The Cycloaddition Reactions of Angle Strained Cycloalkynes. A Theoretical Study

Transcription

1 Journal of the Chinese Chemical Society, 2005, 52, The Cycloaddition Reactions of Angle Strained Cycloalkynes. A Theoretical Study Ming-Der Su* ( ) School of Medicinal and Applied Chemistry, Kaohsiung Medical University, Kaohsiung 80708, Taiwan, R.O.C. The potential energy surfaces for the cycloaddition reactions of angle strained cycloalkynes to ethylene have been studied using ab initio methods. All the stationary points were determined with the MP2/6-311G(d,p) method with some calculations performed at the CCSD(T)/6-311G++G(d,p)//MP2/6-311G(d,p) level. Three kinds of cycloalkyne species, including monocyclic alkynes, bridged bicyclic alkynes, and heterocyclic alkynes, have been chosen in this work as model reactants. Two different reaction pathways have been proposed: (A) 1,2-carbon shift and (B) 1,2-hydrogen shift. That is, reactants [2+1]-TS-1 spirocarbene intermediate (A) TS-A Pro-A or (B) TS-B Pro-B. As a result, it is found that ground-state cycloalkyne appears to react more like a monocarbene than like an alkyne or a vicinal dicarbene as conventionally proposed. Our theoretical investigations also suggest that a cycloalkyne with a small C C C bond angle should be a good candidate for cycloaddition to an olefin. Moreover, in the cycloaddition reaction of a small ( six-membered ring six-membered) ring cycloalkyne, both 1,2-carbon and 1,2-hydrogen migrations will compete with each other. On the other hand, reactions involving larger ( seven-membered) ring cycloalkynes should proceed with a 1,2-carbon shift, leading to the major [2+2] cycloadduct. Furthermore, a configuration mixing model has been used to rationalize the computational results and to develop an explanation for the barrier heights. The results obtained allow a number of predictions to be made. Keywords: Cycloaddition reactions; Cycloalkynes; Configuration mixing model. INTRODUCTION In recent years cycloalkynes have been found to be a highly interesting class of strained organic molecules because of their intriguing physical and chemical properties. 1 Indeed, the chemistry of angle strained cycloalkynes has brought about both novel insights into structural and spectroscopic problems, as well as possibilities for the synthesis of new and interesting systems. 2 Experimental evidence has demonstrated that highly strained cycloalkynes 3 exhibit enhanced reactivity over analogous open chain alkynes, 4 particularly in addition reactions to the triple bond. As such, many reactions can be carried out with angle strained cycloalkynes that are not feasible or at least cannot be carried out under mild conditions with analogous open chain alkynes. The cycloalkynes and related compounds offer an interesting combination. They combine the high reactivity of the triple bond with steric protection of the resulting addition product. They are therefore used for the synthesis of new systems and for the realization of new reaction pathways. Cycloaddition reactions have been used extensively for trapping unstable cycloalkynes. 5 In fact, some typical cycloaddition reactions leading to interesting products were observed only with highly strained cycloalkynes. Of these the [2+2] cycloaddition reactions of cycloalkynes have been known for some time. 6,7 Recently, Laird and Gilbert have investigated the reaction of norbornyne (1) with 2,3-dihydropyran and found a fascinating array of products. 8a See Scheme I. One of the major products (2) is what would be expected from a [2+2] cycloaddition. The most striking result is the adduct 3 which is the most dominant product formed and has no precedence in analogous systems. As suggested by Laird and Gilbert, the reason for producing both major products (2 and 3) can be accounted for by if norbornyne (1) behaves as a dicarbene that initially adds to the alkene to give the spirocarbene shown in 4. 8a That intermediate then can provide the observed products through separate reaction pathways. Moreover, density functional theory (DFT) calculations by Bachrach, Gilbert, and Laird (BGL) support the hypothesis. 8b On the other hand, it was reported the cycloaddition of cyclopentyne (5)toaneth- Dedicated to Professor Ching-Erh Lin on the Occasion of his 66 th Birthday and his Retirement from National Taiwan University

2 600 J. Chin. Chem. Soc., Vol. 52, No. 4, 2005 Su ylene only leads to the 1,2-carbon shift product (6), rather than the 1,2-C-H insertion product (7). 7 See Scheme II. Similar experimental evidence and observations have been reported by several research groups. 6,7 Again, BGL used DFT calculations to demonstrate that cyclopentyne (5) proceeds via a concerted [2+1] cycloaddition with ethylene to form cyclopropylcarbene (8). This carbene (8) then prefers a 1,2- carbon shift over a 1,2-C-H insertion. Nevertheless, the origin of barrier heights for these cycloaddition reactions still remains unknown. Scheme I Scheme II states becomes sufficiently large, only one product will result. Accordingly, knowing the origin of barrier heights of different transition states can help explain why reactions with certain cycloalkynes give only one product isomer, while others give a mixture. The calculation of reaction pathways for cycloalkyne cycloadditions and the location and identification of the structures of the TSs is therefore of great theoretical interest. To the best of our knowledge, until now, no theoretical work has been devoted to the systematic study of the reactivities of cycloalkynes based on a sophisticated theory. It is astonishing how little is known about the cycloadditions of cycloalkynes to olefins, considering their importance to synthetic chemistry 1,2,7 and the extensive research activity on analogous open chain alkynes. 10 It is these unsolved problems that inspired this study. In order to elucidate the mechanisms and barrier heights for the cycloalkyne cycloaddition reactions, we have now undertaken a systematic investigation of the potential energy surfaces of several different kinds of cycloalkyne systems. As a result, three kinds of cycloalkynes, (i) monocyclic alkynes, (ii) bridged bicyclic alkynes, and (iii) heterocyclic alkynes, have been chosen as model systems in this study. Our specific aims are to gain a deeper understanding of the reaction mechanism using ab initio methods, to explain trends in the reactivity with strain angle in the ring, and to bring out factors that control the magnitude of the activation barrier. Moreover, a better understanding of the thermodynamic and kinetic aspects of such cycloalkyne cycloadditions may shed some light on optimal design of further related catalytic processes and chemical synthesis. CALCULATION METHODS Strangely, although the related addition of cycloalkynes to olefins has been intensely studied from a synthetic point of view, important questions about their structures, reactivities, and reaction mechanisms still largely remain unanswered. In fact, the behavior of cycloalkynes has puzzled chemists for years. When cyclopentyne (5) reacts with alkenes, for example, the products appear to result from a concerted [2+2] cycloaddition reaction that violates the Woodward-Hoffmann rules. 9 Besides this, for reactions in which two products are formed, each of the products will form via a specific transition state (TS) characterized by its structure and energy. The TS with lower energy will lead to the more favored product and, when the energy difference between the two transition Geometries of reactants, precursor complexes, transition states, and products were fully optimized by employing second-order Møller-Plesset (MP2) perturbation theory 11 without imposing any symmetry constraints. For triplet cycloalkyne systems, we also carried out second-order unrestricted MP calculations (UMP2) with annihilation of the spin contaminants (PUMP2). 12 A standardized 6-311G basis set 13 was used together with p and d polarization functions. 14 In these calculations inner-shell molecular orbitals were not included for computing electron correlation energies (frozencore approximation). The geometries obtained in this way are denoted by MP2(fc)/6-311G(d,p). Vibrational frequencies at stationary points were calculated at the MP2(fc)/6-311G(d,p) level of theory to identify them as minima (zero imaginary

3 Cycloadditions of Cycloalkynes J. Chin. Chem. Soc., Vol. 52, No. 4, frequencies) or transition states (one imaginary frequency). In order to obtain more reliable energies, coupled cluster calculations with single and double excitations and a perturbative estimate of triple contributions (CCSD(T)) 15 were carried out with the geometries optimized at the MP2(fc)/6-311G(d,p) level. These single-point CCSD(T) calculations were performed using the G(d,p) basis set. Thus, the relative energies given in the text are those determined at CCSD(T)/6-311G++G(d,p)//MP2(fc)/6-311G(d,p) and include zero-point vibration energies (ZPVE) corrections determined atmp2(fc)/6-311g(d,p). All of the MP2 and CCSD(T) calculations were performed with the GAUSSIAN 03 package of programs. 16 RESULTS AND DISCUSSION Geometries and Energetics of Monocyclic Alkynes Before discussing the geometrical optimizations and the potential energy surfaces for the cycloalkyne cycloaddition reactions, we shall first discuss the geometries of the monocycloalkyne reactants. At present, the specific monocyclic alkynes we have investigated are cyclopropyne (3- ring), 17 cyclobutyne (4-ring), 18 cyclopentyne (5-ring), 19 and cyclohexyne (6-ring). 19 For comparison, we also chose the smallest open chain alkyne, acetylene (HCCH), as a reference molecule. The optimized geometries for these cycloalkynes were calculated at the MP2/6-311G(d,p) level of theory and their selected geometrical parameters are collected in Scheme III, where they are compared with previous theoretical calculations. Unfortunately, there are no experimental geometries yet available for the monocycloalkyne species studied in the present work to allow a definitive comparison. Therefore, the reliability of the predicted geometries can only be estimated by comparison between different levels of theory (taking into account the standard errors of each method). As one can see in Scheme III, the prediction of geometric parameters for the 4-ring, 5-ring, and 6-ring species compare well among the various levels of theory. Besides, our calculated C C (1.337 Å) and C C (1.215 Å) bond lengths in ethylene and acetylene Scheme III

4 602 J. Chin. Chem. Soc., Vol. 52, No. 4, 2005 Su agree reasonably well with the experimental data (1.339 Å and Å, respectively). 20 It is therefore believed that the MP2/6-311G(d,p) calculations provide an adequate theoretical level for further investigations of the molecular geometries and energetic features of the cycloalkyne cycloaddition reactions. Two structural features are of critical importance here, the C C bond length and the C C C bond angle, which as noted earlier 3 prefers 180 when given a choice in the matter. The r(c C) values of 3-ring, 4-ring, 5-ring, and6-ring at the MP2/6-311G(d,p) level are seen in Scheme III to be 1.339, 1.274, 1.254, and Å, respectively. Note that these values are significantly longer than the prototype C C distance 20 in HC CH, namely Å. Accordingly, strictly on the basis of bond-length predictions, it would classify these singlet monocyclic alkynes as possessing a weak triple bond. Of necessity, the C C C bond angle in cycloalkyne must be less than the 180 called for by the classic sp hybridization on the acetylenic carbon atoms. At the MP2/6-311G(d,p) level of theory, this angle is predicted to be (average), 94.38, 114.6, and for the 3-ring, 4-ring, 5-ring, and 6-ring species, respectively. These angles are not much wider than the 60, 90, 108, and 120 that would characterize a triangle, square, pentagon, and hexagon, respectively. An interesting trend that can be observed in Scheme III is the decrease in the C C bond length and increase in the C C C bond angle on going from the three-membered to the six-membered ring. For instance, the MP2/6-311G(d,p) results indicate that the C C bond length decreases in the order Å (3-ring) > Å (4-ring) > Å (5-ring) > Å (6-ring), while the C C C bond angle increases in the order (3-ring) < (4-ring) < (5-ring) < (6-ring). It should be noted that the C C bond length and the C C C bond angle for HCCH are calculated to be Å and 180.0, respectively, at the same level of theory. The reason for this can be understood readily using molecular orbital theory. 21 Namely, a better -overlap between adjacent carbon atoms in a linear alkyne leads to a smaller C C bond length than that in an analogous bent alkyne. Moreover, our theoretical calculations suggest that the singlet-triplet energy splitting ( E st =E triplet E singlet )of monocyclic alkynes generally increases as the C C C bond angle is increased. For instance, the MP2/6-311G(d,p) calculations show an increasing trend in the singlet-triplet splitting for 3-ring (16.1 kcal/mol) < 4-ring (19.5 kcal/mol) < 5-ring (29.5 kcal/mol) < 6-ring (60.3 kcal/mol) < HCCH (120 kcal/mol). This is because the stability of the singlet state decreases with decreasing the C C C bond angle. The reason for this can be derived from basic molecular orbital theory. 21 For convenience, we here select acetylene as a model molecule to interpret the origin of the singlet-triplet energy splitting of cycloalkynes. In Scheme IV we show the Walsh diagram for the acetylene frontier orbitals, and *. The H C C angle is varied from linear (180 ) to bent while maintaining a fixed C C length. Bending at the carbon splits the initially degenerate and * orbitals, producing a four-orbital pattern. The and * orbitals perpendicular to the plane bending (labeled b 1 and a 2, respectively, in C 2v symmetry) do not change in energy with bending, because the H s lie in the nodal plane. On the other hand, the remnants of the and * orbitals (labeled a 1 and b 2, respectively) are affected substantially; a 1 goes up in energy and b 2 comes down. This energy change is a result of mixing carbon p y and s atomic orbitals into the and * orbitals. 21 Scheme IV As a result, the above hybridization causes two prominent effects. One is that orbital mixing occurs in such a way as to hybridize the orbitals away from the H s and from the center of the C C axis. It is therefore expected that such hybridization should cause bent acetylenes to be more reactive, since it results in a hybridized HOMO that points away from the two attached H s and leads to a better overlap with an incoming molecule. The other is seen in Scheme IV. In a bent geometry, the LUMO (b 2 ) is greatly stabilized as the H C C

5 Cycloadditions of Cycloalkynes J. Chin. Chem. Soc., Vol. 52, No. 4, angle decreases and the HOMO (a 1 ) increases in energy, resulting in a narrowing of the HOMO-LUMO gap. Accordingly, based on this orbital rationale, it is found that the energy gap between the HOMO and LUMO levels for acetylene species is strongly dependent on the H C C angle as shown in Scheme IV. Namely, the smaller the H C C angle, the smaller the singlet-triplet splitting ( E st ) of the acetylene molecule. Indeed, as demonstrated above, our theoretical results for the relationship between the E st of monocyclic alkynes and the C C C bond angle are nicely in agreement with this model prediction. We shall use the above results to interpret the origin of barrier heights for cycloalkyne cycloadditions in a later section. Before proceeding further, it is necessary to deal with the electron-counting convention from a valence bond perspective. As Scheme IV shows, the valence electronic configuration of closed-shell singlet acetylene is (b 1 ) 2 (a 1 ) 2 (b 2 ) 0 (a 2 ) 0. This strongly suggests that a bent acetylene should adopt a valence bond structure in which a lone pair of electrons are localized on an in-plane orbital of one carbon atom, while that of the other carbon atom is empty. See 9(a)-9(d). Namely, one may consider that a bent acetylene act as a monocarbene species. However, BGL suggested that strained cycloalkynes, such as norbornyne (1) and cyclopentyne (5), 8,22 might behave as though they were vicinal dicarbenes with two lone pairs of electrons centralized on each triply bonded carbon atom as shown in 9(e) or 9(f). Nevertheless, according to Scheme IV, this requires two in-plane orbitals in a bent acetylene to be filled, with other out-of-plane orbitals being empty. That is, the electronic configuration of 9(e) should be (b 1 ) 0 (a 1 ) 2 (b 2 ) 2 (a 2 ) 0, which should be high in energy since two electrons are now moved to excited states. As a result, from the above analysis, one may easily see that the ground-state cycloalkyne species would have monocarbene characters in nature. It is therefore suggested that angle strained cycloalkynes should behave as if they were monocarbenes. We shall see that the calculational results support these predictions. 9(a) 9(b) 9(c) 9(d) 9(e) low in energy 9(f) high in energy etc. Geometries and Energetics of Monocyclic Alkynes + C 2 H 4 During the cycloaddition reaction of a monocyclic alkyne to an ethylene, the system passes through a precursor complex (Cpx) at an early stage before the formation of the spirocarbene intermediate. Like many alkyl-, dialkyl-, and cycloalkylcarbenes, the spirocarbene intermediate will undergo intramolecular rearrangements. 23 A hydrogen or carbon atom will migrate to the carbenic carbon and restore its electron octet. There are two commonly found rearrangement patterns for monocycloalkynes. The first is a 1,2-hydrogen shift which is predominant in most simple alkyl carbenes with an -hydrogen. 24 Instead of a hydrogen, an adjacent carbon can also migrate to the carbenic center. This rearrangement is particularly important in the case of sterically constrained carbenes, 25 such as spirocarbenes (vide infra). As a result, there could be two kinds of reaction pathways for the monocyclic alkyne cycloaddition reactions, i.e., (A) 1,2-carbon shift and (B) 1,2-hydrogen shift. For convenience, cycloaddition reactions considered here all are as follows: reactants (Rea) Cpx TS-1 intermediate (Int) (A) TS-A Pro-A or (B) TS-B Pro-B. Selected geometrical parameters of the stationary point structures along the pathways calculated at the MP2/6-311G(d,p) level are shown in Figs. 1-5 for cyclopropyne (3-ring), cyclobutyne (4-ring), cyclopentyne (5-ring), and cyclohexyne (6-ring), and acetylene (HCCH), respectively. It is noted that this work also constitutes further testing of the validity of using the MP2 theory to treat the present systems. Hence, the energetics of reaction mechanisms were calculated with use of the CCSD(T) theory. The relative energies obtained at the MP2/6-311G(d,p) and CCSD(T)/6-311G++G(d,p)//MP2/6-311G(d,p) (hereafter designed MP2 and CCSD(T), respectively) levels of theory are collected in Table 1. Several noteworthy features from Figs. 1-5 and Table 1 are revealed. As one can see in Figs. 1-5 and Table 1, the calculated relative energies of the stationary points for the 4-ring, 5- ring, and 6-ring cycloadditions compare well between MP2 and CCSD(T) levels of theory, within kcal/mol. In the case of the 3-ring and HCCH systems, the relative energy difference between MP2 and CCSD(T) levels vary more, ranging from 0.2 to 21 kcal/mol. Nevertheless, all the relative energies in Table 1 show that the MP2 trends parallel those from the CCSD(T) calculations, at least qualitatively. We shall therefore use the CCSD(T) energies from now on. The reactions of monocyclic alkynes and ethylene studied at the present level of theory all undergo a [2+1] cycloaddition to give an intermediate, which can then proceed via

6 604 J. Chin. Chem. Soc., Vol. 52, No. 4, 2005 Su Fig. 1. The MP2/6-311G(d,p) geometries (in Å and deg) for the precursor complexes (Cpx), transition states (TS), intermediate (Int), and products (Pro) of cyclopropyne (3-ring) with acetylene. Values in parentheses are the relative energies obtained at the CCSD(T)/6-311G++G(d,p)//MP2/6-311G(d,p) level of theory. The heavy arrows indicate the main atomic motions in the transition state eigenvector.

7 Cycloadditions of Cycloalkynes J. Chin. Chem. Soc., Vol. 52, No. 4, Fig. 2. The MP2/6-311G(d,p) geometries (in Å and deg) for the precursor complexes (Cpx), transition states (TS), intermediate (Int), and products (Pro) of cyclobutyne (4-ring) with acetylene. Values in parentheses are the relative energies obtained at the CCSD(T)/6-311G++G(d,p)//MP2/6-311G(d,p) level of theory. The heavy arrows indicate the main atomic motions in the transition state eigenvector.

8 606 J. Chin. Chem. Soc., Vol. 52, No. 4, 2005 Su Fig. 3. The MP2/6-311G(d,p) geometries (in Å and deg) for the precursor complexes (Cpx), transition states (TS), intermediate (Int), and products (Pro) of cyclopentyne (5-ring) with acetylene. Values in parentheses are the relative energies obtained at the CCSD(T)/6-311G++G(d,p)//MP2/6-311G(d,p) level of theory. The heavy arrows indicate the main atomic motions in the transition state eigenvector.

9 Cycloadditions of Cycloalkynes J. Chin. Chem. Soc., Vol. 52, No. 4, Fig. 4. The MP2/6-311G(d,p) geometries (in Å and deg) for the precursor complexes (Cpx), transition states (TS), intermediate (Int), and products (Pro) of cyclohexyne (6-ring) with acetylene. Values in parentheses are the relative energies obtained at the CCSD(T)/6-311G++G(d,p)//MP2/6-311G(d,p) level of theory. The heavy arrows indicate the main atomic motions in the transition state eigenvector.

10 608 J. Chin. Chem. Soc., Vol. 52, No. 4, 2005 Su Fig. 5. The MP2/6-311G(d,p) geometries (in Å and deg) for the precursor complexes (Cpx), transition states (TS), intermediate (Int), and products (Pro) of acetylene (HCCH) with acetylene. Values in parentheses are the relative energies obtained at the CCSD(T)/6-311G++G(d,p)//MP2/6-311G(d,p) level of theory. The heavy arrows indicate the main atomic motions in the transition state eigenvector.

11 Cycloadditions of Cycloalkynes J. Chin. Chem. Soc., Vol. 52, No. 4, Table 1. Relative energies (in kcal/mol) for singlet and triplet monocycloalkynes and for the process: reactants [2+1]-TS-1 intermediate path(a) TS-A Pro-A or path(b) TS-B Pro-B (1,2) Relative Energies (3) E st (4) E epx 3-ring 4-ring 5-ring 6-ring HCCH (-1.327) (-1.005) (-1.635) (-1.142) ( ) (5) E (-19.69) (-9.547) (5.865) (19.45) (33.88) (6) E (-55.46) (-39.45) (-35.96) (-6.093) (33.88) (5) E A (-53.58) (-35.73) (-31.17) (-2.536) (36.19) (7) H A (-65.75) (-69.85) (-83.46) (-67.38) (-31.28) (5) E B (-33.59) (-28.54) (-25.34) (2.644) (51.82) (7) H B (-90.43) (-97.79) (-94.62) (-69.95) (-23.08) (1) At the MP2/6-31lG(d,p) and CCSD(T)/6-31lG++G(d,p)/MP2/6-31lG(d,p) (in parentheses) levels of theory. The MP2 optimized structures of the stationary points see Figs (2) Energies differences have been zero-point corrected. See the text. (3) Energy relative to the corresponding singlet state. A positive value means the singlet is the ground-state. (4) The stabilization energy of the precursor complex relative to the corresponding reactants. (5) The activation energy of the transition state, relative to the corresponding reactants. E A and E B stand for activation energies for path (A) and (B), respectively. (6) The energy of the intermediate relative to the corresponding reactants. (7) The exothermicity of the product relative to the corresponding reactants. H A and H B stand for enthalpies for path (A) and (B), respectively. two separate pathways to final products (Pro-A and Pro-B). The expected intermediate of the [2+1] cycloaddition reaction is the spirocarbene. The optimized transition state structures (3-ring-TS-1, 4-ring-TS-1, 5-ring-TS-1, 6-ring-TS-1, and HCCH-TS-1) along with the calculated transition vectors at the MP2 level are shown in Figs. 1-5, respectively. The arrows in the figures indicate the directions in which the atoms move in the normal coordinate corresponding to the imaginary frequency. Examination of the single imaginary frequency (cm -1 ) for each transition state (376i for 3-ring- TS-1, 548i for 4-ring-TS-1, 171i for 5-ring-TS-1, 527i for 6-ring-TS-1, and 797i for HCCH-TS-1) provides an excellent confirmation of the concept of an addition process. That is, the vibrational motion for the addition of ethylene to cycloalkyne involves a bond forming between ethylene and one triple-bonded carbon in the monocyclic alkyne. It should be mentioned that both TS-1 and spirocarbene (Int) structural pictures are consistent with the proposal that cycloalkyne itself bears monocarbene characters as discussed above. As seen in Table 1 and Figs. 1-5, the [2+1] transition states 3-ring-TS-1 and 4-ring-TS-1 are predicted to be 20 and 9.5 kcal/mol lower in energy than their respective starting materials. On the other hand, the present calculations predict that the energies of 5-ring-TS-1, 6-ring-TS-1, and HCCH-TS-1 are above those of the corresponding reactants by 5.9, 20, and 51 kcal/mol, respectively. This strongly indicates that the [2+1] cycloaddition of ethylene to cyclopropyne (3-ring) and cyclobutyne (4-ring) should take place more readily favorably than the analogous reactions of cyclopentyne (5-ring), cyclohexyne (6-ring), and acetylene (HCCH). In short, this shows that it is much easier to add ethylene to one of the triple-bonded carbon atoms of a small ring cycloalkane than either to one in a large ring cycloalkane or in an open chain alkyne. Recently, it has been found that the CM model based on Pross and Shaik s work 26 can provide a powerful but simple method to understand a variety of chemical reactions. 27 It appeared to us that this CM model provides the key to understand reactivity in the cycloaddition reactions of strained

12 610 J. Chin. Chem. Soc., Vol. 52, No. 4, 2005 Su cycloalkynes. Below we use this model explicitly to investigate the reason why the barriers for the [2+1] cycloaddition of small ring cycloalkynes are lower than those of large ring cycloalkyne systems. In the cycloaddition reaction, the participants may exist in a number of predetermined states, each of which may be approximated by the appropriate electronic configuration. However, in this case there are only two predominant configurations that contribute considerably to the total wavefunction and, in turn, affect the shape of the potential energy surface. The key electronic configurations for the cycloalkyne cycloaddition are illustrated in Fig. 6. One configuration, labeled [Cycloalkyne] 1 [C 2 H 4 ] 1, is termed the reactant configuration because this configuration is a good description of the reactants; the two electrons on the cycloalkyne are spin-paired to form a lone pair while the two electrons on the C 2 H 4 moiety are spin-paired to form a C C double bond. On the other hand, the spin arrangement of the other dominant configuration is different. The electron pairs are coupled to allow both C C bond formation and simultaneous C C bond breaking. In order to obtain this configuration from the former, each of the two original electron pairs needs to be uncoupled. Namely, the initial changes are [Cycloalkyne] 1 [Cycloalkyne] 3 and [C 2 H 4 ] 1 [C 2 H 4 ] 3. Hence, this configuration is labeled [Cycloalkyne] 3 [C 2 H 4 ] 3. It should be noted that [Cycloalkyne] 3 [C 2 H 4 ] 3 has an overall singlet configuration, despite the fact that it contains within it two local triplets. Consequently, the avoided crossing of these two configurations leads to the simplest description of the ground state energy profiles for the [2+1] cycloaddition reactions of cycloalkynes and ethylene. As demonstrated previously, 26,27 it is clear that the barrier height ( E ) as well as the reaction enthalpy ( H) can be expressed in terms of the initial energy gap between the above two configurations. In other words, the reactivity of such [2+1] cycloadditions will be governed by the singlet-triplet excitation energies for each of the reac- Fig. 6. Energy diagram for an addition reaction showing the formation of a state curve ( ) by mixing two configurations: the reactant configuration and the product configuration. In the reactants, they are separated by an energy gap S. S = E st (i.e., the cycloalkyne singlet-triplet splitting) + E * (i.e., the (C=C) *(C=C) triplet excitation energy for ethylene). Configuration mixing near the crossing point causes an avoided crossing (dotted line).

13 Cycloadditions of Cycloalkynes J. Chin. Chem. Soc., Vol. 52, No. 4, tants, i.e., Est (= E triplet E singlet for cycloalkyne) and E * (= E triplet E singlet for C 2 H 4 ). It must be emphasized that the barrier of the chemical reaction is caused by the promotion energy (S) that is nonzero as shown in Fig. 6. The decrease of S(S= E st + E * ) also stabilizes the product and makes reaction enthalpy H more exothermic. 28 Accordingly, if E * is a constant, then a smaller value of E st leads to (i) reduction of the reaction barrier since the intended crossing of [Cycloalkyne] 1 [C 2 H 4 ] 1 and [Cycloalkyne] 3 [C 2 H 4 ] 3 is lower in energy, and (ii) production of a larger exothermicity since the energy of the product is now lower than that of the reactants. Combining this insight and the previous bond-angle model predictions, one may therefore conclude that the smaller the C C C bond angle, the smaller the E st of a monocyclic alkyne, the lower the barrier height, and the larger the exothermicity, and in turn, the faster the [2+1] cycloaddition reaction should occur. Our model calculations confirm the above prediction. For instance, at the MP2 level of theory, the C C C bond angle and the E st of monocycloalkyne (kcal/mol) increase in the order: 3-ring (65.1, 16.1) < 4-ring (94.4, 19.5) < 5-ring (115, 29.5) < 6-ring (131, 60.3) < HCCH (180, 112). The barrier height for the [2+1] cycloaddition of monocyclic alkyne to ethylene also increases in the order (kcal/mol): 3-ring-TS-1 (-12) < 4-ring-TS-1 (-10) < 5-ring-TS-1 (+8.3) < 6-ring-TS-1 (+24) < HCCH-TS-1 (+57). Likewise, the reaction enthalpy (kcal/mol) for [2+1] cycloaddition parallels the same trend as the barrier: 3-ring-Int (-72) < 4-ring-Int (-38) < 5-ring-Int (-35) < 6-ring-Int (-1.8) < HCCH-Int (+39). These results are also consistent with the Hammond postulate that the activation barrier should be correlated to its reaction enthalpy. 28 As a consequence, our model observations provide strong evidence that an electronic factor resulting from the bending of the C C C bond plays a decisive role in determining the reactivity of monocyclic alkynes. Since the singlet-triplet splitting E st as well as the C C C bond angle of the cycloalkyne strongly correlates with its chemical reactivity, they can both be used as a diagnostic tool for the prediction of the reactivity of monocyclic alkynes. In fact, it was found experimentally that in most cases reactivity in addition reactions to the triple bond increases with increasing C C C bond angle deformation. 1d Accordingly, the reactivity of monocycloalkynes towards olefins is predicted to be in the order three-membered ring > four-membered ring > five-membered ring > six-membered ring > seven-membered ring > eight-membered ring and so on. Note that, as shown in Scheme V (at the MP2/6-311G(d,p) level of theory), the order of the C C C bond angle of monocycloalkyne follows the same trend as that of its E st (kcal/mol), i.e., cyclononyne (163, 130) > cyclooctyne (152, 112) > cycloheptyne (146, 95.5) > cyclohexyne (131, 60.3) > cyclopentyne (115, 29.5) > cyclobutyne Scheme V

14 612 J. Chin. Chem. Soc., Vol. 52, No. 4, 2005 Su (94.4, 19.5) > cyclopropyne (65.1, 16.1). From another point of view, our theoretical results confirm a general belief that the most important influence on the isolability of a monocycloalkyne is its ring size. 1d It has been reported that cyclononyne (9-ring) and cyclooctyne (8-ring) are isolable compounds, but cycloheptyne (7-ring), cyclohexyne (6- ring), and cyclopentyne (5-ring) cannot be isolated. 1d Moreover, experimental evidence for cyclobutyne (4-ring) and cyclopropyne (3-ring) has not been reported yet. 1d Again, we shall apply the above conclusion to other kinds of cycloalkyne systems in a later section. As mentioned earlier, after the formation of the spirocarbene intermediate, it rearranges further either into a [2+2] cycloadduct via a 1,2-carbon shift (i.e., path (A)) or into a spiro cycloproduct bearing a double bond via a 1,2-hydrogen shift (i.e., path (B)). For reaction path (A), we have located the transition state (TS-A) for each monocyclic alkyne species at the MP2/6-311G(d,p) level of theory, along with the imaginary frequency eigenvector (see Figs. 1-5). These reactions appear to be concerted; we have been able to locate only one TS for each reaction and have confirmed that it is a true TS on the basis of frequency analysis. The MP2/6-311G(d,p) frequency calculations for the transition states 3-ring-TS-A, 4-ring- TS-A, 5-ring-TS-A, 6-ring-TS-A, and HCCH-TS-A suggest that the single imaginary frequency values are 114i, 418i, 388i, 256i, and 260i cm -1, respectively. As Figs. 1-5 show, vibrational motion for the 1,2-carbon migration involves bond formation between carbon and carbon in concert with the other C C bond breaking. Moreover, our attempt to locate the transition states for the direct [2+2] cycloaddition of monocyclic alkynes with ethylene to give [2+2] cycloproduct failed. Thus, our theoretical investigations suggest that the [2+2] cycloaddition of cycloalkyne to ethylene proceeds via a two-step process, which does not violate the Woodward-Hoffmann rules. 9 Our model conclusion is in accordance with previous theoretical findings. 8b For reaction path (B), the TS geometries for the 1,2- hydrogen shift are depicted in Figs. 1-5, respectively. All these TSs possess one imaginary frequency and are true first-order saddle points. Our MP2/6-311G(d,p) frequency calculations indicate that the single imaginary frequency values are 783i, 863i, 985i, 912i, and 270i cm -1 for 3-ring-TS-B, 4-ring-TS-B, 5-ring-TS-B, 6-ring-TS-B, and HCCH-TS- B, respectively. As seen in Figs. 1-5, the major component of the TS-B vibrational mode is located at the hydrogen migration. Thus, it is apparent that these transition states connect the corresponding spirocarbene to the spiro cycloproduct with a double bond. From the above discussion, one can easily see that, of the two possible routes for monocyclic alkyne addition reactions, the most promising one is path (A) (1,2 carbon shift), which has a lower activation energy compared to path (B) (1,2 hydrogen shift). For example, at the CCSD(T) level of theory, comparing the barrier heights between path (A) and path (B), we find the following trend: 3-ring-TS-A (2.0 kcal/mol) < 3-ring-TS-B (22 kcal/mol), 4-ring-TS-A (4.0 kcal/mol) < 4-ring-TS-B (11 kcal/mol), 5-ring-TS-A (5.0 kcal/mol) < 5-ring-TS-B (11 kcal/mol), 6-ring-TS-A (3.6 kcal/mol) < 6-ring-TS-B (8.7 kcal/mol), and HCCH-TS-A (2.0 kcal/mol) < HCCH-TS-B (18 kcal/mol). Accordingly, our model calculations strongly indicate that the cycloadditions of monocyclic and open chain alkynes to ethylene should produce the exclusive [2+2] cycloadduct via a 1,2- carbon shift, rather than the spiro cycloproduct via a 1,2- hydrogen shift. This is somewhat surprising, since, for most singlet carbenes possessing -hydrogens, the 1,2-hydrogen migration is the predominant rearrangement path. 29 The difference in the heights of the activation barriers for path (A) and path (B) from carbene intermediates (i.e., 3-ring-Int, 4-ring-Int, 5-ring-Int, 6-ring-Int, and HCCH-Int) may be understood by comparing the structures of the spirocarbenes and of the transition states for both pathways (i.e., TS-A and TS-B). As one can see from Figs. 1-5, the similarity in the structures of the spirocarbene intermediate and TS-A suggests that 1,2-carbon rearrangement is quite facile compared with 1,2-hydrogen migration. As a consequence, it may explain the low barrier of only about kcal/mol for 1,2- carbon shift because the spirocarbene intermediate undergoes only minor geometrical changes to reach TS-A. Onthe other hand, the carbene intermediate has to undergo a large geometrical change before it reaches TS-B. This is the reason why the activation barriers for the 1,2-H migration in the present case study are greater than 9.0 kcal/mol. Our computational results are in good agreement with some theoretical findings, 8,30 in which C-migration was the favored path of rearrangement and led to the major [2+2] cycloadduct. Further, as shown in Figs. 1-5, since the transition states for 1,2-carbon migration and 1,2-hydrogen shift are much lower in energy than those of the [2+1] cycloaddition, it is believed that a monocycloalkyne will undergo cycloaddition to olefin in a concerted manner. As a result, the stereochemistry of its final cycloproducts should be preserved. The experimental observations confirm this prediction. 8a

15 Cycloadditions of Cycloalkynes J. Chin. Chem. Soc., Vol. 52, No. 4, Geometries and Energetics of Bridged Bicyclic Alkynes +C 2 H 4 The six sets of bridged bicyclic alkyne reactants used in the present work are shown in Scheme VI: 10 (bicyclo- [2,1,1])hexyne), 11 (bicyclo[2,2,1])heptyne), 12 (bicyclo- [2,2,2])octyne), 13 (bicyclo[3,2,1])octyne), 14 (bicyclo- [3,2,2])nonyne), and 15 (bicyclo[4,2,1])nonyne). Of these, to our knowledge, only 11, 31 12, 32 and have been generated as reactive intermediates and trapped with appropriate alkenes. Like the case of monocyclic alkynes, the cycloaddition reactions of the bridged bicyclic alkynes studied in this work follow similar reaction paths to those shown earlier. That is, Rea TS-1 Int (A) TS-A Pro-A or (B) TS-B Pro-B. Since, generally speaking, the MP2 values can reproduce the CCSD(T) results well in the monocyclic alkyne systems shown earlier, we thus still use the MP2/6-311G(d,p) method to investigate the reaction mechanisms of the cycloadditions of bridged bicyclic alkynes to ethylene. The potential energy profiles for such cycloadditions of to ethylene at the MP2 level are given in Figs. 7-12, respectively. The singlet-triplet splitting ( E st )of10-15 and the energies relative to the reactant molecules (bridged bicyclic alkynes + C 2 H 4 ) are also summarized in Table 2. There are several important conclusions from these results to which attention should be drawn. First, according to the discussion in the previous section, it is expected that the larger the C C C bond angle ( ), the smaller the C C bond length (r), and the larger the singlet-triplet splitting ( E st =E triplet E singlet ) of the bridged bicyclic alkyne. Our MP2 results are in agreement with this prediction. Namely, as seen in Scheme VI, the magnitude of the E st of the bridged bicyclic alkyne reactant follows the same trend as the acetylenic bond angle : 10 (37.6 kcal/mol, 105 )<11 (45.8 kcal/mol, 110 )<12 (53.9 kcal/mol, 118 )< 13 (63.7 kcal/mol, 128 ) <14 (77.2 kcal/mol, 136 ) <15 (90.6 kcal/mol, 143 ). On the other hand, the trend in the C C bond length r mirrors the trend in E st : 10 (1.273 Å) > 11 (1.267 Å) > 12 (1.257 Å) > 13 (1.245 Å) > 14 (1.237 Å) > 15 (1.232 Å). Later we shall use the above results to explain the origin of barrier heights for their pericyclic chemistry with alkenes. Second, as mentioned above, we can see that the first step in the reaction mechanism is to undergo a [2+1] cycloaddition via a transition state, TS-1. Again, it appears that a bridged bicyclic alkyne reacts more like a monocarbene than like an alkyne. Considering geometrical effects, our theoretical investigations suggest that the cycloaddition reaction of a Scheme VI Fig. 7. Potential energy surfaces for the cycloaddition reactions of bicyclo[2,1,1])hexyne (10)toethylene. The relative energies (kcal/mol) are taken from the MP2 level as given in Table 2. The MP2-optimized structures of the stationary points are available as Supporting Information.

16 614 J. Chin. Chem. Soc., Vol. 52, No. 4, 2005 Su small ring bridged bicyclic alkyne is more facile than that of a bridged bicyclic alkyne bearing a large ring. For instance, at the MP2 level of theory, since the size of the ring of a bridged bicyclic alkyne molecule is in the order: 10 < 11 < 12 < 13 < 14 < 15, the barrier height for its ethylene addition increases in the same order (kcal/mol): [2,1,1]-TS-1 (6.4) < [2,2,1]- TS-1 (8.4) < [2,2,2]-TS-1 (24) < [3,2,1]-TS-1 (30) < [3,2,2]- TS-1 (37) < [4,2,1]-TS-1 (44). In fact, it is readily seen that this result is in accordance with the trend in the E st of the bridged bicyclic alkyne reactants as shown above. Namely, according to the CM model, 26 one predicts that a bridged bicyclic alkyne with a more strained ring would have a smaller E st and a more facile cycloaddition to ethylene than the one possessing a large less strained ring. Thus, like the case of monocyclic alkynes, our theoretical findings suggest that the reactivity of the bridged bicyclic alkynes is strongly correlated with their C C C bond angle as well as with their singlet-triplet splitting E st. Third, as stated previously, after the [2+1] transition state (i.e., TS-1), a spirocarbene intermediate is formed along the reaction coordinate. In addition to the 1,2-hydrogen and 1,2-carbon shifts described earlier, an alternative 1,3-hydrogen shift is possible. This is usually found in carbenes with no -hydrogens and leads to the formation of cyclopropanes. 34 Generally, 1,3-hydrogen migration has a barrier that is slightly higher than that of 1,2-hydrogen migration and should only be observed in cases where the carbon possesses no -hydrogens or the geometry of the carbene favors this rearrangement. 35 Indeed, this is the case in the 10, 11, and 12 systems. That is, only 1,2-carbon shift and 1,3-hydrogen shift were found in these molecules, which are represented by path (A) and path (B), respectively, in this work. On the other hand, both 1,2-carbon shift and 1,2-hydrogen shift were observed in the 13, 14, and 15 systems, which are also denoted by path (A) and path (B), respectively, for convenience (vide infra). Fourth, from Figs. 7-9, it is apparent that the transition states for [2+1] cycloaddition between 10, 11, 12 and ethylene are the only transition states (i.e., [2,1,1]-TS-1, [2,2,1]- TS-1, and [2,2,2]-TS-1, respectively) that lie below the en- Fig. 8. Potential energy surfaces for the cycloaddition reactions of bicyclo[2,2,1])hexyne (11) toethylene. The relative energies (kcal/mol) are taken from the MP2 level as given in Table 2. The MP2-optimized structures of the stationary points are available as Supporting Information. The values in parenthesis are taken from Ref. (8b) at the B3LYP/6-31G(d) (298 K) level of theory. Fig. 9. Potential energy surfaces for the cycloaddition reactions of bicyclo[2,2,2])hexyne (12) toethylene. The relative energies (kcal/mol) are taken from the MP2 level as given in Table 2. The MP2-optimized structures of the stationary points are available as Supporting Information.

17 Cycloadditions of Cycloalkynes J. Chin. Chem. Soc., Vol. 52, No. 4, ergy of the corresponding reactants. This strongly implies that the spirocarbene intermediate (i.e., [2,1,1]-Int, [2,2,1]- Int, and [2,2,2]-Int) are kinetically unstable and may rearrange spontaneously to the more stable minima if they are produced. That is to say, when the [2+1] cycloaddition occurs, it will be easily followed by the generation of either [2+2] cycloadduct (i.e., [2,1,1]-Pro-A, [2,2,1]-Pro-A, and [2,2,2]-Pro-A) or spiro cycloproduct with a cyclopropyl skeleton (i.e., [2,1,1]-Pro-B, [2,2,1]-Pro-B, and [2,2,2]- Pro-B) via 1,2-carbon migration (path (A)) and 1,2-hydrogen migration (path (B)), respectively. The supporting evidence comes from fact that, as mentioned in the Introduction, two major products (2 and 3 for path (A) and path (B), respectively) were observed in the cycloaddition of norbornyne (1) to an ethylene. 8a Moreover, as seen in Fig. 8, since our theoretical calculations show that the barrier to [2,2,1]-Pro-B (2.6 kcal/mol) is smaller than the barrier to [2,2,1]-Pro-A (5.1 kcal/mol), it is anticipated that the yield of the former will be larger than that of the latter. This conclusion is in good accordance with the experimental findings. 8a,36 Similarly, as shown in Figs. 7 and 9, our MP2 results indicate that the activation barriers to [2,1,1]-Pro-B (19 kcal/mol) and [2,2,2]- Pro-B (2.3 kcal/mol) are larger than those to [2,1,1]-Pro-A (7.3 kcal/mol) and [2,2,2]-Pro-A (2.1 kcal/mol), respectively. This would lead to the latter [2+2] cycloadduct in a larger yield than the former spiro cycloproduct. To the best of our knowledge, neither experimental nor theoretical work has been reported on these species. Fifth, on the 13, 14, and 15 potential energy surfaces, the TS-1, Int, TS-A, and TS-B stationary points are all found to lie above the energies of their corresponding reactants at the MP2 level of theory. In addition, as shown in Figs , our theoretical results demonstrate that the 1,2-carbon migration barriers (path (A)) are larger than for 1,2-hydrogen migration (path (B)) for these bridged bicyclic alkynes. It is therefore expected that, from a kinetic viewpoint, the [2+2] cycloadduct (i.e., [3,2,1]-Pro-A, [3,2,2]-Pro-A, and [4,2,1]- Pro-A) should be the only major product in such larger ring systems. As there are no relevant experimental and theoretical data on such systems, the above conclusion is a prediction. Finally, as one can see in Figs. 7-12, the energy of the fi- Fig. 10. Potential energy surfaces for the cycloaddition reactions of bicyclo[3,2,1])hexyne (13)toethylene. The relative energies (kcal/mol) are taken from the MP2 level as given in Table 2. The MP2-optimized structures of the stationary points are available as Supporting Information. Fig. 11. Potential energy surfaces for the cycloaddition reactions of bicyclo[3,2,2])hexyne (14)toethylene. The relative energies (kcal/mol) are taken from the MP2 level as given in Table 2. The MP2-optimized structures of the stationary points are available as Supporting Information.

18 616 J. Chin. Chem. Soc., Vol. 52, No. 4, 2005 Su Table 2. Relative energies (in kcal/mol) for singlet and triplet bridged bicyclic alkynes (10-15) and for the process: reactants [2+1]-TS-1 intermediate path(a) TS-A Pro-A or path(b) TS-B Pro-B (1,2) Relative Energies (3) E st (4) E 1 (5) E 2 (4) E A (6) H A (4) E B (6) H B (1) At the MP2/6-31lG(d,p) level of theory. The MP2 optimized structures of the stationary points (see Supplementary Materials). (2) Energies differences have been zero-point corrected. (3) Energy relative to the corresponding singlet state. A positive value means the singlet is the ground-state. (4) The activation energy of the transition state relative to the corresponding reactants. E A and E B stand for activation energies for path (A) and (B), respectively. (5) The energy of the intermediate relative to the corresponding reactants. (6) The exothermicity of the product relative to the corresponding reactants. H A and H B stand for enthalpies for path (A) and (B), respectively. nal cycloproducts relative to their corresponding reactants are -81.3, ([2,1,1]-Pro-A, [2,1,1]-Pro-B); -81.4, ([2,2,1]-Pro-A, [2,2,1]-Pro-B); -60.7, ([2,2,2]-Pro-A, [2,2,2]-Pro-B); -59.0, ([3,2,1]-Pro-A, [3,2,1]-Pro-B); -51.9, ([3,2,2]-Pro-A, [3,2,2]-Pro-B); and -43.1, kcal/mol ([4,2,1]-Pro-A, [4,2,1]-Pro-B), indicating that these bridged bicycloalkyne cycloadditions are all highly exothermic. Thus, considering both the kinetics and the thermodynamics of the pericyclic reactions, one may conclude that the cycloaddition of bridged bicycloalkyne to ethylene should proceed in a concerted manner. In other words, such cycloadditions of bridged bicycloalkynes should be favored for producing stereoretention products. Some available experimental observations are in good accordance with this conclusion. 8a Fig. 12. Potential energy surfaces for the cycloaddition reactions of bicyclo[4,2,1])hexyne (15)toethylene. The relative energies (kcal/mol) are taken from the MP2 level as given in Table 2. The MP2-optimized structures of the stationary points are available as Supporting Information. Geometries and Energetics of Heterocyclic Alkynes + C 2 H 4 The seven sets of heterocyclic alkyne reactants used in this work are shown in Scheme VII: 1d 16 (3,3,6,6-tetramethylcyclohexyne), 17 (3,3,6,6-pentamethyl-1-oxa-4-cycloheptyne), 18 (1,3,3,6,6-pentamethyl-1-aza-4-cycloheptyne), 19 (3,3,7,7-tetramethylcycloheptyne), 20 (3,3,6,6-tetramethyl- 1-thiacycloheptyne), 21 (1,1,3,3,6,6-hexamethyl-1-silacycloheptyne), and 22 (3,3,8,8-tetramethylcyclooctyne). To the best of our knowledge, only 19, 37,38 20, 37 21, 39 and 22, 40 in which the triple bond is shielded by neighboring methyl

19 Cycloadditions of Cycloalkynes J. Chin. Chem. Soc., Vol. 52, No. 4, Scheme VII groups, have been isolated and characterized. In contrast, the other species (16, 41 17, 42 and ) have been recognized as transient intermediates. For consistency with our earlier work, the following reaction mechanism has been used to explore the cycloaddition reaction of heterocycloalkyne to ethylene: Rea TS-1 Int (A) TS-A Pro-A or (B) TS-B Pro-B. For the systems 16-22, their geometries and energetics have been calculated using the MP2/6-311G(d,p) level of theory. The relative energies of the stationary points for the above mechanism are collected in Table 3 and Figs for the heterocycloalkynes, respectively. The major conclusions drawn from the current study can be summarized as follows. First, as mentioned above, only a small number of substituted heterocycloalkynes have been synthesized and characterized unequivocally Although only a few details concerning their geometrical parameters are as yet available, we may compare some of our results with those obtained for substituted 20, 21, and 22. See Scheme VII. Our calculated C C bond lengths in methyl substituted thiacycloheptyne and cyclooctyne (1.232 Å and Å at the MP2 level, respectively) compare favorably with those determined from X-ray data in 20 (1.21 Å) synthesized by Krebs et al. 37 and in 22 (1.23 Å) reported by Krebs et al. 40 In these compounds the calculated C C C bond angles on 19, 37 20, 37 and (143.1, 146.3, and 159.8, respectively) are in close agreement with the corresponding experimental values (146, 145.8, and 158.5, respectively). In any event, due to the agreement between the MP2/6-311G(d,p) theory and the available experimental data, one would therefore expect that the same relative accuracy should apply to the geometries as well as the energetics predicted for the other heterocyclic alkyne species. Second, as in the case of monocyclic alkynes and bridged bicyclic alkynes, the larger the C C C bond angle ( ), the smaller the C C bond length (r), and the larger the singlet-triplet splitting ( E st ) of the cycloalkyne. Our model calculations for heterocycloalkyne reactants confirm this prediction and suggest a decreasing trend in the C C bond length for 16 (1.244 Å) > 17 (1.236 Å) 18 (1.236 Å) > 19