i=3! u 7...! h .5»... . Tn... .. .3 . . , i: t. f?! .52! .IL: A...) .l l S ll. .3! 39"»: .v .1. . . 23:: in . :5 ‘ul.&. I ”HE-E £Z.I>p.\ll.trnr N f ;: . a E in: .. r. vita—”tr: 9!... 1.. IA... tr .1) ESSLS: .. ‘.. 3: Fit]. . fillvtli 51:2... .5. l 19"..- .(|. AR. 7. 1.. . vfu~uibg pl! . r. i... t... 35.5.». .3. (Virgin! ,.r!rr#.r.. .5. 1.1.5:»: rt...?§.tlao.t .11. l I?! r72. flit. f:..r..u.f. I‘ll 3r»: L. .i , u (r: (.75. 3 .f .31 ...I.olt.r: .5336! .1 I; .551: 2. «it ‘52:»: [Kill- 1‘.‘ {5.1; . .Ar.r1, . 31...: . .. 9...}...7. .1. 5.5.... 1.10: y . . 2- ....ru.‘r\t.:(.. {.1. , . 7 .L .x... ill lllllllllllmIHHII|l||llIlllllHllllHillllllllllllllllllml w 31293 00582 0364 a! 1 LIBRARY Michigan State University This is to certify that the dissertation entitled Experimental and Theoretical Studies on Didehydrobenzenes and Didehydropyridines presented by Hak-Hyun Nam has been accepted towards fulfillment of the requirements for Ph . D degree in Chemistry 4’ Armeszv , Major professor Date April 14, 1989 MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE MSU Is An Affirmative Action/Equal Opportunity institution “A; _ , ,r a, , EXPERIMENTAL AND THEORETICAL STUDIES ON DIDEHYDROBENZENES AND DIDEHYDROPYRIDINES By Hak-Hyun Nan A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Cheniatry 1989 ‘1 {20104 ABSTRACT Experimental and Theoretical Studies on Didehydrobenzenes and Didehydropyridines By Hak-Hyun Nam Experimental and theoretical investigations on the three isomers of didehydrobenzene (BBB) and the six isomers of didehydropyridine (DHP) are reported. While experimental studies were limited to vicinally didehydrogenated benzene (1,2-DHB) and pyridines (2,3- and 3,4-DHP), all possible isomers were subject to theoretical investigation. Infrared spectra of 1,2-DHB were recorded following photolysis (X>210 nm) of matrix isolated phthalic anhydride. Two new vibrational frequencies, at 1355 and 1395 cm’l, which agree well with our normal coordinate analysis, were identified. Successful isolation of 3,4-DHP in Ar or N, matrices via mild photolysis (1)340 nm) of 3,4-pyridine dicarboxylic anhydride (3,4-PDA) provided the first direct evidence of any known didehydroheteroarene. Infrared spectra in the 2000 - 2300 cm'1 region prior to and following controlled photolysis of 3,4-PDA X>340nm l>210nm ( 3,4-PoA-——> 3,4-DHP—-—> c,“2 + HCCCN or C4H2 + HCN ) clearly demonstrate the formation of 3,4-DHP (2085 cm'l) and itS' subsequent decomposition to HCN (2101 cm'l) and diacetylene (2185 cm'l), or cyanoacetylene (2236 cm'l) and acetylene. Ten additional frequencies below 2000 cm“1 were also attributable to 3,4-DHP. The 2,3- isomer could not be isolated under our experimental conditions. The photolysis of 2,3-PDA in N, or Ar matrices leads to rupture of the ring structure, and infrared spectra taken at various time intervals provide no evidence of 2,3-DHP among the photolysis products. Three different levels of ab-initlo calculations (RHF, ROHF and GVB) with a 3-216 basis set were carried out for all DHB and DHP isomers with full geometry optimization at each level. The results for DHBs generally agree with previous GVB calculations by Noell and Newton. In contrast to the semi-empirical calculations, there are three local minima for 1,4-DHB; one diradical and two bicyclic structures. The bicyclic 1,3-DHB is located on an inflection point of the GVB potential curve. The ground states of all DHPs except the 2,6- isomer are predicted to be singlets. The stabilities decrease in the order: 3,4-(S)>2,3-(S)~2,4-(S)>2,6-(T) ~2,5-(S)~3,5—(S)~2,6-(S), where (S) and (T) represent singlets and triplets, respectively. The DHP structures are discussed in terms of the effect of electron correlation between the two radical centers and the interaction between the nitrogen lone pair and the two radical centers. To My Parents and Soak ii ACKNOWLEDGMENTS I am grateful to Dr. G. E. Leroi for his support and thoughtful advice throughout my graduate work. Whenever I ran into difficulties, both in research and in personal matters, he was always willing to help me out with very generous understanding. Without the help of Dr. J. F. Harrison, it would have been impossible to carry out the theoretical calculations presented in this dissertation. I also would like to thank Dr. A. Popov and Dr. J. Allison: they provided me an opportunity to develop my own interest in solution chemistry, signal processing and new matrix isolation techniques. Mr. R. Haas, electronic technician, was of tremendous help whenever the FTIR spectrometer was in trouble. Drs. J. Lopez-Garriga and L.-L. Soong were always willing to share their knowledge in science. Mr. M. Sabo spent his valuable time with me to fix the FTIR spectrometer. I am also thankful to the members of the Molecular Spectroscopy group and Theoretical Chemistry Group for their friendship and iii cooperation. Finally, I wish to thank my wife, Sock, for her patience, encouragement and love throughout my graduate career. It would have been impossible to complete this dissertation without her devotion. Financial assistance from MSU and NSF, and fellowships from SOHIO, MOBAY and H. T. Graham are gratefully acknowledged. iv LIST OF TABLES..... TABLE OF CONTENTS Page I I I I I I I I I I I I I I I I I I I I I I I I I I I I I l I I I I I I I .Viii LIST OF FIGURES I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I x CHAPTER 1 Nomenclature of Didehydrogenated Ring Systems and Methods for their Experimental and Theoretical Studies Introduction............................. Formulas and Nomenclature................ Experimental Method: Matrix Isolation Spectroscopy (MIS)....................... MIS of Chemical Intermediates............ o Photochemical Products of Matrices Containing Reactive SPeCieS.o......o.-.......o.....o... Infrared MIs-ocot-no.0too'ooooouocanoonootho Identification of Photochemically Generated Reactive Species......................... Experimental Section..................... Theoretical Considerations: Ab-Inltio Calculations............................. Effect of Two Electron Interactions in Didehydrobenzenes and Didehydropyridines. Wave Functions for DHB and DHP........... Qualitative Discussions of Wave Functions for 11 15 19 22 23 23 CHAPTER 2 DHB and DHPIOOOOIODIOOOIOo 0.00000 Quantitative Calculation: Ab-Initio Methods...IIIIIIIIIOIIIIIIIIUOIII Computation..... References....................... Didehydrobenzenes IntroductionIIIOI.IOOIIOIIIIIIIII 1,2-Didehydorbenzene (1,2-DHB)... SCF The Ground Electronic State of 1,2-DHB. The Ionization Energies and The Heat of Formation of 1,2-DHBonI0.00.00.00.00..- Electronic Spectra of 1,2-DHB.......... Vibrational Frequencies of 1,2-DHB..... The CEC Stretching Frequency of 1,2-DHB and Other Angle Strained Cycloalkynes.. The Singlet Geometry of 1,2-DHB.. The Triplet Geometry of 1,2-DHB... 1,3-Didehydrobenzene (1,3—DHB).... Experimental Studies............. Theoretical Studies.............. 1,4-Didehdrobenzene (1,4-DHB).... Experimental Studies............. Theoretical Studies.............. Summary and Conclusions.......... References..............o........ Vi I 28 31 33 37 40 43 43 46 52 53 58 61 65 67 67 68 73 73 74 80 84 CHAPTER 3 Didehydropyridines Introduction................................ 3,4—Didehydropyridine (3,4-DHP)............. Infrared Spectrum........................... Theoretical Calculation..................... 2,3-Didehydropyridine (2,3-DHP)............. Infrared Spectra of The Photolyzed Products of 2,3-PDA.................................. Theoretical Calculation..................... 2,4'DidehYdrOPyridine (2,4‘DHP) v t o c v o o I o o o o o I Theoretical Calculation..................... 2,5-Didehydropyridine (2,5-DHP).. Theoretical Calculation..................... 2,6-Didehydropyridine (2,6-DHP)............. Theoretical Calculation..................... 3,5-Didehydropyridine (3,5-DHP)............. Theoretical Calculation..................... Summary and Conclusion....................... References................................... .107 .107 .114 .119 .119 .122 .122 .127 .127 .129 .129 .131 .134 Table 2.1 Page 45 47 48 51 57 64 66 71 72 77 78 79 81 96 100 103 LIST OF TABLES Singlet-Triplet energy separation of 1,2-DHB Experimental IEs of 1,2—DHB Calculated IEs of 1,2-DHB The enthalpy of formation of 1,2-DHB (298K) Infrared spectrum (cm-1) of 1,2-DHB in matrices Geometrical parameters of singlet 1,2—DHB Geometrical parameters of triplet 1,2-DHB Geometrical parameters of 1,3-DHB Singlet-triplet separation of 1,3-DHB Geometrical parameters of 1,4-DHB The overlap population for 1,4~DHB Singlet-triplet separation of 1,4-DHB Relative energies (kcal/mole) of DHBs Infrared bands (cm‘l) resulting from photolysis of 3,4—PDA in an N2 matrix at 13 K Geometrical parameters of 3,4—DHP Geometrical parameters of pyridine Vibrational frequencies of pyridine viii 104 106 109 115 117 120 123 128 130 132 Comparison of calculated (RHF/3-21G) and experimental frequencies (cm'l) for 3,4—DHP Calculated vibrational frequencies (cm-1) of 3,4-DHP Infrared bands (cm-1) resulting from photolysis of 2,3-PDA in an N2 matrix at 13K Geometrical parameters of 2,3-DHP CEC bond length (A) and vibrational frequency of vicinally didehydrogenated aromatic systems Geometrical parameters of 2,4-DHP Geometrical parameters of 2,5-DHP Geometrical parameters of 2,6-DHP Geometrical parameters of 3,5-DHP Relative AET (kcal/mole) values of DHPs with respect to GVB energy of 3,4-DHP ix Figure Page 10 13 18 24 35 60 97 LIST OF FIGURES Typical vacuum system for matrix isolation Closed cycle cryostat Typical infrared matrix isolation experiment Photolysis apparatus for matrix studies Typical characterization route for matrix isolated species Possible configurations arising from the occupancy of two MOs by two electrons Various levels of ab-inltio calculations CEC stretching frequency of cis-bent acetylene calculated at the RHF/3—21G and GVB/3—21G level IR spectra of 3,4—PDA and its photolyzed products in the 2050 - 2300 cm‘1 region in an N2 matrix at 13 K: (a) 3,4—PDA; (b) after 100 min photolysis through water and X>340 nm filter (The peak at 2281 cm'1 is due to 13002); (c) following additional 30 min photolysis with x>210nm. 98 101 108 112 126 Difference spectrum of 3,4-PDA before and ' indicates a band after mild photolysis. to diacetylene. Total atomic charge of 3,4-DHP and 1,2—DHB calculated at the GVB/S—ZIG level. Three types of growth-curve when 2,3-PDA is irradiated with k>300nm light. IR spectra of the photolyzed products of 2,3-PDA in the 2050 - 2300 cm‘1 region in an N2 matrix at 13 K: (a) 10 hour photolysis through water filter and X>340 nm filter; (b) 2 hour photolysis after (a) with X>300 nm filter; (0) 1.5 hour photolysis after (b) with X>210 nm filter. Approximate MOs of 2,5—DHP and 1,4-DHB xi CHAPTER 1 Nomenclature of Didehydrogenated Ring Systems and Methods for Their Experimental and Theoretical Studies Introduction Nomenclature should not only unambiguously define the function and structural formula of a molecule, but also be used consistently. Various names referring to didehydrogenated aromatic or heteroaromatic systems exist in the literature1‘9. However, they are sometimes inadequate to depict their structural formula or inexplicit in characterizing their functions'. Thus it is necessary to establish a consistent nomenclature for didehydrogenated ring systems prior to the main discussion. In the first section of this chapter, merits and disadvantages of widely—used nomenclature for didehydrogenated ring systems will be compared and the nomenclature which will be used in this dissertation will be selected. Since didehydrogenated rings are highly reactive and difficult to isolate for spectroscopic investigationl‘s, few experimental techniques can be used to determine their structure. Matrix isolation, which traps a desired unstable species in a solid inert gas via pyrolysis, electric discharge or photolysis of a precursor prior to or after codeposition with an inert gas for subsequent spectroscopic investigation, is one of the most effective experimental methodsll'15. The second section is devoted to explaining the basic notions of matrix isolation spectroscopy and includes the experimental details which are only relevant to this thesis. While experimental studies in this work are limited to vicinally didehydrogenated benzene and pyridineszl, all their possible isomers are subject to theoretical investigation. These theoretical results provide valuable information regarding the structures and the detailed course of chemical reactions of those reactive intermediates where experimental evidence does not exist or is inconclusivezs. However, depending on the method of calculation, results are of large variance and in some cases they lead to an entirely erroneous conclusionzs. The third section will discuss the reliability and limitations of various methods of ab-initlo molecular orbital calculations for didehydrogenated ring systems. 1.1. Formulas and nomenclature In order to determine the nomenclature of didehydrogenated ring intermediates, two factors may have to be considered: (1) the representative canonical formula of a molecule and (2) the relation to its parent compound. For example, removal of two adjacent hydrogen atoms from benzene may lead to several canonical formulas of 1a - 1g5. Certainly, all of these structures except triplet 1e are contributors to the o 4. 1 ° ' 1 1c 1d 19 However, the name of this compound is heavily dependent on same resonance hybrid. 1. 1b 19 1f which canonical formula is regarded as a representative one. While this compound is conventionally called bonzyne1»4, emphasizing that the parent compound is benzene and the two dehydrogenated centers may form a partial triple bond like 1a, Chemical Abstracts names it as 1,3-cyclohexadiene-5-yne, taking If as its canonical formula. If 1g is selected as a canonical formula, the name would be 1.2.3.5-cyclohexa- tetraene. If any formula from 1b - 1e is chosen, we might have to use other names to represent this compound. Thus naming this compound by a certain canonical formula cannot provide a consistent nomenclature without firm theoretical and experimental justification. Even with this complication, resonance structure 1a or Kékulé structure 1f are the most popular representations of vicinally didehydrogenated benzenez's, and other aromatic and heteroaromatic systems are also depicted in a similar way5‘9. To denote a partial triple bond in these systems, the suffix ’-ene’ of the parent arene or heteroarene is replaced with ’-yne’, namely aryne1 and heteroaryne7. For example, benzene becomes benzyne, pyridine becomes pyrldyne and naphthalene becomes nephthalyne, etc. The suffix ’-yne’ of aryne or heteroaryne is meaningful only when two dehydrogenated centers are vicinal, whereas in much of literature the terms ’aryne’ or ’heteroaryne’ are equivalently used to imply all didehydrogenated arenes or heteroarenes5. For example, 2a or 2b is called 1,3-benzyne or meta-benzyne, 3a or 3b is called 1,4-benzyne or para-benzyne, etc. However, extension of the terms ’aryne’ or ’heteroaryne’ to the non-adjacent didehydrogenated compounds, such as 2a-3b, inadequately relates the nomenclature to the structure of 2a 2b 3a 36 those compounds. The most consistent nomenclature for dehydrogenated ring systems seems to be prefix-didehydro-parent, where the prefix indicates the sites of dehydrogenation5o9. For example, instead of 1,3— benzyne it becomes 1,3-d1dehydrobenzene, 3,4-pyr1dyne becomes 3,4-d1dehydropyr1d1ne, etc. This nomenclature not only eliminates the unwarranted structural implication associated with the ’-yne’ suffix, but also includes all possible canonical resonance formulas. A further advantage of ’—d1dehydro’ nomenclature is that the name clearly indicates the bidentate reactivity of these intermediates. In this dissertation, therefore, the most consistent ’-d1dehydro’ nomenclature will be used to denote didehydrogenated arenes or heteroarenes. For the sake of convenience, resonance or Kékulé canonical structures will represent any didehydroarenes or didehydroheteroarenes without any implication as to bonding, geometry, charge distribution, or electron multiplicity. 1.2. Experimental Method: Matrix Isolation Spectroscopy (HIS) Since the inception of MIS by Pimentel in 1954 to study free radicalsl0, the technique has been widely applied to many areas of chemistry11'15. They include: intermolecular forces, conformational studies, transition metal atom chemistry in relation to its catalytic ability, molecular complexes and free radicals and unstable molecules. Spectroscopic methods used for matrix isolated samples now cover most areas of spectroscopy, namely IR, Raman, ESR, UV, magnetic circular dichroism, Massbauer, luminescence, etc. Among the various applications of M18 listed above, vibrational spectroscopic investigation of chemical intermediates in a matrix will be discussed in this section, with focus on their photochemical generation, analysis of observed frequencies and the determination of their possible structures. 1.2.1. HIS of Chemical Intermedggtes In order to spectroscopically investigate an unstable chemical intermediate, we need to ensure at least two important experimental conditions: (1) the desired species should survive long enough for spectroscopic measurement, and (2) the desired species must interact minimally with its environment. MIS has been demonstrated to be one of the most effective methods to study an unstable species. A very low concentration of a reactive species or its precursor (guest) is frozen with an excess volume of an inert gas (host) on a very cold (4 ~ 20 K) spectroscopic window (e.g. alkali halide for IR, quartz for UV, sapphire rod for ESR, etc.) which maintains its low temperature by thermal contact with the cryogenic device. High host—to-guest ratios (100 ” 10000) are required to isolate the highly reactive intermediates, and low temperatures are necessary to prepare a rigid matrix to prohibit their intermolecular reactions and to stabilize them for routine spectroscopic measurement. There are three basic units of instrumentation required to perform an MIS experiment: (1) vacuum system, (2) cryogenic device and (3) spectrometer. The use of a good vacuum system (10"5 ~ 10'8 Torr) is required to insulate the cryogenic temperature, to protect the matrix from contamination, and to prepare the matrix gas of high purity (Figure 1.1). Three types of cryogenic devices - double dewar cryostat, Joule—Thomson open cycle cryostat and closed cycle cryostat are widely used to maintain the cold temperature of the deposition window, through thermal contact. For most MIS experiments, a closed cycle cryostat is the most popular device because it can afford a wide range of temperature (8 — 300K) without consuming expensive liquid refrigerants (Figure 1.2). Figure 1.3 shows the schematics of an MIS experiment. .nxoooaouo 32.2. S .0 v.3 < 03.00330 c0033 0.3—2, 3) “cos-o coda-Muco— Nu “09:3 03:38.35 3 rouge-I on "39!. 3:9: .3 «so a 8...: non 5.5-! 0 2:2. sou—Tn s. “2;; 0:82. o “0:: 01's m “in... 73:20! on V 30000:". cu a mix. cannot—p ~ “55a queue-cool on a coda-3— x‘gaal tom Ivonne I392, ~33».— .uJ .3... eqxé L .w. .u. a. .n. . ‘I a \I». d F! ‘ V Nu «munchto o—uxo pogo—o N.~ .o—a macaw 302a 03333. coupe—sax 3.01.: h—dflbm seven! toqooo «no sagas: toned o>~o> sweetga poo— :aaoooet 03h I II III. s on _ / _/ // cauuuum anon sweets» ,LIIII 309 at; use 9.5.... on: guano: aoaooccoo some": oaaamoga o: oqaoomOLuuoam ecuunuo uaoo 10 acgugoaxo coda-Jo: 3.5!: pot-Lt: anode»... .m4 .3... ante scootvue 3:0: 5 a: can - - , V<.~.~A lubawuflhn L — V...” .. ,. 93026 7 ’5’ unauoxto . woo 0: . ® ® on: 0355 O «.300»: 3339.50 0: on“? c _ . _ a A. : _ Ann § 3809 an no stuns. . . .- oca_ 590303 Ma 35:25 Loin—£0 admin 0 . .m tops! panacea—u ”no u_ou 11 1.2.2 Photochemical Production of Matrices Containing Reactive Species Commonly used techniques for generation of the reactive species are pyrolysis, microwave discharge or mass spectrometric decomposition of the precursor prior to codeposition with inert hosts, or photolysis of relatively stable precursors after they are isolated at low temperature11:12. In this research we have employed the photolytic method because generation of the transients can be most easily controlled, and the extent of the reaction in a matrix can be most easily followed, with this technique. To prepare the matrix isolated precursor, various sampling techniques can be used. Gaseous molecules may be premixed in the desired ratio with an inert gas in a vacuum system. This mixture is then sprayed via a needle valve onto the cold deposition window. Stable liquid or solid samples that can vaporize without decomposition are supplied to the matrix using suitable heating units. This can be done by heating the material to an appropriate temperature with some type of furnace. The heated material is then swept by an inert carrier gas to the cold deposition window through a single nozzle. Alternatively, two jets consisting of the sample vapor and an inert gas deposit the matrix simultaneously. Organic substances can be evaporated in a simple glass tube 12 with a heating wire; metals and other inorganic substances can be heated by the use of a Knudsen cell or high-powered NszAG laserlz’ls. The photolysis of a precursor can only be efficiently carried out using radiation that is strongly absorbed by the molecule, and that has sufficient energy to break chemical bonds. Except for a few colored precursors, in general ultraviolet radiation is required. As a source of ultraviolet radiation, a high pressure mercury or mercury-xenon lamp with various optical filters is most commonly used. Fig. 1.4 shows the typical photolysis apparatus for matrix studiesll. The photolytic decomposition of a precursor in general produces two or more fragments, and one of them may be the desired product. Identification of the desired species, - however, is often intricate because of the possible isomerization of a precursor or the recombination of fragments which are essentially in contact in the same trapping site11. Careful selection of a precursor, which might produce small atom fragments which may diffuse out of the matrix cage, or which produces reactive fragments but leaving groups which are chemically inert to the desired species, can minimize the complication of recombination reaction. As an example, photolysis of diazoacetonitrile in an N2 matrix produces HCCN free radical and N2 which can l3 «ompaua suguol sow «anaconda mama—cuoca .¢.fi .ouu repeat mmo v—oo screws Nutoau tonnage oco~ o—aiuw 952.030 5 touch Nate: sesame - - Loa~uw caduceuwot co>o to ~oo~wao Logo: -- --- -- -- «Lassa .An "|'r-"-'-' "I'v‘I touduw gowns uo~c~ new egos“ peas 030 team) use OX\o: Nagcac poo: acumo so o—aEm vouaaoaov 14 behave as a new part of the matrix cage (HCNNCN --uv--> HCCN +N2)1°. 1.2.3. Infrared MIS Although many kinds of spectroscopic methods have been applied to matrix isolated species, the majority of investigations continue to rely on infrared absorption as the principal means of detection and characterization14. The absence of hot bands and rotational structure in the vibrational spectrum of most matrix isolated molecules due to the low temperature and rigid environment greatly sharpens the fundamental frequencies and allows vibrational assignments to be made with greater confidence and accuracy13. Even though host-guest interaction causes some shift of fundamental frequencies from those of the gas phase, they are usually in a tolerable range (normally within 1 percent). Uneven multiple trapping sites in a matrix will generally lead to splitting of the fundamentals11. Aggregation of the guest to multimers also complicates the matrix spectra. These problems, however, can be identified by carefully annealing the matrix (raising the temperature within 30% to 50% of the host’s melting point), changing the deposition rates, or varying the host-to-guest ratios of a matrix. Moreover, the recent development of Fourier transform interferometers greatly extends the detection limit for low 15 concentration matrix isolated samples14. 1.2.4 Identification of Photochemlgally Generated anctive species When a matrix isolated precursor is photolyzed with an appropriate wavelength, infrared bands due to the precursor begin to decrease in intensity and new infrared bands due to product species begin to grow, as the extent of reaction increases. The assignment of new bands to a particular reactive species then can be made more or less empirically using the methods described below. (1) Examine the intensity of new bands versus duration of photolytic radiation. This process is particularly important when photolysis of a precursor produces more than one product. If this curve-of—growth analysis is done carefully it is possible to distinguish peaks that vary in the same way from all others. (2) Warm the matrix above 50% of its melting point (e.g. Ar: 42K, N2:32K, etc.) to allow diffusion-controlled reaction. Above about 50% of the melting point the matrix loses its rigidity and diffusion of trapped species will occur. Infrared bands of reactive species, if formed, will then disappear during this diffusion operation, 16 which is followed by recooling of the annealed matrix. (3) Examine the photochemical behavior of the products upon continued irradiation with various wavelengths. Isomerization reactions or further photochemical decomposition of the products can be induced by this process. This photochemical experiment not only helps to assign the new infrared bands to a particular species, but also elucidates the detailed photochemical mechanism of the products. (4) Use isotopically substituted precursors to assign the observed vibrational frequencies to the specific normal modes of the product. (5) Design another experiment that can produce the same reactive species with a different precursor. This may not always be possible, but is necessary to establish irrefutable evidence for the proposed reactive species. Once we classify a group of frequencies as belonging to a certain reactive species based on the above procedures, a number of possible structures may be postulated to explain the observed vibrational frequencies. The most likely structure of the target compound is then deduced from the group frequency analysis, normal coordinate analysis and 17 quantum chemical calculations. Figure 1.5 schematically a typical characterization route for isolated chemical intermediate. summarizes a matrix 18 mo—ooam peas—0am nugaQE tofi enact couuowusouogegu aoouaxh .m.q .o—a hawk—5a uuaoaooA a «gaucou o>es acogowyup can: m—mzuoaoga pose—«coo accuaupsoo couuostta $0 scuvougo> mouooam despu>~pcu on «pace we «cacao—m2 «co—uupcoo somuouomu we 833...; EcuaAa odnoumca so we ecuu~_omu nuquoca , cane—pgoou actouaoa motocou ~oscoz moupaaa «douoau ecuuoacu$coo «vacuum *0 otauuaguw ogauoegvu o—numoom moan» coquugnu> on 53-333 act—£330 ~oo—uogooch socoaootw cacao co—uooog po——0gucoo :o—maquo 19 1.2.5 Experiaental section Our primary vicinally didehydrogenated benzene (1.2-didehydrobenzene) and pyridines (2.3- and 3,4—didehydropyridine)21.22. To these intermediates, anhydrides ( 4, 4 phthalic anhydride their corresponding / \\ N 5 2,3-pyrid1ne dicarboxylic anhydride interest was to isolate and characterize the 5 and 6) were used as precursors. 3,4—pyridine dicarboxylic anhydride Carbon monoxide and carbon dioxide are easily fragmented from a dicarboxylic impact to form didehydrogenated compounds17‘20, and the tendency may also carbon monoxide and dioxide have well characterized bands, the appearance variation of their indicators of photochemical reactions. precursors (Aldrich, anhydride intensities expected upon photolysis. of their infrared bands - 97%) were vacuum sublimation before use. can be further upon pyrolysis or electron used as excellent Commercially obtained generate dicarboxylic same Since both infrared and the purified by 20 Precursors were placed in an L-shaped glass tube sample holder (3/4 inch diameter) wrapped with heating tape; the heating temperature was regulated with a Variac and a thermocouple embedded underneath the heating tape. This glass tube was connected to a specially-designed shutter vacuum flange which can seal the sample holder from the cryostat cell following the deposition (Fig. 1.4). Nitrogen or argon gas (Matheson, 99.999%) was initially collected in a 3 liter bulb which is attached to a vacuum line. The rate of deposition and the amount of host gas deposited were measured by a mercury manometer. In this measurement the ideal gas equation (AP/At)V=(An/At)RT was used, where V is the total volume of that part of vacuum system (V58 of Fig. 1.1) plus the sample outlet line which contains matrix gas during deposition. The rate of sample flow was controlled by means of a micrometer needle valve on the sample outlet line which leads to the cryostat (Air Products 08202 Displex cryostat). The cryostat head consists of two KBr windows for IR measurement and one quartz window perpendicular to the IR windows for photolysis (Fig. 2.3). The vacuum shroud of the cryostat is rotated for an IR measurement after the photolysis through the quartz window. Since precursors were codeposited by sublimation (with nitrogen or argon gas (flow rate 2 mmole/min), exact 21 guest-to-host ratios could not be measured. Photolysis was conducted with a high pressure 200 W Hg/Xe lamp equipped with a water filter and various cutoff filters (Fig. 1.4). Infrared spectra of the precursor and photolyzed products at 13K were recorded with a BOMEM DA3.01 interferometric spectrometer (Fig. 1.3). As summarized in section 1.2.4, growth curves for newly appearing bands, diffusion-controlled reaction studies and photochemical behavior upon‘ continued irradiation with different wavelengths were carefully examined. The results of these experiments and their interpretations will be discussed in the following chapters. 22 1.3. Theoretical Considerations: Ab-Initio Calculations Although we have successfully isolated and characterized some reactive chemical intermediates in inert matrices, the data we can obtain using limited spectroscopic methods may explain only a fraction of their properties. Moreover, many reactive intermediates are too unstable to live sufficiently long for spectroscopic investigation, even with the distinctive advantages of the matrix isolation method. Another difficulty is that it is not always possible to interpret the experimental observations on certain reactive intermediates since they tend to violate the usual concept of chemical bonding. Thus, experimental studies on many chemical intermediates are rather limited or inconclusive. Theoretical methods, such as the various levels of ab-initio molecular orbital calculation utilized in this dissertation, can provide information on various properties of a reactive intermediate beyond the limited scope of experimental observationzs. For instance, geometry, electronic structure and energetics of a reactive intermediate and related transition structures and reaction paths, and molecular properties such as dipole moment, polarizability and vibrational frequencies may all be calculated using these methods. Like all experimental data, however, the reliability of the results from various levels 23 of ab-initio calculation must be cautiously assessed. 1.3.1. Effect of Two Electron Integactions in Didehydro- benzenes and Didehydropyridines Removal of two hydrogen atoms from a parent benzene or pyridine ring systems leaves two electrons and two orbitals coplanar with the parent ring which could conceivably interact in a variety of ways. They may strongly interact to introduce a new bond between the two dehydrogenated carbon centers, or weakly interact through-space or through-bond as either a singlet or a triplet diradical30. Whatever type of interaction is involved between the two odd electrons it will deform and strain the equilibrium ring geometry of the parent compound. The reactivity of a didehydrogenated ring is often measured in terms of this ring strain energy (RSE)27b. Thus it is the interaction between the two odd electrons that primarily accounts for the physical and chemical properties - such as optimum geometry, ground state spin multiplicity and chemical reactivity - of didehydrobenzenes (BBB) and didehydropyridines (DHP). 1.3.2. Have Functions for OMB and DHP If we place all core electrons of DHB and DHP in closed shell configurations while the two odd electrons occupy two 24 M03 which are approximately symmetric and antisymmetric combinations of the two radical lobes n1 and n2 (91 ~11! +112 (1) T2 ~ n, ‘ n2 (2) then there are six possible ways to place the two electrons in the two M08 as shown in Figure 1.6. The relative positions of u, and m, in Figure 1.6 are not intended to imply their actual energy levels. While u, normally has lower energy than $2, owing to positive overlap between two lobe orbitals ( z 0), through-bonding interaction can reverse this order if becomes negative (e.g. 1,4-DHB)3°. |¢2> + + -+— + —_ H- ..,. —+— —+— -+— —+— ++— —— '9152‘5 W139,“ Info,“ |¢13¢2“> W151“) Info,“ Figure 1.6. Possible configurations arising from the occupancy of two "0’s by two electrons. Of the six electronic conformations, proper combination of spatial and spin eigenfunctions give rise to one triplet and three singlet Hartree-Fock (HF) wave functions as follows: 25 fl [{core}¢1¢2(aa)] I3w1> = A [{core}¢1¢2(GB+8G)] (3) A [{core}¢1¢2(88)] 11w2> = a [{core}¢1¢2(aB-Ba)] (4) I163) = A [{core}¢1¢1a8] (5) I1Q4> = a [{core}¢2¢2a8] (6) where A is the antisymmetrizer and the superscripts on W denote the spin multiplicity. Corresponding HF energies of these states are expressed in terms of one-electron energies hi, (=<¢1(1)|h(1)l¢,(1)>), Coulomb repulsion integrals J,J (= (u,(1)¢,(2)|r12-1I¢J(1)¢J(2)>) and exchange integrals Kij (= (WU,(1)Q[1(2)'1P12'1’MWJ(:l)¢1 (2):>)3 3F.1 = Ec + hc11 + h“22 + J12 - K12 (3’) 1E2 = E:c + h‘=11 + 11°22 + J12 4» K12 (4') 1E3 '- EC + 211011 + J11 (5’) 1E4 = Eo + 2h°22 + J22 (6') where closed h°H 5 bn + 2 (2in - Kpi) (7) p and closed C1030d E0 5 2 .2 hpp + Z (2Jpq - qu) (8) p pvq 26 For geometries where |1@3> and I1Q4> are allowed to interact, which is the general case for DHB and DHP, these MOs can be mixed by appropriate linear combinations23’24: llms> Cl|1w3) - 02|1W4> (9) 11w6> 01|1Q3> + c2|1T4> (10) with corresponding energies 1E5 Ec+C12(2h°11 + J11)+C22(2h°22 + J22)-2C1C2K12 (9’) 1E6 Ec+c12(2h°11 + J11)+c22(2h°22 + J22)+2c1c2K12 (10') where c12+c22=1. I1W2> is not incorporated in equations (9) and (10) because it has spin symmetry different from the other two singlets. Equation (9) introduces the correlation between the two electrons (last term in equation (9')) and is often referred to as a two configuration (TC) wave function. This TC wave function can be further factored into 11w5> fl[{core}(c1¢1¢1-c2¢2¢2)«Bl (13) A[{Core}(c11/2¢1+021/2‘P2)(911/2‘91‘C21/2‘P2)(“B'8“)] (13’) which involves two singly occupied, non-orthogonal Mo’s, @1' = (ell/201+c21/2¢2)/(cl+cz)1/2 (14) 27 @2' = (C11/2T1‘C21/2T2)/(C1+Cz)1/2 (15) with overlap integral value S = (cl-c2)/(c1+c2) (16). The M08 in equation (14) and (15) are often referred to as general valence bond (GVB) orbitals25 and equation (13) is their natural orbital expression. Thus, two-term one-pair GVB wave functions and the TC wave function_ are equivalent in their functional form. However, the conceptual basis for TC and GVB wave functions do not coincide, in that the TC wave function is a linear combination of two doubly occupied singlet configurations while the GVB wave function is a product of two localized non-orthogonal MOs: $1' and u;’. From equation (13) written in natural orbital form, we see that another way of obtaining the GVB wave function from the usual HF closed shell wave function i325: ¢1¢1GB ‘ (C1T191' 02¢2¢2)a59 <¢1|¢2> = 0 (17) That is, an electron pair normally described by a HF closed shell orbital is instead described by a geminal expansion consisting of two orthogonal doubly occupied orbitals. This pair correlation description can in fact be extended to an 28 arbitrary length m to include any amount of correlation between the two singlet coupled electronszsz «>wa -° qfl cqwqwan. <¢plq>q> = 8,, (18) If the nitrogen lone pair orbital $3 in DHP plays an important role correlating two singlet coupled orbitals $1 and oz, we may incorporate ¢3 using equation (18), i.e., where c1'2+c2'2+c3'2=1. While c1 and c2 of equation (13) are taken to be positive, coefficients of (18) and (19) can be either positive or negative. Thus, the appropriate wave functions for DHB or DHP are: equation (3) for a triplet, and equations (4), (9) and (10) for three singlet states. In the case of DHP,equation (19) will be further examined in order to account for the role of the nitrogen lone pair orbital in the correlation between the two singlet coupled electrons. 1.3.3, Qualitative Discussion of Nave Functions for OMB and DHP In order to understand the qualitative aspects of DHB or 29 DHP wave functions developed in section 1.3.2, we will consider two restricted cases assuming that the same set of known MOs are used to construct the wave functions given in equation (3) - (10). They are: (1) M03 Q1 and $2 are degenerate with no conceivable interaction between the two radical lobes (i.e., has small values), and (2) M08 $1 and u, are split in energy either by interaction with themselves or with other levels. In the first case, we may predetermine the mixing coefficients c1 and c2 of equation (9) to be equal. Under this condition, the relative energies of wave functions (3), (4), (9) and (10) can be expressed in terms of only Coulomb and exchange integrals: A32, = J,,- K12 (3") A1E2 = J,,+ K12 (4") A1Es = (J11+ J22V2 ' K12: J11" K12 (9") A136 = (J11+ J22V? + K12= J11+ K12 (10") These AE’s then predict that the triplet state is the lowest in energy, obeying Hund’s rule. This is owing to the fact that the Coulomb repulsion integral between two electrons in the same MO is always greater than that between two electrons in different MOs (J,J+|1W4>), will be the 30 highest in energy. However, it is less clear whether (4") is greater or less than (9"). Thus, the lowest singlet state can be specified only after we carry out the calculation for J13 and Kij with known functional form of $1 and $2. It is unlikely that the radical lobes do not interact, as presumed in the previous case. When the two levels are not degenerate, variational optimization of c, and 02 in equation (9’) can bring this singlet energy level down, below that of the triplet. One of the closed shell singlet wave functions (I1w3> or I1W4>) could also be a/ground state if the M03 o, and u, are significantly split in energy. In this case, equation (9) and (10) are not relevant; the mixing coefficient belonging to the higher energy configuration in equation (9’) approaches zero, resulting in a singlet energy of either (5’) or (6'). A sufficient condition for a closed shell singlet ground state is then assured if the low-lying singlet energy (5’) or (6’) is less than triplet energy (3'), i.e., closed hm" h,” > Jn’ 313+ Kij+ 2' {(ZJJp- KJp) - (2J1p- Kip» (20) p where i,j = 1 or 2 and p denotes the Mo’s in core shells. Interpreted physically, equation (20) requires that the difference in one-electron energies be large enough to overcome the greater Coulomb repulsion energy associated with 31 having two electrons in the same MO, rather than in different MOs with parallel spin. In conclusion, if two odd electrons occupy a pair of degenerate or nearly degenerate MOs, the normal consequence is a triplet ground state. If the two levels are significantly split in energy, by interaction with themselves or with other levels, then the possibility of a thermodynamically and kinetically stabilized singlet state arises30. 1.3.4. Quantitative Calculation: aQ-initio SCF Methods In the previous section, we could draw some useful qualitative conclusions from a known set of M08 for all wave functions. However, to obtain these MOs, it is necessary to carry out various level of self-consistent field (SCF) calculations (see Figure 1.7). The restricted Hartree-Fock (RHF) calculation, in which the MOS occupied by electrons of a and 8 spin are restricted to be the same, results in one of the low-lying doubly occupied singlets (equation (5) or (6)) as its solution. The other two singlet states will not be specified by the RHF method. As was discussed in sections 1.3.1 and 1.3.2, a 32 single configuration doubly occupied singlet wave function cannot properly describe the ground singlet state of DHB and DHP. Correlated wave functions for pure spin states require linear combinations of electron configurations. Such multiconfigurational wave functions can be obtained directly by multiconfigurational SCF (MCSCF) calculations, provided one knows for which orbitals electron correlations will be most important. The two configuration wave function, a restricted form of the MCSCF wave function, correctly accounts for the most important correlations between the two odd electrons of DHB and DHP. Thus TCSCF, or equivalently the one-pair GVB calculation, yields an appropriate description of the singlet ground state of DHB and DHP37o23. The energy and wave function of the triplet state can be found by either unrestricted Hartree-Fock (UHF) or restricted open shell Hartree-Fock (ROHF) SCF calculation. While the UHF method, which allows different spatial orbitals for electrons of a and 8 spin, computes a favorable triplet energy in an open shell system, the resultant wave function is mixed with higher spin states (spin contamination). That is, the UHF wave function is not an eigen function of the spin operator. In order to maintain the correct functional form for the triplet state of DHB and DHP (equation (3)), a ROHF 33 calculation is necessary. However, the difference in energy between the two methods is not significant when the spin contamination of the UHF wave function is smallza. The triplets of DHB and DHP can be adequately described with a single configuration ROHF calculation. However, the singlets require correlation of the in-plane electron pair through the TCSCF calculation. Of course, both these calculations will recover only a small fraction of the total electron correlation energy, and yet this singlet TCSCF/triplet ROHF approach should account for the major correlation difference between the singlet and the triplet states’3'27. In order to recover the total electron correlation energy, a configuration interaction (CI) calculation is required. Another popular way of obtaining this correlation energy is the Méller-Plesset (MP) perturbation treatment. However, the CI or MP calculations on DHB or DHP systems could not be done in this dissertation because of the limitations in existing computing facilities. 1.3.5. Computation 34 Three different levels of ab-initio calculations (RHF, ROHF and GVB) have been carried out using the GAUSSIAN86 program29 which is installed on a Micro-VAX II in the Chemistry Department and an IBM 3090 system in the Michigan State University Computer Center. The singlet and triplet energies of all isomers of DHB and DHP have been obtained from fully optimized geometries. In addition, vibrational frequencies have been calculated for vicinally didehydrogenated systems (1,2—DHB, 2,3- and 3,4-DHP), using either the finite difference or the analytic second derivative method at the RHF, ROHF and GVB levels. All calculations are done with the 3-216 basis set. The effect of adding a polarization function to the basis set has not been extensively investigated. In order to examine the characteristics of the strained triple bond in 1,2-DHB and 3,4-DHP, various properties of cis-bent acetylene (O°~80°) have been calculated at the RHF, ROHF and GVB levels. Non-relativistic time independent Schrodinger equation ”I = E! I Born-Oppenheimer approximation fl = H + M i Variational principle el nucl Ee1 = Israels-11 / N's-n z 5. Hartree-Fock SCF method l Roothan—Hall RHF calculation 921-1“ ‘2: PC = 3C8 extend to open shell - ROHF \ I Take correlation energy into account i’21--1°‘ '1", = lexz...xnl , xi = { 9219 HF h x: 3 81x1 K Qt =u£iC”‘iq,‘l l LCAO expansion T K 2 dnl,k 91(“n,k ") ‘PDI gk-l expand Slater type A0 with Gaussian basis set. i Pople-Nesbet UHF calculation 921-13 .21 fie“ = seas.“ FBCB a SCBBB 33111? .- S(S+1) III°> (spin contamination) 36 Eexaot- EMF * Eoorrelation l l l . . I'D‘.‘ HIIIOP-PIOIIOt ; perturbation theory 0. ; if optimized (If) ground state configuration 9“: excited state ' configurations Truncate at 2“ tenn I .2 F l | optimize 0". only optimize D"s and Q"s pull conflgor.tlon Hulti configuration SCF Truncate .g 4*“ e."- interaction (CI) _ ("CSCF) fine include double Include the terms which are expected to- excitatlon terms only be important to overall structure (Brillouin's theorun) CID include single .i.i¢—+.u.n(¢ - p.) and double excitation- ,“m e“ - (cg/1g + oz”’OJ)/(o‘ + 02)”? 6130 '2‘ . (011/2.‘ _ 021/2.J)/(°1 . 02’1/2 Generalized Valence Bond (GVB) theory Special case of MCSCF (2X2 SCF or TCSCF) REFERENCES 10. 11. 37 Roberts, J. D.; Simmons, H. E.; Carlsmith, L. A. J. Am. Chem. Soc. 1953, 75, 3290. (a) Hoffman, R. W. "Dehydrobenzene and Cycloalkynes"; Academic Press, New York 1967.; (b) Hoffman, R. W. In "Chemistry of Acetylenes"; Viehe, H. 6., Ed.; Marcel Deckker, New York 1969. (a) Field, E. K. In "Organic Reactive Intermediates"; McManus, S. P., Ed.; Academic Press, New York 1973; Chapter 7.; (b) Field, E. K.; Meyerson, 8. Adv. Phys. Org. Chem. 1968, 6, 1. (a) Reinecke, M. G. In "Reactive Intermediates, vol 2."; Abramovitch, R. A., Ed.; Plenum Press, New York 1981; Chapter 5.; (b) Reinecke, M. G. Tetrahedron 1982, 38, 427. den Hertog, H. J.; van der Plas, H. C. Advan. Het. Chem. 1965, 4, 121. Kauffmann, Th. Angew. Chem. (Intern. Ed. Engl.) 1965, 4, 543. Kauffmann, Th.; Wirthwein, R. Angew. Chem. (Intern. Ed. Engl.) 1971, 10, 20. van der Plas, H. C.; Roeterdink, F. in "The Chemistry of Functional Groupes, Supplement C: The Chemistry of Triple Bonded Functional Groups" Patai, S.; Rapport, 2., Eds.; John Wiley and Sons, New York 1983. Whittel, E.; Dows, D. A.; Pimentel, G. C. J. Chem. Phys. 1954, 22, 1943. Craddock, S.; Hinchcliff, A. J. "Matrix Isolation"; 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 38 Cambridge University Press, Cambridge 1975. Barnes, A. J.; Orville-Thomas, W. J.; Muller, A.; Gaufres, R. "Matrix Isolation Spectrocopy"; D. Reidl Publishing Co., Dordrecht 1981. Jodl, H. J. In "Vibrational Spectra and Structure, vol. 13" Durig, J. R., Ed.; Elsevier Science Publishers, Amsterdam 1984. Green, D. W.; Reedy, G. T. In "Fourier Transform Infrared Spectrocopy, vol. 3." Ferraro, J. R.; Basile, L. J., Eds.; Academic Press, New York 1982. Knight, L. B. Acc. Chem. Res. 1986, 19, 313. Dendramis, A.; Leroi, G. E. J. Chem. Phys. 1977, 66, 4334. Brown, R. F. C.; Crow, W. D.; Solly, R. K. Chem. Ind. (London) 1966, 343. Cava, M. P.; Mitchel, M. J.; Dejongh, D. C.; van Fossen, R. Y. Tetrahedron Letters, 1966, 2947. Reinecke, M. G.; Newsom, J. G.; Chen, L.-J., J. Am. Chem. Soc. 1981, 103, 2706. Dewar, M. J. S.; Tien, T.P. J. Chem. Soc. Chem. Comm. 1985, 1243. Nam, H.-H.; Leroi, G. E. J. Am. Chem. Soc. 1988, 110, 4096. Nam, H.-H.; Leroi, G. E. J. ”01. Struct. 1987, 157, 301. "Diradicals"; Borden, W. T. Ed.; John Wiley and Sons, 1982.; Chapter 1 and 2. Salem, L.; Rowland, C. Angew. Chem. Int. Ed. Engl. 1972, 11, 92. 25. 26. 27. 28. 29. 30. 39 Bobrowicz, F. W.; Goddard III, W. A. In "Methods of Electronic Structure Theory"; Schaefer III, H. F. Ed.; Plenum Press: New York, 1977.; Chapter 4. Hehre, W. J.; Radom, L.; Schleyer, P. v. R.; Pople, J. A. "Ab-Initio Molecular Orbital Theory"; John Wiley and Sons, 1986. (a) Newton, M. D.; Noell, J. O. J. Am. Chem. Soc. 1979, 101, 51. (b) Newton, M. D. In "Applications of Electronic Structure Theory"; Schaefer III, H. F., Ed.; Plenum Press, New York 1977; Chapter 6. (a) Hiller, I. H.; Vincent, M. A.; Guest, M. F.; von Nissen, W. Chem. Phys. Lett. 1987, 134, 403. (b) Rigby, K.; Hiller, I. R.; Vincent, M. A. J. Chem. Soc. Perkin Trans. II. 1987, 117. Frisch, M. J.; Binkley, J. S.; Schlegel, H. B.; Raghavachari, K.; Melius, C. F.; Martin, R. L.; Stewart, J. J. P.; Bobrowicz, F. W.; Rohlfing, C. M.; Kahn, L. R.; Defrees, D. J.; Seeger, R.; Whiteside R. A.; Foc, D. J.; Fleuder, E. M.; Pople, J. A. "GAUSSIAN 86"; Carnegie-Mellon Quantum Chemistry Publishing Unit, Pittsburg PA, 1984. Hoffman, R. Acc. Chem. Res. 1971, 4, 1. CHAPTER 2 Didehydrobenzenes (DHB) Introduction Since the pioneering experiments of Wittig et al. in the 19403 and of Roberts, Huisgen and their coworkers in the early 19508, a large volume of work has substantiated the intermediacy of 1,2-DHB in many organic reactionsl‘s. This interesting intermediate has attracted widespread attention, as demonstrated by numerous review articleslb'3'7, including a monograph1' and two periodical reportsz. However, although many of its chemical properties are well known, the physical properties of 1,2-DHB, such as its geometry, electronic structure and ionization energy, are still being debated. Related, but much less studied intermediates are the 1,3- and 1,4- isomers. Berry et al. first brought them to attention from their observation of an m/z=76 peak in the time-of-flight mass spectrum when benzenediazonium-S- or -4-carboxylate were decomposed by flash vacuum photolysis‘z. Further experimental evidence of these intermediates has been provided by the groups of Bergman (1,4-DHB)°9-71, Washburn (1,3-DHB)°° and Billups (1,3-DHB)°7. 4O 41 The structures of these DHBs are commonly depicted as follows: (1) (20) (2b) (3a) (3b) Structure (1) represents one possible resonance form of 1,2-DHB. The. 1,3- and 1,4- isomers have two types of structure: diradical (2a) and (3a), and bicyclic (2b) and (3b). Numerous theoretical calculations concerning the geometries, nature of ground electronic states and the relative stabilities of all these DHBs have been published3'23. While the chemistry of the DHBs has been thoroughly reviewed1'7, the physical data, both experimental and theoretical, have not been critically examined to date. Considering the amount of accumulated information regarding 42 the physical properties of these intermediates, it seems appropriate to review them now and to indicate the direction of future research in this area. Included also are the results of our own theoretical calculations (GVB/3-21G level) on all DHB isomers. The C-C bonds of the six-membered ring are specified with the subscripted carbon numbers as shown below (e.g. C1-Cg): where the number 1 always corresponds to the first dehydrocarbon center. 43 2.1. 1,2-Didehydrobenzene (1,2-DHB) 2.1.1. The groung electronic stgte of 1.2-DHB For triplet 1,2-DHB (3B2), a good qualitative description of the electronic structure is proVided by the single configuration (there are 40 electrons in this system and 10a, is the 20th orbital) ~ 2b121a2210a118b21(aa) (1) where the 10a, and 8b2 orbitals approximately correspond to symmetric and antisymmetric combinations of the two in-plane a radical lobes left behind by didehydrogenation of benzene. The triplet wave function can be found by either restricted open-shell or unrestricted Hartree-Fock (ROHF or UHF) methods. Two low energy singlet states are possible9. In the higher energy one the electrons are arranged as in the triplet, but with their spins opposed to form a 182 wave function: ~ 2b121a2210a118b21(a8 - Ba) (2) In the lower energy singlet both electrons are in the same orbital, but a second configuration of this type makes a large contribution. Thus the 1A1 closed-shell SCF function ~ 2b121a2210a,2 (3) is considerably less accurate than the two-configuration SCF 44 (TCSCF) function75 ~ 2b121a22(c1 10a,2 - c2 8b22), (4) where the coefficients c1 and c2 are variationally optimized and their values are often used to describe the relative diradical character of this molecule2°o25’23. Therefore, in order to make a reliable prediction for the 1,2-DHB singlet-triplet energy separation, AE(S-T), the ROHF and TCSCF ( or equivalently one pair GVB) wave functions are required75. Wilhite and Whitten performed rather extensive configuration interaction calculations for the three DHB isomers under the constraint that their carbon skeletons retain a benzene configuration15‘. Even with this severe geometrical limitation they could show that the ground state of 1,2-DHB is a singlet at the two-determinant CI level, while this order is reversed at the single determinant SCF level. Noell and Newton performed GVB calculations for 1,2-DHB with a partially optimized singlet geometry at the RHF/4-31G leve12°. They predicted that the singlet is 28.1 kcal/mole more stable than the triplet. The AE(S-T) values calculated with the fully optimized singlet (GVB/3-21G and GVB/6-31G*) and triplet (ROHF/3-21G and ROHF/6-31G*) geometries are 29.9 (3-21G) and 27.6 kcal/mole (6-31G*), respectively. More extensive CI calculation would increase the AE(S-T) gap furtherls. In general, semi-empirical calculations, such as MINDO/3 and MNDO methods, predict much 45 smaller AE(S-T) values than ab-initio TCSCF/ROHF comparisons (see Table 2.1). Table 2.1 Singlet-triplet energy separation of 1,2-DHB Methods AE(S-T) Comments kcal/mole 2-det. CI15 13.0 benzene geometry many-det. CI15 16.6 GVB/ROHF20 28.1 singlet; RHF/4-3iG partial geometry optimization2° triplet; HINDO/S optimized18 6V6/R0HF** 29.9 6V6/3-21628, ROHF/3-216 6V6/R0HF** 27.6 GVB/6-316*27'23, ROHF/6-316* INDO13 12.5 benzene geometry "mm/318 8.6 MINDO/3 optimized geometries HNDO RHF/ RHF/HE23 3.8 MNDO optimized geometries LNDO/S PERTCI4° 33.2 MNDO optimized geometries23 ** This work Lineberger and coworkers have attempted to measure AE(S-T) from the photoelectron spectrum of the 2E, state of CBH4‘; their result, 37.7 kcal/mole54, is about 30% larger than the predictions from GVB/ROHF calculation820. There have been some attempts to trap the triplet 1,2-DHB in 1,2-cycloaddition reactions with various olefins58‘32. Jones and Levin, however, provided convincing evidence against the 46 possibility of a triplet intermediate from the stereochemical analysis of the [2+2] and [2+4] cycloaddition reactions of 1,2-DHB with appropriately substituted cis- or trans- olefins or with some dieness7. An attempt to generate triplet 1,2-DHB via the photolytic decomposition of triplet phthaloyl peroxide in benzophenone medium failed because of the rapid triplet-singlet interconversion in the intermediate stepsse. From the above evidence, the ground electronic state of 1,2-DHB is undoubtedly a singlet. However, other electronic properties, such as the ionization energy and the relative energy levels of valence shells, are not yet clearly elucidated. 2.1.2. The ionizgtion energy and the hegt of formation of 1,2-DHB The experimental values reported for the ionization energy (IE) of 1,2-DHB are collected in Table 1.2. Three bands, at 9.24, 9.75 and 9.87 eV, were observed in the photoelectron (PE) spectra of the products from the flash vacuum thermolysis of phthalic anhydride or indantrione by Dewar and Tien53. With the help of MNDO calculations, the authors have interpreted these as the first three ionization energies of 1,2-DHB. 47 Table 2.2 Experimental 15s of 1,2-DHB P853 mass48 mass48 9.24 9.75 9.45 9.75 9.87 Before Dewar and Tien’s experiment, the IEs of 1,2-DHB were obtained from mass spectrometry‘se43. Fisher and Lossing pyrolyzed o-diiodobenzene in a reactor coupled to a mass spectrometer and detected 1,2-DHB with a vertical IE of 9.75 eV which is 0.25 eV higher than that of benzene (9.50 eV in their experiment)45. Subsequently, Grfitzmacher and Lohmann reported 9.45 eV as the IE of 1,2-DHB by measuring the appearance energy of the m/z=76 peak resulting from the pyrolysis of bis-2-ioodophenyl43. Rosenstock et al. corrected this value to 8.95 eV based on the fact that the average IE of the linear CSH4 isomers is 9.09:0.02 eV5°; no substantial experimental evidence for this adjustment has been provided. Nevertheless, this correction suggests that the IE of 1,2-DHB might be much lower than the previously reported values from the mass spectrometric measurements. Based on the MNDO calculation, Dewar and Tien assigned the first two bands of the PE spectrum to the a, and b1 orbitals, which are benzene-like n orbitals, and the third to the 48 Table 2.3 Calculated IEs of 1,2-DH8 RHF‘ RHF” OMGF27 2ph-TDA27 c127 moo53 assignment (3-216) (02+?) (02+?) (02+?) (6-316‘) 9.73 9.75 9.55 9.54 9.07 9.33 ie, 9.81 9.77 9.57 9.54 9.06 9.57 261 10.21 10.23 9.77 9.65 9.99 9.93 10.1 ‘ This work in-plane a, orbital which corresponds to the symmetric combination of the two a radical lobes. This assignment, however, has been challenged by Hiller et al. on the basis of ab-initio CI, OVGF (outer valence Green Function method) and extended 2pthDA (two-particle-hole Tamm-Dancoff approximation) calculations”. The first two IEs of 1,2-DHB are predicted to be nearly degenerate in these calculations (see Table 2.3). Since the first IE in the PE spectrum, 9.24 eV, is substantially separated from the other two rather closely spaced IEs, Hiller et al. suggested that the bands at 9.75 and 9.87 eV are attributable to the 2b, and 1a; orbitals of 1,2-DHB and the band at 9.24 eV may originate from some contaminant. Wentrup et al. also seriously questioned the attribution of the PE spectrum to 1,2-DHB41 because other possible intermediates, such as cyclopentadienylideneketene, could have been formed in Dewar and Tien’s experiment53. According to the analysis by Hiller et al., if we discard the first band of the PE spectrum, the first IE of 1,2-DHB is 9.75 eV and this value agrees well with the previous mass 49 spectrometric measurement by Fisher and Lossing. The above MNDO and ab-initio analyses (except the CI calculation) place the in-plane 10a, orbital below the out-of-plane fl-orbitals, 1a, and 2b,. Considering that the vast amount of experimental results have been interpreted on the assumption that the 10s, orbital is the highest occupied MO (HOMO) of 1,2-DHB1'7, the MNDO or ab-initio energy level ordering of the valence shell orbitals disagrees with the conventional description of 1,2-DHB. However, the recent UV/VIS spectrum of 1,2-DHB by Mfinzel and Schweig shows that the in-plane MO 10a, is indeed the HOMO of 1,2-DHB45. This discrepancy arises from the fact the closed-shell SCF procedure overestimates the strength of the 01-02 bond, and thus lowers the energy of the 10a, orbital below those of the out-of-plane n-orbitals. However, inclusion of electron correlation at the CISD level reverses this order27 (see Table 2.3). Thus, a simple application of Koopman’s theorem without considering the effect of electron correlation may not correctly account for the IRS of 1,2-DHB, and it may be premature to conclude that the PE spectrum band at 9.24 eV is not due to 1,2-DHB. On the other hand, Dewar and Tien’s experiments also include some refutable ambiguities as have been pointed out by Wentrup et a141. Thus, in our opinion, the first IE of 1,2-DHB is still not well established. 50 With the known IEs, the heat of formation (AH,) of 1,2-DHB could be derived from the mass spectrometrically determined AH, of the C3H4* ion, which is one of the principal fragments of various phenyl derivatives. Several groups have calculated the AH, of 1,2-DHB from such measurements‘7‘5°»52. Their values range from 100 to 120 kcal/mole, depending on the estimated AH, of the CsH4t ion and the IEs of 1,2-DHB used. These results are collected in Table 2.4. As one may note from the entries, there are large variances in both AH,(CSH‘*) and the IE of 1,2-DHB. Thus, the estimated values of AH,(1,2-DHB) from mass spectrometry should be cited cautiously. Pollack and Hehre employed ion cyclotron resonance (ICR) spectroscopy to determine the AH, of 1,2-DHB51. They measured the proton affinities of the unstable neutral molecule C5D4 (9.5 kcal/mole) by abstracting a deuteron from the phenyl-d5 cation with bases of varying strengths, and combined with these values the AH, of 1,2-DHB (11815 kcal/mole) could be derived from the previously known AH, of Cst+ (270:3 kcal/mole), of H* (367.3 kcal/mole) and the enthalpy of protonation of ammonia (205 kcal/mole). Theoretically calculated enthalpies of formation of 1,2-DHB range from 107 to 138.2 kcal/mole (see Table 2.4). Table 2.4. The enthalpy of formation of 1,2-DHB (298 K) 51 Mass spectrometry source of Cat!" 44, (Cal-14*) IE(C6H‘*) 2H, (1 , 2-DHB) ref . benzene 345 9.75 120 47 electron impact bis-2-iodophenyl 336 9.45 118 48 pyrolysis benzene 318 9.45 100 49 photo ionization benzonitrile 313 8.95 107 50 photo ionization . o-dlbromobenzene 308 8.95 101 52 photo ionization Ion Cyclotron Resonance Spectroscopy (ICR) measurement of electron affinities of C804 118 51 Theoretical calculations HINDO/Z 107 14 RHF/STO-3G 120 17a HINDO/3 118 18 HINDO/3-CI 114 18 MNDO/SCF 138 23 MNDO/3x3CI 126 23 MNDO/UHF 120 23 AH,; kcal/mole, IE; eV 52 However, the quality of each theoretical method may not be judged merely on the basis of numerical coincidence with the experimental AH,(1,2-DHB) values, since in our opinion the reported values have not achieved the desired accuracy. 2.1.3. Electronic spectrg of 1,2-DHB Berry et al. reported the first UV spectrum of gaseous 1,2-DHB42, which showed a broad absorption with the maximum at 243 nm‘ze43. They suggested that the absorption may be due to the transition of an electron either from the in-plane 0 MO (10s,) to the 0* MO (8b,) or from the out-of-plane n MO (1a2) to the 0* MO (8b2). Yonezawa et al. calculated the electronic transition energies of 1,2-DHB in the semi-empirical ZDO approximation; they predicted that the o e 0* transition would lie at longer wavelengths (411 nm) and suggested that the maximum observed by Berry et al. might correspond to the n r fl* transition11. 0n the other hand, Wilhite and Whitten predicted that the 0 ~ 0* transition would occur at shorter wavelengths than the n ~ n* or n ~ 0* transitions, based on ab-initio CI calculationsls. However, this prediction presents some incomprehensible problems because it contradicts the orbital energy levels of 1,2-DHB in their own calculation. 53 Kolc photolyzed the precursor benzocyclobutenedione, isolated in an EPA matrix at 77K, and obtained the UV spectrum of 1,2-DHB in the 270 - 380 nm range44. Only featureless broad absorption, which may correspond to a tail of the previously reported spectrum42, was observed. Recently, Mfinzel and Schweig reported the UV/VIS spectrum of 1,2—DHB in an Ar matrix, obtained via photolysis of either 3-diazobenzofuranone or benzocyclobutenedione‘s. Five bands, at 380, 293, 246, 214 and 199 nm, were attributed to 1,2-DHB, and assigned with the help of LNDO/S PERTCI calculations. The lowest energy band at 380 nm corresponds to a o e a“ transition, and its broad band shape indicates lengthening of the CEC bond following the HOMO-LUMO excitation. The following four bands, corresponding to n d n' transitions, originate from the splitting of the three benzene UV bands (1Lb r 1A, 1Lll r 1A and 1B r 1A) due to the lowered symmetry of 1,2-DHB (02,) from that of benzene (Doh)~ Thus, they concluded that 1,2-DHB can be considered as benzene with an additional 0 bond. 2.1.4. Vibrationgl freguencies of 1,g-D&Q The first IR spectrum of 1,2-DHB was reported by Chapman and coworkers in 197337‘. They photolyzed the precursors 54 phthaloyl peroxide or benzocyclobutenedione, isolated in an Ar matrix at 8K, with UV light to obtain eight bands in the 400 - 1700 cm"1 range which were attributable to 1,2-DHB. One additional acetylenic band at 2085 cm'1 was found two years later. by the same group37b, following short wavelength photolysis of 3-diazobenzofuranone under similar matrix conditions. On the basis of those results, Laing and Berry proposed a cycloalkyne-like structure and a set of force constants for 1,2-DHB35. Badger’s rule and Coulson’s bond-order/bond-length relationship were used to deduce the Cay symmetry geometry, and the complete vibrational spectra of the normal and perdeuterated molecules (C2v symmetry: 9a, + 4a, + 3b, + 8b, normal modes) were calculated with Wilson’s GF matrix method (normal coordinate calculation). Among the significant predictions was the expectation of two 030 ring stretching modes of A1 symmetry in the 2000 - 2500 cm"1 region for each isotopomer. Subsequently, Dunkin and MacDonald obtained improved IR spectra of 1,2-DHB and tetradeuterio-1,2-DHB by UV photolysis of phthalic anhydride and its perdeuterated analog, isolated in N2 matrices at 12 K33. The CBH4 spectrum reported by Chapman et al. was generally confirmed, with the addition of a C-H stretch at 3088 cm"1 and the exception of a band 55 previously observed at 1627 cm'l. Eleven bands were reported for C8D4; agreement between observed and predicted frequencies was quite reasonable below 2000 cm'1, but rather poor in the critical higher wavenumber region. Reevaluation of the 1,2-DHB force field was suggested38 and the normal coordinate analysis by Laing and Berry was criticized in an independent MNDO study by Dewar et alzl. Nam and Leroi indeed found a fundamental mistake in Laing and Berry’s calculation35: the proposed sz planar ring structure was not closed with the given geometrical parameters, which resulted in an incorrect formulation of the G matrix73. This error progressively accumulated throughout their calculation. Hence, Nam and Leroi carried out a new normal coordinate analysis of 1,2-DHB36 with the theoretically calculated structure20 and the additional frequencies of CsD433. From this calculation, the following conclusions were drawn: (1) the bond-length/bond-strength correlation (Badger’s rule) is not applicable to 1,2-DHB, which is explicable in terms of alternating n-electron overlap population around the ring.; (2) only one CEC stretching frequency over 2000 cm'1 is predicted. Two additional IR frequencies (1355, 1395 cm‘l) of 1,2-DHB were reported by Nam and Leroi‘°, which agree well with their previous calculation35. However, attempts to obtain the Raman 56 frequencies of 1,2-DHB have not been successful to date4°. Brown et al. obtained the IR spectra of 1,2-DHB from various precursors39, but no additional peaks were identified. Lineberger and his coworkers reported photoelectron spectra of gaseous C3H4' and C6D4', from which three vibrational intervals in each of the corresponding ground state neutral species were inferred (CBHB: 1860, 1040 and 605 cm'l; CBD4: 1860, 980 and 585 cm‘1)5‘. The highest frequency mode was attributed to the acetylenic CEC bond stretch, which is noticeably smaller than the previously observed values in matrices. They suggested that this mode may have not been detectable in the previous experiments. The recent scaled GVB/3-21G calculation by Rigby et al. obtains 1859 cm‘1 as the CEC stretching frequency of 1,2-DHB. However, as the authors have pointed out, this low frequency is due to an artifact of their scaling method. Thus, there are no available experimental or theoretical results that support a C50 stretching freqency below 2000 cm'l, as suggested by Lineberger et al. Three quantum mechanical calculations of the vibrational spectrum of 1,2-DHB have been published; one MNDO21 and two ab-initio (RHF/3-21G25 and GVB/3-21G23) calculations. The MNDO calculation by Dewar et al. is in poor agreement with the known experimental frequencies, although it correctly 57 Tile 2.5. Infrared spectrum (cm'1) of 1,2-Dl-B ln matrices Wentrup et al. Nam and Dunkin and Chapman et al. Normal mode (phthalic- Leroi ("2,40 MacDonald (61,)” (Ar)37 calculation” anhydride)‘u - 470 472 469 482 (b,) 720 739, 743 735 743 (6,) 815 848 847 848 845 (b1) 1020 1038 1038 1038 1038 (hi) 1045 1055 1055 1053 1052 (.1) 1355 1 1360 (6,) 1395 1391 (a,) 1440 1448 1448 1451 ' 1450 (52) 1588 1588 1807 1589 (a,) t 1627 1657 (6,) 2080 2082 2084 2085 2081 (8,) 3066 3061 (e,) 58 reproduces the CEC calculation by Radom et al. than does the MNDO calculation, mis-assigned to an report4°. The more recent Rigby et al. stretching. infrared scaled GVB/3-21G The scaled RHF/3-21G more closely fits the IR data but two strong IR bands are inactive a2 mode in this calculation by presents no improvement over the RHF results; rather the calculated CiC stretching frequency (1859 cm‘1) is in poorer agreement with the observed frequency (2085 The scaling factors utilized calculated (RHF/3-216) and benzene and were applied to obtained from their GVB/3-ZlG frequencies of 1,2-DHB were coordinate calculation using these Rigby et al. from one molecule to another especially when the RHF scaling factors are cm‘l). were obtained by comparing the experimental frequencies of the force constants of 1,2-DHB calculation. The vibrational then obtained from the normal scaled force constants. pointed out that the transfer of scaling factors may not always be appropriate, applied to the force constants derived from a correlated wave function. 2.1.5. The C§C angle straineg cycloglkynes 1.2-DHB is often referred to as a The physical and stretching frequencies of 1.25958 and other strained cycloalkynel. chemical properties of the angle-strained 59 050 bond in cycloalkynes have been explained with a cis-bent acetylene mode129'34. For example, it was shown that the increased reactivity of 1,2-DHB and other cycloalkynes toward nucleophiles is due to the large decrease in LUMO energy compared to the small increase in HOMO energy upon cis bending of the acetylenic bond29. The expected lowering of LUMO energies with decreasing ring size was indeed observed from the electron transmission spectra of a few selected cycloalkynes30'32. Recent MNDO and MNDOC/BWEN studies on the [2+2] cycloaddition reaction paths of 1,2-DHB with ethylene found no special electronic effects due to the aromatic conjugation of the true cyclic structure in 1,2-DHB relative to the bent acetylene model3‘. In this section, we will also employ the cis-bent acetylene model to explain the CEC stretching frequencies of 1,2-DHB and other cycloalkynes. Figure 2.1 shows the CEO stretching frequencies of cis-bent acetylene (vc.c(a); a denotes the bond angle H-CEC) calculated at the RHF and GVB level with 3-21G basis set. While the GVB Vc-c dramatically shift to lower wavenumber as the angle a decreases (vc.c(180“)-vc.c(120°) ~ 200 cm‘l), the RHF Veac is quite rigid upon bending. The RHF method allows the angle a to decrease less than 60° (i.e. a > 120°) before an imaginary frequency results. On the other hand, cis-bent acetylene suffers bending more resiliently when it is allowed to have diradical character with the GVB method, and all of 60 its normal modes remain real when a > 100°. N _a 8 i l L :3 CC frequency (cm-1) .a L j 17x 1 I 1 I I 180 160 140 120 100 HCC angle (degree) Figure 2.1. 05C stretching frequency of cis-bent acetylene calculated at the RHF/3-21G (x---x) and GVB/3-216 (.——.) level. Meier et al. found a linear correlation between the Vc-c and «(C-CEO) by comparing the experimentally observed "ego of various cycloalkynes with respect to their C-CEC bond angles, and the ”one of 1,2-DHB agrees well with the predicted relationship33. The slope Ave-c/Aa(C-C§C) of this linear correlation (Vcac(180°)’VCsc(130°) a 150 cm'l) is qualitatively concordant with the GVB calculation. The RHF method fails to describe this relationship. These results indicate that as far as the vibrational structure is concerned the C1-02 bond of 1,2-DHB is a true strained triple bond with appreciable diradical character. 61 2.1.6. The singlet geometry of 1,2-DHB Numerous theoretical studies on the geometry of 1,2—DHB have been reported in the last three decadess'28 ; their results are collected in Table 2.6. While all calculations agree that the 01-02 bond is the shortest bond in the ring, the degree of alternation from an equilibrium geometry of benzene in the remaining bonds is still in dispute7. The calculated geometrical parameters are often used to determine the most significant resonance structure of 1,2-DHB, namely the aromatic (1a), cycloalkyne-like (lb) or cumulene-like (1c) structures of 1,2-DHB. GI (13) (1b) (lc) The geometrical parameters from an early n-electron calculation by Coulson8 and a normal coordinate analysis by Laing and Berry35 are consistent with structure (1b). However, the former should be regarded as having at best a crudely qualitative significance as was warned by the author, and the latter does not conform a cyclic structure35. Thus, practically no theoretical calculation provides geometrical 62 parameters suitable to structure (1b)17. The RHF/STO-BG equilibrium geometry is somewhat close to structure (1b), but with a negligible bond length alternation between the Cz-Ca and C3-C4 bonds. The absence of lengthening in the bonds adjacent to C1-02 was explained in terms of an ab-initio hybrid orbital analysis17. 0n the other hand, the total overlap populations do alternate in the sense of structure (1b)17. EHT calculations on 1,2-DHB employing the benzene geometry suggest a sizable contribution from structure (1c), i.e., the total overlap population for the C4-Cs bond is larger than that for the adjacent Ca-C4 and Cs‘Cs bonds, and the n-electron overlap population alternates around the ring in the sense of (1c)9. Haselbach supported this prediction with the MINDO/2 calculationll. However, these calculations are at too rudimentary a level to draw any definitive conclusion. Most other calculations provide geometries which are characterized by short Cl-Cz acetylenic bond lengths (1.22 - 1.30 A) and progressively increasing C2-C3 (1.38 - 1.39 A), C3-C4 (1.39 - 1.42 A) and C4-C5 (1.40 - 1.43 A) bond lengths around the ring17‘23, structure (1a). Except for the 01-02 bond, the bond lengths are quite close to the benzene value (1.40 A). Through a comparison of calculated bond lengths, n-charge transfer, and n-electron overlap population for 63 1,2-DHB, benzene, and acyclic molecules, Bock et al. concluded that 1,2-DHB possesses a highly strained aromatic structure24. The IR spectrum of 1,2-DHB favors the cycloalkyne-like structure with cumulene-like fl-electron system37‘. On the other hand, the UV/VIS spectrum clearly shows that the n e n" transitions are from the benzenoid structure‘s. Recently, Brown et al. reported the microwave transitions of 1,2-DHB from which rotational constants were derived (A: 6990, B: 5707 and C: 3140 MHz)55. The calculated rotational constants using the HF 6-31G* optimized geometry by Bock et al. show a good agreement with the observed valuesss. Combining all currently available experimental and theoretical results, we conclude that the true structure of 1,2-DHB is an aromatic system with an acetylenic C1-Cz bond. There is no substantial experimental or theoretical evidence supporting a cumulene-like structure. A cycloalkyne-like structure seems to be the most significant resonance contribution, as is reflected by the n-electron population17iz°o24, but the absence of bond length alternation around the ring is inconsistent with this structure. At this time, the geometry optimized at the GVB/6-31G* level by Rigby et al. seems to be the best , since among all structures in Table 2.8, its rotational constants 64 .«zeOeuwua .2znoauvna . 02.5223 00 0 .022 0.022 0.402 004.2 500.2 204. 2 040.2 .938 22:02 00.50 0.022 0 .002 0.202 4.022 5.002 050.2 050.2 404.2 000.2 000.2 000.2 4020-0\0>0 40 0.022 0.002 4.002 0.022 0.502 050.2 050.2 024.2 200.2 000.2 000.2 4020u0\h!¢ 40 0.022 0.002 4.002 0.022 0.502 050.2 000.2 024.2 000.2 000 .2 000.2 0200\2 00 0.022 0.402 0.002 0.022 0..502 000.2 000.2 004.2 000.2 000.2 000.2 020.4?!3 50 0.022 0.002 0.202 0.022 0.002 050.2 000.2 004.2 000.2 000.2 200.2 020i0\0>0 00 .00 0.022 0.002 4.002 0 .022 0.502 050.2 000.2 004.2 000 .2 000 . 2 000.2 020-0\..-.& .8 .52 0 .022 4 .502 0 .002 0 .002 ,0 .502 400 .2 400 .2 204 . 2 000 . 2 000 . 2 020 . 2 00.6.5503. 00 004 . 2 024 . 2 000 . 2 200 . 2 223002.. 00 004.2 004.2 200 . 2 040.2 200x0\0006.. 00 004 . 2 004 . 2 200 . 2 040 . 2 mum\0092 m0 004.2 204.2 2cm.2 mm~.2 2oxooz: 20 0.022 0.502 0.002 0.502 0.002 000.2 000.2 404.2 404.2 000.2 000.2 02 02 0.002 0.002 0.502 024.2 524.2 400.2 200.2 2010\0222: 02 .02 0 .002 5 .002 4.502 404.2 024.2 000 . 2 050.2 .0322: 42 0.022 0.002 0.402 0.002 0.002 400.2 400.2 004.2 000.2 440.2 000.2 0\§2: 23:29.8 22 00.2 24.2 00.2 00.2 :25: .000 0 04.2 50.2 44. 2 00.2 cocaue2ei= .0... «0 20 no «a 28 «0 20 40 mm «a 2x 232.. 12.»... l...- 08102 00 accuse-.30 282.5285 .0 .0 02.1» 65 are in best agreement with the experimentally observed values. 2.1.8. The triplet geometry of 112-DHB Since the lowest triplet state of 1,2-DHB is known not to be the ground state and is not likely to be observed experimentally, its structure has attracted less interest. However, our knowledge of 1,2-DHB would not be complete without an adequate description of its triplet state. The geometrical parameters of triplet 1,2-DHB, calculated at the ROHF/3-21G level are listed in Table 2.7 along with the previously reported MINDO/3 results. Both calculations agree that the Cl-Cz bond distance will be considerably elongated in the triplet state, with the consequent loss of triple bond character, but they predict very different bond length alternation for the other three bonds around the ring. While the MINDO/3 geometry is somewhat close to a cumulene-like structure, the ROHF/S-ZIG geometry corresponds to a Kékulé benzene structure. 66 x50! O2Nim 02£h 0.m22 0.0N2 m.2N2 @.w22 0.0N2 Nho.2 250.2 mmm.2 VQ@.2 25m.2 2mm.2 \mIOM 02 0.0N2 0.022 V.NN2 O®O.2 omo.2 604.2 024.2 Vom.2 Ohm.2 m\OoZHZ 305205 .WOM «Q 2& ”8 N5 25 Nm 2” V“ m“ NM 2“ Low-“COG 05.0.2 6.023: to 232528 2823288 .5.~ «3.: 67 2.2. 1.3-Didehydrobenzene (1.3-DHB) 2.2.1. Experimental studies Berry et al. observed a transient species at m/z=76 by means of time-of-flight mass spectrometry following the flash vacuum photolysis of benzenediazonium-3-carboxylate33'. The kinetic UV spectrum of this compound shows a continuous absorption from 220 to 290 nm and perhaps two maxima at 328 and 339 nm. They attributed this spectrum to 1,3-DHB whose structure is represented by either 1,3-diyl (2a) or bicyclo- [3.1.0]hexatriene (2b). Experimental evidence for the intermediacy of both (2a) and (2b) has been obtained. Washburn and his coworkers firmly established the involvement of (2b) as a reaction intermediate in the dehalogenation of exo,exo-4,6-dibromo bicyclo[3.1.0]hexene and related compounds with lithium dimethylamide at -75°C“3. Experimental results by Billups et al. (isolation of the substituted naphthalenes from the dehydrohalogenation of benzo-6,6-dichloro-5-bromobicyclo [3.1.0]hex-2-ene)57, by Bergman et al. (high temperature pyrolysis of disubstituted 1,5-hexadiyne-3-enes)88 and by Gavifia et al. (thermal decomposition of diaryliodonium-3- carboxylate with trapping resins)74 strongly support the intermediacy of (2a) in their reaction mechanisms. 68 These experimental results indicate that the occurrence of (2a) or (2b) depends largely on the method of generation. At low temperature, from a precursor that already contains the 1-3 bridge bond, bicyclic (2b) can be generated and trapped by very reactive reagents before ring opening to diradical. On the other hand, high temperature thermal rearrangement or decomposition produces diradicals (2a), which subsequently exhibit free radical properties. No direct spectroscopic observation has yet been reported for this interesting isomer. 2.2.2. Theoreticgl stugles The results from various theoretical calculations for the 1,3- isomer are as controversial as those from experimental studies. Hoffman et al. pointed out that there exists substantial 1-3 bonding in 1,3-DHB, based on the EHT population analysis9. Hess and Schaad estimated the REPEs (resonance energy per fl-electron) for the bicyclic, diradical and zwitterionic structure of 1,3-DHB, from which they found that the bicyclic structure (2b) (0.0558) retains the most aromatic character (e.g. benzene; 0.0658)°5. Subsequently, based on the MINDO/3-SCF and MINDO/3-CI calculations, Dewar 69 and Li predicted that 1,3-DHB exists as a singlet bicyclic structure with ground state energy comparable to 1,2-DHB13. Washburn also reported that the RHF/STO-3G geometry optimization leads to a closed-shell bicyclic structure (2b)66a . Noell and Newton performed a limited geometry optimization for 1,3-DHB at the GVB level with a 4-31G basis set2°. They concentrated their efforts on the CI-Ca separation while each C-C bond length in the six-membered ring was held fixed at the corresponding MINDO/S-SCF equilibrium value. From this calculation, they concluded that 1,3-DHB exists as a singlet diradical. However, no attempt was made to ascertain if the GVB level would yield a local minimum for a bicyclic geometry. Thiel carried out MNDOC/SCF, MNDOC/ZxZCI and MNDOC/BWENl calculations on all three DHB isomers 22. It was found that the SCF geometry of 1,3-DHB corresponds to a closed-shell bicyclic structure, and the 2x2CI and BWENl to a diradicaloid. Dewar et al. reexamined the structures and energetics of DHBs using MNDO/SCF, MNDO/BXBCI and UMNDO calculationsz3. While the MNDO/SCF method failed to obtain the diradical structure, both structures were MNDO/3x3CI at the local minima with (2a) being more stable than (2b) by 5.7 kcal/mole. Since the ab-lnitio calculations on 1,3-DHB by Noell and 70 Newton were somewhat limited, we have reexamined the structures and energetics of this molecule with the fully optimized geometry at the RHF and GVB levels with a 3-ZlG basis set. The results of these calculations are collected in Table 2.8 along with the previously reported results. Except the C-C bond length alternation around the ring, the results are in good accord with those from the GVB/4-3lG calculations. The bicyclic structure is found to be a local minimum on the RHF potential surface, but is located on an inflection point of the GVB potential curve (ETGVB vs. Rei-ca) at a C1-03 distance of 1.54 A, which lies 33.2 kcal/mole above the equilibrium diradical structure. The MNDOC /2xZCI and /BWEN1 geometry optimization also failed to find the bicyclic minimum for 1,3-DHB. On the other hand, the barrier to conversion of (2b) to (2a) is substantially large (~ 9 kcal/mole) in the MNDO/3x3CI calculation, which indicates that (2b) must exist as a stable reaction intermediate. If the failure to describe bicyclic 1,3-DHB at the GVB/3-ZIG level stems from the overestimation of ring strain energy by the use of a small split basis set, this problem may be corrected by employing a larger basis set including polarization functions. However, as discussed further in the next section, this is not likely to be the case because the GVB/3-21G calculation could successfully optimize the 71 9.30335 02205020.... .5358 5x59... .5 e2. e5 3 55.5.... 52.. 5.2.3556 5:53.... e .55 5.422 5.552 5.452 2.522 555.5 524.2 555.2 555.2 525-455255 5.65 5.55 5.552 5.522 5.552 4.522 555.5 555.2 455.2 555.2 525-5\5:oe 52 5.552 5.522 555.5 555.2 555.2 .5\552.2 555 .55 5.522 5.522 5.552 5.552 555.5 524.2 555.2 555.2 525.e\5>5 5.6: .255 5.522 5.522 4.552 5.552 545.5 555.2 555.2 555.2. 525-5\5>5 ..5.65 5.5» 4.522 5.552 5.552 2.55 425.2 524.2 255.2 555.2 525-55555 ...55 4.222 5.552 5.552 5.55 555.2 555.2 255.2 455.2 55-5»555:5 5.552 555.5 524.2 255.2 255.2 ..55 5.55 555.2 554.2 555.2 555.2 .5.552: 55 5.452 552.5 524.2 555.2 555.2 2zmznxuoaz: 55 5.552 455.5 554.2 555.2 555.2 2555555552: ..55 4.55 555.2. 554.2 555.2 555.2 55555552: 52 5.522 2.522 5.452 2.252 552.5 224.2 555.2 555.2 25-555522: 52 5.522 5.522 4.542 5.25 555.2 524.2 555.2 555.2 555522: .<. .0... vs 5.5 «6 2a 45. 50 5a 2a 2.3 52.5. 10.5 III 9310.2 00 9.35.852. 232328.50 .0.0 28... 72 extremely strained bicyclic ring structure of 1.4-DHB (butalene). Thus, our GVB calculations on 1,3-DHB suggest that the bicyclic form (2b) is not a stable intermediate, but exists as a transient species. Early INDO and ab-initio CI calculations on 1,3-DHB with the equilibrium geometry of benzene predict that the ground electronic state of 1,3-DHB is a triplet. More recent calculations, however, show that the singlet state is 5 - 18 kcal/mole below the triplet state (see Table 2.9). Table 2.9. Singlet-Triplet separation of 1.3-DHB method AE(S-T) comments ref. INDO -26.S benzene geometry 13 2-Det. CI - 1.8 benzene geometry 15 ROHF triplet many-Det. CI - 1.6 benzene geometry 15 MINDO/a 6.2 "INDO/3 singlet 18 GVB/ROHF 17.8 4-316, partially 20 optimized geometry GVB/ROHF 10.5 3-21G This work 73 2.3. 1,4-Dldehydrobenzene (1.4-DHB) 2.3.1. Experimentgl studies Like the 1,2- and 1,3- isomers, Berry et a1. ascribed the species at m/z=76 obtained during the flash vacuum photolysis of benzenediazonium-4-carboxylate to 1,4-DHBB3. Based on the appearance and decay patterns of this species measured by kinetic UV and time-of—flight mass spectrometry, they proposed two most likely structures, (3a) and (3b), for 1 , 4—DHBO Bergman and his coworkers reported that the gas phase thermal equilibration of cis—1,6- and cis-3,4-dideuterio-1,5- hexadiyne-3-ene takes place via (3a)59o71. When these molecules were heated in solution, aromatic products consistent with the trapping of the (3a) were obtained. Recently, Gavifia et al. found evidence of (3a) as a reaction intermediate in the thermal decomposition of diaryliodonium- 4-carboxylate with trapping resins7‘. 0n the other hand, Breslow et al. presented evidence for the intervention of (3b) as a stable intermediate in the reaction of lithium dimethylamide with 1-chlorobicyclo[2.2.0]hexa-2,5-diene72. The spin state of 1,4-DHB has been inferred as a singlet from the fact that no ESR signal could be observed for matrix 74 isolated 9,10-didehydroanthracene ( )70. From CIDNP (chemically induced dynamic nuclear polarization) and cage-escape reaction studies on diradical 2,3-dipropyl- 1,4-DHB produced in the solution thermolysis of Z-4,5- diethynyl-4-octene, Bergman and his coworkers concluded that the reactive state of the diradical is a singlet71. Thus, the experimental results indicate that 1,4-DHB can exist in both diradical (3a) and bicyclic (3b) forms, and that the most probable ground electronic state is a singlet diradical. As in the case of 1,3-DHB, no direct spectroscopic observations of 1,4-DHB have been reported. An attempt to obtain the IR spectrum of 1,4-DHB in an N2 matrix via UV photolysis of the precursor 1,4-diiodobenzene has not been successful73. 2.3.2. Theoretiqgl studies In contrast to previously published resultsls-22:23, the GVB/S-ZIG geometry optimization for singlet 1,4-DHB finds three local minima; two bicyclic and one diradical structure. The first bicyclic structure (3b) (hereafter simply termed butalene-A) lies 81.3 kcal/mole above the lowest diradical structure with long transannular bond (Cl—C4; 1.698 A). The second bicyclic structure (3b') (butalene-B) lies 10.7 kcal/mole above butalene-A (92.0 kcal/mole above the 75 diradical 1,4-DHB) with a rather short Cl—C4 bond distance (1.454 A). The energy differences for these two bicyclic structures with respect to the lowest diradical 1,4-DHB (3a) are much higher than those from the semi-empirical calculations. As Noell and Newton pointed out2°, these values may be too large by as much as a factor of 2. The detailed structures of these three 1,4-DHBs are listed in Table 2.10 along with the previously reported values. :l 1 (3b’) (3c) (3d) Based on the EHT population analysis, Hoffman suggested that the resonance structure bisallene (3c) may contribute significantly to the diradical 1,4-DHB9; the total overlap population for the 01-02 (equivalently C3-C4, 04-05 and C8-C1) bond is larger than that for the C2-Ca (or Cs-Cs) bond, and the overlap population for C1-C4 is a negative value. All semi-empirical methods and RHF/3-ZlG yield diradical geometries which seem to reflect this tendency. Recently, Jhonson proposed that the non-planar bisallene (3d) be considered as a possible structure of 1,4-DHB7. Even though the GVB/3-216 equilibrium geometry also reflects a 76 resonance contribution from a bisallene, the bond length and the n-electron overlap population alternation around the ring are much smaller than those predicted from other calculations. That is, the GVB equilibrium geometry retains the most aromatic character. The presence of butalene-A has been reported by Dewar and Li13. The results of their MINDO/3 calculation show that butalene-A lies 38.5 kcal/mole above the diradical 1,4-DHB with a 4.6 kcal/mole barrier to conversion of (3b) to (3a). Subsequent theoretical results from the MNDOC calculation by Thiel’z, and from the MNDO calculation by Dewar et 5123. are qualitatively similar to those from the MINDO/3 calculation. A limited geometry optimization at the GVB/4-3lG level starting with the MINDO/3 equilibrium geometry yielded an inflection point at the C1-C4 distance of 1.67 A20, which suggests that a complete optimization would lead to butalene-A. As was mentioned earlier, a complete geometry optimization at the GVB/3-21G level indeed obtains butalene-A, and to our surprise, a new species butalene-B. Perhaps the most interesting comparison between butalene-A and butalene-B structures is the reversal of the C1-C4 and C2-C3 bond distances; while the Cl-C4 bond length of butalene-B is considerably shortened from that of butalene-A (1.698 ~ 1.454 A), the opposite C2-C3 bond is noticeably elongated (1.437 r 1.482 A). 77 Teble 2.10. Geometricel peremetere of 1.4-DHB peremeter epeclee eethoe R1 R1 R3 a1 a: ref. (3e) "INDO/3 1 . 359 1 . 465 2. 728 117. 7 126. 6 18 rqu/a-CI 1 . 373 1. 439 2. 724 117. 3 124. 3 18 "DOC/SCF 1. 344 1.466 2. 604 22 ”DOC/2x26! 1. 370 1.420 2. 624 22 "DOC/REM ‘ 1. 363 1.453 2. 653 22 1100/3130! 1. 37 1.43 2. 55 23 Inf/3416- 1. 324 1. 510 2.674 116. 1 127. 9 " GVB/3416 1. 369 1.403 ‘ 2. 678 117. 7 124.3 " GVB/4416 1. 373 1.439 2. 701 117.4 125. 3 20 (fi) HIhDO/3-CI 1 . 393 1 . 426 1. 667 95. 0 170. 0 18 mac/set 1 . 388 1 . 442 1 . 604 22 WOO/2x261 1 . 389 1. 444 1. 599 22 WOO/DEM 1. 397 1. 453 1. 613 22 FIDO/3:361 1 . 39 1 . 45 1. 62 23 GVB/3416 1 . 391 1 . 437 1 . 688 95. 4 169. 2 " (a: ') GVB/3416 1 . 389 1 . 482 1 . 454 89. 4 181 . 2 ‘ triplet HIM/3 1. 386 1.410 115. 7 128.6 18 Raf/M316 - 1. 373 1.439 2. 677 116. 8 126.4 23 Raf/341G 1. 374 1. 394 2.654 117. 3 125. 4 " R1-C1C2. R2-C2C3. R3-c1c3. a1-4C1C2C3. G§IIC3C4C5 '3 This work 78 The 0- and n-electron overlap population for butalene-A and butalene-B alternate markedly around the two fused rings, which is indicative ofcomplete loss of aromaticity. While butalene-A and -B form a strong transannular bond, the n-electrons in this bond are quite repulsive. As we squeeze the C1-C4 distance of diradical 1,4-DHB to form butalenes, the opposite Cz-Ca bond length is elongated. 0n the other hand, the total overlap population for this bond also increases with the elongation; this surprising result -may explain how this extremely strained molecule can retain its ring structure. These results are summarized in Table 2.11. 'able 2.11. The overlap population for 1.4-DHB overlap population overlap population comer bond n(o) n(n) leaner bond n(o) n(n) (35) 010, 0.671 0.299 (3b') 0102 0.406 0.353 6263 0.559 0.277 02C3 0.962 0.180 0164 -0.013 -0.025 0104 1.033 -0.237 (3b) clc2 0.468 0.331 benzene cc 0.650 0.299 0203 0.876 0.217 c,c4 0.949 -0.169 _ Table 2.12 lists the singlet-triplet separations calculated for 1.4-DHB at various levels of theory. In accord with the GVB/4-31G calculation3°, we predict that triplet 79 1,4-DHB lies 1.3 kcal/mole below the singlet. While the C-C bond lengths of the triplet 1,4-DHB hexagon are less distorted from an equilibrium benzene geometry, the 01-04 transannular bond is shorter than that of the singlet structure. This may be explained in terms of the electronic configuration of triplet 1,4-DHB. The HOMO (5blu) of diradical 1,4-DHB approximately corresponds to the antisymmetric combination of the two radical lobe orbitals and the LUMO (Gag) to the’symmetric combination. Thus, the electronic configuration of triplet 1,4-DHB (~ 5b1016a91) increases the bonding character between the two radical centers. Table 2.12. Singlet—Triplet separation of 1.4-DHB method AE(S-T) comments ref. INDO -42.7 benzene geometry 13 2-Det. CI - 5.7 benzene geometry 15 ROHF triplet many-Det. CI — 3.5 benzene geometry 15 "INDO/3 -10.8 18 GVB/ROHF 1.4 4-316, partially 20 optimized geometry GVB/ROHF 1.3 3-216 This work 80 2.4. Summary and conclusion The relative energies of three DHBs (1a, 2a and 3a) are in the order: 1,2- (0.0 kcal/mole)<1,3- (16.0 kcal/mole)<1,4- (23.9 kcal/mole) at the GVB/3-216 level. This is in good accord with the GVB/4-31G calculation20. In contrast to the ab-initio calculations, at the 2x2CI level, the semi-empirical MINDO/3, MNDO and MNDOC methods predict that singlet 1.3-DHB is the most stable isomer13922o23. Only the MNDOC/BWENI gives the same order as the GVB results”. Dewar et al., however, suggested that! their MINDO/3 and MNDO results should be reinterpreted based on the following rule-of-thumb23: if the difference between the MINDO/3 or MNDO/SCF and their CI energies for a molecule is less than 15 kcal/mole, the former is then to be preferred; if greater, the CI values plus 15 kcal/mole should be taken. According to this rule, the relative energies of three DHBs are given as follow; 1,2— = 1,3-<1,4- (14 kcal/mole) at the MINDO/3 levells, and 1,2-<1,3-(8 kcal/mole)<1,4- (9 kcal/mole) at the MNDO leve123. The main conclusions drawn are that the 1,3- isomer has similar stability to that of 1,2-DHB, and that bicyclic 1,3- and 1,4-DHB are stable intermediates. However, the greatest uncertainty in the semi-empirical results for DHBs arises from the fact that a direct comparison of tabulated numerical values at a given level of calculation (see Table 2.13) does not lead to the same conclusion. As a result, there 81 has been considerable confusion amongst other researchers concerning the appropriate choice of theoretical values. Table 2.13. Relative energies (kcalflmoie) 0f DHBs N (1) <24) <20) (94) (ab) (30') ref. INDO 0.0 + 6.50) + 9.8(T) 19 25‘... CI 0.0 410.8(1‘) + 9.90) 15 many-Det. CI 0.0 +11.0m +10.9m 15 Hum/9.6a: 0.0 - 0.7 +1s.6 f 16" "INDO/3.211261 0.0 - 7.1 +. 2.6 was 16" mac/2x201 0.0 - 6.9 - 9.5 426.2 22 mace/2x201 0.0 410.4 - 1.9 +91.3 22 moo/am 0.0 + 7.4 +17.9 +41.7 22 . 1600/66? 0.0 + 8.5 +3s.7. +26.6 23" mac/35901 0.0 + 5.2 +10.9 + 5.9 +91.9 * 29" unoo 0.0 + 6.6 - 2.5 +3s.7 23" GVB/9-216 0.0 +16.0 +46.0' +29.9 +1os.2 +11s.9 This work GVB/4-316 0. 0 +14. 5 +29. 9 +100. 5' 20 * ' Not a local minimum. " See text for reinterpreted values. (T) indicate the ground state is a triplet. Since the established theoretical relative experimentally, it is results stability of three DHBs has not been difficult Judge which are correct. However, the merits of the 82 ab—initio predictions over the semi-empirical results are in their consistency at a given level of calculation. Various experimental and theoretical results suggest that 1,2-DHB has dual nature in conventional interpretation; while it is an aromatic system with respect to its electronic structure (section 2.1.3 and 2.1.6), its 01-02 bond closely resembles the cis-bent acetylenic bond (sections 2.1.4 — 2.1.6). All known results generally agree that the ground electronic state of 1,2-DHB is a singlet (section 2.1.1). On the other hand, as discussed in section 2.1.2, the first ionization potential and the heat of formation for 1,2-DHB are not yet definitively determined. While both bicyclic and diradical structures of 1,3-DHB are predicted to correspond to local minima on the MNDO/3x3CI potential surface23, only the diradical structure is found at the GVB/3-21G level. The bicyclic structure is located on an inflection point of the GVB potential curve, at a C1-Ca distance of 1.54 A, thus suggesting that it may not be a stable reaction intermediate. There are three isomeric species for 1,4-DHB; one diradical and two bicyclic structures. The GVB/3-216 geometry optimization obtains a modestly distorted diradical structure from the equilibrium geometry of benzene, but the extent of 83 this distortion is much less than those obtained by semi-empirical optimizations. The strained bicyclic structures retain their equilibrium geometries by balancing the lengthening of the Cz-C3 bond with the increase in its total overlap population. The fl-electron overlap population for each C-C bond of bicyclic 1,4-DHBs alternate markedly around the two fused four-membered rings, indicating that the bicyclic 1,4-DHBs are anti-aromatic in character. REFERENCES 10. 84 (a) Hoffman, R. W. "Dehydrobenzene and Cycloalkynes"; Academic Press: New York, 1967.; (b) Hoffman, R. W. In "Chemistry of Acetylenes"; Viehe, H. G., Ed.; Marcel Deckker: New York, 1969; p 1063. Various authors "Arynes, Carbenes, Nitrenes, and Related Species" in Annu. Rep. Progr. Chem., Sect. 8. from 1972 (vol 68) to 1985 (vol 81). (a) Field, E. K. In "Organic Reactive Intermediates"; McManus, S. P., Ed.; Academic Press: New York, 1973; Chapter 7.; (b) Field, E. K.; Meyerson, S. Adv. Phys. Org. Chem., 1968, 6, 1. Levin, R. H. React. Intermed. (Hiley) 1978, 1, 1; ibid. 1981, 2, 1; ibid. 1985, 3, 1. Reinecke, M. G. In "Reactive Intermediates, vol 2."; Abramovitch, R. A., Ed.; Plenum Press: New York, 1981; Chapter 5. Gilchrist, T. L. In "The Chemistry of Triple-Bonded Functional Groups, Supplement C:" Patai, S.; Rappoport, 2., Eds.; Wiley: New York, 1983; Chapter 11. Johnson, R. P. In "Molecular Structure and Energetics, vol. 3."; Liebman, J. F.; Greenberg, A. Eds; VCH: Deerfield Beach, 1986; Chapter 3. Coulson, C. A. Chem. Soc. Spec. Publ. 1961, 12, 100. Hoffmann, R.; Imamura, A.; Hehre, W. J. J. Am. Chem. Soc. 1968, 90, 1499. Gheorghiu, M. D.; Hoffmann, R. Rev. Roum. Chin. 1969, 14, 11. 12. 13 14. 15 16. 17. 18. 19. 20. 21. 22. 23. 85 947. Yonezawa, T.; Konishi, H.; Kato, H. Bull. Chem. Soc. Jap. 1969, 42, 933. Atkin, R. W.; Claxton, T. A. Trans. Faraday Soc. 1970, 66, 257. Olsen, J. F. J. Mol. Struct. 1971, 8, 307. Haselbach, E. Helv. Chim. Acta 1971, 54, 210. (a) Wilhite, D. L.; Whitten, J. L. J. Am. Chem. Soc. 1971, 93, 2868.; (b) Whitten, J. L. Acc. Chem. Res. 1973, 6, 236. Mille, P; Praud, L.; Serre, J. Int. J. Quantum Chem. 1971, 4, 187. (a) Newton, M. D.; Fraenkel, H. A. Chem. Phys. Lett. 1973, 18, 244. (b) Newton, M. D. In "Applications of Electronic Structure Theory"; Schaefer, H. F., III, Ed.; Plenum Press: New York, 1977; Chapter 6. (a) Dewar, M. J. S.; Lee, W.—K. J. Am. Chem. Soc. 1974, 96, 5569. (b) Dewar, M. J. S. Pure and Appl. Chem. 1975, 44, 767. Yamaguchi, K.; Nishio, A.; Yabushita, S.; Fueno, T. Chem. Phys. Lett. 1978, 53, 109. Newton, M. D.; Noell, J. O. J. Am. Chem. Soc. 1979, 101, 51. Dewar, M. J. S.; Ford, G. P.; Rzepa, H. S. J. Hal. Struct. 1979, 51, 275. Thiel, W. J. Am. Chem. Soc. 1981, 103, 1420. Dewar, M. J. S.; Ford, G. P.; Reynolds, C. H. J. Am. Chem. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 86 Soc. 1983, 105, 3162. Bock, C. W.; George, P.; Trachtman, M. J. Phys. Chem. 1984, 88, 1467. Radom, L.; Nobes, R. H.; Underwood, D. J.; Lee, W.-K. Pure and Appl. Chem. 1986, 58, 75. Apeloig, Y.; Arad, D.; Halton, B.; Randall, C. J. J. Am. Chem. Soc. 1986, 108, 4932. Hiller, I. H.; Vincent, M. A.; Guest, M. F.; von Nissen, W. Chem. Phys. Lett. 1987, 134, 403. Rigby, K.; Hiller, I. H.; Vincent, M. A. J. Chem. Soc. Perkin Trans. II. 1987, 117. Rondan, N. G.; Domelsmith, L. N.; Houk, K. N.; Bowne, A. T.; Levin, R. H. Tetrahedron Lett. 1979, 3237. Ng, L.; Jordan, K. D.; Krebs, A.; Rfiger, W. J. Am. Chem. Soc. 1982, 104, 7414. Schmidt, H.; Schweig, A.; Krebs, A. Tetrahedron Lett. 1974, 1471. Krebs, A.; Wilke, J. Top. Curr. Chem. 1983, 109, 189. Meier, H.; Peterson, H.; Kolshorn, H. Chem. Ber. 1980, 113, 2398. Maier, W. F.; Lau, C. G.; McEwen, A. B. J. Am. Chem. Soc. 1985, 107, 4724. Laing, J. W.; Berry, R. S. J. Am. Chem. Soc. 1976, 98, 660. Nam, H.-H.; Leroi, G. E. Spec. Chim. Acta 1985, 41A, 67. (a) Chapman, 0. L.; Mattes, K.; McIntosh, C. L.; Pacansky, J.; Calder, G. V.; Orr, G. J. Am. Chem. Soc. 1973, 95, 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 87 6134.; (b) Chapman, 0. L.; Chang, C.-C.; Kolc, J.; Rosenquist, N. R.; Tomioka, H. ibid. 1975, 97, 6586. Dunkin, I. R.; MacDonald, J. G. J. Chem. Soc. Chem. Comm. 1979, 772. Brown, R. F. C.; Browne, N. R.; Coulston, K. J.; Danen, L. B.; Eastwood, F. W.; Irvine, M. J.; Pullin, D. E. Tetrahedron Lett. 1986, 27, 1075. Nam, H.-H.; Leroi, G. E. J. Moi. Struct. 1987, 157, 301. Wentrup, C.; Blanch, R.; Briehl, H.; Gross, G. J. Am. Chem. Soc. 1988, 110, 1874. (a) Berry, R. S.; Spokes, G: N.; Stiles, M. J. Am. Chem. Soc. 1962, 84, 3570.; (b) Schafer, M. E.; Berry, R. S. ibid. 1965, 87, 4497. Porter, G.; Steinfeld, J. I. J. Chem. Soc. (A) 1968, 877. Kolc, J. Tetrahedron Lett. 1972, 5321. Mfinzel, N.; Schweig, A. Chem. Phys. Lett. 1988, 147, 192. Fisher, I. P.; Lossing, F. P. J. Am. Chem. Soc. 1963, 85, 1018. Natalis, P.; Franklin, J. L. J. Phys. Chem. 1965, 69, 2935. Grfitzmacher, H. F.; Lohmann, J. Justus Liebigs Ann. Chem. 1967, 705, 81. Rosenstock, H. M.; Larkins, J. T.; Walker, J. A. Int. J. Mass. Spectrom. Ion Phys. 1973, 11, 309. Rosenstock, H. M.; Stockbauer, R. L.; Parr, A. C. J. Chin. Phys. 1980, 77, 745. Pollack, S. K.; Hehre, W. J. Tetrahedron Lett. 1980, 21, 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 88 2483. Moini, M.; Leroi, G. E. J. Phys. Chem. 1986, 90, 4002. Dewar, M. J. S.; Tien, T.P. J. Chem. Soc. Chem. Comm. 1985, 1243. Leopold, D. G.; Miller, A. E. S.; Linegerger, W. C. J. Am. Chem. Soc. 1986, 108, 1379. Brown, R. D.; Godfrey, P. D.; Rodler, M. J. Am. Chem. Soc. 1986, 108, 1296. Luibrand, R. T.; Hoffman, R. W. J. Org. Chem. 1974, 39, 3887. Jones, M.; Levin, R. H. J. Am. Chem. Soc. 1969, 91, 6411. Gassman, P. G.; Benecke, H. P. Tetrahedron Lett. 1969, 1089. Wasserman, H.; Soldar, A. J.; Keller, L. S. Tetrahedron Lett. 1968, 5597. Jones, M.; Levin, R. H. Tetrahedron Lett. 1968, 5593. Tabushi, I.; Oda, R.; Okazaki, K. Tetrahedron Lett. 1968, 3743. Campbell, C. D.; Rees, C. W. Chem. Comm. 1965, 192. (a) Berry, R. S.; Clardy, J.; Schafer, M. E. Tetrahedron Lett. 1965, 1003.; (b) 1965, 1011. Evleth, E. M. Tetrahedron Lett. 1967, 3625. (a) Hess, B. A.; Schaad, L. J. Tetrahedron Lett. 1971, 17.; (b) Hess, B. A.; Schaad, L. J. J. Am. Chem. Soc. 1971, 93, 305. (a) Washburn, W. N. J. Am. Chem. Soc. 1975, 97, 1615.; (b) 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 89 Washburn, W. N.; Zahler, R. ibid. 1976, 98, 7827.; (c) 1977, 99, 2012.; (d) Washburn, W. N.; Zahler, R.; Chen, I. ibid. 1978, 100, 5863. (a) Billups, W. E.; Buynak, J. D.; Butler, D. J. Org. Chem. 1980, 45, 4636.; (b) ibid. 1979, 44, 4218. Jhonson, G. C.; Stofko, J. J.; Lockhart, T. P., Brown, D. W.; Bergman, R. G. J. Org. Chem. 1979, 44, 4215. Bergman, R. G. Acc. Chem. Res. 1973, 6, 25. Chapman, 0. L.; Chang, C.-C.; Kolc, J. J. Am. Chem. Soc. 1976, 98, 5703. (a) Lockhart, T. P.; Comita, P. B.; Bergman, R. G. J. Am. Chem. Soc. 1981, 103, 4082.; (b) Lockhart, T. P.; Bergman, R. G. ibid. 1981, 103, 4091. (a) Breslow, R.; Napierski, J.; Clarke, T. C. J. Am. Chem. Soc. 1975, 97, 6275.; (b) Breslow, R.; Khanna, P. L. Tetrahedron Lett. 1977, 3429. Nam, H.-H.; Leroi, G. E. unpublished result Gavifia, F.; Luis, S. V.; Safont, V. S.; Ferrer, P.; Costero, A. M. Tetrahedron Lett. 1986, 27, 4779. Borden, W. T. "Diradicals"; Borden, W. T. Ed; John Wiley and Sons: New York, 1982; Chapter 1. Wilson, E. B. Jr.; Decius, J.; Cross, P. "Molecular Vibrations"; McGraw Hill: New York, N. Y., 1955. CHAPTER 3 Didehydropyridines (DHP) Introduction Didehydroheteroarenes (DHHA) have been proposed as likely intermediates in many organic reactions, principally those involving cycloaddition or cine-substitution. These interesting intermediates have attracted widespread attention, as demonstrated by the many publications, including several reviews, on this subject in the last two decades1’7. Often DHHAs have been generated by dehalogenation of the corresponding halogenated heteroarenes in the presence of strong bases. Dicarboxylic anhydrides, aminotriazoles or diazonium carboxylates of heteroarenes are also well-known precursors of DHHAs. The observation of cine-substituted products, Diels-Alder adducts with some dienes, or tele- substituted products has been taken as evidence of intervening DHHA species. However, this kind of indirect evidence, based on trapping experiments to verify the intermediacy of DHHAs, possesses severe limitations. Other mechanisms, e.g. addition-elimination, transhalogenation, or ANRORC can also account for the formation of the observed 90 (I 91 products from a given precursorses. Mass spectrometric analysis following the electron impact or the pyrolytic fragmentation of some heteroarene dicarboxylic anhydrides has been used to conjecture the structure of DHHAs corresponding to certain m/z peaksg’lz. However, no structural information could be obtained from the mass spectrometric analysis. To date, the only direct evidence of any DHHA is the infrared spectrum of 3,4—didehydropyridine (3,4-DHP) generated via near uv photolysis of 3,4-pyridine dicarboxylic anhydride (3,4-PDA) isolated in N2 or Ar matrices at 13K19. Several theoretical papers have been published which predict the geometries, electronic structures, or heats of formation of some simple DHHAs (didehydro- pyridines, pyridazines, pyrimidines, pyrazines and thiophenes)13'13. Some of these calculations are at a rather rudimentary level (EHT17, semi-empirical M015o13); others at a more sophisticated level (MNDO14v15, ab-initiol3) fail to report some important information, such as the full structural parameters of optimized geometries, the effect of electron correlation between the two radical centers and the role of the heteroatom lone pair electrons. Thus, more detailed theoretical analyses are necessary to adequately understand the nature of DHHAs. Didehydropyridines (DHP) have been the most studied of all 92 known DHHAsl'7. There are six possible isomers of DHPs depending on the sites of dehydrogenation. Among the six DHPs, the 3,4- isomer has the most convincing evidence, but the evidence for other isomers is somewhat inconclusive as has been discussed in the previous sectionsos. Infrared matrix isolation spectroscopy could be a very useful method for the identification of vicinally didehydrogenated pyridinesl9, such as 3.4- and 2,3-DHP, since the two dehydrocarbons are expected to form a partial triple bond which has a characteristic vibrational frequency above 2000 cm-1. However, it is quite difficult to identify other isomers by vibrational spectroscopy. Theoretical calculations can provide valuable information regarding the structures and the energetics of DHPs, especially for the isomers whose experimental evidence is inconclusive13'18. EHT (Extended Hfickel Theory) calculations on the six possible DHPs have been carried out and the result of their stability sequence is 3,4->2,4->2,5->2,3->3,5->2,6-DHP17. However, this calculation used assumed geometries for the DHPs and made no distinction between singlet and triplet states. Energy calculations in the MNDO approximation suggest that all DHPs have singlet ground states with relative energy ordering of 3,4->2,4->2,5->2,6->2,3->3,5-DHP14. From these calculations, 3,4-DHP emerges as the most stable isomer and 2,3-DHP is expected to be much less stable than the 2,4- or 93 2,5- isomer. Early HUckel MO calculations predicted that 2,3-DHP is more stable than the 3,4- isomer because the nitrogen lone pair would be delocalized over the two adjacent radical centersls. This argument has not been supported by higher level calculations. EHT calculations showed that the nitrogen lone pair substantially destabilizes the adjacent radical lobes. Recent ab-initio CASSCF/3-21G calculations on 3,4- and 2,3-DHP with the RHF/3-ZlG optimized geometries favored 3,4-DHP over the 2,3- isomer by 13.9 kcal/mole13. However, the influence of the nitrogen lone pair on the equilibrium geometries of DHPs and on the correlation between the two singlet-coupled radical electrons has not been accounted for at a sufficiently high level of theory that reliable predictions can be made. In this chapter, matrix isolation studies on 3,4- and 2,3-DHP are presented. Theoretical calculations on all six DHP isomers have been carried out at the RHF, GVB and ROHF levels with a 3-210 basis set; these results are also reported and discussed in this chapter. 94 3.1. 3,4-Didehydropyridine (3,4-DHP) 3,4-DHP is the first DHHA to be proposed in modern times, and the evidence supporting its existence is the most convincing of any DHHA“. For example, diazabiphenylene, the dimer of 3,4-DHP, was identified in the time-of-flight mass spectrometric and kinetic uv spectroscopic analysis of the products formed by flash photolysis of pyridine-3-diazonium- 4-carboxylate12. Our subsequent success in isolating 3,4-DHP in argon or nitrogen matrices has established that this molecule is a true reactive intermediate, not just a transition state19. 3.1.1. Infrared Spectrum As summarized in Scheme I, mild irradiation (1)340nm and less than 100 min duration) of 3,4-PDA in N, or Ar matrices at 13 K readily fragmented the precursor to form CO, 002, and 3,4-DHP, which has a strong peak at 2085 cm‘1 diagnostic of carbon-carbon triple bond formation. Subsequent irradiation with X>210nm light immediately decomposed 3,4-DHP into HCN, diacetylene, acetylene, and cyanoacetylene as a result of alternative two-bond scissions. The infrared spectrum in the 2050 - 2300 cm'1 region prior to and following controlled photolysis (Figure 3.1) clearly demonstrates the formation of 3,4-DHP and its subsequent decomposition. The peak due to 95 3,4-DHP at 2085 cm"1 disappears upon shorter wavelength irradiation, and new peaks at 2101 cm'1 (HCN), 2181 cm‘1 (diacetylene) and 2236 cm"1 (cyanoacetylene) begin to grow. Ten additional peaks below 2000 cm"1 show the same growth and decay pattern as the 2085 cm'1 band and are also attributable to 3,4-DHP (Table 3.1). The IR frequencies of 3,4-DHP indicate that this molecule is remarkably similar to 1,2-didehydrobenzene in character. However, 3,4-DHP decomposes much faster. The wavenumbers observed for both 1,2-DHB and 3,4-DHP in N2 matrices are collected in Table 3.1. Schmmel " +c0+co2 "\ X:>2101m1 V PRN++HOGXM:m-Homi+lfix3N 96 Table 3.1. Infrared Bands (cis't) Resulting from Photolysis of 3, 4-PDA in an N, matrix at 13 K 1>940 nm‘ »210 mu" Photolyzed products 1.2-01-8' 2238 cyanoacetylene - 2181 diacetylene 2101 HCN 2085 3.4-DHP 2082 1558 3.4-DHP 1598 1448 1387 ‘ 3,4-0HP 1395 1355 3.4-0HP 1355 1260 polymer 1216 3.4-DHP 1055 3,4-DHP 1055 1038 998 3.4-0HP 853 3,4—0HP2 .848 3,4-0HP . 848 802 3.4-0HP 751 acetylene 744 744 acetylene 703 polymer 739 673 cyanoacetylene 648 648 diacetylene 635 ' 635 diacetylene 489 3,4-OHP 470 ‘ Photolysis of 3,4-PDA (100 min). ° Reference 19 3 Additional 30 min photolysis after a. 96a Fig.3.i IR spectra of 3,4-PDA and its photolyzed products in the 2050-2300 cm"1 region in an N2 matrix at 13 K: (a) 3,4-PDA; (b) after 100 min photolysis through water and x>340nm filter (The peak at 2281 cm'1 is due to 13C02); (6) following additional 30 min photolysis with x>210nm. Fig.3.2 Difference spectrum of 3,4-PDA before and after mild photolysis. * indicates a band due to diacetylene. 97 CIEoV 503E355; 005nm .. 005NN 00_NN 0052 N 01...: 51 1111111 I \iliiilllla 1...... Dos /\ p s I ”m 2:55 a .2 _ 2 _ c a“ 5:45 5 6:5 .55.: 55 55;. 56.33023 .0 5:5 552x 5.5:: 85545. £3. 955.8555 .5 505.5055 .0 .................... ’ ‘ ‘fifl “' a." 00__.N 000m 00.0 eouoqiosqo 98 CIEoYmQEncgo; on? om: 83 0»? com 9% o? p n p b p b b. p b n . OFonll T .8... v m fl 4 18.0 * H . r N15 . .8 .86 SE 82 .03.: 5.5% £3, mazsofi . Lmto vco 83mg <0ml¢.m “.0 £3.50QO 853:5 . eouquosqo cusp 99 3.1.2. Theoretical Calculgtlon The predicted geometries of singlet and triplet 3,4-DHP are given in Table 3.2 with their corresponding electronic energies. At the single-determinant SCF level the lowest singlet 3,4-DHP lies 1.6 kcal/mole higher in energy than a triplet. However, consideration of correlation between the two unpaired electrons through a GVB calculation reverses this order, and 3,4-DHP has a singlet ground state with a singlet-triplet energy difference, AE(S-T), of 30.7 kcal/mole, which is very similar to that of 1,2-DHB (29.9 kcal/mole) when it is calculated at the same level of theory. A comparison between the RHF and the GVB optimized structures of singlet 3,4-DHP shows only minor differences, the most significant change in bond distances being the lengthening of the C3C4 triple bond (1.229 r 1.263 A) as a consequence of the participation of the in-plane n” anti-bonding character in the GVB description. The occupation number of the acetylenic in-plane n* natural orbital in the GVB/3-ZlG optimized structure is 0.11. The same trends are found in the calculation of 1,2-DHB: the same magnitude of C102 triple bond lengthening (1.225 r 1.261 A), and the same in-plane n“ occupation number (0.11). This similarity implies that the physical properties of the bond between the two 100 Teble 3.2. Geometrical parameters of 3,4-DHP singlet (IA’) triplet (3A') RHF/3-218 GVB/3-218 ROHF/3-216 N10, 1.337 1.334 1.340 czc3 1.388 1.383 1.372 83c, 1.229 1.283 1.330 8485 1.378 1.380 1.371 8586 1.403 1.335 1.383 can1 1.341 1.340 1.327 4118,83 113. s 115.5 121. 1 28,8,c4 124.2 123.2 113.3 2838,85 127.0 124.8 120.0 2C4C3C. 103.0 114.4 117.1 28,9331 124.5 124.0 123.0 2c,N,c, 121.8 118.1 113.5 C2H1 1.085 1.088 1.083 c5112 1.087 1.087 1.070 8,143 1.070 1.070 1.070 m1c2H1 119.2 118.8 121.1 2£4C5H2 . 128.3 128.3 118.8 2859,143 118.3 113.8 120.2 E, -243.996328 -244.047831 -243.998843 Unlt: bond length (A): bond angle (degree); energy (Hertree) 101 dehydrocarbon centers are almost identical for both 1,2-DHB and 3,4-DHP. On the other hand, as depicted in Figure 3.3, the total atomic charges on the dehydrocarbon centers of 3,4-DHP are quite different from those of 1,2-DHB. While the two dehydrocarbon centers of 1,2-DHB have the same negative charges (-0.002), those of 3,4-DHP have opposite charges (C3:-0.09, C4:+O.07). This unequal charge distribution in 3,4-DHP would explain the unequal rates of nucleophilic addition at the two ends of the dehydro-bond7. -0.350 —0.269 0.114 / 0.068 -o.234 / -o.002 -0.645 N\ _0.090 -0.234 \ -0.002 0.092 -0.269 Fig. 3.3 Total atomic charges of 3,4—DHP and 1,2-DHB calculated at the GVB/3-216/lGVB/3-216 level. Prior to the calculation of the vibrational frequencies of 3,4-DHP at RHF/3-21G level, the performance of the theory was monitored by comparing the calculated and the observed vibrational frequencies of pyridine. The structural parameters of pyridine are listed in Table 3.3, and the calculated and observed vibrational frequencies are listed in 102 Table 3.3 Geometrical parameters of pyridine 3-213' 4-216" 4-318° 8-318" Experimental" ‘ N1C1 1.331 1.333 1.323 1.321 1.338 C203 1.383 1.382 1.382 1.385 1.334 c3c4 1. 384 1. 384 1. 383 1. 332 1. 332 2141853 122.8 122.8 122.4 123.8 28,83c4 118. 8 118. 5 118. 8 118.5 .463C4C5 113.0 113. 0 113. 0 118. 4 zCsN1C2 118. 8 118.4 118.0 118. 3 (22141 1. 070 1.071 1. 070“ 1. 088 83H2 1 . 070 1. 070 1 . 070 1. 082 84113 1 . 072 1. 072 1 . 072 1 . 081 211182141 118.7 118.4 118.4 118.3 420,112 120.7 120.4 120.3 120.1 423124113 120.5 120.5 120.5 120.8 ' This work; 5 Ref. 21; ° Ref. 22; d Ref. 13; ' Ref. 23 Table 3.4, from which can be seen that the calculated in-plane (A1, 8,) and out-of-plane pyridine are overestimated by 10.5% and 20.5% on the average, respectively. factors vibrations) results Thus, (0.905 are Bi) vibrations we applied two different scaling for in-plane and 0.830 for out-of-plane the calculated vibrations of 3,4-DHP. These summarized in Table 3.5. The vibrational 103 Table 3.4. Vibrational frequencies of pyridine symetry RHF/3-216 Experimental Exp/Cale A1 689 601 0.87 1083 991 0.92 1138 1032 0.91 1199 1072 0.89 1354 1218 0.90 1654 1483 0.90 1749 1583 0.91 3364 3030 0.90 3378 3402 3094 0.91 Ave 2 0.90 81 486 403 _ 0.83 822 700 0.85 875 744 0.85 1118 937 - 0.84 1213 1007 0.83 Ave 2 0.84 82 749 652 0.87 1158 ' 1079 0.93 1191 1143 0.96 1327 1227 0.93 1529 1362 0.89 1607 1442 0.90 1742 1581 0.91 3370 3042 0.90 3393 3087 0.91 Ave = 0.91 A2 458 373 0.81 1039 871 0.84 1190 966 0.81 Ave = 0.82 ‘ This work; 5 Reference 22 frequencies for the GVB wave functions are given in Table 3.6. However, the RHF scaling factors are not applied to Ithe GVB frequencies since the transferability of an empirical factor between the two different levels of calculation does not seem reasonablez‘. 104 Table 3.5 Comparison of calculated (RHF/3-21Gl/RHF/3-21G) and experimental vibrational frequencies (cm’l) for 3,4-DHP symmetry RHF/3-216 scaled Experimental A” 484 i 333 A’ 535 487 483 A” 553 484 A’ 745 878 A” 758 827 A” 351 783 802 A’ 352 888 848 A" 1047 883 853 A’ .1073 382 338 A" 1115 325 A’ 1148 1043 1055 A’ 1254 1141 ‘ A’ 1238 1173 A' 1385 1241 1218 A’ 1502 1387 1355 A’ 1525 1388 1385 A’ 1578 1434 1558 A’ 2189 1332 2085 A’ 3332 3088 A’ 3433 3124 A’ 3441 3131 105 In general, the agreement between the scaled and the observed frequencies is quite satisfactory. However, the scaled vibrational frequency of the C3C4 stretch, 1992 cm'l, becomes too low compared to the experimental value, 2085 cm'l. The same tendency also has been found for the C1C2 stretch (RHF/3-ZIG; 2209 cm‘l, scaled (0.91): 2010 cm'l) of 1,2- DHB. The GVB method calculates the C304 stretching at 1930 cm'l, which is very close to the C102 stretch of 1,2-DHB, 1931 cm'l, obtained at the same level of calculation. These data suggest that the bonding between the two dehydrocarbon centers of 1,2-DHB and 3,4-DHP are essentially the same. The harmonic vibrations of the 3A’ state are also listed in Table 3.6. The most noticeable change between the singlet and triplet occurs at the C3C4 stretching vibration (1703 cm’l), and its magnitude tells us that the triple bond character of C3C4 is essentially lost in the triplet state. 106 Table 3.6 Calculated vibrational frequencies (cm'l) of 3,4-DHP symmetry GVB/3-216 (lA’) ROHF/3-216 (3A') A” 474 483 A” 558 515 A’ 874 888 A’ 734 728 A” 738 808 A” 354 837 A' 1022 1067 A” 1030 1080 A’ 1031 1123 A’ 1123 1152 A“ 1157 1184 A’ 1284 1134 A’ 1232 1243 A’ 1371 1384 A’ 1517 1530 A’ 1584 1573 A’ 1537 1878 A’ 1330 1703 A’ 3331 3373 A’ 3427 3396 A’ 3431 3402 107 3.2. 2,3-Didehydropyridine (2,3—DHP) Evidence supporting the generation of 2,3-DHP is neither as extensive nor as convincing as that for 3,4-DHP5. For example, the photolysis of 2,3-PDA in N2 or Ar matrices leads to rupture of the ring structure, and the infrared spectra of the products taken at various time intervals during photolysis provide no evidence for the intermediacy of 2,3-DHP19. This isomer seems particularly interesting, however, because of the various possible interactions among three adjacent lobe orbitals. 3.2.1. Infrgred Spectrg of the photolyzed progpcts of 2,3-PDA When matrix isolated 2,3-PDA was irradiated with l>300nm light for 14 hours, the plot of a growth-curve for each newly appearing band revealed that at least three products have been formed in the matrix. As shown in Figure 3.4, the first type of compound (2) grows continuously and remains as the most stable product. The second type of compound (Y) grows faster than the compound 2, but after a certain period of time ( ~ 330 min ) it slowly begins to decay. The third type of compound (X) is very similar to Y, but it begins to decay earlier than Y (~ 210 min). The bands of X, Y and Z are listed in Table 3.7. 108 0e4'1 H ”Wan-1 (z) ‘ H 2095cm-1 (X) e—e 21340111-1 (Y) 0.3-1 3? 00 C or cell 5 .3 at! 2 D I! Irradiation time (minute) Fig. 3.4 Three types of growth-curve when 2,3—PDA is irradiated with X>300nm light. As summarized in Scheme II and in Figure 3.5, the compounds X, Y and Z are also distinguished by their photochemical behavior. The spectra (a), (b) and (c) in Figure 3.5 have been obtained respectively from the photolysis of 2,3-PDA in an N2 matrix with X>340nm filter for 10 hours, subsequently 2 hours with l>300nm filter and 1.5 hours more with X>210nm filter. The same spectra could be observed when the samples were separately irradiated over each wavelength region mentioned above. While the bands due 109 Table 3.7 Infrared bands (cm‘l) resulting from photolysis of 2,3—PDA in an N2 matrix at 13 K compound irradiation observed frequencies X l>340nm 2185, 2095 X>300nm Y X>300nm 2154, 1015, 876 I t Z X>340nm 3322, 3317, 3110, 2968, 2257, 2235, 2130, 2118, x>300nm 2109, 1628, 1607, 1038, 896, 763, 730 X>210nm * Visible when irradiated with x>210nm light to compound Z appear in all spectra, those due to compound Y are evident only in the spectrum (b). The bands due to compound X occur in both spectra (a) and (b). However, the bands of X and Y completely vanish upon shorter wavelength (1)210nm) irradiation. On the other hand, the bands due to X and Y do not change their shapes upon annealing of the matrix. The photoproduct Y does not interconvert to X photochemically when the l>300nm filter is replaced with X>340nm filter. Thus, the compounds X and Y are stable photochemical intermediates of Z. However, it is not certain whether compound Y is independently formed from a precursor or photochemically converted from compound X. The compound 2 is readily identifiable as B-ethynyl- acrylonitrile, (4) in scheme II. Compound X, which has a 110 Scheme II 2 0 C . ass 0 a. .0 .0 w . so as .00 0.00 es m m 2 w 0 HM . ............. iv l>340nm -602 m n 0. 0. so > .A X>210nm 111 characteristic =C=C=O band at 2095 cm-1 and -NEC band at 2185 cm-1, may be attributable to (1)25. The interpretation of compound Y is rather difficult. If it has been formed directly from a precursor, (2) is the plausible structure’s. On the other hand, if it has been formed from compound X, it might have structure (3). In any event, the presence of compounds X and Y preclude the intermediacy of 2,3-DHP in the formation of final product Z. 111a Fig. 3.5 IR spectra of the photolyzed products of 2,3-PDA in the 2050-2300 cm'1 region in an N2 matrix at 13 K; (a) 10 hour photolysis through water and X>340nm filter; (b) 2 hour photolysis after (a) with X>300nm filter; (8) 1.5 hour photolysis after (b) with X>210nm filter. 112 31580528355; CONN DmNN CONN Om PN 00 —N onON — b — — b 0.6 D 1. <11 411' . a 23 _ _. x .. Noo . .36 N H N T l¢.O /\ «B O aouoqiosqo oo _ m -wd N AIIIECENIal .. N + > + x Alliecoonslu we; N + x AIITEEETI: 3.70m . . 24.3328 822.: 8:8 os comm ‘ owmu comm on 5 8 5 once P _ _ _ _ . 0.0 O I (fill . . . rNd l ‘0. O souoqiosqo 114 3.2.2. Theoreticgl Calculation The total electronic energy of singlet 2,3-DHP at the GVB/3-21G level lies 7.4 kcal/mole higher than that of 3,4-DHP, but 17.5 kcal/mole lower than that of its lowest triplet state. The optimized geometries and total electronic energies of singlet and triplet 2,3-DHP are listed in Table 308. Early Hfickel MO calculations predicted that 2,3-DHP is more stable than the 3,4- isomer because the nitrogen lone pair would be delocalized over the two adjacent radical centersls. The RHF/3-210 optimized geometry indeed indicates that there is a substantial delocalization of the nitrogen lone pair over the two dehydrogenated carbon centers. This widens the bond angle at CZ (146.4) to allow a maximum overlap among the three lobe orbitals centered on N1, C2 and C3 at the cost of increased angle strain at N1 (107.0) and C3 (108.1). Thus, the delocalization of the nitrogen lone pair not only negatively contributes to the total electronic energy of 2,3-DHP as was pointed out by Hoffman, but also significantly increases the ring strain. In order to examine the influence of polarization functions on the equilibrium geometry of 2,3-PDA, the structure has been optimized with a 6-3lG* basis set. As 115 Table 3.8 Geometrical parameters of 2,3-DHP singlet (1A') triplet (3A‘ .RHF/3-2iG RHF/6-3IG‘ GVB/3-21G R0H843-218 N102 ' 1.253 1.244 1.313 1.285 0203 1.253 1.275 1.285 1.333 83c4 1.421 1.432 1.383 1.388 c4cs 1.403 1.402 1.338 1.402 8388 1.333 1.388 0 1.337. 1.375 can1 1.375 1.353 1.351 1.348 4410203 148.4 155. 5 131.3 122.3 28,8304 108.1 37.8 120.8 113.4 2838485 118.8 122.5 112.8 118.0 28,850, 121.5 113.7 121.1 118.8 285cc», 120.2 113.7 123.2 121.3 268N102 107.0 104.3 110.4 120.2 c4111 1.070 1.077 1.083 1.071 csH2 1.070 1.074 1.070 1.070 can3 1.088 1.072 1.083 1.088 2£3C4H, 122.4 118.3 125.3 121.1 2£4C5H2 119.8 120.5 120.0 120.8 28,0,H, 123.3 123.7 120.6 121.8 E -243.987656 -245.381877 -244.035964 -244.008141 116 noted from Table 3.3, the polarization function preferentially strengthens the NC bond. The same effect is found in the RHF/6-31G* optimized geometry of 2,3-PDA; the N102 bond length is considerably shortened with significant increase in its adjacent bond .angle (155.5), and the calculation diminishes the interaction between the two radical centers as is reflected by an enlongated C2C3 bond length (1.275 A). The resultant geometry is much more strained than that from the RHF/3-21G calculation, which suggests that the use of a polarization function does not necessarily provide improved results for the geometries of DHPs. The RHF HOMO and LUMO of 2,3-DHP are approximately expressed as linear combinations of three lobe orbitals centered on N1, CZ and C3. When these two orbitals are taken as a pair of initial GVB natural orbitals, the GVB SCF procedure minimizes the magnitude of the nitrogen lobe orbital, and leads to symmetric and antisymmetric combinations of the two radical lobe orbitals. Consequently, the bond length N102 substantially increases from 1.259 A to 1.313 A while the bond angle at CZ decreases by 14.5°. The GVB optimized geometry of 2,3-DHP is then rather close to other vicinally didehydrogenated aromatic systems. The entries in Table 3.9 show the similarity of the bonds between the two vicinal dehydrocarbon centers of 1,2-DHB, 3,4-DHP and 117 2‘, 3-DHPe Table 3.9 CaC bond length (A) and vibrational frequency of vicinally didehydrogenated aromatic systems singlet” tripletn Vcsc (GVB) Vcac (exp.) 1,2-DHB 1.281 1.383 1331 cm-1 2082 cm-1 3,4-DHP 1.283 1.330 1330 2085 2,3-DHP 1.265 1.333 1884 * GVB/3-216; ** ROHF/3-216 Since the GVB method may underestimate the contribution of the nitrogen lobe orbital, we have examined the three-term separated-pair wave function which explicitly includes the nitrogen lobe orbital as the third term in the correlation between the two singlet coupled electrons. However, it turns out that the contribution from the third term to an overall wave function is negligible (”0.04%). Thus, the lone pair electrons are confined to the nitrogen and the bonding between CZ and C3 is determined primarily by the coupling of the two odd electrons. The RHF/3-216 0203 harmonic stretching frequency is predicted to be at 1936 cm'l. Considering that RHF/34216 harmonic frequencies are overestimated by about 11%23, an empirical' correction of 11% would reduce the calculated 0203 stretching frequency in 2,3-DHP to 1742 cm‘l. This value suggests that the C203 bond in the RHF optimized geometry of 118 2,3-DHP is essentially a strained allenic bond. The GVB/3-21G harmonic frequency of the 0203 stretching in 2,3-DHP is calculated to be 1894 cm'l. Since the GVB method underestimates the vibrational frequency of the strained triple bond in 1,2-DHB or 3,4-DHP by 7.3% (see Table 3.9), the stretching frequency of 0203 in 2,3-DHP may be up-scaled to about 2044 cm‘l. Thus, we expect that the CEC stretching frequency of 2,3-DHP, if it is ever formed in a matrix isolation experiment, would be observed at least 40 cm"1 below that of 3,4-DHP. 119 3.3. 2,4-Dldehydropyridine (2,4-DHP) 2,4-DHP has been proposed as a likely intermediate in the tele-substitution of 6-substituted-Z-bromopyridines with KNH227. Two previous theoretical calculations predict that 2,4-DHP is the second most stable isomer of DHP15'17. The MNDO calculation estimates that the heat of formation of 2,4-DHP is only 0.5 kcal/mole higher than that of 3,4-DHP, but 8.6 kcal/mole lower than that of 2,3-DHP. No theoretical calculation has been published for 2,4-DHP at the ab-initio level. 3.3.1. Theoretical calculation At the GVB/3-21G level, the lowest singlet state of 2,4-DHP lies 9.6 kcal/mole higher in energy than that of 3,4-DHP. Contradictory to previous calculationsls»17, 2,4-DHP is less stable than 2,3-DHP by 2.2 kcal/mole. The ground electronic state of 2,4-DHP is a singlet with AE(S-T) of 14.0 kcal/mole. The optimized geometries of singlet and triplet 2,4-DHP and their corresponding energies are given in Table 3.10. The RHF calculations on 1,3-DHB yield a bicyclic structure with a C1-Ca separation of 1.514 A. Since the two dehydrocarbons of 2,4-DHP are also meta, one might expect the 120 Table 3.10 Geometrical parameters of 2,4-DHP singlet (lA') triplet (34') RHF/3-216 GVB/3-21G ROHF/3-216 N102 1 . 220 1. 280 1. 234 8283 1.357 1.404 1.385 cac, 1. 414 1. 382 1. 371 C405 1.411 1.373 1.382 csc6 1.381 1.385 1.332 can1 1.372 1.343 1.347 21416203 142. 8 123.4 128.0 2£20304 102.0 108.5 113.8 2cac4cs 123.2 128.8 123.6 2640508 122.2 117.1 118.2 2c5c8N1 123.8 113.0 122.0 286N1c2 115.8 113.3 118.5 03111 1.080 1.083 1.088 85112 1.071 1.083 1.083 CBH3 1.067 1.083 1.069 28,83111 129.8 124.2 122.8 28405112 113.5 121.8 122.5 2£5C8H3 128.4 122.4 121.3 E, -243.957776 -244.032503 -244.010254 Czc4 distance: RHF; 2.153, GVB; 2.245, ROHF; 2.308. PYridine(RHF/3-216);2.378 121 RHF calculation to predict a bicyclic structure for 2,4-DHP. However, as we have observed from the example of 2,3-DHP, at the RHF level the nitrogen lone pair tends to delocalize over the adjacent dehydrocarbon lobe orbital, which considerably weakens the 02-04 interaction. The total overlap population between the two radical centers is negative, which is indicative of no effective bonding between them. The GVB SCF procedure restricts the delocalization of nitrogen lone pair to maximize the correlation between the two odd electrons. A comparison between the RHF and the GVB optimized structures of singlet 2,4-DHP shows large differences in the N1~Cz bond length (RHF: 1.220, GVB:1.280 A) and its adjacent bond angle (RHF:142.6, GVB:129.4). The resultant GVB geometry is less strained than the RHF structure with much lower energy (AET = 46.9 kcal/mole). We have examined the three term separate-pair wave function for 2,4-DHP in the same manner as 2,3-DHP. The third term approximately corresponds to the nitrogen lone pair orbital. However, the weight of the third term is less than 0.03%. Thus, the role of the nitrogen lone pair in the correlation between the two odd electrons is negligible. 122 3.4. 2,5-Didehydropyridine (2,5-DHP) The heat of formation of 2,5-DHP calculated by MNDO is 133.5 kcal/molels, which is only 1.6 kcal/mole higher than that of 3,4-DHP. To our knowledge, no definitive experimental evidence of 2,5-DHP has been reported. 3.4.1. Theoretlcgl cglculgtion RHF geometry optimization of 2,5-DHP leads to the B-ethynylacrylonitrile structure by relocating the bonding electrons of Cs-Nl to the bonds N1-02 and Cs‘Cs (Scheme III). Scheme III “ § . H On the other hand, GVB geometry optimization does yield a closed ring structure for 2,5-DHP. At the GVB level, the total electronic energy of singlet 2,5-DHP is 9.5 kcal/mole higher than that of 2,4-DHP; singlet 2,5-DHP lies 4.9 kcal/mole lower in energy than the triplet state. Table 3.11 shows the optimized geometry of singlet and triplet 2,5-DHP from our calculation. 123 Table 3.11 Geometrical parameters of 2,5-DHP singlet (lA’) triplet (3A’) RHF/3-21G GVB/3-2IG ROHF/3-2iG N102 1.140 1.278 1.235 czc3 1.423 1.380 1.388 cac4 1.325 1.400 1.385 0485 1.424 1.375 1.382 csc6 1.183 1.358 1.388 can1 4.258 1.371 1.343 2N1C2C3 173.2 120.0 127.2 28,8304 123.3 128.0 115.8 2c3c4cs 124.4 115.4 118.3 2£4CSC° 173.3 118.2 123.5~ 285cc»:1 122.3 118.8 288N1c2 117.5 117.3 8,111 1 071 1.068 1.088 04H: 1.073 1.071 1.071 csua 1.051 1.066 1.067 28203111 118.0 123.0 121.8 21:304142 113.0 113.0 121.8 2cscsn3 173.8 125.3 123.2 E, -244.073108 -244.017220 -244 003355 CZCs distance: GVB; 2.638, ROHF; 2.613, pyridine(RHF/3-21G);2.715 124 The REF method fails to describe the ring structure of 2,5-DHP- because the RHF method tends to strongly bind the nitrogen atom (N1) and its adjacent dehydrocarbon(Cz) while it underestimates the interactions between the two dehydrocarbon centers. Nonetheless, the open structure obtained by the .RHF calculation corresponds to the global minimum of 2,5-DHP. As we have observed from the examples of 2,3- land 2,4-DHP, -the GVB method limits the interaction between N, and C, to maximize the interaction between the two odd electrons, which results in a local minimum cyclic structure for 2,5-DHP. However, the fact that cyclic 2,5-DHP is a relative minimum at the GVB level tells us nothing about its depth. If the minimum is rather shallow, the cyclic 2,5-DHP will undergo a facile ring opening to form) B-ethynylaorylonitrile which corresponds to the global minimum. This conclusion is supported by a related experimental resultz'; the formation of compound (8) from (5) is explained by an unusual ring-opening-cyclization process. / . / 1.1 / / o ' Li -LiBr ‘ 02° . a.- \N e ' 8r\N R \ N 9' 00 \N e (5) (6) (7) (8) The triplet structure of 2,5-DHP is quite unusual compared to all other isomers of DHP. In general, the triplet DHP ring 125 structure is less distorted from an equilibrium pyridine geometry than the singlet structure. However, the bond angles at the dehydrocarbon centers of triplet 2,5-DHP (02‘ 127.2°, Cs: 123.5“) are noticeably larger than those of the singlet (02: 120.0°, Cs: 118.2°), which narrows the 02-05 separation (singlet: 2.638 A, triplet: 2.613 A). The same tendency is found for 1,4-DHB, but the extent of distortion for 1,4-DHB (singlet: 124.3° e triplet: 125.4°) is much less than that of 2 ,5’DHB. Figure 3.8 shows approximate MO diagrams for 2,5-DHP and 1,4-DHB. The combinations of the two dehydrocarbon lobe orbitals are antisymmetric for HOMOs and symmetric for LUMOs. Thus, the electronic configuration of the triplet, [{core}(HOMO)1(LUMO)1], increases the bonding character between the two dehydrocarbon centers, which would narrow the distance between them. In the cases of 2,3-, 2,4— and 2,6-DHP, the addition of a nitrogen lone pair orbital to a pair of GVB wave functions (three term separate-pair wave function) slightly lowers the GVB energy by 0.2 - 0.6 kcal/mole; the weight of the newly added term is negligibly small (less than 0.04%). The geometry optimization with this wave function converges to the same structure as with the two-configuration (TC) wave function. On the other hand, in the case of 2,5-DHP, the 126 2,5-DH’ 1.4-0'9 Fig. 3.6. Approximate M05 of 2,5-DHP and 1,4-DHB addition of the nitrogen lone pair orbital considerably destabilizes the system; it increases the GVB energy by 0.8 kcal/mole and the geometry optimization fails to converge. This result suggests that the cyclic structure of 2,5-DHP is not as rigid as that of the other DHP isomers. 127 3.5. 2,6-Dldehydropyridlne (2,6-DHP) The intermediacy of 2,6-DHP has been proposed for the resin formation which is obtained when 2-halogenopyridines are treated with lithium piperidinez; however, the evidence is rather questionable“. It has been suggested that 2,6-DHP should be the most stable DHP isomer due to favorable overlap among the three lobe orbitals centered on Cz-Nl-C513. However, the results of MNDO and EHT calculations do not support this suggestion. They predict that the 2,6- isomer would have much lower or even the lowest stability15o17. 3.5.1. Theoretical cglculgtlon Among the six possible isomers of DHP, only the 2,6- isomer has a triplet ground state, which lies 2.4 kcal/mole lower in energy than the lowest singlet. The details of the optimized structures at the RHF, GVB and ROHF level are summarized in Table 3.12. Similar to the results for the 2,3- isomer, the RHF method delocalizes the nitrogen lone pair onto the adjacent dehydrocarbons, which shortens the bond between N, and 02 or equivalently Cs (1.227A), and widens the bond angle at N, (150.2°) to allow the maximum overlap among three lobe orbitals centered on Cz-Nl-CB. This causes a large angle 128 Table 3.12 Geometrical parameters of 2,6-DHP ((32v structure) singlet (1A1) triplet (332) RHF/3-21G GVB/3-2lG ROHF/3-2lG N1C2 1.227 1.329 1.302 C2C3 1.436 1.375 1.378 C3C4 1.398 1.330 1.332 4614182 150.8 111.4 117.3 2N102C3 108.8 128 3 124.3 26203C4 118.1 116.9 118.0 2C304C5 124.1 118.2 120.3 83H1 1.085 1.068 1.088 C4H2 1.072 1.072 1.072 2c2c3H1 120.7 120.3 122.0 253C4H2 118.0 120.9 119.8 ET -243.942669 -244.013945 -244.017725 strain at 02 or OS (106.6°) and destabilizes the system. Since the GVB method restricts the delocalization of the nitrogen lone pair, the GVB optimized geometry of 2,6-DHP is rather close to the equilibrium geometry of pyridine. The three-term separate—pair wave function insignificantly improves the GVB energy (0.2 kcal/mole), and the weight of the third term, the nitrogen lone pair orbital, is less than 0.02%. 129 3.6. 3.5—Didehydropyrldlne (3.5-DHP) No experimental evidence for 3,5-DHP has been reported. MNDO calculations on 3,5-DHP predict that it would be the least stable DHP isomerls. 3.6.1. Theoreticgl cglculgtion At the GVB level, 3,5-DHB has a singlet ground state with a predicted AE(S-T) of 10.0 kcal/mole. Its stability is comparable to singlet 2,5-DHP or triplet 2,6-DHP. Details of the optimized structures are collected in Table 3.13. 3,5-DHP is remarkably similar to 1,3-DHB in many respects. The RHF geometry optimization yields a bicyclic equilibrium structure (02,) with a 03—05 separation of 1.540 A. The positive 03-05 overlap population indicates the formation of weak bonding between them. The GVB optimized structure increases the 03-05 distance to 2.229 A, which is 0.061 A shorter than that of pyridine. The overlap population becomes negative, which is indicative of no effective bonding between the dehydrocarbon centers in the GVB singlet structure. 130 Table 3.13 Geometrical parameters of 3.5-DHP (sz structure) singlet (141) triplet (382) RHF/3-21G GVB/3-216 ROHF/3-216 N1C2 1.357 1.338 1.337 82c3 1.381 1.372 1.375 03c4 1.343 1.378 1.375 266N102 110.8 116.1 120.1 2N1C203 110.0 121.1 120.0 28283c4 153.8 128.8 122.5 2cac4c5 88.8 108.1 115.0 82H1 1.082 1.088 1.088 C4H2 1.083 1.085 1.071 2N1C2H1 121.8 117.7 117.4 2C304H2 145.2 125.3 122.5 E, -243.940822 —244.017937 -244.002274 C36s distance: RHF;1.540, GVB; 2.229, ROHF; 2.319, pyridine; 2.290 C3C5 overlap pop: RHF; +0.10, GVB; -O.18 131 3.7. Conclusion and Summary The results of primary interest are the relative energetics among the six cyclic isomers of DHP. These are summarized in Table 3.14. At the RHF level, the ground states of all DHPs except for the 2,5- isomer are predicted to be triplets. However, at the GVB level the limited electron correlation between the two odd electrons lowers the RHF energies of DHPs by 30 - 48 kcal/mole and the ground state of all DHPs except the 2,6- isomer are predicted to be singlets. Considering the limitations of the small basis set (3-21G) used in this calculation, and attributing little significance to total energy differences less than 5 kcal/mole29, the difference in ground state energy between the 2,3- and 2,4- isomer, and between the 2,6- (T) and 2,5- or 3,5- isomers are too small to determine their relative ordering with confidence. However, the splitting between the 3,4- and 2,3-, and between the 2,4- and 2,6- (T) isomers are sufficiently large (7.4 and 9.3 kcal/mole) that their relative stability is probably independent of the level of the computations. Thus, the ground state stabilities of the DHPs decrease in the order: 3,4-(S)>2,3-(S)~2,4-(S)>2,6-(T)~2,5-(S)~3,5-(S)~2,6-(S), where S and T in parentheses represent singlets and triplets, respectively. 132 Table 3.14 Relative AE (kcal/mole) values of DHPs with respect to the ground state energy of 3,4-DHP RHF (singlet) GVB (singlet) ROHF (triplet) 3.4- 32.2 0.0 30.7 2,3- 37.7 7.4 I 24.8 2,4- 58.4 8.5 23.5 2.5- -15.8* 13.1 24.1 2,6- 65.9 21.2 18.8 3,5- 67.1 19.1 28.5 * B-ethynylacrylonitrile, the global minimum of 2,5-DHP The GVB optimized geometries of 2,4-, 2,5-, 2,6- and 3,5-DHP strongly suggest that their singlet structures are not bicyclic, but rather are monocyclic with significant diradical character. All attempts to find bicyclic geometries for those compounds at the GVB level have failed. On the other hand, MNDO calculations predicted bicyclic s ructures for these isomers. Does this mean that the strained bicyclic structures cannot be dealt with by the GVB method, or that they exist because the MNDO method underestimates their ring strain energies? No definitive answer can be given at this time. The structure of a DHP isomer depends critically on the extent of interaction between the nitrogen lone pair and the 133 two odd electrons on dehydrocarbon centers. When the dehydrocarbons are non-adjacent to the nitrogen atom (N1), such as 3,4- or 3,5-DHP, the nitrogen lone pair does not play a significant role in determining the structures. When one of the dehydrocarbon centers is adjacent to N1, we have two contradictory descriptions: at the RHF level the nitrogen lone pair and the adjacent radical lobe are delocalized; on the other hand, at the GVB level they are localized on their own atoms and the nitrogen lone pair has limited interaction with the adjacent radical center. This difference between the two methods is well illustrated in the optimized structures of the 2,n-DHPs (n=3,4,5,6); while the RHF method strongly binds N1 and C2 and obtains an extremely strained ring or open acyclic structure, the GVB method obtains modestly distorted cyclic structures. When the nitrogen lone pair orbital is included as the third term of the GVB wave function the total electronic energies are slightly improved (by 0.2 — 0.6 kcal/mole), except for 2,5-DHP for which the energy is increased by 0.8 kcal/mole. The weight of the third term is less than 0.04% for all 2,n-DHPs. Geometry optimization of each 2,n-DHP with this wave function converged to the same geometry as with the TC wave function, except for the 2,5- isomer which did not retain its cyclic structure. Thus, it is apparent that the interaction between the nitrogen lone pair and the adjacent radical center contributes negatively to the stability of these systems. REFERENCES 10. 11. 12. 13. 134 den Hertog, H. J.; van der Plas, H. C. Advan. Het. Chem. 1965, 4, 121. Kauffmann, Th. Angew. Chem. (Intern. Ed. Engl.) 1965, 4, 543. Hoffmann, R. W. "Dehydrobenzene and cycloalkynes" Academic Press, New York 1967. Kauffmann, Th.; Wirthwein, R. Angew. Chem. (Intern. Ed. Engl.) 1971, 10, 20. Reinecke, M. G. in "Reactive Intermediates, vol. 2" Abramovitch, R. A. ed., Plenum Press, New York 1981. Reinecke, M. G. Tetrahedron 1982, 38, 427. van der Plas, H. C.; Roeterdink, F. in "The Chemistry of Functional Groupes, Supplement C: The Chemistry of Triple Bonded Functional Groups" Patai, S.; Rapport, Z. eds. John Wiley and Sons, New York 1983. Brown, R. F. C.; Crow, W. D.; Solly, R. K. Chem. Ind. (London) 1966, 343. Cava, M. P.; Mitchel, M. J.; Dejongh, D. C.; van Fossen, R. Y. Tetrahedron Letters, 1966, 2947. Reinecke, M. G.; Newsom, J. G.; Chen, L.-J., J. Am. Chem. Soc. 1981, 103, 2706. 810, F.; Chimichi, S.; Nesi, R. Hetercycles, 1982, 19, 1427. Krammer, J.; Berry, R. S. J. Am. Chem. Soc. 1972, 94, 8336. Radom, L.; Nobes, R. H.; Underwood, D. J.; Lee, W.-K. Pure 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 260 135 & Appl. Chem. 1986, 58, 75. Dewar, M. J. S.; Kuhn, D. R. J. Am. Chem. Soc. 1984, 106, 5256. Dewar, M. J. 8.; Ford, G. P. J. Chem. Soc. Chem. Comm. 1977, 539. Yonezawa, T.; Konishi, H.; Kato, H. Bull. Chem. Soc. Jap. 1969, 42, 933. Adam, W.; Grimison, A.; Hoffmann, R. J. Am. Chem. Soc. 1969, 91, 2590. Jones, H. L.; Beveridge, D. L. Tetrahedron Letters 1964, 1577. H.-H. Nam; Leroi, G. E. J. Am. Chem. Soc. 1988, 110, 4096. Theoretical calculations (RHF, GVB and ROHF with 3-21G basis set) on 1,2-, 1,3-, and 1,4-DHB have been carried out independently for this dissertation. Pang, F.; Pulay, P.; Boggs, J. E. J. Moi. Struct. (THEOCHEH) 1982, 88, 79. Wiberg, K. B.; Walters, V. A.; Wong, K. N.; Colson, S. D. J. Phys. Chem. 1984, 88, 6067. Sorensen, G. 0.; Mohlen, L.; Rastrup-Andersen, N. J. Hal. Struct. 1974, 20, 119. Rigby, K.; Hiller, I. H.; Vincent, M. J. Chem. Soc, Perkin Trans. II 1987, 117. Wentrup, C.; Lorenéak, P J. Am. Chem. Soc. 1988, 110, 1880. Rapi, G.; Sbrana, G. J. Am. Chem. Soc. 1971, 93, 5213. 27. 28. 29. 136 den Hertorg, H. J.; Boer, H.; Streef, J. W.; Vekemans, F. C. A.; van Zoest, W. J. Rec. Tray. Chim. Pays-Bas. 1974, 93, 195. Newkome, G. R.; Sauer, J. D.; Staires, S. K. J. Org. Chem. 1977, 42, 3524. Noell, J. 0.; Newton, M. D. J. Am. Chem. Soc. 1979, 101, 51.