Toward Understanding the Decomposition of Carbonyl Diazide (N3

Jul 28, 2015 - Carbonyl diazide, (N3)2CO (I), is a highly explosive compound. The isolation of the substance in a neat form was found to provide uniqu...
5 downloads 11 Views 2MB Size
Download PDF
Article pubs.acs.org/JPCA

Toward Understanding the Decomposition of Carbonyl Diazide (N3)2CO and Formation of Diazirinone cycl-N2CO: Experiment and Computations Hongmin Li,† Dingqing Li,† Xiaoqing Zeng,*,† Kun Liu,¶ Helmut Beckers,‡ Henry F. Schaefer, III,§ Brian J. Esselman,*,⊥ and Robert J. McMahon⊥ †

College of Chemistry, Chemical Engineering and Materials Science, Soochow University, Jiangsu 215123, P. R. China College of Chemistry, Tianjin Normal University, Tianjin 300387, P. R. China ‡ Institut für Chemie und Biochemie, Freie Universität Berlin, D-14195 Berlin, Germany § Center for Computational Quantum Chemistry, University of Georgia, Athens, Georgia 30677, United States ⊥ Department of Chemistry, University of Wisconsin, 1101 University Avenue, Madison, Wisconsin 53706-1322, United States ¶

S Supporting Information *

ABSTRACT: Carbonyl diazide, (N3)2CO (I), is a highly explosive compound. The isolation of the substance in a neat form was found to provide unique access to two other high-energy molecules, namely, N3−NCO (III) and cyclN2CO (IV), among the decomposition products of (I). To understand the underlying reaction mechanism, the decomposition reactions including the thermal conversion of two conformers of (I) were revisited, and the potential energy surface (PES) was computationally explored by using the methods of B3LYP/6-311+G(3df) and CBS-QB3. The most stable syn−syn structure (I) readily converts into the syn−anti conformer (ΔHexptl = 1.1 ± 0.5 kcal mol−1), which undergoes decomposition in two competing pathways: a concerted path to N3−NCO (III) or a stepwise route to (III) via the nitrene intermediate N3C(O)N 1(II). The calculated activation barriers (Ea) are almost the same (∼33 kcal mol−1, B3LYP/6-311+G(3df)). Further decomposition of (III) occurs through a concerted fragmentation into 2 N2 + CO with a moderate Ea of 22 kcal mol−1, and this process is compared to the isoelectronic species N3−N3 → 3 N2 (Ea = 17 kcal mol−1) and OCN−NCO → N2 + 2 CO (61 kcal mol−1). No low-energy pathway leading to (IV) was found on the singlet PES. However, the intervention of triplet ground-state 3(II) from the initially generated 1(II) through an intersystem crossing (ISC) offers a likely approach to (IV); that is, 3(II) can decompose in a concerted process (Ea = 30 kcal mol−1) by eliminating one N2 to yield the disfavored OCNN 3(VI). A careful intrinsic reaction coordinate analysis and a combined energy scan of the N−C−N angle reveals a bifurcation point on this triplet PES, which allows a spin crossover to the singlet PES along the reaction coordinate and eventually leads to the formation of the metastable diazirinone (IV).

■

polyethylene glycol8 were recently reported. Similar to other carbonyl azides, the synthesis of carbonyl diazide will greatly aid further studies of its rich chemistry. Carbonyl diazide has a melting point of 16 °C. The solid sublimes easily in vacuo, and its vapor pressure at ice−water temperature is 5.6 mbar. The neat substance revealed, however, an unexpected kinetic stability, as no decomposition was observed with the neat liquid as well as the gaseous substance at room temperatures. Thermal decomposition of the azide in the gas phase occurred at ∼500 °C. In addition to products N2 and CO, a small cyclic molecule bearing novel structural and

INTRODUCTION The first synthesis of the high-energy carbonyl diazide, (N3)2CO (I), was reported in the 1920s.1,2 However, its extremely explosive nature discouraged further attempts to isolate and study the properties of the neat substance.3,4 In 2010, neat carbonyl diazide was obtained as a byproduct from the reaction of FC(O)Cl and NaN3, which mainly yields the monoazide FC(O)N3.5 Two years later, pure diazide was prepared quantitatively from the solvent-free reaction of gaseous FC(O)N3 with solid NaN3.6 However, these synthetic routes using FC(O)Cl as starting material have some serious disadvantages, as FC(O)Cl is a highly toxic gas at room temperature and not easily accessible in ordinary laboratories. Alternative synthetic routes from commercially available triphosgene, OC(OCCl3)2, with tetra-n-butylammonium azide (TBAN3) in ether solution7 or sodium azide in nonvolatile © 2015 American Chemical Society

Received: May 13, 2015 Revised: June 30, 2015 Published: July 28, 2015 8903

DOI: 10.1021/acs.jpca.5b04586 J. Phys. Chem. A 2015, 119, 8903−8911

Article

The Journal of Physical Chemistry A

species including N3C(O)N (II), N3−NCO (III), and cyclN2CO (IV) are discussed.

bonding properties, namely, diazirinone (cycl-N2CO, IV), was being identified among the flash pyrolysis products.6,9 Photolytic decomposition of (N3)2CO in solid noble gas matrices has also been studied.10 A stepwise decomposition of the azide via a triplet nitrene intermediate N3C(O)N 3(II) was observed upon UV irradiation with light of λ = 255 nm (Figure 1). The nitrene rearranges into N3−NCO (III) under visible

■

EXPERIMENTAL SECTION Caution! Carbonyl diazide, (N3)2CO, was found to be an extremely explosive and shock-sensitive compound in the liquid and solid states. Safety precautions must be taken, including face shields, leather gloves, and protective leather clothing. Work only with small quantities of material. Ignoring safety precautions can lead to serious injuries. Sample Preparation and Matrix Infrared Spectroscopy. Carbonyl diazide, (N3)2CO, was prepared from FC(O)Cl and NaN3 according to the procedures of reference.5,6 Matrix IR spectra were recorded on a Fourier transform infrared spectrometer as described in the literature.6,9,10 Briefly, the gaseous sample was prepared by passing argon gas through a cold U-trap (−65 °C) containing ca. 10 mg of (N3)2CO. The resulting mixture ((N3)2CO/inert gas ≈ 1:1000 estimated) was passed through a heated quartz furnace with a nozzle (i.d. 1 mm) at a flow rate of 2 mmol/h of argon or neon, followed by deposition on the matrix support (argon 16 K; neon 6 K). Details of the matrix apparatus have been described elsewhere.23 Photolysis experiments were performed using an ArF excimer laser (Lambda-Physik). Quantum Chemistry. All calculations were performed using the Gaussian 03 software package. 24 Geometry optimizations were performed by using the B3LYP method25 at the 6-311+G(3df) basis set. The complete basis set composite method (CBS-QB3)26 was used for further calculation of thermochemical properties. The CBS-QB3 procedures include geometry optimization and frequency calculations at the B3LYP/6-311G(d, p) level followed by CCSD(T)/6-31+G(d′), MP4SDQ/6-31+G(d, p), and MP2/6311+G(2df, 2p) single-point energy calculations with CBS extrapolation. Local minima were confirmed by harmonic vibrational frequencies, which also provided zero-point vibrational energy (ZPVE) corrections. The transition states were characterized by a single imaginary frequency, and the connection of each transition state was checked by intrinsic reaction coordinate (IRC) calculations at the B3LYP/6311+G(3df) level.27,28 Temperature corrections were all made at 298.15 K. Calculations performed using B3LYP/6-31G(d) or CCSD/cc-pVDZ afforded similar results to those described herein.29 There have been considerable efforts for developing efficient minimum energy crossing point (MECP) optimization algorithms.30−34 In this work, the MECP is located at the B3LYP/6-311+G(3df) level using the Newton−Lagrange method, which was introduced by Koga and Morokuma.30 The algorithm is based on the minimization of the Lagrangian function L(Rλ) = E1(R) − λ[E1(R) − E2(R)], where R is nuclear coordinates, E1(R) and E2(R) are the energies of the states presently considered as functions of R, and λ is Lagrange multiplier. The energies, energy gradients, and Hessian matrixes of both of the two states need to be calculated, and the stationary point is found on the seam of intersection. These calculations are treated using a homemade program LookForMECP (version 1.0). This program can be obtained from the authors upon request. The early version of this program has been used successfully to search the MECP.35−38 According to Morokuma’s work,30 there are three characteristics of the MECP between two potential surfaces. (1) There is an equal energy of the two states; (2) There is a same geometry

Figure 1. Observed photolysis (blue) and thermal (red) decomposition routes starting from (N3)2CO.6,9,10 Irradiation conditions used in photolysis experiments and temperatures at which the thermal decay occurs are indicated. N2CO isomers include ONCN, ONNC, and NOCN.

irradiation of λ > 455 nm. Further decomposition of (III) to CO + 2 N2 was observed with subsequent irradiation of λ > 335 nm. In contrast to the pyrolysis experiment, diazirinone (IV) was not found among the photolysis products of (I). Similar to the isoelectronic analogues OCCO11,12 and NNNN,13 cycl-N2CO (IV) is a small molecule of fundamental importance in chemistry.14 Quantum-chemical calculations predict a moderate barrier for dissociation of singlet diazirinone (IV) to CO and N2 of ∼25 kcal mol−1, allowing its experimental observation.15−17 The lowest triplet state is sufficiently higher in energy to prevent singlet−triplet crossing. Previous attempts to generate metastable (IV) from organic precursors failed, however, and only decomposition products N2 and CO were observed.18−22 The detection of (IV) among the flash pyrolysis products of carbonyl diazide (I) encouraged us to isolate small amounts (∼5 mg) of neat (IV) as a brownish-yellow solid at low temperatures after tedious work with the highly explosive diazide precursor.6 The yield of (IV) was estimated to be merely 1% based on the diazide precursor. Surprisingly, at room temperature this solid evaporates into a violet gas, which in the dark slowly decomposes into N2 + CO with a half-life time of ∼30 h in a clean quartz container. Upon UV laser photolysis, the NN bond in (IV) is cleaved, and the ring converts into the chain isomers ONCN, ONNC, and NOCN, while visible light irradiation (λ > 530 nm) leads to decomposition into N2 and CO.6 The potentially rich fundamental and applied chemistry of (IV) prompted us to probe the mechanism of its formation from (I), so as to improve its synthesis. Herein, we revisited the flash pyrolysis experiment of (N3)2CO (I) and computationally explored the potential energy surface (PES) for its decomposition by using B3LYP/6-311+G(3df) and CBS-QB3 methods. The mechanisms for the formation of intermediate 8904

DOI: 10.1021/acs.jpca.5b04586 J. Phys. Chem. A 2015, 119, 8903−8911

Article

The Journal of Physical Chemistry A

When the temperature of the furnace was increased to 400 °C, the abundance of syn−anti (I) was estimated to be ∼34%. However, decomposition of (I) also occurred at this temperature, as indicated by the appearance of weak IR bands of CO (2140 cm−1) and cycl-N2CO (2034 cm−1).6 Further increasing the furnace temperature to 500 °C caused more decomposition and also an increase of the abundance to 39%. Assuming an equal decomposition rate of the two conformers at higher temperatures, a van’t Hoff plot was made for the equilibria of the two low-energy conformers. The linear fit is poor; nevertheless, an enthalpy difference of 1.1 ± 0.5 kcal mol−1 between the two conformers was roughly estimated. This value is close to the 1.8 kcal mol−1 predicted at the CCSD(T)/ccpVTZ level.7 Decomposition of (N3)2CO (I). Thermal decomposition of (I) occurs at ∼400 °C, and the main products are CO and CO2, with traces of HNCO, N3, OCNN3 (III), and cycl-N2CO identified among the flash vacuum pyrolysis products.6,9 A typical IR spectrum of the matrix-isolated pyrolysis products of (I) in neon dilution (ca. 1:500) at 500 °C is compared with that of the diazide precursor and shown in Figure 3. The yield of

structure of the two states; (3) The ratios of the energy gradients between the two states are a constant. The third characteristic can be used to testify that the obtained MECP is a real minimum and not common crossing point (CP) on the crossing seam hypersurface.

■

RESULTS AND DISCUSSION Conformers of (N3)2CO (I). From IR spectroscopy studies,5,7 two conformers of (I) were detected in the roomtemperature gas-phase spectrum, denoted as syn−syn and syn− anti depending on the relative conformation (syn or anti) of the CO and two N3 groups. The abundance of the higher-energy syn−anti conformer in the gas phase at room temperature was estimated to be 12%.5 Recently,7,39,40 the PES for the interconversion between the conformers of (I) was studied using the high-level ab initio CCSD(T)/cc-pVTZ method, which predicts an energy difference of 2 kcal mol−1 between these two conformers with a barrier (Ea) of ∼10 kcal mol−1 for the conversion from syn−syn to syn−anti. A third nonplanar anti−anti conformer of (I) with C2 symmetry, was also found to be a true minimum.7,39 However, because it is much higher in energy than syn−syn (I; ΔE = 12 kcal mol−1), its contribution to the room-temperature spectrum of the diazide is almost negligible and therefore is not further considered in the present work. The well-resolved CO stretching bands for the two conformers of (I) in the matrix IR spectrum (Figure 2) can be

Figure 3. (upper trace) Ne-matrix (6 K) IR spectrum of (N3)2CO (I); (lower trace) Ne-matrix IR spectrum of pyrolysis (500 °C) products of (I). The IR bands of syn−syn (I) and syn−anti (I′) conformers of (N3)2CO (I), OCNN3 (III), and cycl-N2CO (IV) are labeled, and the bands of H2O are marked by asterisks.

cycl-N2CO (IV) was improved by increasing the pyrolysis temperature from 400 to 500 °C, and neon was found to have better performance than argon as the dilution gas for the production of (IV). In contrast, the yield of OCNN3 (III) decreased at higher temperatures, while the yields of HNCO and N 3 increased. The formation of CO 2 from the decomposition products of (I) is ascertained by using a 13Clabled (N3)2CO, with which 13C-labled CO2 was obtained (Figure S1 in the Supporting Information), implying the occurrence of bimolecular reactions under the pyrolysis conditions. Photolytic decomposition of matrix-isolated (I) under UV light (255 nm) irradiation yielded a mixture of triplet nitrene N3C(O)N (II), OCNN3 (III), CO, and NOCN, no cycl-N2CO (IV).10 A recent study of the photochemistry of (IV) in solid noble gas matrices showed that (IV) was quickly destroyed, producing N2 and CO under visible irradiation. In contrast, ArF excimer laser (193 nm) irradiation caused transformation into the open-chain isomers ONCN, ONNC, and NOCN.6 Therefore, the question arises as to whether (IV) was initially

Figure 2. Ar-matrix IR spectra (1755−1705 cm−1) of the IR bands at 1747.4 (normalized) and 1713.0 cm−1 of (N3)2CO (I) in the temperature range of 298 (black)−773 K (yellow). (inset) Van’t Hoff plot obtained from the ratios of the integrated areas for the two IR bands.

used to estimate its conformational equilibrium. A mixture of gaseous (I) diluted in argon (ca. 1:1000) was heated by passing through a heated quartz furnace (i.d. 1.0 mm), and the resulting mixture was immediately quenched as solid matrices. The IR spectra obtained at different furnace temperatures are shown in Figure 2, which reveals a significant increase in the relative band intensity of the CO stretching band (1747 cm−1) of the higherenergy syn−anti conformer at the higher temperature. This increase corresponds to a change of its relative abundance from 12% at 25 °C to 27% at 101 °C, as estimated from integrated absorbances and computed IR intensities for the two CO stretching vibrations located at 1747 (syn−anti) and 1713 cm−1 (syn−syn). 8905

DOI: 10.1021/acs.jpca.5b04586 J. Phys. Chem. A 2015, 119, 8903−8911

Article

The Journal of Physical Chemistry A

Figure 4. Computed relative energies (kcal mol−1) of the stationary points and transition states (TS) on the singlet PES of (I) at the B3LYP/6311+G(3df) level (CBS-QB3 energies in parentheses). The intersystem crossing (ISC) of N3C(O)N (II) is indicated by a blue arrowhead.

formed from the photolytic decomposition of (I) and then converted into NOCN under further irradiation. To answer the question, an ArF excimer laser (193 nm, 1.5 mJ, 3 Hz) was applied to the matrix-isolated (N3)2CO. The same decomposition products were obtained, but the yield of the nitrene intermediate (II) was significantly increased, enabling the enrichment of the Curtius-rearrangement41 product (III) in matrices under subsequent irradiation of >335 nm. Further photolysis of (III) using the laser furnished only the open-chain isomers rather than (IV). Theoretical Decomposition PES of (N3)2CO (I). The complete decomposition pathway from (N3)2CO (I) to 3 N2 + CO via N3C(O)N (II) and N3−NCO (III) was explored on the singlet PES at the B3LYP/6-311+G(3df) and CBS-QB3 levels of theory, and the results are summarized in Figure 4. Optimized structures for the minima and transition states (TS) on the singlet PES are given in Figures 5 and 6,

Figure 6. Optimized structures of transition states (TS) on the singlet PES with the B3LYP/6-311+G(3df) method. Bond lengths are given in angstroms. Nitrogen, carbon, and oxygen atoms are shown in blue, gray, and red, respectively. Molecular symmetries are shown in parentheses. The corresponding bond angles are given in Figure S3 in the Supporting Information.

calculated by using the more demanding ab initio methods CCSD/cc-pVDZ and CCSD(T)/cc-pVTZ, respectively. Decomposition of carbonyl azides RC(O)N3 has been the topic of several computational studies.42−44 Generally, two nearly degenerate conformers of carbonyl azides were found to be true minima on the PES, and the lower-energy syn structure favors a concerted Curtius-rearrangement to the corresponding isocyanate RNCO without involvement of the carbonyl nitrene intermediate RC(O)N. In contrast, the anti conformer often exhibits a higher decomposition barrier that first forms RC(O) N upon N2 elimination, and the nitrene is susceptible to rearrangement to RNCO. Steric factors can probably explain the different pathways for the two conformers, as the migrating substituent R in the syn azide can more favorably attack the αnitrogen during the concerted fragmentation of the Nα−Nβ bond. The latter attack is sterically hindered in the anti-azide conformer, in which an intramolecular interaction between the carbonyl oxygen and Nα instead is preferred, and this interaction leads to the formation of a singlet carbonyl nitrene intermediate with an oxazirine-like structure.45

Figure 5. Optimized structures of minima on the singlet PES at the B3LYP/6-311+G(3df) level of theory. Bond lengths are given in angstroms. Nitrogen, carbon, and oxygen atoms are shown in blue, gray, and red, respectively. Molecular symmetries are shown in parentheses. The corresponding bond angles are given in Figure S2 in the Supporting Information.

respectively. Consistent with recent computational studies,7,39 the global minimum syn−syn conformer of (I) is lower in energy than syn−anti by ΔE = 2 kcal mol−1, and the activation barrier (Ea, TS1) for the conversion from syn−syn to syn−anti was predicted to be 9 kcal mol−1 at the B3LYP/6-311+G(3df) level of theory (Figure 4). Both values are close to those (ΔE = 2 kcal mol−1, Ea = 10 kcal mol−1)7 and (ΔE = 1.8 kcal mol−1)8 8906

DOI: 10.1021/acs.jpca.5b04586 J. Phys. Chem. A 2015, 119, 8903−8911

Article

The Journal of Physical Chemistry A

Figure 7. Computed PES for the decomposition of N3−N3, N3−NCO, and OCN−NCO. Relative energies (in kcal mol−1) and structures (key bond lengths in angstroms) of the transition states as obtained at the B3LYP/6-311+G(3df) level (CBS-QB3 energies in parentheses) are indicated. Nitrogen, carbon, and oxygen atoms are shown in blue, gray, and red, respectively.

is in agreement with the results of previous studies.46,47 Interestingly, while both molecules prefer a concerted decomposition pathway, the structural changes between their nonplanar transition state and the trans-bent minimum are quite different (Figure 7). In the transition state for fragmentation of N3−N3, the central N−N bond (1.295 Å) is significantly shortened at the expense of the two neighboring N−N bonds (1.431 Å), which are elongated in comparison to the respective bonds in the planar N3−N3 minimum (1.439 and 1.238 Å, respectively). In contrast, the decrease of the central N−N bond length in the fragmentation transition state for N3− NCO (TS 1.296 Å, minimum 1.403 Å) is associated primarily with the elongation of the neighboring N−N bond (TS 1.648 Å, minimum 1.244 Å), whereas the C−N bond length (TS 1.225 Å, minimum 1.218 Å) is increased only slightly. These distinct structural changes suggest that N3−N3 dissociates by simultaneous elimination of both terminal N2 fragments, while the TS for concerted fragmentation of N3−NCO is mainly associated with N2 elimination. This finding is consistent with previous computational predictions of a barrierless fragmentation of singlet OCNN chain.15,16 A similar fragmentation has also been predicted for the isoelectronic open-chain singlet species OCCO11,12 and NNNN.48 Similar to OCNN,49 openchain structures for these species may only exist on the triplet PES. A concerted fragmentation of OCN−NCO (Figure 7) is mainly associated with the elongation of one N−C bond. The corresponding barrier of 61 kcal mol−1 is significantly higher than that for N3−NCO (22 kcal mol−1), and even higher than the N−N bond dissociation energy (55 kcal mol−1) in OCN− NCO. Hence, OCN−NCO is a kinetically viable molecule,50 which thermally decomposes into two NCO radicals rather than by fragmentation into one N2 and two CO. This prediction is consistent with its previous synthesis by flash pyrolysis of N3C(O)NCO, where the CO and NCO radicals were detected only as minor byproducts.51 OCN−NCO has a half-life of ∼10 min in an argon-diluted gas mixture at room temperature. However, OCN−NCO polymerizes rapidly in condensed phases, even at 130 K.51 Because of the lower dissociation barrier, the capture of N3−NCO should be more difficult, and our attempts to isolate neat N3−NCO by trapping the pyrolysis products of (N3)2CO at low temperatures failed. Formation Mechanism of cycl-N2CO (IV). As shown in Figure 4, the formation of cyclic (IV) from nitrene (II) needs to surmount a barrier of 35 kcal mol−1 (TS5), which is much

Interestingly, all attempts to locate a TS for the decomposition of the syn−syn conformer (I) failed. Instead, two transitions states were found for the syn−anti conformer, namely, TS2 and TS3 (Figure 4), which correspond to concerted and stepwise decompositions of the diazide, respectively. TS3 is computed to be slightly higher (1 kcal mol−1) above TS2 at the B3LYP/6-311+G(3df) level. However, at the CBS-QB3 level TS3 was found to lie lower than TS2 by 1 kcal mol−1. The fairly close energies of TS2 and TS3 suggest that both processes can occur concurrently during the thermal decomposition of (I). For the rearrangement of the oxazirene-like singlet carbonyl nitrene N3C(O)N (II) to N3− NCO (III), a small barrier (TS4, 8 kcal mol−1) is predicted. The computational results are consistent with the experimental observation of (III) among the pyrolysis products of (I). By contrast, the triplet ground state carbonyl nitrene intermediate 3 (II) has been obtained only by UV photolysis of (I) isolated in noble gas matrices.10 The molecular structure of TS2 is similar to that of TS1, in the sense that one of the N3 ligands is orientated nearly perpendicular to the O−C−Nα plane. The most striking differences are related to the C−Nα and Nα−Nβ bond lengths (Figure 6), and the ∠NαCN′α bond angles. In TS2 ∠NαCN′α (92.0°) is much smaller than that in TS1 (109.6°), facilitating a concerted migration of the N3 ligand to form (III). In TS3, the related ∠NαCN′α angle is much larger (126.0°) and will be further increased to 136.4° when singlet nitrene (II) is formed, which is stabilized by an intramolecular Nα···O interaction (∠NαCO = 90.4°). The decomposition barrier from (III) to two N2 and CO is 22 kcal mol−1. It is slightly lower than that for cycl-N2CO (25 kcal mol−1), implying that this open-chain nitrogen-rich molecule would quickly decompose at room temperature. To get more insight into the thermal decomposition pathway, the energy surfaces for the unimolecular decompositions of two isoelectronic analogues, namely, N3−N3 and OCN−NCO, were also explored at the same level of theory, and the results are compared in Figure 7. In their most favorable conformations these two molecules adopt planar trans-bent structures like that of (III; Figure 5). The PES for the yet unknown N3−N3 molecule is quite similar to that of (III). Both molecules exhibit low barriers for fragmentation into the diatomic molecules but higher central N−N bond dissociation energies. In this series N3−N3 shows the lowest fragmentation barrier of 17 kcal mol−1. This barrier 8907

DOI: 10.1021/acs.jpca.5b04586 J. Phys. Chem. A 2015, 119, 8903−8911

Article

The Journal of Physical Chemistry A higher than that (8 kcal mol−1, TS4) for the competing Curtiusrearrangement of (II) into (III). Such a high barrier would rule out the generation of (IV) by pyrolysis of (I). Considering the rather small energy difference between the two transition states for the concerted (TS2) and stepwise (TS3) decomposition of (I), and also the small singlet−triplet energy gap (ΔEST = 8 kcal mol−1) for the nitrene intermediate (II), an alternative mechanism for the formation of cyclic (IV) via the triplet nitrene 3(II), can be envisioned. The small energy difference between TS2 and TS3 implies that the singlet nitrene 1(II) should be formed upon pyrolysis of (I). The small ΔEST value and the closed-shell electron configuration of 1(II) promote its rapid decay to the triplet ground state 3(II) through a spin− orbit allowed mechanism.52,53 As mentioned above, the closedshell configuration of 1(II) is stabilized by an intramolecular O → N interaction, as indicated by the small OCN angle of 90.4° (Figure 5), a common structural feature found for other singlet carbonyl nitrenes.54,55 In fact, formation of triplet carbonyl nitrenes by pyrolysis of the corresponding azides has been demonstrated for FC(O)N356,57 and alkoxycarbonyl azides.58,59 The PES for the decomposition of 3(II) was studied using the same theoretical methods. The results are shown in Figure 8, and molecular structures of the corresponding minima and transition states are reported in Figure 9.

As reported for other carbonyl azides,42−44 the low-energy conformer syn 3(II) favors a concerted decomposition to yield open-chain OCNN 3(VI), with a moderate activation barrier of 30 kcal mol−1. In contrast, for the higher-energy anti 3(II), a higher barrier of 35 kcal mol−1 is predicted. Anti 3(II) decomposes via an elusive triplet oxazirine intermediate cyclC(N)ON in a shallow minimum (3(V), Figure 9) to triplet nitrosyl cyanide ONCN, which is indeed observed among the photolysis products of (I).10 The intrinsic reaction coordinate analysis shows that the decomposition of syn 3(II) toward OCNN 3(VI) proceeds via cyclic 3(IV; Figure S5, Supporting Information), for which the computed IR spectrum shows one imaginary frequency. 3(IV) collapsed into 3(VI) when the C2v symmetry constraint was removed. Cyclic (IV) exhibits a positive vertical singlet−triplet energy gap, ΔEST = −31 kcal mol−1 (Figure 8), whereas the relative energies of the transition states for the decomposition of syn (II) on the singlet and triplet PES are reversed. That is, TS12 (triplet PES) lies ∼33 kcal mol−1 below the related TS6 (singlet PES). As a consequence, these two energy surfaces are expected to cross on the reaction pathway. Intersystem crossings between the lowest singlet and triplet energy surfaces have been studied for the isoelectronic systems N446 and C2O2,12 where it was found that the spin-forbidden decay resulting from the crossing between two states of different multiplicity may significantly reduce the predicted decomposition barrier, compared to when only a spin-allowed singlestate decay channel was considered. A thermal reaction that involves a spin crossover along the reaction coordinate may provide an unanticipated low-energy route to synthetically difficult intermediate species. As a further evidence for the expected singlet−triplet crossings energy scans of cyclic (IV) (varying the N−C−N angle in both the singlet and triplet states) were performed at the same theoretical level, and the results are shown in Figure 10. As shown in Figure 10, the two PESs cross several times. The most relevant spin crossover (C1) occurs at an NCN angle of ∼80°, where it is expected that a spin-forbidden ISC may become important and the reaction coordinate proceed downhill toward singlet 1(IV) rather than the spin-allowed

Figure 8. Relative energies (UB3LYP/6-311+G(3df), kcal mol−1) of the stationary points and transition states (TS) on the triplet PES of N3C(O)N 3(II) (CBS-QB3 energies in parentheses). The decomposition of cycl-N2CO (IV) on the singlet PES is indicated by red dashed lines.

Figure 9. Optimized structures of stationary points and transition states (TS) on the triplet PES of N3C(O)N 3(II) at the UB3LYP/6311+G(3df) level of theory. Bond lengths are given in angstroms. Nitrogen, carbon, and oxygen atoms are shown in blue, gray, and red, respectively. Molecular symmetries are shown in parentheses. The corresponding bond angles are given in Figure S4 in the Supporting Information.

Figure 10. Relaxed energy scans for cyclic (IV) for both the singlet (black curve) and the triplet state (blue curve) by varying the N−C− N angle at the B3LYP/6-311+G(3df) level of theory. Relevant potential crossings between these lowest-energy surfaces are indicated by C1 to C3. 8908

DOI: 10.1021/acs.jpca.5b04586 J. Phys. Chem. A 2015, 119, 8903−8911

Article

The Journal of Physical Chemistry A route to 3(VI). In fact, the NCN angle in the TS12 (97.1°) is close to that required for C1 (80°). The higher energy of the competing transition state TS11 (Figure 8), which requires a widening of the NCN angle to yield 3(V), excludes this elusive intermediate being formed during the thermolysis of (N3)2CO (I). Note that singlet (V) is not a minimum on the PES (Figure 10). To verify the existence of this crossing point (C1), the singlet−triplet vertical energies (ΔEST) for the species with structures around C1 were calculated, and the results clearly show a change of the sign for ΔEST by varying the NCN angle from 85 to 75°. To obtain the exact molecular structure at C1, the MECP is located at the B3LYP/6-311+G(3df) level using the Newton−Lagrange method (Table S1 in the Supporting Information).30 The NCN angle in this C2v symmetric MECP (C1) was calculated to be 80.2° with C−N and C−O bond lengths of 1.299 and 1.276 Å, respectively. Therefore, thermal decomposition of syn 3(II) provides a unique approach to cyclic (IV). However, the competing Curtius rearrangement of syn 1 (II) and the thermal decomposition of (IV) under the pyrolysis conditions account for the low yield of (IV) that can be obtained from the pyrolysis of (I). Spin−orbit coupling may also reduce the thermal stability of both the cyclic 1(IV) and open-chain 3(VI) structures. As shown in Figure 10 the two lowest surfaces cross near the TS9, the transition state for the spin-allowed thermal decomposition of 1(IV). This crossing may result in a slightly reduced barrier for the thermal decomposition of 1(IV), which may instead proceed via initial spin crossing at C2. The existence of this MECP (C2) is also ascertained by locating its structure (rOC = 1.155 Å, rCN = 1.290 Å, rNN = 1.338 Å, ∠NCN = 49.2°, Table S1 in the Supporting Information). The NCN angle is larger than that in the transition state TS9 (42.9°), and the MECP is about merely 2 kcal mol−1 lower in energy. This means that the ring-opening decomposition of 1(IV) might provide access to the short-lived 3(VI)49 through the surface crossing. The possible existence of a third crossing point (C3) was predicted by the relaxed energy scans of the NCN angle (Figure 10), however, further theoretical study is required for locating its structure.

decomposition of 3(II) and energy scans for the ring-opening of (IV) in both singlet and triplet states suggest the likelihood of an important spin crossover from triplet to the singlet PES. That is, first 3(II) eliminates one N2 molecule and forms an elusive oxazirine intermediate cycl-C(N)ON 3(V), then the triplet PES crosses to the singlet on the way to open-chain NNCO 3(VI), by reducing the N−C−N angle, and this crossing leads to the formation of cycl-N2CO (IV).

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpca.5b04586. Experimental matrix IR spectra, calculated bond angles and IRC pathway, energy gradients of the minimum energy crossing points, atomic coordinates, energies, and IR frequencies for all species discussed in the paper. (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. Phone: (86)512-65883583. (X.Q.Z.) *E-mail: [email protected]. (B.J.E.) Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundation of China (Nos. 21372173, 21422304, and 21172173), the Project of Scientific and Technologic Infrastructure of Suzhou (No. SZS201207), the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD), and the Beijing National Laboratory for Molecular Sciences (No. 20140128). This work at the Univ. of Wisconsin was supported by the U.S. National Science Foundation (Nos. CHE-1011959 and CHE-1362264) and NSF support for chemistry department computational facilities (No. CHE0840494). X.Z. gratefully acknowledges Prof. H. Willner for the generous support of research stay in Wuppertal. We gratefully acknowledge Dr. A. B. Alekseev for helpful discussions, Dr. S. Tong for help with calculations, Prof. T. Benter for using the ArF excimer laser. The authors are very grateful to the reviewers’ invaluable comments and suggestion.

■

CONCLUSIONS The decomposition of the high-energy carbonyl diazide, (N3)2CO (I), was experimentally revisited, and the corresponding PESs were computationally explored. According to the DFT computations, the most stable syn−syn conformer of (I) first isomerizes to the syn−anti conformer, and their energy difference (ΔH) was experimentally found to be 1.1 ± 0.5 kcal mol−1. This higher-energy conformer initiates decomposition by losing one N2 molecule in either a concerted or stepwise pathway via a nitrene intermediate N3C(O)N (II). Both conformers yield N3−NCO (III), and the activation barriers were calculated to be rather close, that is, 32 and 33 kcal mol−1, respectively. Further decomposition of metastable (III) into 2 N2 and CO was found to proceed stepwisely with a moderate activation barrier of 22 kcal mol−1. No low-energy (