Gas phase $C6H6−$ anion: Electronic stabilization by opening of the benzene ring

Benzene with its planar hexagonal form of D_6h symmetry represents one of the most iconic structures in chemistry. An interesting question arises on how this structure changes upon varying its charge state. The benzene radical cation, $C6H6+$⁠, is Jahn–Teller active and thus has two structural forms—the compressed and elongated rings. They are energetically very close to each other, and their abundance depends on the elementary environment of the cation.¹ Even more interesting is the benzene dication, $C6H62+$⁠, which has a number of stable isomers with the most stable structure being the pyramidal one with a C₅H₅ base and a CH group at the apex.² Nonetheless, independent of these structural peculiarities, both the cation and dication are directly observed, e.g., in the standard electron impact mass spectra of benzene.

The situation is very different for the benzene radical anion, $C6H6−$⁠, which so far has not been detected in the isolated form. The reason is that anionic states of benzene are embedded in the continuum and the electron thus undergoes autodetachment on an ultrafast timescale. For example, the lowest anionic state, which is an e_2u shaped resonance, lies 1.12 eV above the neutral benzene,³ and its electronic width of 0.12 eV corresponds to the autodetachment lifetime of 27 fs. Upon ring distortion, the energy of this resonant state slightly decreases;⁴ however, no configuration has been found where this anion state is below the neutral, i.e., the gas phase radical anion is not electronically stable. However, this situation changes dramatically upon solvation. The benzene anion $C6H6−$ is electronically stabilized and readily observed in polar solvents, e.g., in liquid ammonia⁵ and in ammonia clusters.⁶ 

The above considerations govern the outcome of reactions of benzene-containing species with free electrons. For a gas phase benzene molecule, the formation of low-lying shape resonances merely leads to a considerable vibrational excitation.³ For high lying core-excited resonances, the dissociative electron attachment channel opens, which leads to hydrogen abstraction; the $C6H5−$ cross section peaks at 8 eV with a shoulder of up to 10 eV.⁷ The second fragment, observed at a similar energy range, is $C2H2−$⁠.⁷ Upon electron attachment to large benzene clusters at low energies (∼0.3 eV), benzene cluster anions $(C6H6)n−$ were detected for sizes n ≥ 53.⁸ This was interpreted as the formation of a solvated electron in benzene, which was supported by photoelectron spectroscopy. No negatively charged benzene clusters were observed for smaller sizes at these low electron energies. However, at much higher electron energies of hundreds of eV, cluster anions with n ≤ 52 (down to n = 2) were observed through fragmentation of large $(C6H6)n−$ clusters. Mitsui et al.⁸ noted that the structures and energy states of small benzene clusters, e.g., the dimer $(C6H6)2−$⁠, remain elusive theoretically. In this paper, we report the first observation of a stable isolated monomer $C6H6−$ anion with the lifetime longer than ten microseconds and we attribute its formation to the opening of the benzene ring and the formation of chain-like structures.

We generated benzene clusters by the continuous supersonic expansion of benzene vapor with an argon buffer gas. The gas was expanded through a conical nozzle (90 μm diameter, 30° full opening angle, and 2 mm length), which was attached to the benzene containing reservoir placed in the source vacuum chamber. The reservoir and nozzle temperature of T_R ≈ 300 K and T₀ = 333 K, respectively, were kept constant by independent heating. The Ar stagnation pressure was P₀ = 3 bar, i.e., the benzene concentration in the gas mixture with Ar was about 5%. These expansion conditions result in (C₆H₆)_N with the mean neutral cluster size of $N̄≈$ 300. This cluster size was determined in our recent pickup experiments.⁹ 

In the present case of relatively large neutral (C₆H₆)_N clusters, the measured cluster ion fragment (positively or negatively charged) mass spectra do not fully reflect the neutral cluster size distributions. First, they suffer from the cluster fragmentation upon ionization, and second, our time-of-flight mass spectrometer in perpendicular arrangement has a limited mass range (m/z ≈ 700), within which the spectra can be obtained without any mass discrimination. The upper mass limit can be pushed to much higher masses using deflection voltages; nevertheless, the mass range still remains discrimination-free within the mass interval Δm/z ≈ 700. Thus, the pickup experiment represents a more reliable method to determine the mean neutral cluster size. A detailed discussion can be found in Ref. 9. Nevertheless, in the present investigation of the negatively charged benzene clusters, we concentrate on the mass region below m/z ≤ 700.

After the expansion, the clusters pass through a skimmer followed by three differentially pumped vacuum chambers, and they are detected in a fourth one by using a reflectron time-of-flight mass spectrometer (TOF), first implemented and described elsewhere.^10,11 Either it can detect positive ions^10,11 or it can work in the negative ion mode^12,13 employed in the present experiments. The clusters are ionized by an electron beam pulsed at a frequency of 10 kHz. The pulse width is 5 µs, and the extraction and acceleration voltages of 4 and 8 keV, respectively, are applied with a delay of 0.1 µs to prevent any unwanted ionization with electrons accelerated by the extraction voltage. After the flight path of ∼95 cm in the reflectron TOF, the ions are detected with a microchannel plate detector and the mass spectra are recorded. The electron energy is scanned between 0 and 12 eV in steps of 0.2 eV to obtain electron energy dependent mass spectra and an energy-dependent negative ion yield. However, due to the very low abundance of the $(C6H6)n−$ ions, some of the presented spectra are recorded for a very long time (several repeated days) at a single electron energy of 8 eV, where the maximum of $(C6H6)n−$ ion signal occurs. The energy scale calibration is done using the 4.4 and 8.2 eV resonances in the electron attachment of CO₂ molecules.

All electronic structure calculations of the $C6H6−$ structures and their neutral counterparts were carried out using the ORCA 5.0.1 program package.¹⁴ Geometries of the anionic structures were first pre-optimized at the second-order Møller–Plesset perturbation (MP2)^15,16 theory level employing the cc-pVDZ basis set and then evaluated at the complete active space self-consistent field¹⁷ (CASSCF) level using the cc-pVTZ basis set.^18,19 Note that the CASSCF calculation always used MP2-based natural orbitals (NOs) as an initial guess calculated with the same basis set. The size of the active space was chosen based on the NO analysis such that orbitals with occupation numbers between 0.075 and 1.95 were considered as the active ones. This resulted in using the CASSCF-(7-6) method for the anionic structures and CASSCF-(6-6) for the neutral ones.

Final single-point energies were then calculated at complete active space perturbation theory¹⁷ (CASPT2) with the aug-cc-pVDZ basis starting again from natural orbitals calculated at the MP2/aug-cc-pVDZ level. Vertical binding energies (VBEs) were subsequently obtained as the difference between the single-point energy of the anionic and neutral structures at the geometry of the anionic one. Note that VBE is equivalent in absolute value but opposite in sign to the vertical detachment energy (VDE), i.e., negative VBE denotes a stable anion. Additional benchmarks concerning the convergence of the VBE with respect to the basis set and size of the active space are presented in the supplementary material.

Figure 1 shows the mass spectrum of negatively charged benzene clusters after the electron attachment to the pure (C₆H₆)_N clusters. The negative ions have extremely low intensities; therefore, the spectrum was recorded and accumulated for a long time (continuous 2–6 h measurements were repeated several times) at a constant electron energy of 8 eV. This energy roughly corresponds to the maximum signal for the $(C6H6)n−$ cluster ions as shown below. The major peaks in the spectrum correspond to the $(C6H6)n−1C6H5−$ and $(C6H6)n−$ series. For the first members of the series n = 1, the fragment $C6H5−$ ion at m/z = 77 is slightly dominating over m/z = 78; however, for n ≥ 2, $(C6H6)n−$ dominate over the neighboring $(C6H6)n−1C6H5−$ fragments. Apart from these, there is a weaker series starting at m/z = 110 (orange triangles) assigned to $(C6H6)nO2−$ due to a small oxygen impurity deposited in the clusters from the vacuum background, which is discussed below. Finally, there is a very weak series of mass peaks displaced by Δm/z = −15 from the $(C6H6)n−$ peaks (green diamonds) tentatively assigned to $(C6H6)n−1C5H2−$⁠, i.e., dissociation of CH₃ from benzene clusters upon the electron attachment.

First, we discuss the $(C6H6)nO2−$ series with the oxygen impurity. The presence of the oxygen containing clusters may be surprising since the background pressure in all chambers, which the clusters pass through, is in the range of 10⁻⁶–10⁻⁹ mbar and we do not see any evidence for the oxygen contaminated benzene clusters in the positively charged ion spectrum. The expected amount of oxygen picked up by the $(C6H6)n−$ clusters from the background is very small. We can make the following estimate—from previous experiments,⁹ we know the mean size of our neutral (C₆H₆)_N clusters of $N̄≈$ 300, which corresponds to the mean cluster radius of 2.2 nm. Thus, we can calculate the mean cluster geometrical cross section of about σ_g ≈ 1500 Ų. The clusters fly about 1.5 m in the vacuum of 10⁻⁶ mbar or lower. From this, we deduce that the uptake probability of a single molecule by the average cluster is ≤0.5, i.e., less than half of the clusters can pick up an O₂ molecule, and the uptake of more than one molecule is probable only for the largest, less abundant, clusters. However, oxygen molecules have large electron attachment cross sections.²⁰ Thus, the electrons attached temporarily to benzene clusters in a metastable state can be stabilized by the oxygen molecule and we may thus see the oxygen containing clusters in the negative ion spectrum despite their low abundance. To check this hypothesis, we deliberately doped the (C₆H₆)_N clusters with more O₂ molecules by filling the oxygen gas into a pickup chamber at the pressure of 1 × 10⁻⁴ mbar, increasing the mean number of collisions of an average cluster with O₂ molecules about 100-times, i.e., each cluster should collide with many (∼50) O₂ molecules. This experiment resulted in the spectrum completely dominated by the $(C6H6)nO2−$ series, while the $(C6H6)n−$ intensities remained unchanged by oxygen doping (Fig. S3 in the supplementary material).

As outlined already in the Introduction, the most interesting observation is the appearance of the $(C6H6)n−$ ions and of the $C6H6−$ monomer, in particular. However, one has to be careful here since there are isotope contributions from the $(C6H6)n−1C6H5−$ ions at the corresponding $(C6H6)n−$ ion masses (i.e., from the ¹³C isotope). Therefore, we show the spectrum of Fig. 1 in more detail in Fig. 2, where the orange crosses correspond to the calculated isotope contribution to the $(C6H6)n−$ ion peaks based on the $(C6H6)n−1C6H5−$ abundances. The isotope contribution clearly cannot account for the entire intensity of the $(C6H6)n−$ peaks. Nevertheless, we have made another test with deuterated benzene C₆D₆ to prove the presence of the $(C6H6)n−$ ions, including the monomer. Figure 3 shows the $C6D6−$ ion peak unambiguously and demonstrates that the isotope contribution of $C6D5−$ at m/z = 83 is small. Thus, all the experimental evidence proves unambiguously the generation of the $C6H6−$ ion.

To explore the origin of the $(C6H6)n−$ ions further, we show the energy-dependent ion yields of these ions in Fig. 4. Note that the top spectrum (green circles) shows the dependence for the $C6D6−$ ion offset by a factor of 10 in the y axis. The $(C6H6)n−$ and $(C6H6)n−1C6H5−$ ion yields are shown by the red squares and blue diamonds, respectively. The ions exhibit very low intensities, and thus, we have integrated several n = 1–4 ion peaks to obtain energy dependencies with somewhat reasonable signal-to-noise ratio. All the spectra exhibit a maximum around 8 eV. Despite the low signal, the peak position agrees well with the previously measured dissociative electron attachment (DEA) spectrum of benzene, yielding $C6H5−$⁠, where a single peak at 8 eV with a weak shoulder extending to 10 eV was observed.⁷ However, it should also be noted that $C2H2−$ ion fragments from the DEA of benzene molecule exhibited very similar energy dependence.⁷ 

We consider six linear structures that can form upon the benzene ring opening due to the electron attachment—a direct product of the ring opening carrying one hydrogen on each carbon atom (structure 1 in Fig. 5) and five structures with one shifted hydrogen atom (structures 2–5 in Fig. 5). Geometries of these radical anions were optimized according to the protocol described above, and their energies were then obtained at the CASPT2-(7-6)/aug-cc-pVDZ level. These energies, referenced to the total energy of neutral benzene at the CASPT2-(6-6)/aug-cc-pVDZ level, are represented in Fig. 5. It can be seen from Fig. 5 that energies of all the structures lie only up to 5 eV above the energy of neutral benzene, rendering the ring-opening event thermodynamically feasible as the energy of the incident electrons is about 8 eV.

Table I presents VBEs of the six linear structures. Unlike the benzene radical anion, which is electronically unstable in the gas phase (having a non-negative VBE),⁵ it can be seen from Table I that all the considered linear radical anion structures are electronically stable, possessing sizable negative VBEs.

The mass spectra in Figs. 2 and 3 show that the $C6H6−$ and $(C6H6)n−$ radical anions are generated in the process of attachment of slow (∼8 eV) electrons to large (⁠$N̄≈300$⁠) benzene clusters. Their abundance is low, yet their appearance is unambiguous. Experiments with the deuterated benzene provided further evidence that the monomeric signal indeed corresponds to the $C6H6−$ radical anion and not just to the isotope contribution of $C6H5−$⁠, which is a rather common product of the DEA process. The formed negative anions must be stable at least on the timescale of their flight time in our TOF spectrometer, which is about 10 µs for the $C6H6−$ monomer and even longer for $(C6H6)n−$ clusters.

Large $(C6H6)n−$ cluster anions with n ≥ 53 were reported in Ref. 8 to be formed in abundance via the generation of the solvated electron. We cannot compare yields of our smaller $(C6H6)n−$ clusters to the yields of the large ones since such large clusters cannot be detected in our TOF spectrometer due to its perpendicular arrangement. All the clusters generated in supersonic expansions attain essentially the same velocity with a very narrow distribution. When the ionized clusters are extracted in the TOF region perpendicularly to the beam velocity, they keep the component of velocity parallel to the beam, and thus, the heavier ones, which spend longer time in the TOF region, can escape the detection. We can partly compensate for that effect by applying a deflection voltage against the beam direction; however, we still cannot detect clusters of m/z ≥ 4000. A more detailed discussion of this effect in our TOF can be found elsewhere.²¹ 

Results shown in Fig. 4 suggest that all observed negative ions are generated by attachment of an electron with an energy around 8 eV. Note that at this energy, the DEA process in an isolated benzene molecule yields $C6H5−$ and $C2H2−$⁠,⁷ while we clearly see that the former anion (and its cluster complexes with benzene) $C2H2−$ is absent within our detection limits. This leads us to the conclusion that the core-excited resonance, formation of which triggers this channel, also leads to ring opening in the gas phase. There, however, the excess energy in the transient chain-like $C6H6−$ is too high, so it decays by further bond breaking to $C2H2−$ (or, possibly, by losing the excess electron). In the cluster, the excess energy is carried away by the environment and $C6H6−$ is stabilized. This cooling clearly works even if all intact benzene molecules are evaporated, which leads to the detection of the monomeric anion. The ring-opening by an electron attachment has been previously observed in a number of molecular targets.^22–25 The unique effect observed here is that the cluster cools down the open structure, which is thermodynamically unstable in the gas phase.

This effect is similar to the three-body stabilization of transient anions known from swarm physics.^26–29 It typically concerns anions with decay lifetimes longer than nanoseconds. At intermediate gas pressures, where such a lifetime is comparable to collision frequency with other gas molecules, the excess energy can be quenched in collisions before the anion decays. The resulting electron attachment rates and anion fragmentation patterns are then a sensitive function of the gas pressure. In clusters, the collision partners are immediately available in the vicinity of a transient negative ion and thus anions with extremely short lifetimes can be stabilized.

In conclusion, we have shown that electron attachment to benzene clusters leads to the formation of the $C6H6−$ anion, which is detected mass spectrometrically. The structure of this anion does not correspond to any distortion of the benzene ring. Instead, we show that there exists a number of linear structures with negative vertical binding energies. These structures lie considerably below the energy of the resonance, which triggers the ring opening. The energy released upon the ring opening causes that the $C6H6−$ anion is not observed in the electron attachment to isolated benzene, where the chain breaks into smaller fragments (with the anionic fragment being $C2H2−$⁠). In clusters, the excess energy is quenched and a “cold” $C6H6−$ anion is produced.

This finding should be taken into account when considering the chemical transformation of organic ring-compounds in environments with a high abundance of free electrons (i.e., in plasma or matter exposed to high-energy irradiation). The aggregation level, or even a local pressure, influences the quenching of the excess energy and thus determines the reaction outcome. We show that this happens even in an iconic compound, such as benzene.

See the supplementary material for benchmarks of the electronic structure calculations and additional mass spectra of benzene clusters doped with oxygen molecules.

The experimental group acknowledges the support from the Czech Science Foundation (GAČR) under Project No. 21-07062S (I.S.V. and M.F.) and from the European Regional Development Fund (project “CARAT” No. CZ.02.1.01/0.0/0.0/16_026/0008382) (A.P. and J.F.). The research stay of S.B. in Prague was supported by the Austrian Science Fund, FWF, Project No. W1259. P.J. acknowledges the support from the European Regional Development Fund (Project ChemBioDrug No. CZ.02.1.01/0.0/0.0/16_019/0000729). T.N. acknowledges the support from the University of Chemistry and Technology Prague where she is enrolled as a Ph.D. student. V.K. acknowledges the support from the Faculty of Science, Charles University, where he is enrolled as a Ph.D. student. T.N. and V.K. also acknowledge the support from the IMPRS for Many Particle Systems in Structured Environments.

The authors have no conflicts to disclose.

Andriy Pysanenko: Data curation (equal); Formal analysis (equal); Investigation (equal); Validation (equal); Visualization (equal). Ivo S. Vinklárek: Data curation (equal); Formal analysis (equal); Investigation (equal). Juraj Fedor: Funding acquisition (equal); Investigation (equal); Validation (equal); Writing – original draft (equal). Michal Fárník: Conceptualization (lead); Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (lead); Supervision (equal); Visualization (equal); Writing – original draft (lead). Stefan Bergmeister: Data curation (equal); Formal analysis (equal); Writing – review & editing (equal). Vojtech Kostal: Data curation (equal); Formal analysis (equal); Investigation (equal); Software (equal); Validation (equal); Visualization (equal); Writing – review & editing (equal). Tatiana Nemirovich: Data curation (equal); Formal analysis (equal); Software (equal); Visualization (equal); Writing – review & editing (equal). Pavel Jungwirth: Funding acquisition (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Validation (equal); Writing – original draft (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.