The structure and bonding of H2NNO, the simplest N-nitrosamine, and a key intermediate in deNOx processes, have been precisely characterized using a combination of rotational spectroscopy of its more abundant isotopic species and high-level quantum chemical calculations. Isotopic spectroscopy provides compelling evidence that this species is formed promptly in our discharge expansion via the NH2 + NO reaction and is collisionally cooled prior to subsequent unimolecular rearrangement. H2NNO is found to possess an essentially planar geometry, an NNO angle of 113.67(5)°, and a N–N bond length of 1.342(3) Å; in combination with the derived nitrogen quadrupole coupling constants, its bonding is best described as an admixture of uncharged dipolar (H2N–N=O, single bond) and zwitterion (H2N+=N–O, double bond) structures. At the CCSD(T) level, and extrapolating to the complete basis set limit, the planar geometry appears to represent the minimum of the potential surface, although the torsional potential of this molecule is extremely flat.

Because nitrogen oxides such as NO and NO2 are critical precursors in the formation of compounds such as ozone1 and acid rain2 that have deleterious health effects, there is great interest in developing efficient processes to remove NOx from combustion sources. Effective NOx conversion is particularly challenging for lean-burning gasolines such as diesel3 and in stationary combustion reactors, where O2 is typically present in large excess relative to NOx. In these situations, one of the most commonly used methods to control NOx levels is thermal deNOx4–7 in which NOx is selectively reduced to N2 without the use of a catalyst by adding NH3.8 It is widely accepted that thermal deNOx is initiated by the conversion of NH3 to the amino radical (NH2).9 This species then reacts with the NO radical to produce a reactive but strongly bound H2NNO adduct that eventually rearranges and decomposes to yield N2 + H2O as indicated in Fig. 1. For this reason, the NH2 + NO reaction has been extensively studied by direct kinetic measurements10–16 and a variety of theoretical approaches.8,17–20 In addition to N2 + H2O, however, two other product channels are possible, resulting in NNH + OH and NNO + H2, respectively. Although NNH + OH is approximately thermoneutral, it is the only one of the three that is chain branching, in which the tenuously stable NNH radical subsequently dissociates on the time scale of 1 μs or less21 to N2 + H. Among the two chain terminating reactions, the NNO + H2 channel is thought to be relatively unimportant relative to N2 + H2O because it is significantly less exothermic (by at least 320 kJ mol−1, using 0 K heats of formation from the Active Thermochemical Tables (ATcT)22,23). Regardless, because the fate of NOx depends on the relative importance of the various exit channels, knowledge of the branching ratio between the three channels is highly relevant; indeed, this has been the main motivation for many kinetic studies. Although a wide scatter in the branching ratio has been reported, temperature is known to play a significant role in determining the relative ratios, with chain branching becoming more important at higher temperatures. Owing to the complexity of the reaction pathways and the presence of barrierless radical-radical recombination reactions, it has proven challenging to accurately predict the temperature dependence and branching ratios. Nevertheless, good agreement has been achieved using transition state theory in combination with the electronic structure and Rice-Ramsperger-Kassel-Marcus (RRKM) theories.20,24

FIG. 1.

Energetics of the important reactive pathways for the NH2 + NO reaction. Values are adapted from previous determinations using theoretical and experimental methods; those indicated with the superscript a are from Diau and Smith using modified Gaussian-2 (G2M) theory,17 while those with the superscript b are experimental values from JANAF Thermochemical Tables;78 all energies are relative to that of H2NNO. Blue horizontal lines represent minimum energy structures, while red horizontal lines denote barrier heights between minima. The various conformers of HNNOH are represented by the average conformer energy for brevity.

FIG. 1.

Energetics of the important reactive pathways for the NH2 + NO reaction. Values are adapted from previous determinations using theoretical and experimental methods; those indicated with the superscript a are from Diau and Smith using modified Gaussian-2 (G2M) theory,17 while those with the superscript b are experimental values from JANAF Thermochemical Tables;78 all energies are relative to that of H2NNO. Blue horizontal lines represent minimum energy structures, while red horizontal lines denote barrier heights between minima. The various conformers of HNNOH are represented by the average conformer energy for brevity.

Close modal

In this context, a variety of theoretical methods have been used to calculate the potential energy surface for the NH2 + NO reaction.8,17,20 The consensus that has emerged from this body of work is that there may be as many as nine intermediates, including five isomers that are located in deep, well-defined minima, and are separated by sizable barriers to isomerization or dissociation. A schematic of the energy surface is shown in Fig. 1. Once the H2NNO adduct is formed, the internal H rearrangement can yield HNNOH in as many as four distinct conformers because both the N–N and N–O bonds are predicted to possess significant double bond character and are therefore fairly rigid.

Although theoretical studies find that H2NNO plays a central role as the initial intermediate in the NH2 + NO reaction, this species has never been isolated nor spectroscopically characterized in the gas phase, its existence has been inferred from neutralization-reionization mass spectrometric studies,25 and separately from mass spectroscopic studies alone26 or in combination with ab initio calculations.27 The only spectroscopic evidence to date is provided by measurements in very low temperature inert matrices by two groups. In the first study, Crowley and Sodeua28 detected infrared absorption bands of H2NNO and two of its isotopic variants in Ar and N2 matrices at 4.2 K, with the measured frequencies in good agreement with predictions. These findings were confirmed and extended in a later study by Jacox and Thompson,29 who measured up to five vibrational frequencies for as many as six isotopic species in a Ne matrix.

In addition to its purely mechanistic importance, questions still persist regarding the geometry of H2NNO (Fig. 2), a factor which might affect the rate of isomerization to the chemically activated HNNOH in the trans, cis configurations (Fig. 1). This isomer undergoes further rearrangement to the cis, trans conformer before dissociating via one of the three exit channels.17,30–34 The calculated ground state geometry is sensitive to the level of theory, with the literature somewhat divided as to whether H2NNO adopts a planar or non-planar structure. Density functional theory [e.g., B3LYP/6-311G(d,p)], for example, predicts a planar structure, but wavefunction-based calculations [i.e., MP2 and complete active space self-consistent field (CASSCF)] generally predict non-planar structures, with dihedral HNNO angles ranging from a few degrees to nearly 20°. Walch,31 for example, has suggested that the inclusion of polarization functions results in a non-planar equilibrium structure, with the planar structure as a saddle point on the potential energy surface.

FIG. 2.

Experimental (r0) structure of H2NNO derived from a least-squares fit to the rotational constants of the normal, five singly substituted isotopic species, and D2NNO, assuming a planar geometry. Hc and Ht refer to cis and trans positions of the two H atoms with respect to the O atom. Numbers in parentheses are 1σ uncertainties in units of the last digit (see Table V).

FIG. 2.

Experimental (r0) structure of H2NNO derived from a least-squares fit to the rotational constants of the normal, five singly substituted isotopic species, and D2NNO, assuming a planar geometry. Hc and Ht refer to cis and trans positions of the two H atoms with respect to the O atom. Numbers in parentheses are 1σ uncertainties in units of the last digit (see Table V).

Close modal

Irrespective of its involvement in the deNOx process, H2NNO is the simplest N-nitrosamine, an important set of compounds that can be produced in food by nitrites but which are commonly carcinogenic to humans.35–38 Substituted derivatives of the form R1R2NNO, such as N-nitrosodimethylamine (CH3)2NNO, are generally more stable and their biological activity has been linked to molecular structure; the accessibility of the electron-rich nitrogen center by nucleophiles plays an important role in determining the efficiency of decomposition and enzyme activity.39,40 Another aspect of N-nitrosamine structure that is related to bioactivity is the correlation between the barrier to rotation about the N–N bond41 and the ability to act as a NO or nitrosonium ion (NO+) donor.42,43

The structure of H2NNO and larger N-nitrosamines can be represented by the two resonance structures shown in Fig. 3. The important distinction between these two idealized Lewis structures is the hybridization of the central nitrogen atom. This atom in the uncharged dipolar structure (I) is sp3 hybridized, which should result both in a non-planar geometry of the C2NNO atoms in larger alkyl-substituted nitrosamines, and a quadrupole coupling constant of zero at this atom; in contrast, this same atom has sp2 hybridization in the zwitterion structure (II), yielding a planar C2NNO geometry and a large, positive quadrupole coupling constant. For N-nitrosodimethylamine, both crystalline44 and gas-phase structures45 are consistent with a planar arrangement of the C2NNO atoms, in agreement with the 1H-NMR spectrum.46 These findings suggest that the bonding in this molecule is closer to the zwitterion structure.

FIG. 3.

Two possible resonance structures of N-nitrosamines.

FIG. 3.

Two possible resonance structures of N-nitrosamines.

Close modal

Because H2NNO is the basis for understanding how specific substituents subtly change the structure of the NNO group in larger nitrosamines, there is much to be gained by characterizing its chemical bonding and geometry in the gas phase. At the high spectral resolution (sub-ppm) that is commonly achieved by rotational spectroscopy, for example, the nitrogen quadrupole hyperfine structure is readily resolved, allowing coupling constants to be precisely determined at each nitrogen atom. This information can in turn be used to infer bond orders. Additionally, very accurate experimental bond lengths and angles can be derived for H2NNO, provided that the rotational spectra of multiple isotopic species are measured.

By means of Fourier transform (FT) microwave spectroscopy in combination with millimeter-wave/microwave double resonance techniques, the rotational spectrum of H2NNO and six of its rare isotopic species have been measured for high accuracy in a supersonic jet between 17 and 92 GHz. For each isotopic species, all three rotational constants have been determined to fractional accuracy of better than 1 ppm, and the diagonal elements of the quadrupole tensor at both nitrogen atoms have been determined to 2%. From the eQq(N) constants inferred along the bond axis, it is possible to the establish the N–N bond order and therefore assess if H2NNO is better described as an uncharged dipolar or zwitterion structure, or some combination of the two. The structure derived from the isotopic analysis provides a complementary but independent gauge of the N–N bond order, while the large number of isotopic species provides multiple constraints for planarity of the ground state through determination of inertial defects. Finally, CCSD(T) calculations have been performed with the goal of accurately predicting the equilibrium structure and torsional H2N-X potential of H2NNO, as well as to assess the question of quasiplanarity of the dynamic molecular structure.

FT microwave spectroscopy in combination with a supersonic jet was used to detect and measure the rotational spectrum of H2NNO and its rare isotopic species in the centimeter-wave band.47,48 Although the spectrometer operates between 5 and 43 GHz, transitions lying above this frequency can be measured by millimeter-wave/microwave double resonance,49,50 provided that a transition sharing either a common upper or lower level with the one sought falls within its operating range. Because the jet expands along the axis of the large Fabry-Peŕot cavity, linewidths are sharp (5 kHz FWHM) and frequencies are routinely measured to an uncertainty of 2 kHz or better for closed-shell molecules. For double resonance measurements, line positions can be determined to an uncertainty which is about an order of magnitude larger (25 kHz), but even here, the spectral resolution is still very high (1 ppm). With a sensitive receiver, molecules present at the ppb level can be detected.

H2NNO and its isotopic variants were generated by adding an electrical discharge source to the end of a standard pulsed valve. The electrical and geometrical characteristics of this source have been empirically refined by trial and error48 to produce many different types of reactive molecules, ranging from long carbon chains,51 free radicals,52 carbenes,53 and molecular ions.54,55 Reactive molecules are produced by expanding gas precursors heavily diluted to 1% or less in an inert buffer gas through this discharge source, prior to adiabatic expansion. Enough collisions with electrons occur in the throat of the expansion to break chemical bonds, whereby a wide range of new species are rapidly formed before the kinetic and rotational temperature of the fast moving jet drops to a few kelvin at the center of the Fabry-Peŕot cavity. As the molecules traverse the center of the cavity, they are excited with a short pulse (1 μs) of resonant radiation; the FT of the free induction decay yields the power spectrum. Since cavity FT microwave spectroscopy is a highly resonant technique, the instantaneous bandwidth of each measurement is only about 0.5 MHz. To achieve frequency agility and wide spectral coverage, the applied frequency and cavity mirrors are synchronously adjusted using computer control.

Lines of H2NNO were first found during an unsuccessful search for HNOO near 24 GHz. During that search, a closely-grouped, unidentified set of lines characteristic of a molecule possessing two or more nitrogen atoms was observed through a gas discharge consisting of a dilute mixture of NH3 and O2 in Ne. Subsequent tests demonstrated that these lines required the presence of both precursors, and double resonance experiments quickly established that the carrier was a nearly prolate asymmetric top, since a transition with an equally complex line pattern was found at nearly twice this frequency. Because additional lines were not found at any of the obvious sub-harmonic frequencies (e.g., 1/3, 1/5, etc.), the inferred rotational constant is only consistent with a heavy atom backbone containing three atoms, two of which must be nitrogen. The lines are also unaffected by the presence of a small magnetic field, an indication that the molecule has a closed-shell electronic configuration. Since the lines do not arise from NNO, the carrier must therefore contain an even number of hydrogen atoms. When N2 was used instead of NH3, the line intensity decreases considerably, but replacing O2 with NO at the same level of dilution yielded 20 fold stronger lines. Additionally, the lines are quite faint when starting from NNO and H2, strongly suggestive of a molecule that possesses a NNO heavy atom backbone but where both hydrogen atoms are bound to the terminal N atom.

The elemental composition and connectivity of H2NNO were ultimately established by isotopic substitution in which commercial samples enriched in 15N (15NH3 or 15NO) and D (ND3) were used instead of NH3 or NO; for H2NN18O, a gas mixture of NH3 and 18O2 was used. Although equal amounts of NH3 and ND3 in combination with NO were found to produce H2NNO and D2NNO almost exclusively, the two singly substituted deuterium species were also observed at a low level, at roughly 1/50th the intensity of either H2NNO or D2NNO, but still a sufficient abundance to observe their spectra with good signal-to-noise ratios (SNRs) as Fig. 4 demonstrates.

FIG. 4.

The fundamental a-type (10,100,0) rotational transition of d1,cis-HDNNO near 23.07 GHz, in which an electrical discharge was applied to a gas mixture consisting of equal quantities of NH3 and ND3 (0.4% each), and NO (0.2%) heavily diluted in Ne, in the throat of supersonic nozzle source. Nitrogen hyperfine structure from the two I = 1 nuclei is well resolved, but deuterium hyperfine structure, of order 10 kHz, is only partially resolved even at a spectral resolution of 0.1 ppm, as the inset illustrates. Each hyperfine transition is split into two Doppler components because the supersonic jet expands along the axis of the Fabry-Peŕot cavity; the rest frequency of each transition is simply the arithmetic average of these two frequencies, while the frequency separation between the doublets is proportional to the velocity of the jet. The spectrum is a concatenation of 12 individual scans, each of 0.4 MHz; the total integration time was approximately 12 min. The asterisk indicates the strong hyperfine transition that was monitored in the DR scan in Fig. 5.

FIG. 4.

The fundamental a-type (10,100,0) rotational transition of d1,cis-HDNNO near 23.07 GHz, in which an electrical discharge was applied to a gas mixture consisting of equal quantities of NH3 and ND3 (0.4% each), and NO (0.2%) heavily diluted in Ne, in the throat of supersonic nozzle source. Nitrogen hyperfine structure from the two I = 1 nuclei is well resolved, but deuterium hyperfine structure, of order 10 kHz, is only partially resolved even at a spectral resolution of 0.1 ppm, as the inset illustrates. Each hyperfine transition is split into two Doppler components because the supersonic jet expands along the axis of the Fabry-Peŕot cavity; the rest frequency of each transition is simply the arithmetic average of these two frequencies, while the frequency separation between the doublets is proportional to the velocity of the jet. The spectrum is a concatenation of 12 individual scans, each of 0.4 MHz; the total integration time was approximately 12 min. The asterisk indicates the strong hyperfine transition that was monitored in the DR scan in Fig. 5.

Close modal

The strongest lines of H2NNO were observed in NH3 and NO diluted in Ne, with a discharge potential of 900 V applied to the electrode furthest from the faceplate of the valve at a stagnation pressure 2.5 kTorr behind the valve. Under these conditions, the strongest hyperfine component in the fundamental rotational transition was observed with a SNR in excess of 100 in 1 min of integration at a collection rate of 5 Hz. By comparing line intensities to that of a calibrated sample of dilute carbonyl sulfide (OCS) in Ar, taking into account differences in the partition function and dipole moment, we estimate that of order 1013 H2NNO molecules are produced per gas pulse.

The quantum chemical calculations in this work were performed using the CFOUR suite of electronic structure programs56 and employed coupled-cluster methods with single, double, and perturbative triple excitations [CCSD(T)]57–59 in the frozen-core approximation. To investigate the influence of basis and electron correlation on the equilibrium structure, we performed systematic geometry optimizations on planar and non-planar forms of H2NNO using Hartree-Fock (HF) and CCSD(T) methods with correlation-consistent basis sets (cc-pVXZ, where X = T,Q,5). The energetic difference between the two structures were scrutinized at the CCSD(T) level; to account for basis truncation, the comparison was made using extrapolated complete basis set (CBS) limit energies,

ECBS=ESCF+ΔEcorr,
(1)

where ESCF and ΔEcorr are the extrapolated HF and electron correlation contributions to the energy, respectively. The HF and correlation energies are extrapolated separately;60 the extrapolated SCF component is obtained by using the formula

ESCF=EXSCF+aexp(bX),
(2)

and the extrapolated correlation contribution via

ECCSD(T)=EXCCSD(T)+cX3,
(3)

where EXSCF/CCSD(T) refers to either the HF or CCSD(T) correlation energy at basis cardinal number X, and a, b, and c are determined from calculations done with finite basis sets (cc-pVXZ, where X = T,Q,5). Additionally, the quadrupole moments of both forms were calculated at the CCSD(T) level to compare with the experimentally derived values for the nitrogen and deuterium atoms. Calculations performed with cc-pVQZ and cc-pV5Z quality bases indicated that the moments were relatively insensitive to basis size.

Because the NNO backbone is significantly bent (115°), H2NNO is an asymmetric top near the prolate limit [κ = (2BAC)/(AC) = −0.949 40], but with non-zero dipole moments along its principal (a) and intermediate (b) inertial axes. As the molecule is relatively light, only its fundamental a-type rotational transition falls in the operating range of our spectrometer; however, through double resonance measurements, one additional a-type and at least two b-type transitions have been detected. Owing to the quadrupole hyperfine structure from the two nitrogen atoms, each transition consists of many sub-components. In total, more than 45 well-resolved hyperfine lines were measured for H2NNO (see Table S1 of the supplementary material) from four low-J rotational transitions.

The low-J rotational spectrum of H2NNO can be reproduced by fitting several free parameters in a standard asymmetric top Hamiltonian61 that includes hyperfine interactions to the measured frequencies. In addition to the three rotational constants, one fourth-order centrifugal distortion term (ΔJ), and two hyperfine terms (χaa and χbb) for each nitrogen nucleus are required to reproduce the measurements to the experimental uncertainty (Table S7 of the supplementary material). Despite the limited data set, the rotational constants have been determined to better than 1 ppm and the fractional accuracy of the four χ(N) parameters is a few percent. The derived inertial defect, a rough indicator of planarity, is 0.007 45 amu Å2; its sign and magnitude are typical for either a planar molecule or a nearly planar molecule whose defects are very often small and positive owing to zero-point vibrational motion, largely arising from the contributions of low-frequency bending modes.62 

Rotational spectra of D2NNO and all five singly substituted isotopic species of H2NNO have been observed (Tables S2–S6 of the supplementary material). Because of the heavier mass and therefore somewhat smaller rotational constants of D2NNO, a fifth rotational transition (21,2 – 11,1) was measured for this species. Detection of the two single 15N species was extremely helpful in disentangling the complicated hyperfine structure of normal H2NNO. As illustrated in Fig. 4, from partially resolved hyperfine structure in the fundamental rotational line of the two single D isotopic species, the deuterium quadrupole coupling constant χaa(D) could be derived, but owing to the ten times poorer resolution of DR spectroscopy, it was not possible to resolve hyperfine splittings that are smaller than 25 kHz and consequently to determine χbb(D). Nevertheless, spectra having very high SNR and well-resolved nitrogen structure are still readily observed, as the spectrum in Fig. 5 indicates. For D2NNO, the line structure is extremely dense owing to hyperfine interactions from the four inequivalent I = 1 spins. Because the two χaa(D) constants are similar in magnitude but opposite in sign, however, the hyperfine structure from these nuclei is poorly resolved in its two a-type lines, and consequently no attempt was made to assign individual hyperfine components at this level of detail. Instead, well-resolved hyperfine structure from the two nitrogen atoms was assigned, and a slightly larger measurement uncertainty (5 vs. 2 kHz) associated with its features at 21.3 and 40.9 GHz.

FIG. 5.

The fundamental b-type (11,100,0) transition of d1,cis-HDNNO measured by DR spectroscopy in which the intensity of the hyperfine components at 23 069.57 MHz [see Table S5 of the supplementary material, denoted with a green asterisk (*) in Fig. 4] was monitored as radiation from a second frequency source was swept near 77 135 MHz. The step size was 25 kHz, and the total integration time was approximately 2 min. Nitrogen hyperfine structure is well resolved but that due to the deuteron is not observed at this instrumental resolution.

FIG. 5.

The fundamental b-type (11,100,0) transition of d1,cis-HDNNO measured by DR spectroscopy in which the intensity of the hyperfine components at 23 069.57 MHz [see Table S5 of the supplementary material, denoted with a green asterisk (*) in Fig. 4] was monitored as radiation from a second frequency source was swept near 77 135 MHz. The step size was 25 kHz, and the total integration time was approximately 2 min. Nitrogen hyperfine structure is well resolved but that due to the deuteron is not observed at this instrumental resolution.

Close modal

Spectroscopic constants of the six rare isotopic species were derived in the same manner as those of the normal species. Best-fit constants of these and the normal species are summarized in Table S7 of the supplementary material; an abbreviated summary of the rotational constants and the corresponding inertial defects is provided in Table I. As with the normal species, four rotational transitions are sufficient to precisely determine the three rotational constants, and these agree to be better than 1% with the best predictions achieved by scaling theoretical values. As a consistency check, we note that the nitrogen quadrupole coupling constants derived from the two single 15N species and H2NN18O agree to better than 2% with those of the normal species, and similarly good agreement is found for the quartic centrifugal distortion constant ΔJ. With respect to the inertial defects, those for the three heavy atom substitutions are essentially identical to one another, differing by at most 0.008 amu Å2 from the normal species, but D atom substitution results in a small but discernible decrease in the defect, with the substitution at the Htrans atom (−0.025 amu Å2, relative to the normal species) nearly three times than that derived for the substitution at the Hcis atom (−0.008 amu Å2); for D2NNO, the defect increases additively relative to the two single D species.

TABLE I.

Experimental rotational constants and both experimental and theoretical inertial defects of isotopic H2NNO.

Constanta,bH2NNOH215NNOH2N15NOH2NN18Od1, trans-HDNNOd1, cis-HDNNOD2NNO
A0 80 993.1 80 603.7 78 044.9 79 842.5 79 336.5 66 591.4 64 889.0 
B0 12 876.5 12 513.3 12 874.8 12 201.7 11 795.1 12 525.5 11 532.5 
C0 11 108.4 10 831.8 11 051.2 10 584.0 10 273.8 10 544.4 9 799.3 
Δ0 0.007 45 −0.000 24 0.001 86 0.000 97 −0.025 48 −0.008 47 −0.037 69 
Δ0 − Δ0,nd  −0.007 69 −0.005 59 −0.006 48 −0.032 93 −0.015 92 −0.045 14 
Δec −0.111 6 −0.113 1 −0.111 7 −0.111 7 −0.163 3 −0.136 0 −0.187 5 
Δe − Δe,nd  −0.001 4 0.000 0 0.000 0 −0.051 7 −0.024 4 −0.075 8 
Constanta,bH2NNOH215NNOH2N15NOH2NN18Od1, trans-HDNNOd1, cis-HDNNOD2NNO
A0 80 993.1 80 603.7 78 044.9 79 842.5 79 336.5 66 591.4 64 889.0 
B0 12 876.5 12 513.3 12 874.8 12 201.7 11 795.1 12 525.5 11 532.5 
C0 11 108.4 10 831.8 11 051.2 10 584.0 10 273.8 10 544.4 9 799.3 
Δ0 0.007 45 −0.000 24 0.001 86 0.000 97 −0.025 48 −0.008 47 −0.037 69 
Δ0 − Δ0,nd  −0.007 69 −0.005 59 −0.006 48 −0.032 93 −0.015 92 −0.045 14 
Δec −0.111 6 −0.113 1 −0.111 7 −0.111 7 −0.163 3 −0.136 0 −0.187 5 
Δe − Δe,nd  −0.001 4 0.000 0 0.000 0 −0.051 7 −0.024 4 −0.075 8 
a

The complete set of spectroscopic constants for each isotopic species is given in Table S7 of the supplementary material. Rotational constants have been rounded to the nearest tenth of a MHz here.

b

Rotational constants are in units of MHz and inertial defects in amu Å2.

c

Derived from the CCSD(T)/cc-pV5Z equilibrium, non-planar structure at CCSD(T)/CBS limit.

d

Δ0,n and Δe,n refer to the inertial defect of the normal isotopic species, either that derived from the measurements or that calculated from the equilibrium structure, respectively.

The work presented here suggests that H2NNO is formed from the reaction NH2 + NO in our discharge nozzle source, as evidenced from (1) its high yield (1013 molecules/gas pulse, Sec. II) when starting with NH3 and NO, whereas other N and O sources such as NNO, N2, and O2 were tested and proved much less efficient in generating this species; (2) the selective production of H215NNO and H2N15NO using precursor samples enriched in 15N, specifically 15 NH3 or 15NO, respectively; and (3) comparably selective production—at the 98% level—of only H2NNO and D2NNO when using an equal mixture of NH3 and ND3 (in combination with NO) with only trace amounts of the two single D species found under these conditions. Taken together, these observations provide compelling evidence that H2NNO is formed promptly, presumably via a bimolecular reaction, with very little atom scrambling prior to molecular formation and subsequent cooling. Other examples of species formed by a simple and prompt reaction in our discharge source include the HOCO radical,63 Si(H)SiH,64 and the recently reported c-SiO2H2.65 Like c-SiO2H2, there are apparently enough collisions in the jet expansion to stabilize and cool H2NNO once formed in a deep potential well on the potential energy surface shown in Fig. 1, prior to subsequent rearrangement and product formation via one of the two exit channels.

The nitrogen quadrupole tensor element χaa(N) can be used to infer the N–N bond order in H2NNO, and its magnitude and sign can be compared to that found in closely related molecules such as NNO66 and HNNO+,67 on the assumption that the N–N bond is closely aligned to the a-inertial axis. Because the N–N bond is only rotated by 30° from the a-inertial axis, and because the hybridization of Na [the outer (amine) nitrogen atom using the notation H2NaNbO] is different in the two idealized Lewis structures in Fig. 3 (sp3 vs. sp2), the value of χaa at this atom provides a simpler and more sensitive indication of the bond order compared to that at Nb, the inner nitrogen atom. As a consequence of the significant NNO bending angle (115°), there is a large component of the quadrupole tensor along the b-inertial axis at Nb [χbb = −6.07(3) MHz; see Table II] and furthermore its hybridization is unchanged (sp2) in the two Lewis structures (Fig. 3); as a result, χaa at this atom is less sensitive to the relative admixture of uncharged dipolar and zwitterion structures.

TABLE II.

Quadrupole coupling constants and heavy atom bond lengths in H2NNO in comparison with other N-bearing species of similar size.

χaa(N) orN–NN–O
MoleculeeQq(N)/MHza(Å)(Å)CommentReferences
N2 −5.01(8) 1.098  Triple bond 70 and 71  
NO   1.154  72  
NNO −3.330(4) 1.127 1.185  66  
HNNO+ 2.737(4) 1.145   67  
HNNH 4.43b 1.252  Double bond 73  
H2NNO 2.28(2) 1.342(3) 1.217(3) This work  
H2NNO2 4.310 1.381 1.232  74  
H2NNH2  1.449  Single bond 75  
χaa(N) orN–NN–O
MoleculeeQq(N)/MHza(Å)(Å)CommentReferences
N2 −5.01(8) 1.098  Triple bond 70 and 71  
NO   1.154  72  
NNO −3.330(4) 1.127 1.185  66  
HNNO+ 2.737(4) 1.145   67  
HNNH 4.43b 1.252  Double bond 73  
H2NNO 2.28(2) 1.342(3) 1.217(3) This work  
H2NNO2 4.310 1.381 1.232  74  
H2NNH2  1.449  Single bond 75  
a

For molecules that contain two inequivalent N atoms, the value quoted is either for the outer atom (NNO), or the N atom to which the H atom(s) are attached, e.g., HNNO+.

b

Derived for the trans isomer.

If H2NNO adopts a strictly uncharged dipolar resonance structure (Fig. 2), eQq is predicted to be zero at Na, using the Townes-Dailey model.68 If H2NNO has a zwitterion structure with sp2 hybridization at Na instead, eQq = +3.73 MHz, on the assumption that the value of eQq for a 2p electron in atomic nitrogen is −11.2 MHz.69 Since the Townes-Dailey model assumes a bond axis system, a rigorous comparison between the experimental and the two idealized values requires projecting χaa, determined using the inertial frame, along the N–N bond axis (owing to nonlinearity of the heavy atom backbone); doing so increases the magnitude by roughly 10%, to 2.6 MHz. This value, although intermediate between the two extremes, is somewhat closer to that expected for the double-bonded N=N zwitterion structure.

The experimental value of χaa(Na) = 2.283(18) MHz is in good agreement with our best ab initio calculation: at the CCSD(T)/cc-pV5Z level, an equilibrium value of 2.36 MHz is predicted. Table II compares χaa(Na) for a number of small N-bearing species. The value derived here for H2NNO is similar to that found in other chains or asymmetric tops near the prolate limit, in which bonding to the nitrogen atom is thought to have significant sp2 character.

1. Ab initio calculations

To help rationalize the experimental observations, a set of extensive ab initio calculations were carried out on the molecular geometry of H2NNO. Particular emphasis was placed on the effect of the basis size and dynamic correlation on the structure and stability of the planar and non-planar forms. Geometry optimization of the two structures suggests that they differ little in terms of bond lengths and angles, with the exception of the two hydrogen atoms which are out of the NNO plane in the non-planar case. For brevity, we show only the structural parameters of the non-planar form in Table III.

TABLE III.

Optimized equilibrium structural parameters for non-planar H2NNO at each respective level of theory. Bond lengths are provided in Å, and angles in degrees.

HF/cc-pVXZCCSD(T)/cc-pVXZ
Parametercc-pVTZcc-pVQZcc-pV5Zcc-pV6Zcc-pVTZcc-pVQZcc-pV5Z
rN−O 1.1751 1.1746 1.1749 1.17485 1.2163 1.2151 1.2153 
rN−N 1.3097 1.3063 1.3049 1.3047 1.3458 1.3350 1.3315 
rN−H(trans) 0.9965 0.9957 0.9956 0.9956 1.0132 1.0120 1.0119 
rN−H(cis) 0.9894 0.9885 0.9883 0.9883 1.0050 1.0034 1.0031 
NNO 114.98 115.11 115.17 115.18 113.43 113.63 113.70 
NNH(trans) 118.27 118.76 119.00 119.03 116.68 117.86 118.34 
NNH(cis) 116.05 116.39 116.56 116.58 114.79 115.79 116.17 
HNNO(trans) 10.17 8.38 7.25 7.11 13.75 10.10 8.08 
HF/cc-pVXZCCSD(T)/cc-pVXZ
Parametercc-pVTZcc-pVQZcc-pV5Zcc-pV6Zcc-pVTZcc-pVQZcc-pV5Z
rN−O 1.1751 1.1746 1.1749 1.17485 1.2163 1.2151 1.2153 
rN−N 1.3097 1.3063 1.3049 1.3047 1.3458 1.3350 1.3315 
rN−H(trans) 0.9965 0.9957 0.9956 0.9956 1.0132 1.0120 1.0119 
rN−H(cis) 0.9894 0.9885 0.9883 0.9883 1.0050 1.0034 1.0031 
NNO 114.98 115.11 115.17 115.18 113.43 113.63 113.70 
NNH(trans) 118.27 118.76 119.00 119.03 116.68 117.86 118.34 
NNH(cis) 116.05 116.39 116.56 116.58 114.79 115.79 116.17 
HNNO(trans) 10.17 8.38 7.25 7.11 13.75 10.10 8.08 

The planarity of the molecule can be described using the dihedral angle HNNO(cis). For the planar structure, this angle was constrained to zero degrees, while the angle is small for the non-planar structure as seen in Table III, but nonzero. At the HF/cc-pV6Z level, which is assumed to have converged to the HF limit, the angle is 7.1°. The inclusion of electron correlation increases the angle and the molecule becomes less planar by a few degrees. Calculations performed with augmented basis sets (aug-cc-pVXZ, where X = T,Q) did not appear to alter the results significantly; the dihedral angle increases by 1° relative to the unaugmented basis sets, and these results are excluded for brevity.

At the highest level considered here [CCSD(T)/cc-pV5Z], the corresponding inertial defect Δe is calculated to be −0.1116 amu Å2 (Table I) consistent with a structure that is only slightly non-planar. The structural parameters of both forms of H2NNO at the CCSD(T)/cc-pV5Z level are summarized in Table V.

Regarding the relative stability of the two structures, harmonic frequency analysis of the two structures at each level of theory shows that the planar form is a saddle point; while all the vibrational frequencies for the non-planar form are real, the planar form possesses an imaginary frequency (200i cm−1) corresponding to the out-of-plane motion of the molecule. Taking the optimized geometries at the CCSD(T)/cc-pV5Z level, the relative electronic energies of the two forms are compared in Table IV. With smaller basis sets, the non-planar structure is the lower energy of the two. As the basis is expanded, the difference in the energy decreases, and at the CBS limit, the relative energies swap with the planar structure now the lower energy form by 5.5 cm−1. The energetics therefore suggest that—within the accuracy of the methods used in this work—the out-of-plane torsional motion associated with the conversion between the planar and non-planar forms is described by a very flat potential energy surface; a large change (10°) in the geometry is accompanied by a small change in the electronic energy. This is considerably less than the CCSD(T)/cc-pV5Z harmonic zero-point energy calculated with the non-planar structure (7134.2 cm−1) and, more significantly, even that associated with only the mode that corresponds to this coordinate (200 cm−1). The conclusion drawn is in accord with the experimental observations, and for these reasons, it seems very likely that the true vibrationally (zero-point) averaged structure is planar.

TABLE IV.

The CCSD(T) electronic energy and the energy of the planar form relative to the non-planar at each respective basis size calculated with the CCSD(T)/cc-pV5Z optimized geometries. The electronic energy is given in Hartrees, and the energy difference is given in cm−1. The CBS energy is calculated by the sum of the extrapolated HF and CCSD(T) correlation energies at the cc-pVXZ (X = T,Q,5).

Basis sizePlanarNon-planarEnergy difference
cc-pVTZ −185.585 8409 −185.585 9991 34.7 
cc-pVQZ −185.644 3850 −185.644 4301 9.9 
cc-pV5Z −185.663 6469 −185.663 6489 0.4 
CBS −185.685 7129 −185.685 6730 −5.5 
Basis sizePlanarNon-planarEnergy difference
cc-pVTZ −185.585 8409 −185.585 9991 34.7 
cc-pVQZ −185.644 3850 −185.644 4301 9.9 
cc-pV5Z −185.663 6469 −185.663 6489 0.4 
CBS −185.685 7129 −185.685 6730 −5.5 

A further comparison can be made with CH2OO, the simplest Criegee intermediate, which is structurally similar to H2NNO; Li et al.76 have computed a global explicitly correlated potential energy surface at the CCSD(T)-F12a/aug-cc-pVTZ level and have found that the same out-of-plane motion translates to 350 cm−1. Thus, while being structurally similar, the potential energy surface describing CH2OO is comparatively more rigid than H2NNO.

2. The r0 structure

From the rotational constants of the normal and six rare isotopic species, it is possible to derive an experimental (r0) structure for H2NNO, in which the four bond lengths and three bond angles have been least-squares optimized to best reproduce the moments of inertia of each isotopic species, assuming either a planar or non-planar geometry, but one in either case that is not affected by isotopic substitutions. Since there is no a priori knowledge of how the vibrational corrections are partitioned along the three inertial axes, all rotational constants rather than a specific subset (e.g., A and C) for each isotopic species were used in the fit, i.e., 21 constants in total.

The best-fit values and associated statistical uncertainties from the two structural analyses are summarized in Table V. The inclusion of an eight structural parameter, the dihedral angle, only marginally improves the fit rms (0.0060 vs 0.0076), and the best-fit value is both close to zero and poorly determined. For these reasons, we find no compelling evidence that H2NNO is non-planar, and this structure was not be considered further.

TABLE V.

Experimental (r0) and CCSD(T)/cc-pV5Z equilibrium (re) structural parameters of H2NNO (bond lengths in Å and bond angles in degrees).

Experiment (r0)aTheoretical (re)
ParameterPlanarNon-planarPlanarNon-planar
rN−O 1.217(3) 1.217(2) 1.216 1.215 
rN−N 1.342(3) 1.342(2) 1.328 1.331 
rN−H(trans) 0.991(4) 0.991(4) 1.002 1.003 
rN−H(cis) 1.010(3) 1.010(2) 1.011 1.012 
NNO 113.67(5) 113.66(4) 113.73 113.70 
NNH(trans) 116.3(4) 116.3(4) 116.91 116.17 
NNH(cis) 117.5(3) 117.5(2) 119.08 118.34 
HNNO(cis) 0b 3.5(6) 0b 8.07 
Experiment (r0)aTheoretical (re)
ParameterPlanarNon-planarPlanarNon-planar
rN−O 1.217(3) 1.217(2) 1.216 1.215 
rN−N 1.342(3) 1.342(2) 1.328 1.331 
rN−H(trans) 0.991(4) 0.991(4) 1.002 1.003 
rN−H(cis) 1.010(3) 1.010(2) 1.011 1.012 
NNO 113.67(5) 113.66(4) 113.73 113.70 
NNH(trans) 116.3(4) 116.3(4) 116.91 116.17 
NNH(cis) 117.5(3) 117.5(2) 119.08 118.34 
HNNO(cis) 0b 3.5(6) 0b 8.07 
a

Numbers in parentheses are 1σ statistical uncertainties in units of the last digit.

b

Fixed.

The seven structural parameters derived assuming a planar geometry reproduce all 14 B and C rotational constants to better than 3 MHz and the seven A constants to no worse than 120 MHz. Because the inertial defects are generally close to zero and relatively insensitive to substitution, the two N–N–H bond angles have been determined to be better than 1°, while the heavy atom angle is roughly 10 times more precise. The statistical uncertainties of the four bond lengths are uniform and small; each has been determined to be better than 0.005 Å. At this precision, it is possible to establish that the N–H bond to the Hcis atom is slightly longer (0.02 Å) than that to the Htrans atom. The same behavior is also found in CH2OO, in which the C–Hcis bond length is 0.004 Å longer than the C–Htrans bond length.77 

In addition to the quadrupole coupling constant, Table II also provides a comparison of the N–O and N–N band lengths of H2NNO relative to other N-bearing chains. The best-fit N–O bond length [1.217(3) Å] is 0.03 Å longer than that found in NNO and about 0.07 Å longer than that in free NO (1.15 Å). This length is quite close, however, to that of H2NNO2, an indication that additional bonds to the N atom weaken and therefore elongate the N–O bond. Most significant is the length of the N–N bond, 1.342(3)Å, which is intermediate between a typical N–N double (1.21 Å) and triple bond (1.449 Å). Here too, there are similarities to H2NNO2, which has a N–N bond (1.381 Å) of comparable length.

The present work provides another vivid illustration that a supersonic jet combined with an electrical discharge source can efficiently “trap” highly energetic intermediates in a bi-molecular (radical-radical) reaction, provided there are sizable barriers to subsequent unimolecular isomerization—even if these intermediates lie several hundred kJ mol−1 above the products (Fig. 1). The detection of specific H2NNO isotopic species, with minimal isotopic scrambling when using isotopically enriched precursors, further corroborates that their formation must be prompt. Although the lifetime of H2NNO has previously been estimated to at least 0.1 msec in the gas phase by mass spectrometry,26 the time scale in our experiments is roughly ten times longer, implying this species is relatively stable once collisionally cooled. The effect of collisional deactivation is therefore expected to be an important factor in determining the reactive kinetics of H2NNO, which should be considered in subsequent kinetics modeling.

Owing to efficient formation of H2NNO under our experimental conditions, and the availability of isotopically enriched samples, an important by-product of the work here is the determination of the precise r0 geometry shown in Fig. 2. The ab initio work conducted here support the experimental observations; although the CCSD(T) optimizations predict a non-planar structure, the CCSD(T)/CBS energies using the optimized structures suggest the planar form is the lower energy structure. Geometry optimization combined with basis extrapolation techniques would likely yield a planar minimum energy structure. Even though the electronic structure calculations show that the torsional potential is extremely flat, the best-fit structure reproduces the 21 experimental rotational constants of all seven isotopic species to better than 0.2% and yields N–H bond lengths in good agreement with those calculated at the CCSD(T)/CBS limit (Table V). Although indirect, the change of the experimental inertial defects with deuterium substitution (yielding Δ0<0) are consistent with a molecule possessing a shallow torsional potential that involves the H atoms.

Now that the first intermediate in the deNOx reaction has been detected and spectroscopically characterized at high spectral resolution, searches for the four remaining cis, trans-HNNOH conformers are worth undertaking. Like H2NNO, these isomers are calculated to lie in deep minima on the potential energy surface, and their fundamental a-type rotational transitions are predicted to lie near 22 GHz; if found, b-type transitions can likely be observed by DR spectroscopy. In addition to a dilute mixture of NH3 and NO, other combination of gases, such as NNO/H2 and N2/H2O, are worth testing as precursors. NNO/H2 appears particularly promising because the barrier height (91.2 kJ mol−1; Fig. 1) separating NNO + H2 and the various HNNOH conformers is comparable to the typical electron energy (a few eVs) in our discharge source.

Finally, analogous gas-phase studies of larger nitrosamines R1R2NNO may prove useful to clarify how substituents affect the N-N bond length and planarity. An obvious target is N-Nitrosodimethylamine, (CH3)2NNO because this molecule has also been the subject of prior structural studies44–46 and high-level quantum chemical calculations,33 all of which are consistent with a planar geometry. It is also commercially available, so microwave studies could be performed relatively quickly and easily.

See supplementary material for tables of measured frequencies of H2NNO and its isotoplogues.

The work in Cambridge is supported by NSF Grant Nos. CHE-1058063 and CHE-1566266. J.F.S. acknowledges the US National Science Foundation under Grant No. CHE-1361031. We thank the anonymous reviewers for helpful suggestions.

1.
D. J.
Jacob
,
E. G.
Heikes
,
S.-M.
Fan
,
J. A.
Logan
,
D. L.
Mauzerall
,
J. D.
Bradshaw
,
H. B.
Singh
,
G. L.
Gregory
,
R. W.
Talbot
,
D. R.
Blake
 et al,
J. Geophys. Res.: Atmos.
101
,
24235
, doi:10.1029/96jd00336 (
1996
).
2.
Clean Air Technology Center (MD-12) Information Transfer and Program Integration Division Office of Air Quality Planning and Standards U.S. Environmental Protection Agency, Technical Report T. B. 456/F-99–006R, EPA,
1999
.
3.
M.
Piumetti
,
S.
Bensaid
,
D.
Fino
, and
N.
Russo
,
Catal., Struct. React.
1
,
155
(
2015
).
4.
R. K.
Lyon
,
Int. J. Chem. Kinet.
8
,
315
(
1976
).
5.
R. K.
Lyon
and
J. E.
Hardy
,
Ind. Eng. Chem. Fundam.
25
,
19
(
1986
).
6.
J. A.
Miller
,
M.
Branch
, and
R. J.
Kee
,
Combust. Flame
43
,
81
(
1981
).
7.
J. A.
Miller
,
Symp. (Int.) Combust.
26
,
461
(
1996
).
8.
D.
Sun
,
W. F.
Schneider
,
J. B.
Adams
, and
D.
Sengupta
,
J. Phys. Chem. A
108
,
9365
(
2004
).
9.
S. J.
Klippenstein
,
L. B.
Harding
,
P.
Glarborg
, and
J. A.
Miller
,
Combust. Flame
158
,
774
(
2011
).
10.
J. A.
Silver
and
C. E.
Kolb
,
J. Phys. Chem.
86
,
3240
(
1982
).
11.
E. W.
Diau
,
T.
Yu
,
M. A. G.
Wagner
, and
M. C.
Lin
,
J. Phys. Chem.
98
,
4034
(
1994
).
12.
B.
Atakan
,
A.
Jacobs
,
M.
Wahl
,
R.
Weller
, and
J.
Wolfrum
,
Chem. Phys. Lett.
155
,
609
(
1989
).
13.
V. P.
Bulatov
,
A. A.
Ioffe
,
V. A.
Lozovsky
, and
O. M.
Sarkisov
,
Chem. Phys. Lett.
161
,
141
(
1989
).
14.
J. W.
Stephens
,
C. L.
Morter
,
S. K.
Farhat
,
G. P. G. P.
Glass
, and
R. F.
Curl
,
J. Phys. Chem.
97
,
8944
(
1993
).
15.
M.
Wolf
,
D. L.
Yang
, and
J. L.
Durant
,
J. Phys. Chem. A
101
,
6243
(
1997
).
16.
J.
Park
and
M. C.
Lin
,
J. Phys. Chem.
100
,
3317
(
1996
).
17.
E. W.-G.
Diau
and
S. C.
Smith
,
J. Chem. Phys.
106
,
9236
(
1997
).
18.
Z.
Chen
and
T. P.
Hamilton
,
J. Phys. Chem. A
103
,
11026
(
1999
).
19.
J. A.
Miller
and
S. J.
Klippenstein
,
J. Phys. Chem. A
104
,
2061
(
2000
).
20.
D.-C.
Fang
,
L. B.
Harding
,
S. J.
Klippenstein
, and
J. A.
Miller
,
Faraday Discuss.
119
,
207
(
2002
).
21.
S. F.
Selgren
,
P. W.
McLoughlin
, and
G. I.
Gellene
,
J. Chem. Phys.
90
,
1624
(
1989
).
22.
B.
Ruscic
,
D.
Feller
, and
K. A.
Peterson
,
Theor. Chem. Acc.
133
,
1
(
2013
).
23.
B.
Ruscic
,
Int. J. Quantum Chem.
114
,
1097
(
2014
).
24.
E. W.-G.
Diau
and
S. C.
Smith
,
J. Phys. Chem.
100
,
12349
(
1996
).
25.
H.
Egsgaard
,
L.
Carlsen
,
T.
Weiske
,
D.
Sülzle
, and
H.
Schwarz
,
Chem. Phys. Lett.
199
,
643
(
1992
).
26.
M.
Farber
and
S. P.
Harris
,
J. Phys. Chem.
88
,
680
(
1984
).
27.
D. N.
Shin
,
M.
Freindorf
,
T. R.
Furlani
,
R. L.
DeLeon
, and
J. F.
Garvey
,
Int. J. Mass Spectrom.
255–256
,
28
(
2006
).
28.
J. N.
Crowley
and
J. R.
Sodeau
,
J. Phys. Chem.
94
,
8103
(
1990
).
29.
M. E.
Jacox
and
W. E.
Thompson
,
J. Chem. Phys.
123
,
064501
(
2005
).
30.
C. J.
Casewit
and
W. A.
Goddard
 III
,
J. Am. Chem. Soc.
104
,
3280
(
1982
).
31.
S. P.
Walch
,
J. Chem. Phys.
99
,
5295
(
1993
).
32.
L. A.
Baker
and
S.
Su
,
Chem. Phys.
228
,
9
(
1998
).
33.
S. A.
do Monte
,
E.
Ventura
,
T. F.
da Costa
, and
S. R.
de Santana
,
Struct. Chem.
22
,
497
(
2011
).
34.
J. A.
Harrison
,
R. G. A. R.
Maclagan
, and
A. R.
Whyte
,
Chem. Phys. Lett.
130
,
98
(
1986
).
35.
K. D.
Brunnemann
,
S. S.
Hecht
, and
D.
Hoffmann
,
J. Toxicol., Clin. Toxicol.
19
,
661
(
1982
).
36.
A. R.
Tricker
and
R.
Preussmann
,
Mutat. Res.
259
,
277
(
1991
).
38.
D. L.
Sparks
(
Academic Press
,
2013
), Vol. 119.
39.
S.
Ponnusamy
,
L.
Sandhiya
, and
K.
Senthilkumar
,
Int. J. Chem. Kinet.
49
,
339
(
2017
).
40.
G. A.
Digenis
and
C. H.
Issidorides
,
Bioorg. Chem.
8
,
97
(
1979
).
41.
H.
Roohi
,
A.
Ebrahimi
,
B.
Mackiabadi
, and
M.
Hagealirezahi
,
J. Mol. Struct.: THEOCHEM
778
,
63
(
2006
).
42.
M.
Miura
,
S.
Sakamoto
,
K.
Yamaguchi
, and
T.
Ohwada
,
Tetrahedron Lett.
41
,
3637
(
2000
).
43.
T.
Ohwada
,
M.
Miura
,
H.
Tanaka
,
S.
Sakamoto
,
K.
Yamaguchi
,
H.
Ikeda
, and
S.
Inagaki
,
J. Am. Chem. Soc.
123
,
10164
(
2001
).
44.
B.
Krebs
and
J.
Mandt
,
Eur. J. Inorg. Chem.
108
,
1130
(
1975
).
45.
P.
Rademacher
and
R.
Stolevik
,
Acta Chem. Scand.
23
,
660
(
1969
).
46.
C. E.
Looney
,
W.
Phillips
, and
E.
Reilly
,
J. Am. Chem. Soc.
79
,
6136
(
1957
).
47.
M. C.
McCarthy
,
M. J.
Travers
,
A.
Kovacs
,
C. A.
Gottlieb
, and
P.
Thaddeus
,
Astrophys. J., Suppl. Ser.
113
,
105
(
1997
).
48.
M. C.
McCarthy
,
W.
Chen
,
M. J.
Travers
, and
P.
Thaddeus
,
Astrophys. J., Suppl. Ser.
129
,
611
(
2000
).
49.
W.
Jager
and
M. C. L.
Gerry
,
J. Chem. Phys.
102
,
3587
(
1995
).
50.
Y.
Sumiyoshi
,
H.
Katsunuma
,
K.
Suma
, and
Y.
Endo
,
J. Chem. Phys.
123
,
054324
(
2005
).
51.
M. C.
McCarthy
,
J.-U.
Grabow
,
M. J.
Travers
,
W.
Chen
,
C. A.
Gottlieb
, and
P.
Thaddeus
,
Astrophys. J. Lett.
494
,
L231
(
1998
).
52.
C. A.
Gottlieb
,
M. C.
McCarthy
,
M. J.
Travers
,
J.-U.
Grabow
, and
P.
Thaddeus
,
J. Chem. Phys.
109
,
5433
(
1998
).
53.
A. J.
Apponi
,
M. C.
McCarthy
,
C. A.
Gottlieb
, and
P.
Thaddeus
,
Astrophys. J.
530
,
357
(
2000
).
54.
C. A.
Gottlieb
,
A. J.
Apponi
,
M. C.
McCarthy
,
P.
Thaddeus
, and
H.
Linnartz
,
J. Chem. Phys.
113
,
1910
(
2000
).
55.
H.
Gupta
,
S.
Brünken
,
F.
Tamassia
,
C. A.
Gottlieb
,
M. C.
McCarthy
, and
P.
Thaddeus
,
Astrophys. J. Lett.
655
,
L57
(
2007
).
56.
CFOUR, Coupled-Cluster techniques for Computational Chemistry, a quantum-chemical program package by
J. F.
Stanton
,
J.
Gauss
,
M. E.
Harding
, and
P. G.
Szalay
, with contributions from
A. A.
Auer
 et al For detailed information, seewww.cfour.de.
57.
R. J.
Bartlett
,
J. D.
Watts
,
S. A.
Kucharski
, and
J.
Noga
,
Chem. Phys. Lett.
165
,
513
(
1990
).
58.
K.
Raghavachari
,
G. W.
Trucks
,
J. A.
Pople
, and
M.
Head-Gordon
,
Chem. Phys. Lett.
157
,
479
(
1989
).
59.
J. F.
Stanton
,
Chem. Phys. Lett.
281
,
130
(
1997
).
60.
T.
Helgaker
,
W.
Klopper
,
H.
Koch
, and
J.
Noga
,
J. Chem. Phys.
106
,
9639
(
1997
).
61.
J. K. G.
Watson
,
Vibrational Spectra and Structure
(
Elsevier
,
1977
), Vol. 6, Chap. 1.
62.
T.
Oka
,
J. Mol. Struct.
352–353
,
225
(
1995
).
63.
M. C.
McCarthy
,
O. M.
, Jr.
,
B. A.
McGuire
,
K. N.
Crabtree
,
M.-A.
Martin-Drumel
, and
J. F.
Stanton
,
J. Chem. Phys.
144
,
124304
(
2016
).
64.
M.
McCarthy
and
P.
Thaddeus
,
J. Mol. Spectrosc.
222
,
248
(
2003
).
65.
M. C.
McCarthy
and
J.
Gauss
,
J. Phys. Chem. Lett.
7
,
1895
(
2016
).
66.
K.
Casleton
and
S.
Kukolich
,
J. Chem. Phys.
62
,
2696
(
1975
).
67.
M. C.
McCarthy
,
O.
Martinez
,
K. N.
Crabtree
,
V.
Lattanzi
,
S. E.
Novick
, and
S.
Thorwirth
,
J. Phys. Chem. A
117
,
9968
(
2013
).
68.
C. H.
Townes
and
A. L.
Schawlow
,
Microwave Spectroscopy
(
McGraw-Hill
,
1955
).
69.
A.
Schirmacher
and
H.
Winter
,
Phys. Rev. A
47
,
4891
4907
(
1993
).
70.
A.
Halkier
,
S.
Coriani
, and
P.
Jørgensen
,
Chem. Phys. Lett.
294
,
292
(
1998
).
71.
K. P.
Huber
and
G.
Herzberg
,
Molecular Spectra and Molecular Structure IV. Constants of Diatomic Molecules
(
Van Nostrand Reinhold Company
,
1979
).
72.
K. P.
Huber
and
G.
Herzberg
, “
Constants of diatomic molecules
,” in
Molecular Spectra and Molecular Structure
(
Springer, Boston, MA
,
1979
).
73.
J.
Demaison
,
F.
Hegelund
, and
H.
Bürger
,
J. Mol. Struct.
413-414
,
447
(
1997
).
74.
A.
Häußler
,
T. M.
Klapötke
, and
H.
Piotrowski
,
Z. Naturforsc., B
57
,
151
(
2002
).
75.
K.
Kohata
,
T.
Fukuyama
, and
K.
Kuchitsu
,
J. Phys. Chem.
86
,
602
(
1982
).
76.
J.
Li
,
S.
Carter
,
J. M.
Bowman
,
R.
Dawes
,
D.
Xie
, and
H.
Guo
,
J. Phys. Chem. Lett.
5
,
2364
(
2014
).
77.
T. L.
Nguyen
,
M. C.
McCarthy
, and
J. F.
Stanton
,
J. Phys. Chem. A
119
,
7197
(
2015
).
78.
M. W.
Chase
,
J. L.
Curnutt
,
J. R.
Downey
,
R. A.
McDonald
,
A. N.
Syverud
, and
E. A.
Valenzuela
,
J. Phys. Chem. Ref. Data
11
,
695
(
1982
).

Supplementary Material