Threshold photoelectron spectroscopy of small organo-selenium radicals

Abstract

We present threshold photoelectron spectra of several reactive selenium-containing intermediates generated by pyrolysis of dimethyldiselenide, (CH₃)₂Se₂. The detected species include CH₃Se₂, Se₂ and CH₃Se radicals. Short-lived intermediates were produced in situ using a heated microreactor and characterized via mass-selected threshold photoelectron spectroscopy, using synchrotron radiation at the Swiss Light Source and a photoelectron–photoion coincidence (PEPICO) detection scheme. The data yield ionization energies for the transition into the ionic ground state of 8.28 eV ((CH₃)Se₂), 8.85 eV (Se₂) and 8.92 eV ((CH₃)Se). The values are accurate within ±10 to ±20 meV. In addition, transitions into excited electronic states of the ion were also investigated. Simulations of the vibrational structure based on quantum chemical calculations provide further insight into structural changes upon ionization. By directly comparing these results to previously studied sulfur and oxygen analogues, we identify both similarities and key differences in the bonding and fragmentation behavior of group 16 radicals.

---

Introduction

Until recently, the chemistry and biology of selenium-containing species were widely neglected. It was not until the 1950s that scientists found out that selenium is essential for animals and humans,^1,2 a discovery that resulted in increased interest in the biology and chemistry of selenium. Today, selenium-containing molecules represent an important and growing area of research, particularly in the context of cancer therapy.^3–6 Another growing research field involves selenium catalysis. The main applications involve organoselenium catalyzed oxygen-transfer reactions^7–10 as well as heteroatom oxidation of sulfur, selenium and nitrogen.^7,11–13 In the context of physical chemistry, there have mainly been NMR and X-ray spectroscopic studies which were used for structural characterization of selenium-containing compounds.^14–16

While the role of selenium in biological and catalytic systems has been widely acknowledged, the gas-phase chemistry of simple organoselenium compounds remains less explored, because so far, the chemistry of selenium and sulfur compounds has only been compared in solution.¹⁷ In particular, insights into the thermal decomposition and the electronic structure of reactive selenium-containing species are still limited. Dimethyldiselenide, (CH₃)₂Se₂, as a representative organoselenium compound, yields various reactive intermediates in a thermal reaction. To study these species, we employed a combination of pyrolysis and mass-selected threshold photoelectron spectroscopy (ms-TPES), which enables high-resolution probing of the electronic states of the resulting radical.^18–22 This study follows our previous investigation of the sulfur analogue, dimethyldisulfide (CH₃)₂S₂, for which we reported vibrationally resolved ms-TPES spectra of the pyrolysis products.²³ By applying the same method to dimethyldiselenide, we aim to compare the gas-phase behavior of sulfur- and selenium-containing species, highlighting similarities and differences in their fragmentation patterns and electronic structures.

One question that we address in our studies of low-valent main group compounds E–R (E = main group element) concerns the bonding motifs and the structural stability of the species formed in thermal reactions. Small radicals of the composition E–CH₃ (E = B,²⁴ Al,²⁵ Ga,²⁵ N,²⁶ P,²⁷ As,²⁸ Sb,²⁸ Bi,²⁹ O,³⁰ and S²³) exhibit distinct bonding preferences that are tied to the size and orbital characteristics of the central element E. For group 15 elements, it has been shown that the heavier members such as antimony and bismuth form σ-bonded methyl species (E–CH₃), while for the lighter congeners, phosphorus and arsenic isomers with E=C double bonds and a terminal hydrogen (H₂C=E–H) are also formed.^28,29 Finally, for E=N, only H₂C=N–H is observed.²⁶ This trend is attributed to the better orbital overlap in lighter elements, which allows for the formation of π-bonding interactions with carbon. A similar pattern was observed in studies of group 13: for boron, the only species detected was the π-bonded HB=CH₂, while methylboranes such as B–CH₃ were not observed.²⁴ In contrast, for the heavier homologues aluminum and gallium, only the E–CH₃ fragments were detected.²⁵ In this study, we extend this comparison to group 16. For sulfur, both CH₃S and CH₂SH have been found. Here, we will search for the equivalent Se isomers. Using the pyrolysis of dimethyldisulfide in combination with ms-TPES, we aim to determine whether selenium behaves similar to sulfur or follows the heavier-element trend, favoring simple σ-type methyl radicals.

Methods

Experimental

The experiments were conducted at the vacuum ultraviolet (VUV) beamline of the Swiss Light Source (SLS) at the Paul Scherrer Institute, Villigen, Switzerland. We utilized the CRF-PEPICO double imaging photoelectron–photoion coincidence setup. Since the setup has been described in detail elsewhere,^31–34 only a concise summary is provided here. Synchrotron radiation generated by a bending magnet was first collimated using a mirror and then vertically dispersed by a 150 lines per mm plane grating. To suppress higher-order harmonics, the beam was passed through a Ne/Ar/Kr gas filter. Photon energy calibration was achieved using the 11s′–13s′ autoionizing transitions of argon, measured in both the first and second diffraction orders. This setup enabled a spectral resolution of 5 meV at a photon energy of 7.882 eV.

Dimethyldiselenide (Se₂(CH₃)₂), purchased from ABCR, served as the precursor. The reactive intermediates were formed in a resistively heated Chen-type pyrolysis reactor made from silicon carbide,³⁵ with the precursor vapor seeded in argon. The pyrolysis products passed through a 2 mm skimmer into the main chamber, where they intersected the synchrotron beam and were photoionized. The resulting electrons and ions were extracted in opposite directions by a uniform electric field of 218 V cm⁻¹. Electrons were detected using a Roentdek DLD40 delay-line detector, while cations were analyzed using a Wiley–McLaren time-of-flight (TOF) spectrometer. To analyze only threshold photoelectrons, contributions from hot electrons were removed from the central region of the Newton sphere.³⁶ Detection of photoions and threshold electrons was carried out in coincidence in a multiple-start/multiple-stop configuration.³⁷ Following the method of Hemberger et al.,³⁸ velocity map imaging (VMI) was used to discriminate between rethermalized and effusive beam ions. Only rethermalized ions, i.e., those undergoing wall collisions, were considered in the final analysis. Consequently, the TPE spectra show thermal conditions near room temperature, have minimal hot-band interference, and benefit from enhanced resolution.

Theory

Calculations of ionization energies, optimized geometries, and vibrational frequencies were carried out using the ORCA³⁹ and Gaussian 16⁴⁰ software packages. Calculations were performed using methods like DFT (density functional theory), CCSD (coupled cluster including single and double excitations), CCSD(T) (triple excitations are perturbatively treated), CASSCF (complete active space self-consistent field) or CASPT2 (second-order complete active space perturbation theory), depending on the complexity of the given molecule. Information on the details of the computations, like the chosen functional and basis sets, is provided along with the results in the SI. Adiabatic ionization energies (IEs) were obtained as total energy differences between the cationic and neutral species, both in their vibrational ground states, incorporating zero-point energy corrections. Franck–Condon (FC) simulations at 300 K were performed using ezSpectrum based on the computed geometries and harmonic frequencies.⁴¹ The stick spectra were convolved with Gaussian functions to generate the FC simulations shown below. Taking into account both photon (5 meV) and electron energy resolution (10 meV), the combined experimental uncertainty is approximately 11 meV. This is generally smaller than the vibrational bandwidths. The band envelope is well described by a Gaussian function. Its half-width at half-maximum (FWHM/2) was used to estimate the error margin of the IEs. Typically, a FWHM of 25 meV resulted in IE uncertainties of ±0.013 eV.

Results and discussion

Mass spectra

Mass spectra are recorded to gain information on the dissociation pathways at different temperatures and hence analyze reaction products. Fig. 1 shows TOF mass spectra (TOF-MS) of (CH₃)₂Se₂ at a photon energy of 9.5 eV and at three different pyrolysis temperatures. At this photon energy, the precursor does not show dissociative photoionization (DPI), while reaction products that are generated through pyrolysis can still be ionized.⁴² Selenium has several naturally occurring isotopes, of which ⁸⁰Se is the most common, accounting for roughly 50% of the signal. Below, we will hence only refer to the ⁸⁰Se isotopologue when discussing the mass spectra.

Fig. 1 TOF mass spectra at a photon energy of 9.5 eV, recorded at reactor temperatures of (a) 300 K, (b) 600 K and (c) 800 K. At room temperature, only the precursor (CH₃)₂Se₂ and a contamination from previous experiments (asterisk) are present. At higher pyrolysis temperatures, CH₃Se₂, Se₂ and CH₃Se are detected as well. The asterisk marks a contamination from a previous experiment.

Trace (a) shows the TOF-MS spectra at room temperature. Here, only the precursor with m/z 190 and a contamination from previous experiments with m/z 94 are detected. The energetically lowest DPI channel to C₂H₅Se with m/z 109 has an appearance energy (AE) of 10.09 ± 0.06 eV at 0 K.⁴² At 300 K, no ion signal was detected at this m/z. However, small amounts of this species may be present at 600 and 800 K due to the redshift of the AE at higher temperatures.

Trace (b) shows the pyrolysis products at 600 K. Here, small amounts of m/z 175 (CH₃Se₂) are formed, which indicate a first methyl abstraction. A consecutive methyl cleavage can lead to Se₂ with m/z 160. Two more mechanisms are possible, but less likely for Se₂, simultaneous abstraction of two methyl groups from the precursor or a concerted abstraction of ethane. Another intermediate is CH₃Se with m/z 95. This species can be formed either by homolytic bond cleavage of the precursor or through a Se loss in CH₃Se₂.

Trace (c) was recorded at a pyrolysis temperature of 800 K. The precursor is almost completely pyrolyzed and Se₂ becomes the dominant signal. Small amounts of m/z 220 ((CH₃)₄Se₂) and 270 ((CH₃)₂Se₃) – both not shown – are detected in the reactor at this temperature, indicating that a small amount of bimolecular reactions takes place. m/z 220 could be formed through reactions of the precursor and ethane or two methyl groups. Both ethane and methyl are present and were detected at higher photon energies. Another possible reaction could be dimerization of (CH₃)₂Se. However, the mass of this molecule (m/z = 110) was not observed. For m/z 270 CH₃Se + CH₃Se₂ seems to be the most reasonable reaction pathway.

While mass spectrometry is limited to revealing the elemental composition of pyrolysis products, TPE spectroscopy provides deeper insights into the molecular structures and their electronic states. In the next sections, the TPE spectra of CH₃Se₂, Se₂ and CH₃Se are analyzed. As the TPE spectrum of the precursor shows no vibrational structure, it is only shown in the SI Fig. S1.

ms-TPES of CH₃Se₂

Fig. 2 shows the TPE spectrum of m/z 175, CH₃Se₂. In the neutral ground state, it is a radical with an electron in the A″ π*-like HOMO. The TPES spectra exhibit four characteristic bands. The first band from 8.2 eV to roughly 8.5 eV is broad and unstructured and is identified as the X^{+ 1}A′ ← X²A″ transition. It is due to ejection of an electron from the antibonding HOMO. The calculated IE is 8.50 eV at the DFT/ωB97X-D3/cc-pVTZ level and 8.35 at the CCSD/cc-pVTZ level, with the latter agreeing quite well with the experimental IE of 8.28 ± 0.012 eV. The ionization leads to a reduction of the Se–Se bond length from 2.23 Å to 2.13 Å. This results in a FC simulation (blue trace) where mostly only the Se–Se stretching mode ν⁺₁₀ = 420 cm⁻¹ is excited. Furthermore, torsional motion of the CH₃ group leads to signifcant broadening of the band. Combination modes of ν⁺₁₀ and the CH₃ torsion ν⁺₁₂ (200 cm⁻¹) are also present.

Fig. 2 ms-TPE spectrum of m/z 175, which corresponds to the molecule CH₃Se₂⁺. It was recorded at a pyrolysis temperature of 730 K. From the neutral ²A″ ground state, transitions into the cationic ground state X^{+ 1}A′ (blue FC simulation) and three electronically excited states are observed, including the a^{+ 3}A″ (red), A^{+ 1}A″ (purple) b^{+ 3}A′ (orange).

The second band starts at 8.6 eV and has a partially resolved vibrational progression up to roughly 9 eV. Here, the HOMO−1 is ionized. This can lead to either the ³A″ triplet state or the ¹A″ singlet state. As the calculated IE to the ³A″ state of 8.63 eV (DFT) or 8.53 eV (CCSD) is in better agreement with the experimental IE of 8.69 ± 0.012 eV, we assigned this band to the transition into the first triplet state, a^{+ 3}A″ ← X²A″. The main feature of this band is the double peak at 8.70 and 8.74 eV, followed by smaller not well resolved signals around 8.85 eV. According to our calculations, the geometry does not change much upon ionization. The biggest changes in geometry are a reduction of the C–Se–Se angle by 2° and a reduction of the C–Se bond length by 0.02 Å. Thus, it is not surprising that the TPES shows an intense origin transition. According to the FC simulation (red trace), the second rather intense peak at 8.74 eV is due to the change in the C–Se–Se bond angle. However, the computed wavenumber of 160 cm⁻¹ for the C–Se–Se bending mode ν⁺₁₁ is considerably smaller than the experimental value of +320 cm⁻¹. Most likely, a very flat potential along this low-energy bending mode is responsible for the deviation between the experimental and computed wavenumber.

Ionization from the HOMO−1 can also yield the A^{+ 1}A″ singlet state of the cation. For the transition into this state, we calculated an IE of 9.27 eV by TD-DFT. The experimental IE is found to be 9.10 ± 0.012 eV. Similar to the transition into the triplet state, the main feature is a single band, without visible bending mode progression. The C–Se–Se bond angle only changes by 1° according to our calculations, in agreement with the absence of a progression. The signal/noise ratio inhibits the observation of low intensity transitions visible in the simulation (purple trace). The ratio of the integrated intensity between the singlet and triplet state is 3.09, very close to the ideal ratio of 3.

A fourth band appears around 9.7 eV, resulting from the ionization of an electron from the HOMO−2. The b^{+ 3}A′ ← X²A″ transition was calculated to be 9.78 eV. The experimental IE is hard to determine since the band is broad and the signal is small. The best match with the simulation is achieved with an IE of 9.88 eV. Upon ionization, the main changes in the geometry involve a 0.04 Å increase in the Se–Se bond and a reduction in the H–C–Se angle of 5°, resulting in a H–C–Se–Se angle of 180°. Compared to the other cationic states, the geometry change is larger and therefore leads to a broad and unresolved band that is also reflected in the simulation (orange trace). The dominant vibrations in this transition are the Se–Se–C bending mode ν⁺₁₂ = 250 cm⁻¹, the Se–Se bending mode ν⁺₁₁ = 380 cm⁻¹ and the CH₃ torsion mode ν⁺₁₀ = 560 cm⁻¹. Combination bands of these fundamental modes are present, which contribute to the unresolved structure.

The calculated and experimental IEs of CH₃Se₂ and its cations as well as of all other investigated species are summarized in Table 1. Details of the computed geometries and frequencies of the investigated species are given in SI Tables S1–S30.

Table 1 Experimental and calculated ionization energies of investigated selenium species and a comparison with the lower sulfur homologue

Species	State	IEexp/eV (lit.)	IEcalc/eV	Sulfur species
a Calculated at the CCSD/cc-pVTZ level. b Calculated at the DFT/ωB97X-D3/cc-pVTZ level. c Calculated at the CCSD(T)/cc-pVTZ level. d Calculated at the CASPT2/cc-pVTZ level.
CH3Se2	^1A′	8.28	8.35^a	8.61^23
			8.50^b
	^3A″	8.69	8.53^a	9.17^23
			8.63^b
	^1A″	9.10	9.27^b	9.65^23
	^3A′	9.88	9.78^b	—
Se2	^2Π1/2	8.85 (8.70)^43	8.84^c	9.371^44
	^2Π3/2	9.12 (8.94)^43	9.09^c	9.428^44
	^4Πu	10.4 (10.68)^43	10.29^c	11.595^44
CH3Se	^3A2	8.92	8.50^d	9.25^23
			8.84^b
	^1E	9.91	9.74^d	—
	^1A1	10.89	10.68^d	10.22^23

---

ms-TPES of Se₂

Cleavage of the residual methyl group from CH₃Se₂ yields Se₂ with m/z = 160, which is formed at elevated temperatures around 800 K. A conventional photoelectron spectrum with similar appearance but only partial vibrational resolution has been recorded previously, yielding a less precise IE and limited information on the vibrational states of the ion.^43,45 In the TPES spectra, two well-resolved bands can be observed, the first one from 8.8 to 9.3 eV (Fig. 3a) and the second one from 10.3 to 11 eV (Fig. 3b). In its neutral ground state, Se₂ has an electronic configuration of …(σ_g)² (σ_u)² (σ_g)² (π_u)⁴ (π_g)², resulting in an X ³Σ_g⁻ ground state, similar to O₂ and S₂.

Fig. 3 Ms-TPE spectra of m/z 160 corresponding to Se₂⁺: (a) the spin–orbit split X^{+ 2}Π_g ← X ³Σ⁻_g transition along with its Franck–Condon (FC) simulation (blue) and (b) the a^{+ 4}Π_u ← X ³Σ⁻_g transition with its corresponding FC simulation (red). The yellow trace is a FC simulation with an adjusted bond length of 2.40 Å.

Upon ionization of an electron from the HOMO, a degenerate X^{+ 2}Π_g ionic ground state is obtained, which is split due to spin–orbit coupling into an X^{+ 2}Π_1/2,g and an X^{+ 2}Π_3/2,g component. The first band is composed of these two transitions. The IEs are 8.85 ± 0.012 eV (Ω_1/2) and 9.12 ± 0.012 eV (Ω_3/2), respectively. Hence, a spin–orbit splitting of 0.27 ± 0.025 eV was measured. These values are in very good agreement with our calculated data. The X^{+ 2}Π_g ← X ³Σ_g⁻ transition is found to occur at 8.84 eV at the CCSD(T)/cc-pVTZ level and the spin–orbit splitting was found to be 0.25 eV at the CASPT2/cc-pVTZ level. The blue line represents a simulation that is the sum of the individual contributions (purple and orange lines, respectively), with an experimental intensity ratio of 2.2 : 1. Due to the presence of spin–spin and spin-rotation coupling in the neutral ³Σ state, transition probabilities for the two spin–orbit components in the cation are difficult to predict. The FC simulation fits the ms-TPES spectra very well; only small deviations are observed. In the experimental data, a small difference in the Se–Se stretching mode wavenumber in the two spin–orbit states can be observed, which is ν⁺(²Π_1/2) = 450 cm⁻¹ and ν⁺(²Π_3/2) = 400 cm⁻¹. According to our calculations, the neutral ground state has a wavenumber of ν(³Σ_g⁻) = 390 cm⁻¹, a value very similar to that of the cations, even though the bond length is reduced by 0.07 Å by ionization.

When the electron is removed from the π_u HOMO−1, five different final electronic states are possible: ⁴Π_u, ²Φ_u and three ²Π_u states. Of these states, only the first was observed in the energy range investigated in the present work, as concluded from computations. Fig. 3(b) shows the ⁴Π_u ← ³Σ_g⁻ transition from 10.3 to 10.9 eV. The corresponding FC simulation is given as a red line. The experimental data show a vibrational spacing of 320 cm⁻¹ and an IE of around 10.4 eV. This is in very good agreement with our calculations, which also predict a spacing of 320 cm⁻¹ and an IE of 10.29 eV at the CCSD(T)/cc-pVTZ level. The calculation indicates a significant bond length increase of 0.15 Å upon ionization, which is also the reason for the long progression and the reduced vibrational wavenumber. However, the intensity of the high energy bands is not described well in the simulations. One explanation for the discrepancy is a change in the bond length which is even larger than that calculated. To check this hypothesis, we performed simulations for longer bond lengths. The best fit was found for a bond length of 2.40 Å, which is depicted as a yellow trace. The frequency was not adjusted. The stick spectrum is not shown for clarity. As can be seen, this simulations fits the spectrum better than the one based on the computationally optimized bond length (red line), indicating an even larger geometry change than obtained through our calculations. Alternatively, the discrepancy might be attributed to the presence of transitions into the spin–orbit components of the ⁴Π_u state.

ms-TPES of CH₃Se

The ms-TPES spectra of m/z 95 are given in Fig. 4. Two isomers could potentially contribute to the TPES spectra, CH₃Se and CH₂SeH. DFT calculations reveal that the IEs of CH₂SeH into the singlet and triplet state of the cation are 7.56 and 9.44 eV, respectively. Since there is no ion signal at m/z 95 before 8.8 eV and almost no TPE signal at 9.44 eV, we conclude that CH₃Se is the sole species present, even when taking into account that adiabatic IEs computed by DFT can deviate by several tenths of an eV. A prior photoelectron spectrum of CH₃Se was reported by Sun et al.; however, only the first band is in accordance with our ms-TPES spectra.⁴⁶ The other two bands at 9.26 and 9.71 eV do not match the values of 9.91 and 10.89 eV obtained by us. As our photoelectron spectra are ion-mass selected, they are less prone to interference from side products and thus more reliable.

Fig. 4 Ms-TPE spectrum of m/z 95 (CH₃Se⁺). The blue FC simulation shows the X^{+ 3}A₂ ← X ²A′ transition, while the red FC simulation describes the transition into the degenerate a^{+ 1}E electronic state. The orange trace simulates the transition into the excited b^{+ 1}A₁ singlet state.

CH₃Se, like CH₃S, possesses D_3h symmetry and is subject to a Jahn–Teller (JT) effect, which leads to a distortion of the ground state geometry. As a consequence, symmetry is reduced to C_s and an X ²A′ ground state results. Our computations yield a barrier of 260 meV to the undistorted D_3h symmetry. After ionization of an electron from the HOMO, a (π*)² occupation remains, giving rise to three electronic states, ³A₂, ¹E and ¹A₁. For all states, our CASPT2 calculations reveal C_3v symmetry with degenerate π* orbitals. The first band, from 8.9 to 9.2 eV, was assigned to the X^{+ 3}A₂ ← X ²A′ transition. The symmetry increase leads to an excitation of the degenerate CH₃ tilting mode ν⁺_7,8 = 830 cm⁻¹. Additionally, since an antibonding π*-orbital is depopulated, the Se–C bond length decreases by 0.04 Å and the Se–C stretching mode ν⁺₉ = 590 cm⁻¹ is also excited. The FC simulation is depicted as a blue trace in Fig. 4. Interestingly, an IE of 8.50 eV was calculated using CASPT2, which is considerably lower than our experimental IE of 8.92 ± 0.012 eV. In contrast, DFT computations yield a better value of 8.84 eV for the IE. However, a better fit was obtained with a simulation based on the CASPT2 geometries and frequencies. As visible, the overall shape of the band is reproduced without explicitly taking the Jahn–Teller effect into account. However, the band intensities are represented less well, indicating that Jahn–Teller distortion is not negligible.

The second band from 9.9 to 10.3 eV is assigned to the transition into the a^{+ 1}E state. Two electrons remain in one π* orbital, and according to our computations, the cation still maintains C_3v symmetry. The excellent agreement between the experiment and simulation (red trace) confirms the computations. When forcing the cation into an A′ structure without degenerate orbitals, the state is destabilized by 1.06 eV (CASSCF) or 0.69 eV (CASPT2). A FC simulation of the transition into an artificially distorted a^{+ 1}A′ state does not match the experiment. This confirms that JT distortion is small. In the ms-TPES of this transition, an IE of 9.91 ± 0.012 eV and a vibrational spacing of 640 cm⁻¹ are observed. Similar to the transition into the X^{+ 3}A₂ state, the calculated IE of 9.74 eV underestimates the experimental value. The FC simulation (red) reveals that the observed vibrational structure is due to activity in the Se–C stretching mode, with ν⁺₉ = 540 cm⁻¹. This mode is active even though the Se–C bond length only increases by 0.02 Å. The visible deviation between the simulation and experiment for n > 2 indicates that the computations underestimate the wavenumber of the mode and the influence of the JT effect cannot be completely ignored. Despite the change in symmetry upon ionization, no other modes are visible in the TPES or the simulation.

The third band is assigned to the b^{+ 1}A₁ ← X ²A′ transition and yields an IE of 10.89 ± 0.012 eV. The CASPT2 value of 10.68 eV underestimates the experimental value only slightly. This electronic state is formed by ionization of the electron from the HOMO−1 orbital and results in an electronic configuration where both π* orbitals are singly occupied by electrons with antiparallel spin. The ms-TPES reveals a vibrational progression with a spacing of 640 cm⁻¹, a value similar to the one observed for the ¹E state. However, the geometry change upon ionization differs slightly. According to our calculations, the Se–C bond length increases by 0.09 Å, while the geometry of the CH₃ group is similar to that observed for the ¹E state. The FC simulation (orange trace) also shows only the Se–C vibration at ν⁺₉ = 720 cm⁻¹. This agrees well with the observed spacing of 640 cm⁻¹. Higher quanta than n > 2 are not observed because of the low intensity of the transition.

Discussion

One goal of this line of our research is to rationalize the observed species and their TPE spectra by the position of element E in the periodic system. Below, we will therefore compare the results obtained for Se species not only with those observed for their sulfur congeners²³ but also with previously investigated oxygen-centered species. Also, a comparison with the results of group 13 and 15 compounds is given.

Some differences are expected and quite easy to understand. For example, in the pyrolyis, CH₃Se appears already at 600 K, while the appearance of CH₃S required heating to 800 K. This suggests that (CH₃)₂Se₂ dissociates more easily, due to a weaker Se–Se bond compared to the S–S bond. In fact, our calculations give 217.5 kJ mol⁻¹ for the loss of Se from (CH₃)₂Se₂, while experimental data report 284 kJ mol⁻¹ for S-loss from (CH₃)₂S₂.⁴⁷

The slow photoelectron spectrum of S₂, the lighter analogue of Se₂, was previously studied by Hrodmarsson et al.⁴⁴ As expected, the spin–orbit splitting is much smaller in S₂ (57 meV) than in Se₂ (270 meV), consistent with the heavy atom effect. This is in agreement with the group 15 analog. There, As₂⁺ has a spin–orbit splitting of 240 meV,²⁸ Sb₂⁺ has a spin–orbit splitting of 388 meV²⁸ and P₂⁺ has a spin–orbit splitting of only 26 meV.⁴⁸ As observed for the other selenium species, the ionization energies of Se₂ are slightly lower than those of S₂. Additionally, in the spectrum of S₂, more vibrational quanta are excited upon ionization in both ²Π_g bands, implying a larger geometry change between the neutral and cationic states. Interestingly, this trend reverses in the ⁴Π_u band: while the S₂ spectrum shows only six quanta of vibration, the selenium spectrum reveals up to eleven. For O₂, the ⁴Π_u state shows an even larger progression with up to 21 excited quanta.⁴⁹

The appearance of radicals of the composition CH₃E₂ is unique to 6th main group elements. Further comparisons can be made between CH₃Se₂ and its lighter homologues CH₃S₂ and CH₃O₂. The electronic structure of CH₃Se₂ and CH₃S₂ is broadly similar: the order of the ionic states remains the same, and only the ionization energies shift to lower values in the selenium compound. This shift is attributed to the larger and more diffuse orbitals of selenium, which require less energy to ionize. A detailed comparison of ionization energies is provided in Table 1. Additionally, both species show similar vibrational patterns. The X^{+ 1}A′ ← X ²A″ transition is broad in each case, and higher-lying bands exhibit the strongest intensity at the 0–0 transition, followed by a steep drop in the signal. Despite minor differences in individual vibrational features, the overall spectral shape highlights the similarity between the two compounds. For the lighter homologue CH₃O₂, only the lowest two cationic states were studied.⁵⁰ Here, the triplet state X^{+ 3}A″ is lower in energy with a calculated IE of 10.21 eV, while the higher lying singlet state a^{+ 1}A′ has an IE of 10.37 eV. In contrast to the higher homologues, these two states overlap in CH₃O₂ due to their broad vibrational progressions. Hence, the sequence of the cationic states changes from oxygen to sulfur, but remains constant from S to Se.

The structure of species with the composition CH₃E is central to our investigation of main group element radicals. Both CH₃S and CH₃Se show a similar structure for the X^{+ 3}A₂ ← X² A′ and b^{+ 1}A₁ ← X ²A′ transitions. The IE for transition into the ³A₂ state in CH₃Se follows the general trend observed for selenium compounds: the band is shifted to lower energies due to ionization from more diffuse orbitals. Interestingly, the IE into the ¹A₁ state is higher compared to that of its sulfur analogue. However, a significant difference arises in the ¹E state. In CH₃S, a strong (JT) interaction leads to a strong structural distortion and as a result, the ¹E state is not observed in the spectrum; instead, its isomer CH₂SH appears.²³ In contrast, in CH₃Se, the transition into the ¹E state of the cation is identified, suggesting a much smaller distortion.²⁸ No evidence for CH₂SeH has been observed. This is in contrast to main group 15. For E = As, which is in the same period, both isomers, CH₃As and CH₂AsH, have been identified. The lightest homologue of group 16 is CH₃O and its isomer CH₂OH. Both were observed in their cationic ground state with an IE of 10.701 ± 0.005 eV³⁰ and 7.56 ± 0.01 eV,⁵¹ respectively. The latter is also more stable in its neutral form by roughly ≈1.2 eV,^52–57 while in its sulfur analogue, the CH₃S isomer is the more stable one.

While photoelectron spectra of CH₂O^58,59 and CH₂S²³ have been reported, no CH₂Se has been observed in the present work. This is again in contrast to our study on main group 15 radicals, which identified CH₂As.²⁸ Interestingly, in group 15, similar radicals were observed for P and As and a significant difference was found to occur between As and Sb. In group 16, on the other hand, a change in the observed structural variety is evident between S and Se.

Species such as Se₂H were not generated via pyrolysis under the applied conditions, while their sulfur and oxygen analogues S₂H²³ and O₂H⁶⁰ were readily observed under similar conditions. However, changes in the pyrolysis conditions can lead to a decrease of bimolecular reactions, which might contribute to the absence of Se₂H in the present work.

Conclusions

Threshold photoelectron spectra of small reactive organo-selenium species were recorded using synchrotron radiation. All species were generated by pyrolysis of dimethyldiselenide and analyzed using the i²-PEPICO setup. Accurate ionization energies were derived for transitions into the ground and excited states of CH₃Se₂, Se₂ and CH₃Se. Experimental and calculated IEs are summarized in Table 1. The experiments were supported by Franck–Condon simulations, which helped to assign the vibrational structure in the spectra. The accuracy of the IEs of Se₂ and CH₃Se were improved in regard to earlier work performed on these species.^43,45,46 In addition, the vibrational structure was resolved. A photoelectron spectrum of CH₃Se₂ was hitherto unknown. We also compared the generated selenium species with their lower sulfur homologous species and found similarities in the vibrational structure. However, IEs were lower for the selenium compounds due to more diffuse orbitals. For CH₃Se, we found no HSeCH₂ isomer, in contrast to E=O and S, where both CH₃E and CH₂EH have been identified. This suggests that selenium prefers to form σ-bonds with carbon atoms, but not σ-bonds with hydrogen or π-bonds with carbon. A comparable trend can be seen among the group 15 elements: interestingly, arsenic forms species in which arsenic is directly bonded to hydrogen like HAsCH₂, whereas no such species is observed for antimony. In the chalcogens, this trend already appears to set in during the third period, when going from sulfur to selenium.

Conflicts of interest

There are no conflicts of interest to declare.

Data availability

The experimental and simulated threshold photoelectron spectra of CH₃Se₂, CH₃Se and Se₂ are available in the PhotoElectron PhotoIon Spectral COmpendium (PEPISCO) database, https://pepisco.psi.ch/.⁶¹

Supplementary information (SI) containing a TPES of the precursor as well as computed geometries and frequencies. See DOI: https://doi.org/10.1039/d5cp03519h.

Acknowledgements

The experiments were performed at the VUV beamline of the Swiss Light Source, located at the Paul Scherrer Institute (PSI). This work was financially supported by the Deutsche Forschungsgemeinschaft (DFG), contract FI575/19-1. DS acknowledges a fellowship by the Fonds der Chemischen Industrie (FCI).

References

1. E. Patterson, R. Milstrey and E. Stokstad, Proc. Soc. Exp. Biol. Med., 1957, 95, 617–620.
2. K. Schwarz, J. G. Bieri, G. M. Briggs and M. L. Scott, Proc. Soc. Exp. Biol. Med., 1957, 95, 621–625.
3. K. El-Bayoumy, Cancer Res., 1985, 45, 3631–3635.
4. J. Brozmanová, D. Mániková, V. Vlčková and M. Chovanec, Arch. Toxicol., 2010, 84, 919–938.
5. P. Guo, P. Zhao, J. Liu, H. Ma, J. Bai, Y. Cao, Y. Liu, H. He and C. Qi, Biol. Trace Elem. Res., 2013, 151, 301–306.
6. T. Adimulam, T. Arumugam, A. Foolchand, T. Ghazi and A. A. Chuturgoon, Int. J. Mol. Sci., 2021, 22, 12952.
7. A. Madabeni, M. Bortoli, P. A. Nogara, G. Ribaudo, M. Dalla Tiezza, L. Flohé, J. B. Rocha and L. Orian, Chem. – Eur. J., 2024, 30, e202403003.
8. L. Yu, J. Wang, T. Chen, Y. Wang and Q. Xu, Appl. Organomet. Chem., 2014, 28, 652–656.
9. L. Sancineto, F. Mangiavacchi, C. Tidei, L. Bagnoli, F. Marini, A. Gioiello, J. Scianowski and C. Santi, Asian J. Org. Chem., 2017, 6, 988–992.
10. R. L. de Carvalho, G. A. Jardim, A. C. Santos, M. H. Araujo, W. X. Oliveira, A. C. S. Bombaça, R. F. Menna-Barreto, E. Gopi, E. Gravel and E. Doris, et al. , Chem. – Eur. J., 2018, 24, 15227–15235.
11. H. Wojtowicz, G. Soroko and J. Młowski, Synth. Commun., 2008, 38, 2000–2010.
12. C. Francavilla, F. V. Bright and M. R. Detty, Org. Lett., 1999, 1, 1043–1046.
13. B. Priewisch and K. Rück-Braun, J. Org. Chem., 2005, 70, 2350–2352.
14. M. S. Silva, D. Alves, D. Hartwig, R. G. Jacob, G. Perin and E. J. Lenardão, Asian J. Org. Chem., 2021, 10, 91–128.
15. N. V. Dolgova, S. Nehzati, S. Choudhury, T. C. MacDonald, N. R. Regnier, A. M. Crawford, O. Ponomarenko, G. N. George and I. J. Pickering, Biochim. Biophys. Acta, Gen. Subj., 2018, 1862, 2383–2392.
16. H. Duddeck, Annu. Rep. NMR Spectrosc., 2004, 52, 105–166.
17. L. A. Wessjohann, A. Schneider, M. Abbas and W. Brandt, Biol. Chem., 2007, 388, 997–1006.
18. P. Hemberger and A. Bodi, Chimia, 2018, 72, 227.
19. I. Fischer and P. Hemberger, ChemPhysChem, 2023, 24, e202300334.
20. I. Fischer and S. T. Pratt, Phys. Chem. Chem. Phys., 2022, 24, 1944–1959.
21. J. Dyke, N. Jonathan and A. Morris, Int. Rev. Chem. Phys., 1982, 2, 3–42.
22. J. M. Dyke, Phys. Chem. Chem. Phys., 2019, 21, 9106–9136.
23. E. Karaev, M. Gerlach, D. Schaffner, S. E. Dutton, M. D. Phillips, P. Hemberger, A. K. Vasiliou and I. Fischer, Phys. Chem. Chem. Phys., 2024, 26, 28939–28946.
24. D. Schleier, M. Gerlach, D. Schaffner, D. P. Mukhopadhyay, P. Hemberger and I. Fischer, Phys. Chem. Chem. Phys., 2023, 25, 4511–4518.
25. E. Karaev, D. Schaffner, M. Gerlach, P. Hemberger and I. Fischer, J. Phys. Chem. A, 2025, 129, 6004–6010.
26. F. Holzmeier, M. Lang, K. Hader, P. Hemberger and I. Fischer, J. Chem. Phys., 2013, 138, 214310.
27. D. P. Mukhopadhyay, M. Gerlach, S. Hartweg, I. Fischer and J.-C. Loison, Phys. Chem. Chem. Phys., 2022, 24, 10993–10999.
28. E. Karaev, M. Gerlach, L. Faschingbauer, J. Ramler, I. Krummenacher, C. Lichtenberg, P. Hemberger and I. Fischer, Chem. – Eur. J., 2023, 29, e202300637.
29. D. P. Mukhopadhyay, D. Schleier, S. Wirsing, J. Ramler, D. Kaiser, E. Reusch, P. Hemberger, T. Preitschopf, I. Krummenacher and B. Engels, et al. , Chem. Sci., 2020, 11, 7562–7568.
30. X. Tang, X. Lin, G. A. Garcia, J.-C. Loison, C. Fittschen, X. Gu, W. Zhang and L. Nahon, J. Chem. Phys., 2020, 153, 031101.
31. M. Johnson, A. Bodi, L. Schulz and T. Gerber, Nucl. Instrum. Methods Phys. Res. A., 2009, 610, 597–603.
32. A. Bodi, M. Johnson, T. Gerber, Z. Gengeliczki, B. Sztáray and T. Baer, Rev. Sci. Instrum., 2009, 80, 034101.
33. A. Bodi, P. Hemberger, T. Gerber and B. Sztáray, Rev. Sci. Instrum., 2012, 83, 083105.
34. B. Sztáray, K. Voronova, K. G. Torma, K. J. Covert, A. Bodi, P. Hemberger, T. Gerber and D. L. Osborn, J. Chem. Phys., 2017, 147, 013944.
35. D. W. Kohn, H. Clauberg and P. Chen, Rev. Sci. Instrum., 1992, 63, 4003–4005.
36. B. Sztáray and T. Baer, Rev. Sci. Instrum., 2003, 74, 3763–3768.
37. A. Bodi, B. Sztáray, T. Baer, M. Johnson and T. Gerber, Rev. Sci. Instrum., 2007, 78, 084102.
38. P. Hemberger, X. Wu, Z. Pan and A. Bodi, J. Phys. Chem. A, 2022, 126, 2196–2210.
39. F. Neese, F. Wennmohs, U. Becker and C. Riplinger, J. Chem. Phys., 2020, 152, e1606.
40. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. Petersson, H. Nakatsujiet al., Gaussian 16, 2016.
41. S. Gozem and A. I. Krylov, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2022, 12, e1546.
42. S. Borkar, B. Sztáray and A. Bodi, J. Electron Spectrosc. Rel. Phenom., 2014, 196, 165–172.
43. D. Streets and J. Berkowitz, J. Electron Spectrosc. Relat. Phenom., 1976, 9, 269–287.
44. H. R. Hrodmarsson, G. A. Garcia, L. Nahon, J.-C. Loison and B. Gans, J. Mol. Spectrosc., 2021, 381, 111533.
45. A. W. Potts and I. Novak, J. Electron Spectrosc. Relat. Phenom., 1983, 28, 267–277.
46. Q. Sun, Z. Li, X. Zeng, W. Wang, Z. Sun, M. Ge, D. Wang, D. K. W. Mok and F.-T. Chau, ChemPhysChem, 2005, 6, 2032–2035.
47. C. W. Bookwalter, D. L. Zoller, P. L. Ross and M. V. Johnston, J. Am. Soc. Mass Spectrom. Weitere Fragen, 1995, 6, 872–876.
48. P. Carroll and P. Mitchell, Proc. R. Soc. London, Ser. A, 1975, 342, 93–111.
49. O. Edqvist, E. Lindholm, L. Selin and L. Åsbrink, Phys. Scr., 1970, 1, 25.
50. K. Voronova, K. M. Ervin, K. G. Torma, P. Hemberger, A. Bodi, T. Gerber, D. L. Osborn and B. Sztáray, J. Phys. Chem. Lett., 2018, 9, 534–539.
51. J. Dyke, A. Ellis, N. Jonathan, N. Keddar and A. Morris, Chem. Phys. Lett., 1984, 111, 207–210.
52. M. A. Haney and J. Franklin, Trans. Faraday Soc., 1969, 65, 1794–1804.
53. K. Hiraoka and P. Kebarle, J. Chem. Phys., 1975, 63, 1688–1690.
54. M. A. Haney, J. C. Patel and E. F. Hayes, J. Chem. Phys., 1970, 53, 4105–4106.
55. M. Munson and J. L. Franklin, J. Phys. Chem., 1964, 68, 3191–3196.
56. F. Lossing, J. Am. Chem. Soc., 1977, 99, 7526–7530.
57. K. M. Refaey and W. A. Chupka, J. Chem. Phys., 1968, 48, 5205–5219.
58. D. W. Turner, Philos. Trans. R. Soc., A, 1970, 7–31.
59. S. G. Lias, J. E. Bartmess, J. Liebman, J. Holmes, R. D. Levin and W. G. Mallard, J. Phys. Chem. Ref. Data, 1988, 17, 1–861.
60. M. Litorja and B. Ruscic, J. Electron Spectrosc. Relat. Phenom., 1998, 97, 131–146.
61. P. Hemberger, Z. Pan, X. Wu, Z. Zhang, K. Kanayama and A. Bodi, J. Phys. Chem. C, 2023, 127, 16751–16763.

---

Footnote

† Present address: HFML-FELIX, Radboud University, Toernooiveld 7, 6525 ED Nijmegen, The Netherlands.

---