Introduction

Molecular anions have been discovered in dark clouds, prestellar cores, protostellar and circumstellar envelopes, Titan’s atmosphere, and cometary comas, and they are involved in the substance evolutions of interstellar medium1,2. Although the detection of new molecular anions in space stagnates recently, laboratorial efforts are continuously being put into exploring anionic properties and identifying potential targets for astronomical observations3. Radiative electron attachment (REA), dissociative electron attachment (DEA), and ion-pair or polar dissociation are recognized as the typical pathways to produce anionic species, while a rapid anion–neutral reaction usually leads to the secondary anion1,2,3,4,5. As one of the smallest diatomic anions with the astronomical interest, cyanide anion (CN−), was proposed to exist in Titan’s ionosphere6, subsequently detected at an altitude of 1015 km7 and discovered in the carbon-rich star envelope IRC + 10216 (ref. 8). Previously, its presence in the coma of comets was also mentioned2,9. Its productions by the REA to CN radical, the DEA to cyanopolyynes HCnN (n = 1–3) or MgCN/MgNC, and the H− + HCnN reactions were evaluated with modeling calculations1,4,5,6,7,8,9,10. However, the role of REA process, initially proposed as the dominant mechanism4, is still open to the debate8,9,10,11,12,13, in particular, the CN− abundance in IRC + 10216 predicted with statistics theory calculations indicated a significant deviation from the observation8,13. Recently, the contribution from the DEA to H2CN in IRC + 10216 was eliminated either14.

CN− at the electronically ground state X1Σ+ is known to be a stable anion because of the large electron affinity (EA = 3.862 ± 0.004 eV) of the neutral15. Furthermore, the CN−(X1Σ+) can be stabilized in various vibrational states (ν) up to ν = 17, while that in the higher ν-state is believed to quickly decay via vibration-induced electron detachment16. Therefore, the vibrationally or electronically excited states above the electron autodetachment threshold of CN−, namely, those locate 3.862 eV (EA) higher than X1Σ+ (ν = 0), are assumed to have no contributions to the CN− abundance and some chemical reactions in the interstellar space1,4,5,6,7,8,9,10,11,12,13. However, a long-lived anion in high-lying excited states can be observed, if its vibration-induced electron detachment proceeds in a non-negligible time, for instance, on a timescale comparable to the lifetimes (such as the nanoseconds to microseconds of LiH− and OH−, ref. 16). By the same token, the CN− could be long-lived in the vibrational states ν ≥ 18 or electronically excited states above the electron autodetachment threshold, but such a conjecture is subject to experimental validation.

Seeking the long-lived anion in the high-lying states is continued with great enthusiasm until now. The long-lived (milliseconds) NC4N−* (ref. 17) and (microseconds to seconds) SF6−* (refs. 18,19) were found to be populated in the vibrational states above their respective electron autodetachment thresholds. A high-lying spin-state 6Π of CO−, as a metastable anion, was predicted20, which potentially elucidated the observation of long-lived (10−5 s) CO− detected in the mass spectrometry experiment21. As illustrated in Fig. 1, various resonant states (M*−) of a polyatomic anion can be formed in electron attachment, and they maintain the equilibrium structure of the neutral and locate energetically above the neutral state M22. The resonant states (M*−) with various lifetimes (denoted with different energy widths for shape or Feshbach resonant states) decay quickly through dissociation or electron autodetachment22. Even if the bound states M−* and M−** exist in theory, the transformation from (M*−) to M−*/M−** is hardly accomplished by the structural relaxation in dozens of picoseconds, due to its much faster competitive processes such as dissociation and electron autodetachment of (M*−). This is the primary reason that M−* or M−** is scarcely observed in the electron attachment experiment. Here superscripts −** (or −*) and *− represent the structurally relaxed (but in an excited state) and the resonant-state (within the neutral structure) anions, respectively; while M− is the anionic ground state. With reference to the neutral superexcited state M** which lies energetically above the cationic ground state M+, M−**, as an anionic state above the ground state M, is named as the anionic superexcited state. A comprehensive theoretical study of the (CN*−) resonant states was reported by Harrison and Tennyson23, indicating an energy sequence of resonant states 3Π < 3Σ+ < 3Σ–; while the CN−** superexcited bound states, the singlet Σ-, Π-, and Δ-symmetric bound states24, were predicted to be in an energy order of 3Σ+ < 3Π < 1Π (ref. 25). Meanwhile, the equilibrium bond lengths of CN−** anions were close to that of the ground-state CN− or CN24,25. Therefore, once (CN*−) is produced, its structural relaxation in the transformation (CN*−) → CN−** would be considerably fast; alternatively, the CN−** superexcited bound states could be directly accessed by the electron attachment to CN radical. These two cases exhibit some possibilities to observe metastable CN−** in experiments. Here we show experimental evidence of CN−** produced in the DEA to cyanogen bromide (BrCN) and emphasize its profound roles in astrochemistry.

Fig. 1: Comparison between neutral and anionic states.
figure 1

Left: M, M*, M+, and M** represent the ground, excited, ionized, and superexcited states of a neutral molecule, respectively; Middle: (M*−) at anionic resonant state is formed by electron attachment to M; Right: Stable M− and M−* can be formed by the structural relaxation from (M*−), while a superexcited-state M−** (lying above M) is scarcely observed in experiments. Different widths of the energy levels of M**, (M*−), and M−** correspond to their lifetimes, due to the couplings with the electron continuum background (in black color).

Full size image

Results and discussion

In this work, we report an experimental evidence that the CN−** species are produced in the DEA to BrCN,

$${\mathrm{e}}^ - + {\mathrm{BrCN}} \to {\mathrm{Br}}({\mathrm{P}}_{3/2})/{\mathrm{Br}}^* ({\mathrm{P}}_{1/2}) + {\mathrm{CN}}^ - /{\mathrm{CN}}^{-**}$$
(1)

where the CN− yield is populated in the vibrational states (ν ≤ 17) at the low electron attachment energy (Ee), while the CN−** produced at the high Ee value is in the vibrationally or electronically superexcited states. In the previous studies, the CN−** species cannot be identified in the CN− production efficiency curve26 and the low vibrational-state (ν < 9) CN− yields were found in the lower attachment energy range (Ee = 1.07–1.97 eV)27. Using the high-resolution time-sliced velocity map imaging (VMI) apparatus27,28,29,30, which was developed on the basis of our previous one31,32, we record the velocity images of the CN−/CN−** yields in the higher Ee range from 3.57 to 6.57 eV, and the results are shown in Fig. 2.

Fig. 2: Time-sliced velocity images of CN−.
figure 2

These CN− ions are produced in dissociative electron attachments to BrCN at the attachment energies of 3.57 (a), 4.57 (b), 5.07 (c), 5.27 (d), 6.07 (e), and 6.57 (f) eV. a The electron incident direction is denoted with an arrow in the middle of image, which defines the forward- and backward-scattered distributions of CN− yields. c The internal energy of the CN−** yields distributed in the inside region of a broken circle embedded on the image is higher than the electron autodetachment threshold. These long-lived metastable CN−** ions are also observed in (d–f). The anionic intensity in each sliced image is normalized with the weighing factors of different-sized Newton spheres, and each image is plotted independently in a relative intensity scale.

Full size image

According to the energy conservation, the internal energy (Eint) of CN− is determined with,

$${{E}}_{{\mathrm{int}}}\left( {{\mathrm{CN}}^ - } \right) = {{E}}_{\mathrm{e}} - {{E}}_{{\mathrm{th}}} - {{E}}_{\mathrm{k}}$$
(2)

where Eth is the DEA threshold [−0.13 eV for Br + CN−(X1Σ+, ν = 0); 0.33 eV for Br* + CN−(X1Σ+, ν = 0)] and Ek is the total kinetic energy release of the DEA products. In a velocity image, the anion having a small kinetic energy locates a position close to the image center, and vice versa. To our surprise, besides the outside strong signals of the fast CN−, some slow anions emerge in the inside region at Ee = 5.07 eV and their intensities are gradually enhanced as Ee increases. According to Eq. (2), these slow CN− must reserve the higher Eint. For example, a broken circle embedded on the image of Fig. 2c denotes the velocity position of the CN−(X1Σ+) at the vibrational state ν = 17, thus the inside signals demonstrate the existence of the superexcited-state CN−** yields. As mentioned above, these superexcited states should be CN−(X1Σ+) in the vibrational states ν ≥ 18 or that in the high-lying electronic states24,25.

On the other hand, the velocity distributions of the inside anions diversify remarkably with the Ee increase. From Fig. 2c to f, the inner ring-like distribution becomes most distinct at Ee = 5.27 eV and it is weakened subsequently; an additional smaller ring appears around the image center at 6.07 eV and turns to be prominent at 6.57 eV. Besides the forward-backward scattered distributions of the outside anions, the anions with the inner ring-like distribution show the relatively strong intensities in the sideward and backward directions (Fig. 2d), then their intensities in the sideward are slightly enhanced (Fig. 2e, f). Although the potential energy surfaces of BrCN− in the low-lying states have been successfully evaluated in our recent study27, those in the high-lying states are still unavailable to date. Moreover, an explicit theory to describe the nuclear-electron motion couplings is required to understand the above variances of the CN− angular distributions. Therefore, we will focus only on the Eint distributions of the CN− yields and identify their internal states by the thermodynamics analyses.

Figure 3 exhibits the CN− kinetic energy distributions, in which each point (solid circle) is obtained by summarizing the ion intensity for a common velocity within whole scattering angles. In the present Ee range, the dissociation channels leading to CN− and Br or Br* are accessible. According to Eq. (2) and the momentum conservation, different CN− quantum states for the channels to CN− + Br and CN− + Br* are assigned in Supplementary Fig. 1 of Supplementary Note 1, while only the former channel is presented in Fig. 3 for the purpose of clear visibility. As shown in Fig. 3a, b, the vibrational states ν ≤ 17 of the ground-state CN−(1Σ+) are assigned and the states of ν = 9 (at Ee = 3.57 eV) and 13 (at Ee = 4.57 eV) are responsible for the maxima of the profiles.

Fig. 3: Kinetic energy distributions of CN−.
figure 3

These CN− ions are produced in dissociative electron attachments to BrCN at 3.57 (a), 4.57 (b), 5.07 (c), 5.27 (d), 6.07 (e), and 6.57 (f) eV. The vibrational states of the ground (1Σ+) and superexcited (3Σ+, 3Π) states are assigned for Br(2P3/2) + CN−/CN−**, and two regions, the superexcited-state one (shaded in yellow) and the low vibrational state region of the ground-state 1Σ+ (shaded in cyan), are divided with a vertical dashed line which lies at the left side of 1Σ+ (ν = 17) indicates the threshold of the electron autodetachment of CN−.

Full size image

At the higher Ee values, some fine structures are observed at the left sides of Fig. 3c–f. They, together with a left part of the big band, are attributed to the superexcited-state CN−** yields (shaded in yellow). Considering the contributions of 3Σ+ and 3Π states, we reproduced the potential energy curves of these two and 1Π superexcited states on the basis of our high-level calculations. The present results are shown in Fig. 4, in agreement with the previous ones25. Using the potential energy curves, we further derived the vibrational energy levels of 3Σ+, 3Π, and 1Π states (more details can be found in Supplementary Tables 1 and 2 of Supplementary Note 3). As depicted in Fig. 3c, d, the low vibrational states ν = 1 or 2 of 3Σ+ should be responsible for the small peak around the kinetic energy of 0.40 eV. At Ee = 6.07 and 6.57 eV, two small peaks at 0.05 and 0.40 eV are attributed to ν = 7, 10 (Fig. 3e) and 9/10, 13 (Fig. 3f) of 3Σ+, respectively. Meanwhile, the contributions from the specific vibrational states of 3Π are possible. Note that the highest intensity (shaded in yellow) of the big band in Fig. 3e (or 3f) may be owed to the superexcited vibrational states 18 ≤ ν ≤ 23 (or 26) of 1Σ+, the lower vibrational states of 3Σ+, or both of them. Here we conclude that the superexcited vibrational states are highly preferable because their corresponding part is a portion of this big band, distinctly different from the isolated small peaks observed on the left side. Furthermore, the CN−** yields in 3Σ+ and 3Π states are expected to be directly produced in an impulsive DEA process, leading to the small peaks at 0.05 and 0.40 eV and their anisotropic distributions (see Fig. 3c–f). On the other hand, as an example, the anisotropic angular distributions in Supplementary Fig. 2 of Supplementary Note 2 show the remarkable differences among the CN−** at 3Σ+/3Π state, the CN−** at the superexcited vibrational states (ν ≥ 18) of 1Σ+ and the CN− at ν < 17 of 1Σ+. It is beyond the present scope to gain more details about their different DEA dynamics, and the sophisticated theoretical calculations are demanded.

Fig. 4: Potential energy curves of CN(X2Σ), CN−(X1Σ+) and CN−**(3Σ+, 3Π and 1Π).
figure 4

In the inset panel, the horizontal broken line denotes the zero-point vibrational energy level of CN (X2Σ), the vibrational (horizontal parallel lines) and electronic states above these broken lines are the superexcited states and embedded on the yellow background, 1 represents the direct production of CN−** (3Σ+, 3Π) in the DEA process of BrCN or by electron attachment to CN(X2Σ) and 2 is the intersystem transition from the high-lying vibrational states of CN−(X1Σ+) to the superexcited state 3Σ+ of CN−**.

Full size image

Besides the direct production of the 3Σ+- / 3Π-state CN−** (its branching ratios are shown in Supplementary Fig. 3 of Supplementary Note 4), an intersystem transition from the superexcited vibrational state of 1Σ+ to the bound state 3Σ+ or 3Π is feasible. Similarly, an inverse internal conversion from a vibrationally-hot ground electronic state to a bound electronic excited state was observed33; more recently, thermionic emission on a millisecond timescale from the vibrationally-hot anion was also reported34. As illustrated in the inset panel of Fig. 4, the intersystem transition (denoted as pathway 2, while pathway 1 represents a direct pumping to 3Σ+) from the vibrational states above ν ≥ 20 of X1Σ+ to 3Σ+ state is conceptually analogical to the singlet-triplet state intersystem transition of the neutral species. Once the triplet states 3Σ+ or 3Π is populated (regardless of pathway 1 or 2), the fluorescence decay of CN−** is unpermitted due to the spin-forbidden rule. We tried to detect the possible phosphorescence of CN−**(3Σ+ or 3Π) → CN−(X 1Σ+) or fluorescence of CN−**(high-lying singlet states25) → CN−(X 1Σ+) using the spectrometer combined with the present apparatus35 but no photoemissions were detected, which indicates an ultralong lifetime of the triplet-state CN−** and the absence of the superexcited singlet-state CN−**. Considering the flying time in the VMI measurements (see Supplementary Fig. 4 and Supplementary Note 5), we conclude that the lifetime of the superexcited triplet-state CN−** is more than 5 μs.

The lifetime of the resonant-state anion is influenced not only by the electron autodetachment but also the nuclear motions. With the C–N bond elongation (from the neutral equilibrium bond length 1.17 to 1.37 Å), the R-matrix calculation predicted a sharp increase of the lifetime (approximately as ħ/2Γ, where Γ is the energy width) of the resonant state 3Σ+ (ref. 23), indicating again that the structural relaxation does enhance the anionic stability. Despite the lack of a theoretical model of the vibration effect on resonant-state lifetime, the anionic lifetime is potentially enhanced by the molecular vibrations17,18,19 or rotations36,37. A high vibrational-state density17,18,19 is expected to prolong the lifetime38, while the highly rotating anion at the electronic ground state can survive further because an energy barrier impedes the dissociation37. Here some rotational states of the CN−/ CN−** yield may be populated, corresponding to the isotropic background of the outside anionic signals in Fig. 2c–f. Besides the possible roles mentioned above37,38, the long lifetime of the ro-vibrationally superexcited CN−** facilitates, in turn, the intersystem transition 1Σ+ (ν ≥ 20) → 3Σ+. More importantly, spin-orbit coupling in this anionic-state transition could be further strengthened with help of the electron continuum background, thus this intersystem transition could be achieved more efficiently than that of the neutral.

Here we report an experimental evidence that the CN−** anions in the superexcited electronic states 3Σ+ and 3Π and the vibrational states ν ≥ 18 of the ground state X1Σ+ are produced in the DEA to BrCN, and propose that the intersystem transition X1Σ+(ν ≥ 20) → 3Σ+ of the free fragment CN−** is favorable in dynamics. According to the operation condition of our VMI measurements, the CN−** lifetime is expected to be not less than 5 μs. This superexcited-state species is unlikely produced in the DEA to NCCN39 or other polyatomic molecules (such as CH3CN40), since the excess energy is apt to be transformed into the internal or translational energies of the multiple neutral products. The energies of the thermally equilibrium electrons in astrophysical environments are much less than 1 eV, but there are a lot of low-energy electrons, in particular, a peak distribution exhibiting an appealing coverage of 4–6 eV, at the altitude of 1015 km in Titan’s ionosphere7. The free electrons with above energies can be also produced promptly in the irradiation ionizations of the interstellar substances. Therefore, as described in Fig. 4, the superexcited-state CN−** are possibly produced by the electron attachment to CN radical (via pathways 1 and 2), because of many free CN radicals in the interstellar space. Furthermore, the long-term arguments4,10,11,12,13 about the CN− abundance in IRC + 10,216 are hopefully settled down if the contribution of the superexcited-state CN−** is considered. In addition, a CN−**-involved reaction proceeds much more readily than that of the ground-state CN−, for example, NCO− in the L134N dark molecular cloud41,42 could be formed in a barrierless reaction CN−** + O2 → NCO− + O. The present finding arouses attention on the potential roles of the superexcited bound states of anionic species in astrochemistry.

Methods

Experiments

In our laboratory, anionic high-resolution velocity (or momentum) imaging has been realized27,28,29,30, owing to the application of a trochoidal electron monochromator. In the experiments, within a crossed-beam arrangement, the monochromatized pulsed electrons (with an energy spread about 100 meV here, along y axis) are guided to the reaction region with a homogenous magnetic field (70 G) which is produced with a pair of Helmholtz coils; then the anionic yields are pushed out, accelerated, and flying (along z axis) through the VMI lens system. In the flight of a given type of anions, different kinetic energies correspond to a set of concentric Newton spheres with different radii. The accelerated anions are detected with double multichannel plates plus a phosphor screen (in x–y plane). The central time-sliced image of the Newton spheres is recorded with a CCD camera mounted behind the phosphor screen and by applying a short high-voltage pulse on the rear multichannel plate. This pulse is also used as the mass gate to selectively detect the anionic yields. The solid sample BrCN is purified with several liquid-nitrogen freeze-pump-thawed cycles before the measurements, then introduced into the chamber with an inlet nozzle. Its volatility at the ice-water mixture temperature is high enough to create a sufficient concentration of target molecules in the gas phase, and the ambient pressure is controlled at ca. 10−6 Torr during the VMI measurements.

Calculations

The potential energy curve calculations are carried out for the neutral and anionic cyano by using the internally contracted multi-reference configuration interaction (icMRCI)43 and the modified aug-cc-pVQZ basis set44 by supplementing one diffuse s-function both on C and N atom. The active space consists of the valence orbitals 3σ–6σ, 1π and 2π, while the core orbitals, 1σ and 2σ, are frozen. The MRCI treatment of electronic states is acquired in the state-averaged complete-active-space self-consistent field calculations on given spin-space symmetry species. More information can be found in Supplementary Note 3.