Volume 29, Issue 35 e202300637
Research Article
Open Access

Bonding in Low-Coordinated Organoarsenic and Organoantimony Compounds: A Threshold Photoelectron Spectroscopic Investigation

Emil Karaev

Emil Karaev

Institute of Physical and Theoretical Chemistry, Julius-Maximilians-Universität Würzburg, Am Hubland, 97074 Würzburg, Germany

Search for more papers by this author
Marius Gerlach

Marius Gerlach

Institute of Physical and Theoretical Chemistry, Julius-Maximilians-Universität Würzburg, Am Hubland, 97074 Würzburg, Germany

Search for more papers by this author
Lukas Faschingbauer

Lukas Faschingbauer

Institute of Physical and Theoretical Chemistry, Julius-Maximilians-Universität Würzburg, Am Hubland, 97074 Würzburg, Germany

Search for more papers by this author
Dr. Jacqueline Ramler

Dr. Jacqueline Ramler

Department of Chemistry, Philipps-University of Marburg, Hans-Meerwein-Str. 4, 35032 Marburg, Germany

Search for more papers by this author
Dr. Ivo Krummenacher

Dr. Ivo Krummenacher

Institute of Inorganic Chemistry, University of Würzburg, Am Hubland, 97074 Würzburg, Germany

Search for more papers by this author
Prof. Dr. Crispin Lichtenberg

Corresponding Author

Prof. Dr. Crispin Lichtenberg

Department of Chemistry, Philipps-University of Marburg, Hans-Meerwein-Str. 4, 35032 Marburg, Germany

Search for more papers by this author
Dr. Patrick Hemberger

Corresponding Author

Dr. Patrick Hemberger

Laboratory for Synchrotron Radiation and Femtochemistry, Paul Scherrer Institut (PSI), 5232 Villigen, Switzerland

Search for more papers by this author
Prof. Dr. Ingo Fischer

Corresponding Author

Prof. Dr. Ingo Fischer

Institute of Physical and Theoretical Chemistry, Julius-Maximilians-Universität Würzburg, Am Hubland, 97074 Würzburg, Germany

Search for more papers by this author
Dedicated to Prof. Dr. Hansjörg Grützmacher in honour of his contributions to p-block chemistry.
First published: 30 March 2023
Citations: 3

Graphical Abstract

Methyl and methylene compounds of arsenic and antimony have been studied by photoelectron photoion coincidence spectroscopy to investigate their relative stability with the help of pyrolysis. While organoarsenic compounds show a variety of methyl and methylene compounds, like HAsCH2, antimony only shows Sb−CH3. In the condensed phase, the precursors E(CH3)3 (E=As, Sb) are relatively poor sources of methyl radicals.

Abstract

Methyl and methylene compounds of arsenic and antimony have been studied by photoelectron photoion coincidence spectroscopy to investigate their relative stability. While for As both HAs=CH2, As−CH3 and the methylene compound As=CH2 are identified in the spectrum, the only Sb compound observed is Sb−CH3. Thus, there is a step in the main group 15 between As and Sb, regarding the relative stability of the methyl compounds. Ionisation energies, vibrational frequencies and spin-orbit splittings were determined for the methyl compound from photoion mass-selected photoelectron spectra. Although the spectroscopic results for organoantimony resemble those for the previously investigated bismuth compounds, EPR spectroscopic experiments indicate a far lower tendency for methyl transfer for Sb(CH3)3 compared to Bi(CH3)3. This study concludes investigations on low-valent organopnictogen compounds.

Introduction

Elements E of the main group 15 (pnictogens, E−N−Bi) favour high oxidation states like −3 (especially for E−N), +3 and +5.1 In particular, the heavier congeners As, Sb and Bi are present in the oxidation state +3 in most of their molecular compounds. Non-stabilised low-valent compounds with E in a low oxidation state such as +1 are generally highly reactive, electron deficient and difficult to isolate.2 However, they can appear as chemical intermediates in numerous reactions (see examples below) and can be expected to differ from their Lewis base-stabilised analogues,3 therefore a characterisation of their electronic structure is of considerable interest. In recent years, we started a systematic investigation of reactive species E−CH3 in the gas phase, using threshold photoelectron spectroscopy (TPES).4 When E−N, a N=C double bond is formed and methanimine HN=CH2 is by far the most stable isomer, while N−CH3 has not yet been observed.5 In contrast, methylbismuth has been characterised as a bismuthinidene biradical Bi−CH3 with a triplet ground state.2f For E−P, the methylene species HP=CH2 is also the most stable isomer, but P−CH3 is only 0.65 eV higher in energy and both have recently been identified in a photoelectron spectrum.2e This raises the question at which position in group 15 the methyl compound (E−CH3) becomes more stable than the methylene species (HE=CH2) and dominates the spectra. To address this question, we synthesised methyl and methylene compounds of arsenic and antimony in situ from their respective E(CH3)3 precursors using a pyrolysis reactor and obtained mass spectra of the decomposition products and high-resolution threshold photoelectron spectra.

Although of fundamental importance for describing bonding in low-valent group 15 compounds, this is not a purely academic question. For example, low-valent organoantimony species have been suggested to be generated as intermediates in the transition metal-catalysed liberation of dihydrogen from stibanes6 and low-valent bismuth compounds have been shown to engage in BiI/BiIII redox cycles for reactions such as the hydrodefluorination of arenes.7 Furthermore, organoantimony compounds are used in semiconductor materials and solar cells.8 Here, the isomer-selective identification of intermediates that are formed in the thermolysis of organo-antimony precursors is crucial for understanding the mechanism of chemical vapour deposition (CVD).8b, 9 So far, the amount of information on reactive Sb and As species is limited. For methylenearsane, an ionisation energy (IE) of 9.7 eV has been determined in a dispersive He (I) photoelectron spectrum,10 in reasonable agreement with computations.11 Not all bands in the spectrum were assigned and vibrational resolution was not achieved. In computations for E−CH3 (E−N, P, As) that included spin-orbit coupling, a triplet ground state was found.12 Bond lengths to carbon were predicted for E−As and Sb by simple Hartree-Fock calculations.13

Due to its sensitivity to the character of electronic states and the isomeric structure, photoelectron spectroscopy is well suited to describe the electronic structure of molecules. Here, we employ threshold photoelectron-photoion coincidence spectroscopy (PEPICO).14 Ion mass-selected threshold photoelectron spectra, ms-TPES, for different species can be extracted from the full data set. The technique is thus well-suited for studying photoelectron spectra of reactive species in complex mixtures.15

Results

Arsenic

To identify the pyrolysis products of As(CH3)3, time of flight (TOF) mass spectra have been recorded at different pyrolysis temperatures and photon energies. In Figure 1, mass spectra of As(CH3)3 recorded at 10 eV are depicted. Without pyrolysis (upper trace) only the precursor m/z 120 is detected, as well as some contaminants from previous experiments (acetone and iodine), thus there is no dissociative photoionisation of the precursor observed, in agreement with the previously reported onset of methyl loss from As(CH3)3 of 10.620 eV.16 By increasing the pyrolysis to 650 °C (lower trace), several fragments from thermal decomposition are observed.

Details are in the caption following the image

TOF mass spectra at 10 eV photon energy of the precursor As(CH3)3 (top) and its pyrolysis products at 650 °C (bottom). Acetone (m/z 58) and I2 (m/z 254) are present as contaminants from previous experiments.

A strong signal of m/z 15 shows the formation of methyl radicals, while the masses at m/z 105 (As(CH3)2), 90 (As−CH3) and 75 (As) indicate a stepwise methyl abstraction [Eq. 1]:

However, a mass spectrum recorded at 12 eV also shows peaks corresponding to C2H6 and C2H4, which indicates the presence of additional concerted reaction pathways associated with loss of ethane or ethene. A signal of m/z 89 suggests decomposition of the species associated with m/z 105 into methane and As−CH2 or H loss from the species associated with m/z 90. A further H-atom loss is indicated by the appearance of As−CH (m/z 88). An expanded view of the mass region between m/z 85 and 95 is given as Figure S1 in the Supporting Information and shows that the mass peaks can be well separated. M/z 103 most likely arises from a consecutive H2 loss of m/z 105 and was already assigned to AsC2H4 with a bridged ethene structure.16 The signal at m/z 76 is likely due to reaction of As with a hydrogen atom and not by methylene loss from the compound detected at m/z 90, since no signal of m/z 14 is observed at higher photon energies. Furthermore, the higher masses m/z 150, 165 and 180 are dimerisation products of two As atoms (m/z 150), As and As−CH3 (m/z 165), and two As−CH3 fragments (m/z 180), respectively. In addition, m/z 300 is observed and corresponds to an As tetramer.

To gain structural information of the pyrolysis products, ion-mass selected threshold photoelectron (ms-TPE) spectra were recorded. Here, we will focus on the ms-TPE spectra of m/z 90 and 89. Dispersive photoelectron spectra have already been reported for group 15 dimers17 and tetramers,18 and no additional information is available from our work. Therefore, their spectra as well as the unstructured ones of the species As2Rx are only given as Figure S2. On the other hand, TPE-spectra of AsH and AsCH will be discussed in a future publication.

Figure 2 shows the ms-TPE spectrum of m/z 90. Three major bands are observed, the first two being associated with a vibrational progression. A simulated spectrum of As−CH3 based on DFT (B3LYP\6-311G(2d,d,p)) computations (green) was found to fit only the low energy part of the first band from 8.5 to 9.2 eV. The computed ionisation energy (IE) of 8.80 eV matches the observed IE of 8.70 eV well. A small band at 8.64 eV is assigned to a hot band transition. As has been pointed out before, As−CH3 has a X 3A2 ground state with a C3v symmetry.12 The two unpaired electrons are located in two p-orbitals (SOMO, singly occupied molecular orbitals) at the As. Although formally non-bonding, removal of an electron from the SOMO leads to shortening of R(AsC) from 1.98 to 1.89 Å (Table 1). Computations yielded a slightly tilted CH3 group relative to the As−C bond in the ion and an elongation of one of the C−H bonds, which causes a symmetry reduction to Cs. However, the calculated pseudo rotational barrier of the CH3 group is 72 meV, while the zero point energy amounts to 918 meV. Therefore, the CH3 group in As−CH3+ can freely rotate and an effective C3v symmetry results. Spin-orbit calculations using NEVPT2 [11,11] revealed an energy difference of 220 meV for the splitting of the cationic 2E state into the 2E1/2 and 2E3/2 components. NEVPT2 (N-electron valence state second order perturbation theory) is an alternative to the well-known CASPT2 method and tends to yield more accurate excitation energies.19 Relativistic effects were included by choosing a relativistic basis set and using the Douglas-Kroll-Hesse (DKH) method,20 The value of 220 meV matches almost exactly the energy difference between the IE of 8.70 and the band at 8.92 eV that is severely underestimated in intensity in the simulation (green). This indicates that the X+ 2E3/2urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0001 X 3A2 progression overlaps with the X+ 2E1/2←X 3A2 (orange) transition. The best simulation of both transitions (red) was obtained when the vibrational band at 8.92 eV was assigned to the origin transition into the upper spin-orbit component and the intensity of transitions into the lower spin-orbit component was weighted by a factor of 1.4. The vibrational progression is dominated by two modes, the CH3-wagging mode urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0002 , computed at 520 cm−1 and the (approximate) As−C stretching mode urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0003 670 cm−1. Note that the geometry of the individual spin-orbit components was not further optimised.

Details are in the caption following the image

Ms-TPES of m/z 90 (black) with FC simulations for comparison. The red line represents the sum of the simulations for the X+ 2E1/2←X 3A2 (green) and X+ 2E3/2←X 3A2 (orange) transitions in As−CH3. The blue line is a fit to the X+ 2A←X 1A1 transition of HAs=CH2. The third band maximising at 10.5 eV is most likely due to a transition into the excited A 2A’ state of HAs−CH2+.

Table 1. Computed minimum-energy geometries of As−CH3 in its X 3A2 and As−CH3+ in its X+ 2E1/2/2E3/2 states.

X 3A2

X+ 2E1/2/2E3/2

R(AsC)/urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0004

1.98

1.89

R(CH)/urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0005

3×1.09

2×1.09

1.11

θ(As−CH)

3×110

2×113

103

θ(HCH)

3×109

2×107

113

The somewhat inferior fit at higher energies indicates that the vibrational modes of the 2E3/2 state have a slightly lower frequency compared to 2E1/2. Table 1 summarises the computed minimum energy geometries of the neutral and the ionic ground state, without considering spin-orbit splitting.

The second band from 9.5 to 10.2 eV is assigned to the constitutional isomer HAs=CH2+. This transition has been reported before10 and an IE of 9.7 eV was derived. However, in the previous work vibrational resolution10 was not achieved, while the present spectrum shows a long and intense progression with a spacing of roughly 700 cm−1. Simple DFT computations indicate that neutral HAs=CH2 is the more stable isomer and 136 meV lower in energy than As−CH3. The FC simulation on the CBS-QB3 level of theory (blue line) fits the experimental data very well. It is based on the computed geometry given in Table 2. The deviations at energies above 10 eV are most likely due to the onset of the third band. The transition at 9.61 eV is assigned to the IE of HAs=CH2, in good agreement with the value of 9.79 eV obtained in CBS-QB3 computations. According to the computations, the neutral ground state is a planar singlet with a double minimum potential along the C−As−H angle θ(HAsC), thus HAs=CH2 is Cs symmetric, similar to methanimine. Upon ionisation, the dihedral angle ϕ(HAs−CH) changes from 0° in the neutral compound to 34° in the cation, thus the cation becomes non-planar. The most important geometry parameters of HAs=CH2 and its cation are listed in Table 2. The ionisation can be described as a D0+←X 1A transition, with an electron being ejected from the urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0006 As−C (A’’) orbital,10 associated with an increase in ΔR(AsC) of +0.03 Å. The observed vibrational progression of ≈700 cm−1 cannot be assigned to a single mode, but is composed of four different low-frequency vibrational modes with computed values of urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0007 790 cm−1, urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0008 710 cm−1 (out of plane bend), urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0009 620 cm−1 (in plane bend), urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0010 500 cm−1 (CH2 torsion). All four vibrations show some As−C stretching character and are depicted in Table S1. Above 10.2 eV, a further unstructured band is apparent that extends to around 11 eV and has also been observed previously.10 It is assigned to the transition into the excited A 2A’ state of HAs−CH2+ due to ionisation of a non-bonding electron in the nAs (A’) in accordance with the prior work and our own calculations that yield a vertical excitation energy in the ion of +0.62 eV. As a different precursor was used by Chrostowska et al., a significant contribution from dissociative ionisation to the band is unlikely (Figure S3). However, a small contribution from the first excited state of AsCH3+ to the band is possible. Based on our computations it is expected to appear at 10.78 eV.

Table 2. Computed minimum energy geometries of HAs−CH2 in its X 1A’ and HAs−CH2+ in its D0+ state.

HAs−CH2

HAs−CH2+

R(AsC)

1.86

1.89

ϕ(HAs−CH)

0

34

θ(As−CH)

124

2×121

119

θ(H−C)

96

94

Figure 3 depicts the ms-TPES of m/z 89, corresponding to As=CH2 at an elevated pyrolysis temperature of around 800 °C. It features two separate transitions between 8.4 and 9.2 eV and from 9.7 to 10.3 eV.

Details are in the caption following the image

Ms-TPES of m/z 89, As−CH2+ (black line). a) complete spectrum with simulations for the X+ 1A1←X 2B2 (red) and A+ 3A2←X 2B2 (blue) transition. The former simulation was adjusted as described in the text. b) Close-up of the X+ 1A1←X 2B2 transition with the calculated (blue) and adjusted (red) geometries.

While trace (a) depicts the complete spectrum, trace (b) is a close-up of the lower energy transition. Again, FC-simulations based on quantum chemical calculations helped to rationalise the spectrum. Neutral As=CH2 is C2v symmetric with a X 2B2 electronic ground state. The intense band at 8.61 eV fits well with the calculated IE for the X+ 1A1←X 2B2 transition of As=CH2 of 8.56 eV at the CCSD(T) level of theory. However, the FC-simulations based on a variety of computational approaches, including the MP2, MP4, CCSD(T), DFT and CASSCF methods deviated from the experiment and the intensity of the bands at higher energy, in particular the band at 1260 cm−1 was significantly underestimated by all methods (Figure 3b (blue) and Table S2). The wavenumber of the symmetric CH2 bending was indeed computed around 1300 cm−1, depending on the method. However, the FC simulations yielded only negligible intensity for this mode. We again included relativistic effects in our calculations by choosing a relativistic basis set and using the DKH method,20 which also had negligible influence on the structure and FC-simulation. Different constitutional isomers, including H2As=C and HAs−CH were calculated using the G4 approach, but neither the computed IEs of 9.47 and 12.9 eV nor their vibrational structure matched the experimental spectrum. Note that the cationic ground state is of singlet multiplicity, so spin-orbit splitting is not relevant.

On the other hand, the FC simulation for the transition into the first excited state A+ 3A2 based on CBS-QB3 computations (blue line) agrees very well with the experiment. An ionisation energy of 9.80 eV is derived from the simulation, in good agreement with the calculated value of 9.89 eV. The band is dominated by the As−C stretching vibration urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0011 710 cm−1 and a feature around 9.70 eV is assigned to a hot band. The electron is removed from a σ orbital with As−C bonding character, which results in an increase of R(AsC) from 1.77 Å in the neutral species to 1.89 Å in the cation and thus explains the activity in the As−C stretch. Since the transition into the triplet state is well simulated, the geometry and force constants of neutral As=CH2 should be correct. Furthermore, transitions from excited states in the neutral compound should not contribute. The reason for the deviations in the transition to the X+ 1A1 state lies therefore in the representation of the ionic state. Also, note that the first excited singlet cationic state lies 1.8 eV above the cationic ground state according to TDDFT calculations.

The first explanation for the initial difficulties to interpret the overall spectrum shown in Figure 3 is perturbation of band intensities by autoionisation. The deviations might then be explained by the enhancement of vibrational transitions, in particular the symmetric CH2-bending mode, due to interactions with resonantly excited autoionising states.21 No peaks were identified in the photoion yield at the corresponding photon energies. Constant ionic state spectra might confirm the presence of autoionisation in the spectrum.22

However, there is also a second possible explanation: a potential energy surface scan revealed a flat potential along the bending coordinate around the minimum energy geometry. We therefore modified the geometry of As−CH2+ in its X+ 1A1 state in order to empirically improve the FC fit of the experimental spectrum and manually varied the angle θ(HCH) as well as R(AsC) until the best fit with the spectrum was obtained. The FC-simulation shown as a red line in Figure 3a and b was obtained with θ(HCH)=107° and R(AsC)=1.68 Å, and fits the experimental spectrum much better. The experimentally observed wavenumbers for the modes urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0012 1260 cm−1 (HCH in plane bend) and urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0013 880 cm−1 (AsC stretch) are slightly lower than the calculated ones of urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0014 1300 cm−1 and urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0015 1000 cm−1, respectively (DFT, urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0016 B97-XD/aug-cc-pVTZ). However, the geometry parameters are substantially different from the computed values for the minimum of θ(HCH)=115–117° and R(AsC)=1.72–1.75 Å. As the energy difference between the best fit geometry and the global minimum was computed to be only 61 meV, the harmonic approximation might be inappropriate for a good fit of the experimental spectrum.

Antimony

For the investigation of antimony compounds, very similar strategies were employed. Trimethylantimony, Sb(CH3)3 has been pyrolysed under very similar conditions as the arsenic compound. Figure 4 shows two mass spectra recorded at 9 eV photon energy. Dissociative photoionisation is not evident in the ion images at 9 eV (see the Supporting Information S3) because the appearance energy AE0K(Sb(CH3)3, Sb(CH3)2+) is 9.9 eV.16

Details are in the caption following the image

TOF mass spectra of the precursor Sb(CH3)3 (top) and its pyrolysis products at 650 °C (bottom) at 9 eV photon energy.

In the upper trace without pyrolysis, only the precursor signal at m/z 166 and 168, Sb(CH3)3 is seen. Note that Sb shows two isotopes, 121Sb (57 %) and 123Sb (43 %), causing a characteristic splitting of the mass peaks. In the lower trace, recorded at a pyrolysis temperature of 650 °C, several mass peaks are observed, with a pattern similar to the one reported above for As(CH3)3 pyrolysis. Again, peaks at m/z 151/153, 136/138 and 121/123 indicate sequential loss of methyl groups. Most likely, a minor concerted pathway associated with ethane or ethene loss is present as well. Here, CH3 is not observed, due to its ionisation energy of 9.84 eV.23 Notably, peaks at m/z 135 and 137 are missing, indicating the absence of SbCH2. Bimolecular reactions in the pyrolysis reactor again lead to species that contain two, three or four Sb atoms. Due to the isotopic ratio, peaks overlap, but Sb2+, Sb3+, and Sb4+ are readily identified by their photoelectron spectra. Several smaller peaks result from multimers of methylated antimony clusters. For example, a peak centred slightly above m/z 300 most likely arises through dimerisation of dimethylantimony to Sb2C4H12+.24

Ms-TPE spectra for all species visible in the mass spectrum were obtained from the PEPICO data set. Apart from the Sb clusters, only the ms-TPE spectrum of SbCH3+ shows a vibrational structure. As photoelectron spectra of Sb clusters have been discussed previously,17, 18 they are only given in Figure S4, together with the unresolved spectra of other compounds (Sb(CH3)2, [Sb(CH3)2]2). Figure 5 presents the ms-TPES of m/z 136, SbCH3+. Three bands can be observed, the first two at 7.9–8.4 eV and 8.4–8.8 eV have a very similar appearance and a well-resolved vibrational structure with a spacing of around 600 cm−1. The third one is broad and unstructured and is assigned to an electronically excited state of SbCH3+, but might contain contributions from dissociative photoionisation of the precursor at higher photon energies.

Details are in the caption following the image

Ms-TPES of m/z 136, Sb−CH3+ (black). The red line represents a FC simulation of the X+ 2E1/2←X 3A2 (green) and X+ 2E3/2←X 3A2 (orange) transitions in Sb−CH3. IE of 8.01 eV (X+ 2E1/2←X 3A2) and 8.43 eV (X+ 2E3/2←X 3A2) are derived.

Bands were simulated based on quantum chemical calculations on different levels of theory. The best simulation was obtained by DFT, using the urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0017 B97-XD functional, a jorge-QZP basis set and the DKH method to include scalar relativistic effects. The calculated geometric parameters are summarised in Table 3. Computations also indicated a symmetry reduction from C3v symmetry to Cs, comparable to As−CH3 (see above). But again, the pseudorotational barrier of the methyl group of 150 meV (CCSD(T)) and the 950 meV zero point vibrational energy suggest that the methyl group can rotate freely, hence assignment as a X+ 2E←X 3A2 transition within the C3v point group remains appropriate.

Table 3. Computed minimum energy geometries of Sb−CH3 in its X 3A2 and Sb−CH3+ in its X+ 2E1/2/2E3/2 states.

X 3A2

X+ 2E1/2/2E3/2

R(SbC)

2.13

2.05

R(CH)

3×1.09

2×1.09

1.10

θ(Sb−CH)

3×110

2×113

103

θ(HCH)

3×109

2×108

112

To obtain the spin-orbit splitting, we then performed NEVPT2 [11,11] calculations at the optimised geometry. Relativistic effects were again included using the DKH method.20 An energy difference of 0.497 eV between the 2E1/2 and 2E3/2 states was calculated. The red line represents a simulation that is the sum of the individual contributions (green and orange line, respectively), assuming an intensity ratio of 1.4 : 1. Although the intensity ratio of the bands of almost 2 : 1 suggests the 2E3/2 state to be the ground state, the calculations show that the 2E1/2 is lower in energy, in accordance with Hund's third rule. Ionisation energies of 8.01 eV and 8.43 eV were derived for the X+ 2E1/2←X 3A2 and X+ 2E3/2←X 3A2 transitions, respectively. Thus, experimental and computed spin-orbit splitting agree within 70 meV.

The vibrational progression is characterised by a single mode that is assigned to the Sb−C stretch, urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0018 . In addition, the CH3-wagging mode urn:x-wiley:09476539:media:chem202300637:chem202300637-math-0019 470 cm−1 appears with significant intensity in the simulation. While a 610 cm−1 spacing is apparent in the 2E1/2 state, it is reduced to 580 cm−1 in the 2E3/2 state. This indicates small differences in the potential energy surface of the two spin orbit states. A full characterisation requires accurate relativistic computations that are beyond the scope of the present work. Note that for its isomer HSb=CH2 a photoionisation transition with an IE around 8.8 eV and a short vibrational progression with Franck-Condon factors comparable to SbCH3+ is predicted. Assuming a similar ionisation cross section, no evidence is visible for the presence of this isomer and its formation in the pyrolysis reactor.

Methyl transfer reactions and pnictinidene transfer reactions

The pyrolysis experiments performed with compounds E(CH3)3 (E−N−Bi) indicate the possibility of up to three homolytic E−CH3 bond dissociations in these species. This is not only relevant for the application of such compounds in metalorganic chemical vapour deposition and related methodologies, but also in the context of methyl transfer reactions and pnictinidene transfer reactions in synthetic chemistry. For the bismuth compound Bi(CH3)3, we have previously shown that the release of methyl radicals is not only relevant under pyrolysis conditions, but can readily be detected by EPR spectroscopy at 60 °C in benzene solution using a nitrone as a spin trap.2f Furthermore, reaction of Bi(CH3)3, with the “bismuthinidene trapping reagent” diphenylsulfide, (PhS)2, gave Bi(CH3)(SPh)2 in high isolated yield, suggesting that BiCH3 might be generated and trapped under these conditions. In the context of this work, we aimed to evaluate the suitability of As(CH3)3 and Sb(CH3)3 as synthetically relevant sources of methyl radicals and methylpnictinidenes (AsCH3 and SbCH3, respectively) in the condensed phase (Scheme 1). In order to test the capability of E(CH3)3 (E−As, Sb) to release methyl radicals, E(CH3)3 was heated to 60 °C in toluene in the presence of phenyl-N-tert-butylnitrone as a radical trap, that is, under conditions analogous to those previously used for Bi(CH3)3 (Scheme 1, top). However, in the case of As(CH3)3 and Sb(CH3)3, only weak signals could be detected (about ten times weaker than in the bismuth case) and a clean methyl radical transfer could not be confirmed (see the Supporting Information). Similarly, no significant conversion could be detected, when As(CH3)3 and Sb(CH3)3 were reacted with (PhS)2 under conditions that gave considerable isolated yields of Bi(CH3)(SPh)2, when the bismuth precursor was used (Scheme 1, bottom). These reactions demonstrate that the chemical robustness of compounds E(CH3)3 increases with decreasing atomic number of the central atom, making the lighter species less suitable for methyl radical transfer and pnictinidene transfer under mild conditions.

Details are in the caption following the image

Reactions of E(CH3)3 with a nitrone (top) and (PhS)2 (bottom) for E=As, Sb (this work) and E=Bi (previous work).2f For E=As, Sb, EPR spectroscopy indicated the presence of two or more closely related radical species (X=CH3 plus at least one unidentified species) with a total intensity that was lower by a factor of approximately 10 compared to the case of E=Bi.

Discussion

With this work, we have now completed our investigation of the methyl and methylene compounds of all group 15 elements E by photoelectron spectroscopy. We will now discuss the trends that are apparent from the experiments. The IEs derived in this and previous work are summarised in Table 4.

Table 4. Ionisation energies of methyl- and methylene-compounds of group 15 elements E observed by threshold photoelectron spectroscopy.

Energy/eV

E

HE=CH

H2E=C

E=CH2

HE=CH2

E−CH3

N5

11.72 (cis)

11.21 (3A”)[a]

12.32 (3A2)

9.99

12.65 (trans)

P2e

8.80 (1B2)

10.07

8.91

As

8.61 (1B2)

9.61

8.70

Sb

8.14

Bi2f

7.88

  • [a] A stable singlet cation has been computed, but not experimentally observed.

From the table, several trends become obvious. As a first-row atom, nitrogen exhibits the largest structural diversity. The species HN=CH (cis and trans) and H2N=C were only observed for nitrogen, not for any of the higher congeners due to their high energy compared to E=CH2. On the other hand, no N−CH3 was observed. The overlap between the p orbitals on N and C permits the formation of stable double bonds, thus the formation of HN=CH2 is strongly favoured. With increasing atomic radius, this overlap deteriorates, and methyl compounds become favoured. For E−P and As, both structures have been observed. In the case of E−P, the HP=CH2 isomer is lower in energy by ΔfH(0 K)=0.65 eV/63 kJ mol−1 and the P−CH3 photoelectron signal has consequently been weak.2e For E−As, the HAs=CH2 isomer is only 136 meV more stable than As−CH3, so both isomers show a comparable stability and appear with comparable intensity in the photoelectron spectrum (Figure 2). Finally, for E−Sb and Bi, only the E−CH3 isomer is visible. All E−CH3 congeners have a triplet ground state and are thus biradicals. Going from As to Sb thus leads to a change in the preferred bonding situation. This is surprising, because the major changes in the electronic structure occur between P and As (3 d shell filled) and Sb and Bi (4 f shell filled).

For all E−CH3 species we find a symmetry reduction upon ionisation in the computations, regardless of the computational level. The CH3 group is slightly tilted, and the C−H bonds become inequivalent. In the related Bi−CH3, the distortion was rationalised by antibonding interactions between the unpaired electron at the Bi centre and the bonding electrons of the two C−H bonds, which are approximately aligned with the singly occupied p-type orbital. The geometric distortion should lead to a splitting of the 2E ionic ground state into 2A” and 2A’ components. In the TPES of Bi−CH3 such a splitting is evident in the spectrum of the lower spin-orbit component, however, there is no evidence in the spectra of the lower congeners. For Sb−CH3 and As−CH3, the zero point vibrational energy suffices to overcome the barrier to internal rotation (see above), so a 2E ionic ground state results and a similar situation can be assumed for E−P.

Increasing the nuclear charge from N to Bi also increases spin-orbit (SO) coupling, which should lead to a SO splitting of the electronic ground state of the ion. While SO splitting is negligible in P−CH3, it is so large in Bi−CH3 that the spin-orbit excited states were outside the investigated energy range. Preliminary computations yielded a value of ΔESO=0.99 eV. This leaves As−CH3 and Sb−CH3 as the most challenging cases. Here, the magnitude of SO splitting permits to observe both SO components in the cation. The experimentally observed splitting is ΔESO=0.20 eV for E=As and ΔESO=0.42 eV for E=Sb. The value for Sb−CH3+ compares well with ΔESO=0.625 eV for I−CH3+,25 whereas the value for As−CH3+ is surprisingly high. According to Hund's third rule, the 2E1/2 state should be lower in energy than the 2E3/2 state, because the degenerate e-orbitals are less than half-filled (one electron in Sb−CH3). On the other hand, due to its twofold degeneracy, the 2E3/2 state should be more intense by a factor of approximately two. In Figures 4 and 5 it is evident that the low energy band is more intense and the intensity ratio would approximately match a 2E3/2 ground state. However, computations show that the ionic ground state is 2E1/2, in accordance with Hund's rules. Although these rules were originally formulated for atoms, the prediction is also valid in our cases. Note that the most important mechanism for intensity perturbations, spin-orbit autoionisation, would increase the intensity of the upper spin-orbit component.

In contrast to E−CH3, the electronic ground state of the HE=CH2 isomer is always a singlet. In HN=CH2, the electron is ejected from a non-bonding orbital located at the nitrogen atom, while in HP=CH2 and HAs=CH2 the ionic ground state corresponds to ionisation from a π-orbital located at the E=C double bond. Consequently, the TPES spectra are dominated by a long bending mode progression in HN=CH2, while E=CH2 stretch and out-of-plane bends dominate in HP=CH2 and HAs=CH2.

The trend in the methylene compounds E=CH2 parallels the one for HE=CH2. Neither Sb=CH2 nor Bi=CH2 have been observed, while P=CH2 and As=CH2 are identified in the spectra. However, isolable compounds, described as stiba-alkenes and bisma-alkenes, have recently been reported using carbodicarbone ligands.26 Interestingly, for E=CH2 the spin multiplicity changes. For N=CH2, a triplet ground state was identified in the cation, while the singlet cation was found to be a transition state to linear HNCH+ on the potential energy surface. For E=P and As, E=CH2+ was identified to be a singlet ground state X+ 1A1. The simulation of the X+ 1A1←X 2B2 transition of As=CH2 underestimated the intensity of several vibrationally excited states in both molecules, while the A+ 3A2←X 2B2 transition was simulated well. A similar effect was observed previously for P=CH2.2e Possible reasons for the deviations between experimental and computed intensities are either an intensity enhancement by autoionisation or the rather flat potential in the ionic ground state.

Conclusion

Threshold photoelectron spectra of reactive low-valent organoarsenic and organoantimony compounds have been investigated by using synchrotron radiation. Vibrationally resolved spectra were recorded for As−CH3, HAs=CH2, As=CH2 and Sb−CH3. The methyl compounds E−CH3 (E−As, Sb) showed spin-orbit splitting in the cationic ground state, values of ΔESO=0.20 and 0.42 eV were derived for E−As and E−Sb, respectively. While ΔESO is in line with that of related compounds for Sb−CH3+, it is surprisingly large for As−CH3+. In both compounds, the 2E1/2 state is lower in energy, in accordance with Hund's rules. Ionisation energies of IE(As−CH3/2E1/2)=8.70 eV and IE(Sb−CH3/2E1/2)=8.01 eV were obtained with error bars of ±20 meV, and energies for the Franck-Condon active modes were determined. While for antimony Sb−CH3 was the only stable isomer, for arsenic we also identified HAs=CH2 with an IE(HAs=CH2)=9.61 eV and the methylene compound As=CH2 with an IE(As=CH2)=8.61 eV. For the latter molecule, simulating the spectrum turned out to be difficult, either due to intensity perturbations by autoionisation or due to a rather flat cationic potential energy surface. This work completes a series of investigations on the reactive methyl compounds of the group 15 elements. While for E−N only species with N=C multiple bond character were observed, for Sb and Bi only E−CH3 in a triplet ground state was identified. For E−P, As, on the other hand, both isomers were observed in the spectra. The increasing bond length R(EC) results in a smaller overlap between the p orbitals of E and carbon, disfavouring the formation of a (formal) double bond. In the condensed phase, neither As(CH3)3 nor Sb(CH3)3 is an efficient methyl transfer or pnictinidene transfer reagent, while for Bi(CH3)3 methyl release was detected in benzene solution by EPR spectroscopy.

These findings contribute to a detailed mapping of the accessibility, geometry, and electronic structure of the isomers of the low-valent, high-energy species ECH3 and ECH2 (E−N−Bi) and will thus be valuable for the design of precursors in materials science as well as methyl and pnictinidene transfer reagents.

Experimental Section

All species were characterised by using the double imaging photoelectron photoion coincidence setup (CRF-PEPICO) at the vacuum ultraviolet (VUV) beamline of the Swiss Light Source (SLS) at the Paul Scherrer Institute, Villigen (Switzerland). A detailed description has been published elsewhere,27 therefore only a brief overview of the setup is given here. Synchrotron radiation provided by a bending magnet is collimated by a mirror and dispersed vertically by a plane grating. The light is guided through a Ne/Ar/Kr gas mixture to suppress higher order radiation. The photon energy was calibrated on the 11 s’–13 s’ autoionisation lines of argon in both first and second order of the grating.

The reactive species were generated by introducing As(CH3)3 or Sb(CH3)3 vapor, seeded in argon through a nozzle with 0.1 mm diameter into a pyrolysis reactor. The gas flow and the precursor concentration in the gas mixture were controlled by two mass flow controllers. The gas mixture was pyrolysed in a SiC tube, resistively heated to 600–700 °C. The pyrolysis products form an effusive molecular beam and enter the experimental chamber through a 2 mm skimmer.28 Here the molecular beam is crossed by the synchrotron radiation for ionisation. The resulting photoelectrons and photoions were extracted in opposite directions in a constant 218 V cm−1 electric field. Photoelectrons were velocity-map-imaged onto a Roentdek DLD40 delay line detector while the cations were detected in a Wiley-McLaren-type time-of-flight (TOF) spectrometer. Threshold photoelectrons were selected in the centre of the detector with an energy resolution better than 4 meV and the contributions of hot electrons were subtracted.29 Photoions and photoelectrons were collected in a multiple-start/multiple-stop coincidence scheme.30 Due to the insufficient expansion cooling, ms-TPES recorded from pyrolysis reactors, mostly suffer from broadening by hot and sequence band transitions. A reduction of these can be achieved by integrating only the room temperature background velocity component in the ions’ VMI as was recently shown.31, 32 The resulting mass-selected threshold photoelectron (ms-TPE) spectra were corrected for photon flux and the ionisation energy in addition corrected for the Stark shift.

As(CH3)3 was synthesised as follows:33 To a three-necked flask equipped with a dropping funnel with pressure compensation and an adapter for inert gas/vacuum, methylmagnesium bromide (1.0 m in nBu2O, 95.3 mL, 95.3 mmol) was added. A solution of AsBr3 (10.0 g, 31.8 mmol) in nBu2O (20 mL) was added dropwise over 2 h at 0 °C. After complete addition, the reaction mixture was stirred for a further 2 h at 0 °C then for 16 h at ambient temperature. The dropping funnel was removed and a Vigreux column (h=19 cm, Ø=3 cm) equipped with a distillation apparatus and a Schlenk tube was added to the three-necked flask. The reaction mixture was distilled at atmospheric pressure and an oil bath temperature of 90 °C. The receiver Schlenk tube was cooled with liquid nitrogen to collect the AsMe3 distillate. Yield: 3.26 g, 27.2 mmol, 85 % (containing <3 mol % nBu2O). Boiling point: 53–55 °C. Analytical data (such as NMR spectra) were in agreement with the literature.33 1H NMR (400 MHz, C6D6): δ=0.76 ppm (s, 9H, CH3).

Sb(CH3)3 was synthesised as follows:33 Methylmagnesium bromide (3.0 m in Et2O, 86.6 mmol, 29.8 mL) was added to a suspension of SbCl3 (6.82 g, 29.9 mmol) in diethyl ether (100 mL) at 0 °C. The reaction mixture was stirred at 0 °C for 15 min, then 1.5 h at ambient temperature. The suspension was filtered, and diethyl ether was removed partially by distillation at 50 °C (oil bath temperature). The residue was distilled at 90 °C and 5×10−3 mbar to give SbMe3 as a colourless oil. Yield: 1.40 g, 8.07 mmol, 27 % (containing 10 mol % Et2O). Analytical data (such as NMR spectra) were in agreement with the literature.33 1H NMR (400 MHz, C6D6): δ=0.60 ppm (s, 9H, CH3).

Ionisation energies (IE), geometries and vibrational frequencies were computed on different levels of theory using ORCA34 and the Gaussian 16 suite of programs.35 IEs were obtained by subtracting the neutral molecule's energy from the cation's energy, including zero point energy contributions. Based on the computed geometries and frequencies, the photoelectron spectra were simulated by ezSpectrum,36 which computes Franck-Condon factors (FCF) for the transitions. The resulting stick spectra were convolved with a Gaussian function of 25 meV full width at half maximum height.

Supporting Information

Additional references are cited within the Supporting Information.37, 38-40

Acknowledgments

The experiments were performed at the VUV beamline of the Swiss Light Source, located at the Paul Scherrer Institute (PSI). The work was financially supported by the Deutsche Forschungsgemeinschaft, contract FI575/13-2 and LI2860/5-1, and by the LOEWE program. This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement no. 946184). Open Access funding enabled and organized by Projekt DEAL.

    Conflict of interest

    The authors declare no conflict of interest.

    Data Availability Statement

    The data that support the findings of this study are available in the supplementary material of this article.