2); pyrazino[6,1,2_cd]pyrrolizine (3); 3-methylcyclopenta[ijlpyridoC2,1,6-delquinolizine (4); cyclopenta[4.5]azepino[7,1.2-cd Jpyrrolizine (5). SimpIicity, and widespread acceptance lead us to adopt the cyclazine notation in this paper.

a’

94

a linear correlation exists, IP,,,, = 0.62 IPcd, + 2.6 eV with standard deviations of slope, intercept and overall of 0.032,0.353 and 0.342 respectively. This suggests that we can construct a correlation between the calculated and observed IP’s, the two ends being fairly firmly established. The more complex central region (12-18 eV) is then assigned by consideration of IP groupings and variations in the series. In this way an overall correlation was obtained Ip, bs = 0.771 IP,& + 1.07 eV, with standard deviations in slope, intercept and overall of 0.011, 0.196 and 0.533 respectively. These correlations are shown inset in Figs. l-5; it will be observed that the principal groupings and spacing are reasonably satisfactory, but less so for 1 and 3 than 2,6 and7. Similar correlations have been obtained for related heterocycles [Sa-6c], and generally the slope increases as the range of IP’s is extended [27]. It is well-known [6a, 12a] that CNDO/B does not give satisfactory groupings of IP’s in the upper valence shell, and that the n-levels are often at too high a binding energy. This is evident in the only case (6) in the present work where semi-empirical orbital energies are available [12a]. Detailed comparison with the present data, and with the PE spectrum of 6 shows that it is not possible to arrive at a satisfactory explanation of the four IP’s at lowest binding energy either in spacing relative to the main u-system, or in order, where the CND0/2 IP order is n < u < 7c< (J< u etc. The photoelectron spectrum of 1,5,9-tridehydro[ 121 annulene (8) was recently reported [ 73, and a number of possible assignments based upon LCBO, SPINDO, MIND0/2, MIND0/3, PPP and X, calculations were given for the first few Ip’s. As shown below we have carried out a number of calculations on 8 owing to its similarity to cyc1[3,3,3] azine (1); the lowest energy state leads to a very accurate correlation with the observed IP’s up to about 17 eV (Table 2), the line being given by IP,b, = 0.816 IP,al, + 0.697 eV (standard deviations in slope, intercept and overall of 0.011, 0.160 and 0.126 respectively). It is worth noting that the present calculations are in almost exact agreement with the orbital ordering from both SPINDO and X, calculations [ 73, but neither of the MIND0 calculations bears much relationship to these, either in ordering or groupings. Although the absolute values for the IP’s calculated by SPEND0 and X, are significantly better than in the present LCGO calculations the groupings are no better, and poorer in the case of the 9.5-10.4 eV region (for SPINDO) and 13.2-13.9 eV region (for X,). The correlation lines are IP,b, = 1.058 -0.869 eV (S-D. as above, 0.023,0.292, 0.211) and IP,,b, = 0.886 IPSPINDO IP, + 2.186 eV (S-D., 0.037, 0.456,0.404).

Population analyses

For all the compounds l-3,6 and 7, nitrogen is an acceptor from carbon, and carbon from hydrogen in total populations (Table 3), as observed previously [SC, 281. For indolizine (6), imidazo[ 1,2-a] pyridine (7) and cyc1[3,2,2] azine (2), the total populations are very similar to those obtained

by superposition of buta-1,3diene with either pyrrole (14) for 6 and 2, or imidazole [28] for 7 respectively. The population deficiencies at the bridgehead atoms in 2,6,7 relative to C-2 in pyrrole or imidazole, arise from the absence of an attached H-atom. Most of the previous calculations cited above were ~-electron, non-iterative HMO types and clearly little direct similarity with an all-electron system can be expected. An exception to this for indolizine (6) is the w-technique iterative HMO method [29] which yields both a wave-function and n-electron distribution very similar to the present work. The valence shell semi-empirical calculations on 6, produce an even closer similarity with the present work, particularly in the n-electron distribution, where often the total population orders are nearly identical [12a, 13, 141. In the aza-indolizine (7), the CNDO/B calculations give [ 13, 141 the two nitrogen atoms, although differing in total density, nearly equal aclensities; in contrast the carbon atoms are often highly polarised in the 7r-system.The present work suggests that the molecule is much closer to the classical structure; both nitrogen atoms are equal in total density, but widely differing in Ir-electron components. Previous population analyses on the cyclazines (1 and 2) are much more fragmentary; the information given from some semi-empirical calculations [ll] is incomplete, but it is of interest to note that the ~-distributions of Dewar and Trinajstic [lo] are not unlike those calculated for 1 and 2 here, except that N-10 for 1 is a much stronger donor to the periphery in the present work (recovering much of this by o-acceptance from the o- and p-positions (C3a and C-2). This is perhaps not unexpected, since the molecule by becoming meso-ionic (an extreme situation) can achieve the stable 14n periphery_ There is considerable variation in populations with geometry change here, the charge separations being smaller in the “classical” alternating species. The same trends as in the pyridine populations (15) are observed: 0, C, > CD> C,; 7~~Cp > C, > C,, but the absolute values are significantly different. The present calculations yield satisfactory values for the molecular dipole moments for 6 and 7; this is also true of the CNDO/B results above, which although leading to larger internal dipoles undergo mutual cancellation. A systematic difference evident in all the comparisons here, is that the bridgehead N atom is a stronger srdonor and weaker o-acceptor than in the present work. Thus it is almost electrically neutral or even positively charged in the CNDO/B calculations [12a, 12b, 13,141, whereas it is always significantly negative in the ab initio calculations.

96

TABLE3 Popdation analysis Cyclf3.3,3]

Is -t2s 2PCI 2Ps Total 1%

azine (regz~lnr, Dab geometly)

C-l 3.0142 1.9990 1.2012 6.2144 H-l 0.8546

c-2 3.0792 2.1561 0.8582 6.0932 H-2 0.8399

C-3a 2.9906 1.9302 0.9112 5.8319

Cycl[3,3,3]azine (alfernafinggeometry) C-l c-2 c-3 1s c 2s 3.0388 3.0765 3.0254 2Pa 2.0267 2.1304 2.0043 l.l.099 0.9056 1.1824 2P?r Total 6.1754 6.1125 6.2081 H-l H-2 H-12 wr 0.8482 0.8442 0.8586

N-10 3.3791 2.4240 1.4855 7.2886

C-9a 2.9921 19139 0.9441 5.8501

N-10 3.3690 2.3663 1.5737 7.3091

Cycl[3,2,2] azine C-l 1s + 2s 3.0298 2pa 2.0538 2Prr 1.0996 Total 6.i831 H-l 1sH 0.8387

c-2 3.0581 2.0878 1.0019 6.1479 H-2 0.8498

C-2a 2.9471 1.8451 1.1318 5.9240 H-5 O-8366

C-4a 2.9524 1.8923 1.0543 5.3989 H-6 0.8482

c-5 3.0573 2.0852 0.9990 6.1414

C-6

3.0682 2.0543 1.0174 6.1399

N-8 3.3453 2.4085 1.5413 7.2951

Pyrazino[G,1,2-cd]pyn C-l 3.0311 1s + 2s 2.0566 2Pa 1.0899 2P* Total 6.1776 H-l 0.8449 W-I

wlizine c-2 3.0585 2.0882 0.9987 6.1446 H-2 0.8342

C-2a 2.9486 1.8464 1.1227 5.9178 H-5 0.8256

C-4a 2.9539 1.8963 1.0590 5.9103

c-5 3.0565 1.9605 0.9998 6.0167

N-6 3.6002 2.6363 1.0435 7.2799

N-8 3.3455 2.4083 1.5407 7.2945

c-2 3.0620 2.0820 1.0332 6.1772 C-8 3.0415 2.1012 1.0212 6.1639 H-2 0.8409

c-3 3.0206 1.8917 1.1480 6.0603 C-8a 2.9762 1.8392 1.0724 5.8879 H-3 0.8262

c-5 3.0110 1.9760 0.9998 5.9868 N-4 3.3667 2.4417 1.5258 7.3343 H-5 0.8184

C-6 3.0446 2.0779 1.0635 6.1860

H-7 0.8344

H-8 0.8306

Indolizine IS + 2s

2Po 2Pn Total Is i- 25

2Fa 2Pz Total

1%

C-l 3.0462 2.0406 1.1322 6.2191 c-7 3.0421 2.1197 1.0038 6.1656 H-l O-8367

H-6 0.8317

97

3 (continued)

TABLE

Imidazo[l,2u]pyridine 1s + 2s 2P0

2P, Total

1s + 2s 2P, 2Pn Total

N-l 3.5521 2.7162

1.1155 7.3835 C-6 3.0480 2.0856 1.0443 H-2

1.5374 7.3355 c-7 3.0355 2-1049 1.0220 6.1624 H-3

0.8328

0.8004

6.1779

IsH

N-4 3.3689 2.4292

c-2 3.0038 1.9409 1.1171 6.0619 C-8 3.0475 2.1078 0.9985 6.1537 H-4 0.8247

c-3 3.0481 1.9810 1.0234 6.0525 C-8a 2.9021 1.7319 1.1115 5.7455 H-5 0.8292

c-5 3.0039 1.9567 1.0305 5.9912

H-6 0.8325

H-7 0.8163

DISCUSSION One of the most obvious features of the ionisation potentials of the series b, 6,7 is the extremely low first IP'sfor the pair of cyc1[3,3,3] azines (1,5.87 eV and 4,6.37 eV) when compared with either the cycl[3,2,2] - or cyc1[4,2,2] azines (2 and 5a, b). These must be among the lowest

1-4,5a,

values recorded for organic molecules; the limiting situation is perhaps represented by graphite whose first IP is 4.9 eV [30], but it is interesting to note that recent values for IP, in the bridged [lo] -annulene (17) [31] and a series of bridged [ 141 annulenes (e.g. 18 [ 321 and 19 [331) have IP’s of 7.90, 7.14 and 6.7 eV respectively. Pyrene itself (cf. 19) has IPI 7.72 eV. The size of the molecules 4, 5a, b has prevented us carrying out computations on these molecules, but the present results suggest that further PE studies of 4nx-periphery condensed systems could lead to even lower IP’s, and possibly to compounds of interest as potential semi-conductors. The recent observation [ 71 that the 1,5,9_cyclododecatriene-3,7,11-triyne (8) has a first IP of 7.69 eV, even though it has a 4rm periphery, at first suggests

16

17

18

98

that the bridging nitrogen atom is critical to the low IP’s in 1 and 4. In order to pursue this situation further we have performed a series of calculations on the tiene-triyne (8). First, two geometries corresponding to the classical

forms 8 and 9 were constructed. The total energy of the acetylenic form 8 was lower than 9 for one electron configuration studied (Table 1). It seemed probable that the reason for the lower IP in 1 than 8 lay in the occupancy of orbital 1~~” (16), whereas an orbital of this symmetry is the lowest unoccupied orbital in 8. Thus two SCF calculations were performed on 8, with the highest occupied orbital la,” or 2uzn; the latter gave the lower energy and thus represents the ground state occupancy; furthermore, the binding energy of la*“, was only 3.9 eV in the former calculation with 2az” only 1 eV above this. This seems conclusive proof of the occupancy of the orbital of ai” symmetry being responsible for the low IP’s; it is of course only

bonding between widely separated atoms (16). As a final proof we considered other molecules where such an orbital might arise. In most of the examples where this might occur, the lower symmetry seemed to preclude an al” type; clearly an even membered ring

with 3 fold symmetry axis would produce the required effect. [12] Annulene, as a regular duodecagon, seemed practicable (Table 1). Although the triplet state had a slightly lower energy than the singlet case, both had very low calculated vaIues (6.43 and 6.27 eV respectively). Inspection of the absolute comparisons for IP, in Table 2, suggests these figures are not too improbable. Finally, the cyclic boron analogue (20) was considered by combining a central boron atom with the hydrocarbon periphery of 13; the lowest energy (singlet state, -483.05411 au.) gave a highest occupied orbital energy of 7.78 eV; the lowest unoccupied orbital was of symmetry la ,” ; it seems unlikely that this (unknown) species would have an especially low first IP. The hydrocarbon (8) has both normal out-of-plane (7~) and acetylenic in plane (n’) orbit&; if the latter are ignored and a direct comparison of the PE spectral data of 1 and 8 performed using the present assignments, then

it is clear that the n-IP’s are significantly shifted to lower binding energy in 1 than in 8 when the nitrogen atom is nodal (le” and 2e” are at 10.3 and 8.26 eV (1) and 11.67 and 9.5 eV (8)). The converse applies for the orbit& of a” symmetry (14.52 and 9.66 eV (1) and 12.4 and 7.69 eV (8)); the reason for this is clear from the orbital population analyses; both a” orbit& in 1 have high lone pair N character and the (2~~)~ electron density arises

from almost equal amounts of the two orbit&. The higher energy one is marked LPN in Table 2. Clearly interaction of 2~~ (of higher binding energy) with linear combinations of 2pc (of a” symmetry) lowers the inner combination (la*“) and raises the antisymmetric outer combination (2~~“). We might expect, therefore, if the present assignments are correct, these resultant orbit& to vary in energy depending upon the 2pN/2pc ratio. The assigned IP’s (and calculated data in parentheses) in ascending proportion of 2pN are (eV): 1,14.52 (17.7); 2,15.56 (18.54); 6, 14.70 (17.63); 3, 15.05 (18.95) and 7,16.02 (18.55), respectively. Clearly cycl[3,2,2] azine (2) lies out of

99

position; no satisfactory reason can be given except to note that, in the strained system, a weakening of the u-bonding might possibly be offset by stronger n-interactions. Finally, we note that the ESR hyperfine splittings (In are nearly equal in the cation of 1 and the anion of 2 at corresponding positions in the 6-membered rings of 1 and 2 [34-361. This is consistent with the close similarities of the HOMO (16) for 1, and the LUMO (21) for 2. CONCLUSIONS

The present calculations suggest that a classical alternating geometry like 13 is probably a better representation than a benzenoid regular one, but the energy difference is sufficiently small that this is still not certain. From the total energies obtamed, the resonance energies for both the cycl[ 3,3,3] azine (1) and cyc1[3,2,2] azine (2) are high, and the difference in stability probably arises from the difference in binding energy of the HOMO rather than to an inherent instability of the [3,3,3] system in total. Thus the very low W’s of 1 and 4 suggest that oxygen is reduced to the radical anion (O,l) by 1; the HOMO of la 1” symmetry is only weakly bonding from its very nature (16), and is not occupied in the closely related, (unknown) boron analog (20) or the 1,5,9-cyclododecatriene-3,7,11-triyne (8) of higher first IP. Both the present calculations and the ESR spectra support the assignment of an orbital of this type to the LUMO in cycl[3,2,2] azine.

TABLE Selected

4 ESR hyperfine

coupling

[3,3,3] Position 1,3,4,6,7,9 2, 5, 8 1,3,4,6,7,9 2, 5, 8 3a, 6a, 9a N [3.2,2} -Anion Position aH 6-Ara[3.2,2]-Anion Position aH

-Cationb

constants

for the cyc1[3,3,3]-

and [3,2,2]azine

[3,3,3]-Anionb

Phenalenyl

(-) 6.45 (+)I78

oH (+)0.05 (-)4.84

=H

UC

ac

*H

(+)9.69 (--)8.71 (-)7.73

radical

-6.29 +1.81

UC

(-)4.47 (+)6.56 (+)I62

aN

ionsa

+9.66 -7.84 -7.84

aN

(+)1.29

(+)6.54

1,4 1.13

2,3 5.34

5.7 6.02

6 1.20

aN

N 0.60

1, 4 1.07

2,3 5.92

5,7 6.22

aN

N-6 2.03

N-8 0.55

aData taken from refs. 20-22.

bSign as required by earlier theoretical

studies [20,

221.

100

ACKNOWLEDGEMENTS We thank the Science hesearch Council for a Scholarship (to J. D. N.) and a Fellowship (to R. W. M.) and the Salters’ Company for a Scholarship (to R. E.). REFERENCES 1 2 3 4a b C

5a b 6a b C d 7

8a b C

d f’ gg 10a b 11 12a b C

d 13 14 15 16 17 18

A. Streitwieser, Molecular Orbital Theory for Organic Chemists, Wiley, New York, 1961. J. Kruszewski and T. M. Krygowski, Can. J. Chem., 53 (1975) 945; I. Gutmen and N. Trinajstic, Can. J. Chem., 54 (1976) 1789 and refs. cited therein. F. Sondheimer, Act. Chem. Res., 5 (1972) 81. D. Farqubar and D. Leaver, Chem. Commun., (1969) 24. M. A. Jessep and D. Leaver, Chem. Commun., (1970) 790. D. Farquhar, T. T. Gough and D. Leaver, J. Chem. Sot. Perkin Trans. 1, (1976) 341. R. J. Windgassen, W. H Saunders and V. Boekelheide, J. Am. Chem. Sot., 81 (1959) 1469. A. Galbraith, T. Small, R. A Barnes and V. Boekelheide. J. Am. Chem. Sot., 83 (1961) 453. M. H. Palmer, A. J. Gaskell and R. H. Findlay, J. Cbem. Sot. Perkin Trans. 2, (1974) 778. M. H. Palmer and R. H. Findlay, J. Chem. Sot. Perkin Trans. 2, (1974) 1885; J. Mol. Strllct., 37 (1977) 229. M. H. Palmer and S. M. F. Kennedy, J. Chem. Sot. Perkin Trans. 2, (1974) 1893; (1976) 81. M. H Palmer, R. H. Findiay. J. N. A Ridyard, A. Barrie and P. Swift, J. Mol. Struct., 39 (1977) 189. J. Wirz, I-IeIv. Cbim. Act-a, 59 (1976) 1647. R. D. Brown and B. A W. Coiler. Mol. Phys., 2 (1959) 158. E. Kleinpeter, R. Borsdorf, G. Fischer and H. J. Hofmann, J. Prakt. Chem., 314 (1972) 515. J. Feitelson, J. Chem. Phys., 43 (1966) 2511. J. A. Berson, E. M. Evleth and S. L. Manatt, J. Am. Chem. Sot., 87 (1965) 2901. W. W. Paudier and H L. Blewitt, Tetrahedron, 21(1965) 353. W. W. Paudler and J. E. Kuder, J. Heterocycl. Chem.. 3 (1966) 33. P. Caries and A JuIg, J. Chim. Phys.Physicochirn. Biol., 65 (1968) 2030. B. A. Hess, L. J. Schaad and C. W. Holyoke, Tetrahedron, 28 (1972) 3635. M. J. S. Dewar and N. Trinajstic, J. Chem. Sot. A, (1969) 1754. M. J. S. I&war and N. Trinajstic, Theor. Cbim. Acta, 17 (1970) 235. W. W. Paudier, R. A. Van Dahm and Y. N. Park, J. Heterocycl. Chem., 9 (1972) 81. V. Galssso, Gazz. Chim. Ital., 99 (1969) 1078. V. Galasso and G. De Aiti, Tetrahedron, 25 (1969) 2259. A. Gamba and G,Favini, Gazz. Cbim. Ital., 98 (1968) 167. C. Aussems, S. Jaspers, G. Iaroy and F. van Remoortere, Bull. Sot. Chim. Belg., 78 (1969) 479. R. J. Pugmire, M. J. Robins, D. M. Grant and R. K. Robins, J. Am. Chem. Sot.. 93 (1971) 1887. W. W. Paudier and J. N. C&man, J. Heterocycl. Chem.. 10 (1973) 499. S. Becker, D. Heidrich, C. Weiss, J. Pancir and R. Zahradnik, Z. Chem., 14 (1974) 440. V. Boekeiheide and R J. Windgasven, J. Am. Chem. 8oc.. 81(1959) 1456. J. D. Bower and G. R Ramage, J. Chem. Sot., (1955) 2834. A W. Hanson, Acta Crystallogr.. 14 (1961) 124.

101 19 20 21 22

23 24

25 26 27 28

29 30 31 32 33 34 35 36

E. Vogel, H. Konigshofer, J. Wassen, K. Mullen and J. F. M. Oth, Angew. Cbem. lnt. Ed. Engl., 13 (1974) 732. T. Nakajima, A. Toyota and H. Yamaguchi, Fortschr. Chem. Forsch., 16 (1971) 103. H. J. Lindner and B. Kitschke, Angew. Chem. Int. Ed. Engl., 15 (1976) 106. See also D. Bertelii, Jerusalem Symp. Quantum Chem. Biol., 3 (1971) 326. M. H. Palmer, R. H. Findlay, A. J. Gaskeii, W. Moyes, S. M. F. Kennedy and J. Nisbet. Quantum Chemistry - the Current State of the Art, Science Research Council, April 1974, (published 1975) p. 229. E. Heilbronner, J. P Maier and E. Haselbach, Physical Methods in Heterocyclic Chemistry, Vol. 6, Academic Press, 1974, p. 1. U. Gel&, C. J. Allan, G. Johansson, H. Siegbahn, D. A. Allison and K. Siegbahn, Phys. Ser., 3 (1971) 237. B. 0. Jonsson and E. Lindholm, Int. J. Mass. Spectrom. Ion. Phys., 3 (1969) 365. C. Fridh, L. Asbrink, B. 0. Jonsson and E. Lindholm, Int. J. Mass. Spectrom. Ion Phys., (a) 8 (1972) 215; (b) 8 (1972) 101. M. H. Palmer, R. H. Findlay and R. Egdell, J. Mol. Struct., 40 (1977) 191. M. H. Palmer, R. H. Findiay and A. J. Gaskell, J. Chem. Sot. Perkin Trans. 2, (1974) 420. M. H. Palmer, The Structure and Reactions of Heterocyclic Compounds, Edward Arnold, 1967, p. 307. J. Burns and E. Yelke, Rev. Sci. Instrum., 40 (1969) 1236. R. Boschi, W. Schmidt and J.-C. Gfeller, Tetrahedron L&t., (1972) 4107. C. Batich, E. Heilbronner and E Vogel, Helv. Chim. Acta, 57 (1974) 2288. V. Boekelheide, J. N. Murrell and W. Schmidt, Tetrahedron Lett., (1972) 575. F. Gerson, J. Jachimowicz. B. Kowert and D. Leaver, Helv. Chim. Acta. 56 (1973) 258. F. Gerson and J. D. W. van Voorst, Helv. Chim. Acta, 46 (1963) 2257. F. Gerson, J. Jachimowicz and D. Leaver, J. Am. Chem. Sot., 95 (1973) 6702.