– A POWERFUL TOOL FOR UNDERSTANDING COMBUSTION PROCESSES

(1)

Fabian Holzmeier

PHOTOIONIZATION OF REACTIVE MOLECULES – A POWERFUL TOOL FOR UNDERSTANDING

COMBUSTION PROCESSES

Dr. Fabian Holzmeier

Politecnico di Milano, Dipartimento di Fisica Piazza Leonardo da Vinci 32, 20133 Milano, Italy E-Mail: fabian.holzmeier@polimi.it

INTRODUCTION

Combustion research is not dead yet! Unfortunately, our energy consumption exceeds still by far our capabilities to provide energy from renewable sources. In 2017, almost 72% of the EU’s primary energy use was based on the combustion of traditional and bio-fuels [1]. More than 60 billion liters of fuel were burned in Germany alone in 2018 [2] to cover our immense demands, especially in transportation. Our everyday life is hence still unthinkable without combustion, and this will not change in the very short term despite all efforts. Therefore, there must still be a high interest in making the energy supply by combustion more efficient for saving natural resources and minimizing its harmful impact on the environment. If we do not make an immense breakthrough in the field of renewable energy sources in the very near future, we will not be able to meet our ambitious climate target without improving combustion processes, too. Besides considering combustion as a declining transient energy source on the way to a 100%

energy supply by renewable sources, it is also explored to produce carbon neutral fuels. The latter ones potentially have no net emission of greenhouse gas or leave a carbon footprint. It is for example sought to produce synthetic fuels from renewable energy sources, such as wind turbines or solar panels [3].

Biofuels produced from biomass, like fuelwood or agricultural waste, do also not lead to a net increase of CO2 [4]. Capturing the CO₂ emissions from fuel powered plants and reusing it for fuel production, e.g., by reaction with solid carbon to produce the fuel gas CO [5], would also mitigate the absolute greenhouse gas emission.

It is therefore necessary to develop better diagnostics for combustion reactions in order to understand the chemistry and finally control and improve the elapsing processes, e.g., by find- ing better fuel blends or additives [6]. In combustion chemistry, reactive molecules, e.g., small open-shell organic radicals and carbenes play a significant role in the decomposition of the fuel molecules and the formation of unwanted side products, such as polycyclic aromatic hydrocarbons leading eventually to soot. The key to understanding the complex processes during combustion is to be able to identify all the intermediates occur-

ring in the reaction and to quantify them at different stages.

This data can then be used to develop mechanistic models, which help understanding combustion and enable to simulate ways to improve combustion efficiency and reduce harmful emissions. However, there is still a lack of accurate base data especially for reactive molecules, since they are challenging to generate and investigate. The increasing demand of biofuels means that not only pure hydrocarbon chemistry is relevant for studying combustion, but also organic molecules containing heteroatoms like oxygen and nitrogen [4]. Characterizing intermediates related to these kinds of fuels is thus highly desired.

In the following, it will be discussed how isolated reactive molecules can be generated in the gas phase and how they can be identified and characterized with a focus on vacuum ultraviolet (VUV) photoionization. It will be demonstrated that the latter technique offers powerful tools for detecting reaction intermediates and products in combustion qualitatively as well as quantitatively. This helps deciphering the fundamental reaction mechanisms in combustion and is therefore of great use for making further advancements in the switch to renewable energies.

GAS PHASE GENERATION OF REACTIVE MOLECULES

Since radicals and carbenes are kinetically unstable, they can only be investigated in an environment, in which reaction times are significantly slowed down. While some spectroscopic methods can be employed when the reactive molecule is stabilized in a cold matrix [7,8], the most versatile approach is to generate the isolated radical or carbene in situ in the gas phase [9]. Here, a stable precursor molecule is transferred into the gas phase and fragmented to the reactive species of interest by electric corona discharge, photolysis, chemical reaction or pyrolysis. Discharge leads usually to many undesired side products and charged particles, while photolysis requires an intense photon source with a photon energy that matches the chemical bonding energy of the precursor and is thus limited in its applicability [10]. Many radicals have been generated suc- cessfully in the gas phase by the chemical hydrogen abstraction with fluoro atoms [11–14]. However, this method is not very selective and is best used when only equivalent hydrogen atoms are present, such as for the amino radical from ammo- nia [15], or when several hydrogen atoms shall be abstracted at once, e.g., for the generation of CH from methane [16]. An extremely versatile method for the generation of radicals is py-

(2)

rolysis, in which the bond cleavage in the precursor molecule is induced by temperature, a parameter that can be easily controlled. Chen and co-workers combined for the first time pyrolysis with a (pulsed) molecular beam [17]. Here, the precursor molecule is diluted in a carrier gas and passed through a short, heated quartz or SiC tube, which is electrically heated up to 1000°C before the gas is expanded into the vacuum. The adiabatic cooling and the short residence time in the heating zone minimizes the production of side products and decreases the rotational and vibrational temperatures of the generated species. All generation methods have in common, however, that the success of the experiment depends very much on a thoughtful choice of the precursor molecule, which leads selectively and ideally quantitatively to the desired species. For the generation of carbon-centered radicals, for example, ha- logenated hydrocarbons have been proven to be very efficient [18–21], since the carbon-halogen bond is usually the weakest one and is thus cleaved first. Release of thermodynamically stable pyrolysis products, e,g., molecular nitrogen from diazo- compounds [22], is another clean source for the generation of isolated reactive intermediates in the gas phase by pyrolysis.

PHOTOIONIZATION FOR IDENTIFICATION AND CHARACTER- IZATION OF RADICALS AND CARBENES

Since open-shell radicals and carbenes are kinetically unstable, an experimental method for their identification and characterization is needed, which is very sensitive to detect even low concentrations. Furthermore, a technique is sought that reveals unique properties of a species for their unambiguous identification, because it is very rare that the reactive molecules is generated without any side products, as explained above. Photoionization is such a highly sensitive method that is perfectly suited for the identification and characterization of radicals and carbenes in the gas phase, since the ionization energy and in particular the threshold photoelectron spectrum is a fingerprint of a molecule. Vacuum ultraviolet (VUV) light is needed to photoionize valence-electrons of atoms and molecules, for which numerous table-top laser sources are available for high-resolution experiments both in the frequency as well as the time-domain. However, for the investigation of reactive species, which must be detected in very low concentrations, VUV synchrotron radiation has the advantage of being easily tunable over a much wider photon energy range than lasers, while offering very high repetition rates (hundreds of MHz) at a high photon flux. Thus, ionization energies can be measured easily employing time-of-flight mass spectrometry and tracking the ion signal of interest as a function of the tuned photon energy. The ionization energy can then be determined by ex- trapolating the obtained photoionization efficiency (PIE) curve of one mass-to-charge ion signal to zero, or curve fitting [23].

The accuracy of this method is however limited with error bars around 100 meV at best. Moreover, it is nearly impossible to detect different isomers simultaneously, since they show only small changes of the PIE curve’s slope and no spectral fea- tures are visible. This drawback can be overcome by detecting electrons and ions occurring from the same ionization event in coincidence. It is necessary that the count rates are kept below one event per light pulse in order to be able to correlate each

electron unambiguously to the corresponding ion. That is why high-repetition rate synchrotron radiation is best employed for coincidence experiments. The PEPICO (photoelectron photoion coincidence) technique thus enables to record mass-selected photoelectron spectra (PES) [24]. In photoelectron spectroscopy the kinetic energy of the released photoelectron is measured at a fixed wavelength, for example with an electrostatic analyzer, and the energy of electronic and vibrational states can be identified by conservation of energy:

E_ion = hv – IE – E_el (1)

Eion is the ion internal energy relative to its ground state located at the ionization energy IE, Eel is the electron kinetic energy and hν is the photon energy. The photoelectron spectra provide accurate ionization energies even if a significant change of the molecule’s geometry occurs upon ionization leading to unfa- vorable Franck-Condon factors and by consequence very shal- low slopes of the PIE curves. In addition, PES resolve in many cases the vibrational structure, which, along with the ionization energy, helps to unambiguously distinguish even between different isomers. Fig. 1 illustrates the difference of PIE curves opposed to the corresponding photoelectron spectrum.

Fig. 1: Illustrative comparison of a photoionization efficiency curve (red) and the corresponding (threshold) photoelectron spectrum (blue [black] curve) of the same molecule. It is much more difficult to determine the ionization energy accurately from the PIE curve than from the photoelectron spectra, which resolve the vibrational structure of the cation.

A further increase in the resolution of the PES can be obtained by scanning the photon energy hν and detecting only so-called threshold photoelectrons, for which Eel=0, in threshold photoelectron spectroscopy (TPES, bottom trace of Fig. 1) [25]. This improves the accuracy of the ionization energy to few meV and the intensity of the resolved vibrational bands follows the Franck-Condon principle. The spectra, which can thus be easily simulated employing quantum chemistry, are specific fingerprints of an isomer. In addition, TPES adds the possibility to investigate also in detail dissociative photoionization (DPI),

(3)

since the ion internal energy can be directly measured with this technique (cf. eq. (1) for E_el=0). A typical setup of such a threshold PEPICO setup coupled with velocity map imaging detectors (i⁽²⁾PEPICO) and a pyrolysis source for the generation of radicals or carbenes, as available at the VUV beamlines of the Swiss Light Source and Synchrotron SOLEIL, is depicted in Fig. 2 [26–28]. The central part of a molecular beam with the in situ generated reactive molecule passes through a set of skimmers into the spectrometer vacuum chamber, where it is crossed with the ionizing VUV synchrotron radiation. Photo- electrons and photoions are accelerated in opposite directions and detected in coincidence on position-sensitive detectors (PSDs) using the velocity map imaging technique [29,30]. Only electrons with zero kinetic energy arriving at the center of the detector are selected and a subtraction scheme for electrons with an initial kinetic energy in direction of the extraction field, which consequently arrive at the center, too, is used [31]. On the ion side, time-of-flight mass spectrometry is employed, while it is also possible to take advantage of the spatial ion image in order to discriminate the signal from the molecular beam against the background gas or the signal of the reactive species against kinetic ions of the same mass produced by dissociative photoionization of the precursor. By correlating the threshold electrons to the respective ions of the same ionization event, the mass-selected threshold photoelectron spectrum is finally obtained.

Fig. 3 shows the TPE spectra of four isomeric radicals of the composition C₆H₆N: the three picolyl (pyridylmethyl) isomers [21] and anilinyl [32]. These radicals are of great interest for combustion research, since pyridinic structures are the principle form, in which nitrogen is bound in biofuels [33]. The initial decomposition step of picoline, i.e, a prototype of a substituted pyridine, is hydrogen abstraction leading to a picolyl radical, an isostere of benzyl. It is assumed that resonance-stabilized radicals occur in high concentrations in reactive environments and represent therefore key branching points in the reaction mechanism influencing the overall kinetics to a high degree. It has been suggested furthermore that N-heterocyclic radicals are involved in the formation of N-containing polycyclic aromatic hydrocarbons (PANHs) in an astrochemical context [34].

This could also be relevant for flames. We therefore wanted to determine accurate ionization energy and obtain unambiguous spectroscopic data of the three picolyl radicals, whose ionization had previously only been investigated by electron-impact ionization mass spectrometry [35]. The picolyl radicals were selectively generated in the gas phase by pyrolysis of the respective (aminomethyl)pyridine precursors (C5H4N-CH2-NH2).

TPE spectra were recorded by imaging PEPICO spectroscopy with VUV synchrotron radiation at the Swiss Light Source. Re- cently, Hrodmarsson et al. complemented the study by the TPE spectrum of the anilinyl radical C6H5NH, also shown in Fig. 3 [32]. In this case, the radical was generated by hydrogen abstraction of aniline with fluorine atoms. TPE spectra were again obtained by imaging PEPICO in an experiment conducted at Synchrotron SOLEIL.

As evident from Fig. 3, the TPE spectra for 2-, 3-, and 4-picolyl are clearly different from each other and the respective ion-

Fig. 2: Experimental set-up of a pyrolysis source for the generation of reactive molecules with a double-imaging PEPICO spectrometer as used at the VUV beamlines at the Swiss Light Source and Synchrotron SOLEIL.

THRESHOLD PHOTOELECTRON SPECTRA

As outlined above, mass-selected threshold photoelectron spectroscopy is an extremely powerful method for the unambiguous identification of species in a complex environment.

The TPE spectra provide the ionization energy and, in many cases, the vibrational structure of the cation of the selected m/z ratio. This is not only restricted to the electronic ground state of the cation, but it is possible to obtain this information also for low-lying, non-dissociative excited states by ionization out of the lower outer-valence orbitals with VUV radiation. By consequence, Since the TPES provides a fingerprint of a molecule, it is even possible to distinguish different isomers of the same mass and atom composition. This is crucial for the identification of the precise isomer, for example in the on-line spectroscopy of flames, which yield experimental evidence of combustion mechanisms.

Fig. 3: Experimental mass-selected threshold photoelectron spectra for the three picolyl isomers and the anilinyl radical. Franck-Condon simulations are shown in blue. While the picolyl isomers have clearly different ionization energies and spectra, 4-picolyl and anillinyl can only be distinguished in the high energy part of the spectrum, where the first excited cationic state occurs.

(4)

ization were determined to be 7.70±0.02, 7.59±0.01, and 8.01±0.01 eV, respectively, in good agreement with computed ionization energies [21]. All three spectra feature a similar and pronounced vibrational progression, which is also reproduced by the shown Franck-Condon simulations. The analysis of these simulations reveal that a single mode dominates in all cases:

the totally symmetric in-plane ring deformation mode, since the in-plane angles of the aromatic ring change significantly upon ionization. The substantial shift of the ionization onset for the 4-picolyl isomer with respect to 2- and 3-picolyl can be rationalized by the electrostatic potentials of the cations. 4-picolyl has a resonance structure, in which the positive charge is located at the electronegative N atom, which destabilizes the cation and leads thus to a higher IE [21].

While the three picolyl isomers can be distinguished easily by their ionization energy and TPE spectra, 4-picolyl and anilinyl possess very similar spectra. The IE of anilinyl was found to be 8.02±0.02 eV, thus within the error bar of 4-picolyl [32]. In addition, the observed vibrational progressions resemble each other, since the same in-plane deformation mode is excited in anilinyl, too. It is therefore not straightforward to differentiate between those two species. However, the spectra seem to devi- ate at higher photon energies. The broad unstructured band for 4-picolyl setting in at around 9 eV was assigned to the lowest cationic triplet state, which was computed to have its onset at 9.08 eV. In anilinyl, by contrast, a very weak signal is observed in the TPES at 9.5 eV only, which was also tentatively ascribed to the first excited triplet state of the cation. However, detecting this small difference in a complex reaction mixture, as typical in combustion, will remain difficult. The example of these four radicals of the composition C₆H₆N illustrates the capabilities of PEPICO spectroscopy for distinguishing different isomers, which would not be possible with simple photoionization mass spectrometry. Mass-selected threshold photoelectron spectroscopy is therefore a powerful experimental technique, well suited for deciphering the composition of complex mixtures in the gas phase as present in combustion.

ABSOLUTE PHOTOIONIZATION CROSS SECTIONS

The next step for decoding combustion mechanisms is to ex- tend the qualitative analysis of the mixtures to a quantitative one. For quantifying the detected ion signal, the absolute photoionization cross section is needed. It is quite challenging to obtain cross sections for reactive molecules, such as radicals and carbenes, in straightforward way, since it is nearly impossible to know their absolute concentration in the gas phase.

However, by choosing the right experimental conditions, there are workaround solutions for the determination of absolute photoionization cross sections of reactive species.

In photoionization the measured ion signal S is directly propor- tional to the photoionization cross section σ_i:

S ∝ P σi [A] l (2) Here, P is the photon flux, [A] the concentration of the molecule of interest and l the interaction length. The latter is in general

a constant for an experimental setup. However, sometimes the spectrometer’s efficiency depends on the mass of the detected particle, introducing the so-called mass discrimination factor, which has to be specified for each setup. Comparing the ion signals in a binary mixture with exactly defined concentrations of an analyte and a reference species with a known absolute photoionization cross section is the easiest way to determine the absolute photoionization cross section of an unknown species as a function of the photon energy. In addition, the signal must be normalized on the photon flux of the light source.

While this approach works very well for stable molecules in the gas phase, it is quite challenging for reactive radicals or carbene. Nevertheless, the generation of radicals and carbenes by pyrolysis can be exploited for the measurement of absolute photoionization cross sections, if the cross sections of side products and/or the precursor molecule are exactly known as well as the stoichiometry of the pyrolysis reactions. Thus, the choice of the precursor is quite delicate and by consequence the σi of only few reactive open-shell species are known so far, e.g., vinyl [36], propargyl [36–38], allyl [39], 2-propenyl [39], phenyl [40], methyl [41,42], and ethyl [43].

We succeeded in adding another combustion-relevant reactive molecule to this list, cyclopropenylidene c-C₃H₂, an aromatic carbene and the most stable one of the three C₃H₂ isomers [44]. Alongside the propargylene isomer HCCCH, c-C₃H2 was identified as an intermediate in a rich cyclopentene flame [45] thanks to the accurate ionization energy as determined by threshold photoelectron spectroscopy [46]. Moreover, it is thought to dominate the dissociation dynamics of the propargyl radical at low combustion temperatures or low pressures [47]. With the knowledge of the absolute photoionization cross section of c-C₃H₂, it would be possible to determine absolute mole fractions in flames and to model the combustion process by deriving the reaction rates. In previous pyrolysis experiment, cyclopropenylidene was generated from chloro-cyclopropene, which loses HCl under pyrolytic conditions [46]. However, the conversion selectivity of this precursor is relatively low, leading to the occurrence of several side products. In addition, the ionization energy of HCl (IE=12.79 eV) and c-C3H2 (IE=9.17 eV) are quite different, which would restrict drastically the region of investigation. An alternative precursor, which circumvents these problems is the quadricyclane spiro(cycloprop[2]ene- 1,3’-tetracyclo[3.2.0.0^2,7.0^4,6]heptane depicted in Fig. 4. Upon pyrolysis, this precursor fragments stoichiometrically to benzene and the desired carbene [48], which have comparable ionization energies. The precursor is therefore a prime candi- date for the determination of the c-C3H2 photoionization cross section. A disadvantage of the quadricyclane is that its synthe- sis is challenging, but an improved route has been developed by A. Krüger and co-workers [49]. Upon pyrolysis of this quadricyclane precursor the ratio of the c-C₃H2 total ion signal to the benzene signal is defined as:

S_C₃_H₂

SC6H6 = [C₃H₂] [C6H6]

σ_i^C³^H² σ_i^C⁶^H⁶

A_C₃_H₂

A_C₆_H₆ (3) As mentioned above, the concentrations [C3H2] and [C6H6] are equal as well as the apparatus function for both species, since we found that the mass discrimination factor is comparable

(5)

for m/z=39 and m/z=78. The equation for the carbene’s cross section reduces hence to:

σ_i^C³^H²= S_C₃_H₂ SC6H6

· σ_i^C⁶^H⁶ (4)

As the picolyl radicals, pyrrolyl c-C4H4N is supposed to play a key role in the combustion of fossil fuels, in which nitrogen is bound to 50-80% in pyrrolic structure [52]. For example, it was identified in the oxidation of pyridine, in which initially pyridyl (C5H4N) is formed, which is then oxidized to pyridoxy (C5H4NO) and further decomposes to pyrrolyl and CO [53]. Moreover, hydrogen abstraction from pyrrole leads directly to the pyrrolyl radical in flames [54]. We generated this radical for the first time in the gas phase by pyrolysis of 3-methoxypyridine [51].

Here, first methyl is cleaved off leading to the pyridoxy radical, which decomposes rapidly in a second step releasing CO under ring contraction. This pyrolysis mechanism is similar to the one of anisole (methoxybenzene) generating the cyclopentadienyl radical c-C5H5 [55–57]. We determined the ionization energy of the two lowest-lying electronic cation states of pyrrolyl by TPE spectroscopy and analyzed the corresponding vibrational structures [51]. Beyond, we were also able to determine the radical’s heat of formation experimentally. For that, we investigated the dissociative photoionization of the precursor molecule, 3-methoxypyridine without pyrolysis. The latter starts to fragment by DPI above 11 eV photon energy, yielding several daughter ions.

One of those fragment ions corresponds to m/z=66, i.e., the pyrrolyl cation. We analyzed the complete fragmentation pat- tern of the precursor molecule up to 13.5 eV photon energy by modelling the unimolecular dissociation rates employing sta- tistical rate theory [58]. We could thus determine accurate appearance energies, i.e., for each fragment the photon energy, at which the dissociation sets in at T=0 K, and our results were found to be in good agreement with quantum chemical calcula- tions. As visible in the cationic potential energy curve along the fragmentation’s reaction coordinate illustrated in Fig. 5, there is no reverse barrier for the formation of the pyrrolyl cation and the two neutral fragments CH₃ and CO from the precursor ion.

Fig. 4: Absolute photoionization cross sections for benzene and the cyclopropenylidene carbene generated by pyrolysis of a quadricyclane precursor.

The absolute photoionization cross section of cyclopropenylidene shown in Fig. 4 was then obtained by normalizing the benzene PIE curve to the absolute photoionization cross section reported in the literature [50]. The latter was in turn determined by referencing it to the one of nitric oxide. This outlines already that the accuracy of the determined cross sections is quite limited, as the cross section of benzene is given with an 20 % error margin. Our experiment adds further, but significantly smaller uncertainties by the integration limits of the mass peak, the subtraction of false coincidences and the assump- tion of a mass-independent apparatus function. Therefore, we indicate an error of ±30 % associated with the absolute photoionization cross section of the cyclopropenylidene carbene.

Nevertheless, this is already a good first starting point for the determination of absolute mole fractions of this reactive molecule in a flame and the derivation of rates for mechanistic models and will therefore contribute to obtain deeper insights into certain combustion processes, in which this reactive small carbene is involved.

THERMOCHEMICAL DATA

The methods discussed above enable to postulate reaction mechanisms in combustion processes, in which the reactive species are involved, and derive reaction kinetics. However, these mechanisms still must obey the laws of thermodynam- ics. Hence, it is necessary to possess thermochemical data on all reaction partners and intermediates to verify that the assumed reaction can indeed take place under combustion conditions. It is therefore of high interest to obtain experimental thermochemical data of reactive molecules, such as the enthalpy of formation. Photoionization experiments on radicals or carbenes generated by pyrolysis, i.e., especially PEPICO experiments can provide access to these data as shown in the following example of the pyrrolyl radical [51].

Fig. 5: Thermochemical cycle for the derivation of the standard heat of formation of the pyrrolyl radical from experimental ionization and appearance energies and the tabulated heats of formation for the precursor and neutral fragments. The potential energy along the reaction coordinate for the decomposition of the precursor (fragmentation of the parent ion) is sketched in solid (dashed) lines.

(6)

The appearance energy AE0K(C6H7NO; C4H4N⁺) corresponds hence to the energy difference between the neutral precursor molecule and the DPI fragments. Furthermore, the ionization energy of the pyrrolyl radical was measured experimentally by means of threshold photoelectron spectroscopy. As shown in Fig. 5, it was then possible to derive the enthalpy of formation for pyrrolyl via a thermochemical cycle, since the heats of formation of the precursor [59] as well as the two neutral fragments methyl and CO [60] are known from the literature. The obtained value of ΔfH0K(C4H4N)=301.1±15.1 kJ mol^-1 is in very good agreement with previous values [61]. This proves that the PEPICO technique is a powerful experiment to determine even thermochemical properties of elusive species.

COMBUSTION PROCESSES IN FLAMES

With an expansive dataset on ionization energies, cross sections and thermochemical properties on hand, one can then pass on to more elaborate experiments and study the combustion in flames on-line. In Germany, especially the groups of Prof.

K. Kohse-Höinghaus at Bielefeld, Prof. T. Kasper at Duisburg- Essen, and of M. Köhler at the German Aerospace Center DLR have been very active on the investigation of flames employing photoionization techniques. A typical experiment uses a flat- flame burner installed in a vacuum chamber and fuelled by a defined mixture of oxygen and the fuel itself [62,63]. A quartz cone is used to sample different zones of the flame (e.g., the reaction zone or the exhaust) by varying the height above the

burner (HAB). A carrier gas is passed through a small aperture of the quartz cone and the intermediates and reaction products are hence picked up and expanded into a second vacuum chamber forming a molecular beam. This preserves also reactive species such as radicals and carbenes formed in combustion. The beam is then intersected with the ionizing radiation and mass spectrometry or PEPICO are employed to probe the qualitative and quantitative composition of the flame [64]. The PIE curves and/or TPE spectra help identifying the reaction intermediates and the mole fraction of each species is derived from the signal intensity and its photoionization cross section as a function of the HAB. The experimental data is then compared to simulations with kinetic combustion mechanisms [65].

In a recent investigation by Kasper and co-workers, a fuel-rich m-xylene flame (7.3% m-C₈H₁₀, 42.7% O₂, 50.0% Ar) was investigated by imaging PEPICO with VUV synchrotron radiation [66].

Typical gasolines contain up to 20% (w/w) of monoaromatic hydrocarbons, such as benzene, toluene or xylenes [67]. Hereof, the latter constitute the largest portion, since they improve anti- knock properties [68]. The scope of the study was to probe the decomposition of the reactive fuel radicals (C8H9) produced in the first step by H-abstraction with special focus on an eventual isomerization of the m-xylyl radical to the ortho- or para-isomer.

This was possible since the TPE spectra of the isolated xylyl radicals, produced by pyrolysis of xylyl bromide precursors, had been recorded previously [69,70]. Fig. 6a shows the experimental mole fraction profiles of the m-xylene fuel as well as for the observed intermediates m-xylyl and p-xylylene and styrene.

From the TPE spectrum corresponding to m/z=105, which were compared to the xylyl TPE spectra, it was clearly shown that the m-xylyl isomer is the dominant fuel radical. While the p-isomer is only detected in low quantities, isomerization to o-xylyl seems to be negligible. The flame TPE spectrum cor- responding to m/z=104, shows contributions from three different species: p-xylylene, styrene and benzocyclobutene. With these data, the initial decomposition steps of m-xylylene could be retraced as summarized in Fig. 6b. The experimental results were compared to simulations based on literature mechanisms. It was stated that the deviations between experiment and simulations are to a large degree due to the lack of photoionization cross sections. Hence, this study demonstrates nicely that fundamental experiments on isolated reactive molecules as described in the previous paragraphs are essential for advancing combustion research.

CONCLUSIONS

Ionization energies and threshold photoelectron spectra are isomer-selective fingerprints, which can be used to unambiguously identify species also in complex mixtures in the gas phase. Valence-shell photoionization experiments, especially mass-selected threshold photoelectron spectroscopy, are a powerful tool to obtain these data. Furthermore, photoionization makes absolute photoionization cross sections and thermochemical properties accessible. Acquiring these character- istics of reactive molecules like organic radicals and carbenes is necessary to investigate combustion processes in flames

Fig. 6: (a) Experimental mole fractions as a function of the height above the burner of selected intermediates and products in the combustion of m-xylylene. The m-xylene curve is scaled by 0.01. (b) Corresponding decom- position mechanism.

(7)

on-line with the same sensitive methods. Hence, reaction mechanisms and kinetics of the combustion can be derived and compared to simulations, which will help improve combustion efficiency and the minimization of harmful side products.

Although combustion research seems to have been sidelined recently with regards to the energy revolution and climate tar- gets, it still offers a large potential for helping to reach those goals as a transition technology as well as a potentially CO₂ neutral energy source. Therefore, it is necessary to advance combustion research taking advantage of the latest tools and techniques available. Not only valence-shell, but also photoionization with soft [71] and hard X-rays [72] have shown its capabilities to serve as diagnostic tools in combustion research.

Since there has been significant progress ultimately on the development of new light sources in the X-ray domain, it will be interesting to see which new insights will be gained for the combustion research with these instruments. PEPICO experiments with VUV synchrotron radiation will meanwhile not lose any significance. Besides combustion, they offer the unique possibilities to explore reaction kinetics [28,73], photoelectron circular dichroism [74], catalysis mechanisms [75,76], or can complement ultrafast experiments, e.g., ones employing high harmonic generation sources [77].

ACKNOWLEDGEMENTS

The author likes to acknowledge the valuable contributions to the presented research of all his co-workers, especially Prof.

I. Fischer, Dr. M. Lang, E. Reusch, Dr. P. Hemberger, and the staff at the Swiss Light Source and Synchrotron SOLEIL stor- age rings.

REFERENCES

[1] https://www.eea.europa.eu/data-and-maps/indicators/primary-energy-consumption-by-fuel-7/assessment.

[2] https://www.transportenvironment.org/what-we-do/sustain- able-finance/fuel-taxes.

[3] J. Kopyscinski, T. J. Schildhauer, S. M.A. Biollaz, Fuel, 2010, 89, 1763–1783.

[4] K. Kohse-Höinghaus, P. Oßwald, T. A. Cool, T. Kasper, N. Han- sen, F. Qi, C. K. Westbrook, P. R. Westmoreland, Angew. Chem.

Int. Ed., 2010, 49, 3572–3597.

[5] P. Lahijani, Z. A. Zainal, M. Mohammadi, A. R. Mohamed, Re- new. Sust. Energ. Rev., 2015, 41, 615–632.

[6] K. Kohse-Höinghaus, Chem. Eur. J., 2016, 22, 13390–13401.

[7] W. Sander, G. Bucher, S. Wierlacher, Chem. Rev., 1993, 93, 1583–1621.

[8] X. Zhang, A. V. Friderichsen, S. Nandi, G. B. Ellison, D. E. David, J. T. McKinnon, T. G. Lindeman, D. C. Dayton, M. R. Nimlos, Rev.

Sci. Instrum., 2003, 74, 3077–3086.

[9] C. Alcaraz, I. Fischer, D. Schröder, Radical Chemistry in the Gas Phase, in Encyclopedia of Radicals in Chemistry, Biology and Materials, C. Chatgilialoglu, A. Studer (Eds.), John Wiley & Sons, Chichester, 2012.

[10] S. Willitsch, J. M. Dyke, F. Merkt, Helv. Chim. Acta, 2003, 86, 1152–1166.

[11] W. E. Jones, E. G. Skolnik, Chem. Rev., 1976, 76, 563–592.

[12] S. J. Dunlavey, J. M. Dyke, N. Jonathan, A. Morris, Mol. Phys., 1980, 39, 1121–1135.

[13] G. A. Garcia, X. Tang, J.-F. Gil, L. Nahon, M. Ward, S. Batut, C.

Fittschen, C. A. Taatjes, D. L. Osborn, J.-C. Loison, J. Chem.

Phys., 2015, 142, 164201.

[14] J. M. Dyke, PCCP, 2019, 21, 9106–9136.

[15] F. Holzmeier, M. Lang, I. Fischer, P. Hemberger, G. A. Garcia, X.

Tang, J.-C. Loison, PCCP, 2015, 17, 19507–19514.

[16] B. Gans, F. Holzmeier, J. Krüger, C. Falvo, A. Röder, A. Lopes, G.

A. Garcia, C. Fittschen, J.-C. Loison, C. Alcaraz, J. Chem. Phys., 2016, 144, 204307.

[17] D. W. Kohn, H. Clauberg, P. Chen, Rev. Sci. Instrum., 1992, 63, 4003–4005.

[18] I. Fischer, Int. J. Mass Spectrom., 2002, 216, 131–153.

[19] P. Hemberger, M. Lang, B. Noller, I. Fischer, C. Alcaraz, B. K.

Cunha de Miranda, G. A. Garcia, H. Soldi-Lose, J. Phys. Chem. A, 2011, 115, 2225–2230.

[20] F. Holzmeier, M. Lang, P. Hemberger, I. Fischer, ChemPhy- sChem, 2014, 15, 3489–3492.

[21] E. Reusch, F. Holzmeier, P. Constantinidis, P. Hemberger, I.

Fischer, Angew. Chem. Int. Ed., 2017, 56, 8000–8003.

[22] E. Reusch, D. Kaiser, D. Schleier, R. Buschmann, A. Krueger, T.

Hermann, B. Engels, I. Fischer, P. Hemberger, J. Phys. Chem. A, 2019, 123, 2008–2017.

[23] P. Hemberger, M. Steinbauer, M. Schneider, I. Fischer, M.

Johnson, A. Bodi, T. Gerber, J. Phys. Chem. A, 2010, 114, 4698–4703.

[24] J.H.D. Eland, Int. J. Mass Spectrom. Ion Phys., 1972, 8, 143–151.

[25] T. Baer, R. P. Tuckett, PCCP, 2017, 19, 9698–9723.

[26] A. Bodi, P. Hemberger, T. Gerber, B. Sztáray, Rev. Sci. Instrum., 2012, 83, 83105.

[27] G. A. Garcia, B. K. Cunha de Miranda, M. Tia, S. Daly, L. Nahon, Rev. Sci. Instrum., 2013, 84, 53112.

[28] B. Sztáray, K. Voronova, K. G. Torma, K. J. Covert, A. Bodi, P.

Hemberger, T. Gerber, D. L. Osborn, J. Chem. Phys., 2017, 147, 13944.

[29] D. W. Chandler, P. L. Houston, J. Chem. Phys., 1987, 87, 1445–1447.

[30] A. T. J. B. Eppink, D. H. Parker, Rev. Sci. Instrum., 1997, 68, 3477–3484.

[31] B. Sztáray, T. Baer, Rev. Sci. Instrum., 2003, 74, 3763–3768.

[32] H. R. Hrodmarsson, G. A. Garcia, L. Nahon, B. Gans, J.-C. Loi- son, J. Phys. Chem. A, 2019, 123, 9193–9198.

[33] A. Doughty, J. C. Mackie, J. Phys. Chem., 1992, 96, 10339–

10348.

[34] D. S. N. Parker, T. Yang, B. B. Dangi, R. I. Kaiser, P. P. Bera, T. J.

Lee, Astrophys. J., 2015, 815, 115.

[35] T. F. Palmer, F. P. Lossing, J. Am. Chem. Soc., 1963, 85, 1733–1735.

[36] J. C. Robinson, N. E. Sveum, D. M. Neumark, J. Chem. Phys., 2003, 119, 5311–5314.

[37] J. D. Savee, S. Soorkia, O. Welz, T. M. Selby, C. A. Taatjes, D. L.

Osborn, J. Chem. Phys., 2012, 136, 134307.

[38] H. Xu, S. T. Pratt, J. Phys. Chem. A, 2013, 117, 9331–9342.

(8)

[39] J. C. Robinson, N. E. Sveum, D. M. Neumark, Chem. Phys. Lett., 2004, 383, 601–605.

[40] N. E. Sveum, S. J. Goncher, D. M. Neumark, PCCP, 2006, 8, 592–598.

[41] B. Gans, L. A. V. Mendes, S. Boyé-Péronne, S. Douin, G. Garcia, H. Soldi-Lose, B. K. Cunha de Miranda, C. Alcaraz, N. Carrasco, P. Pernot, D. Gauyacq, J. Phys. Chem. A, 2010, 114, 3237–

3246.

[42] J.-C. Loison, J. Phys. Chem. A, 2010, 114, 6515–6520.

[43] B. Gans, G. A. Garcia, S. Boyé-Péronne, J.-C. Loison, S. Douin, F. Gaie-Levrel, D. Gauyacq, J. Phys. Chem. A, 2011, 115, 5387–5396.

[44] F. Holzmeier, I. Fischer, B. Kiendl, A. Krueger, A. Bodi, P. Hem- berger, PCCP, 2016, 18, 9240–9247.

[45] C. A. Taatjes, S. J. Klippenstein, N. Hansen, J. A. Miller, T. A.

Cool, J. Wang, M. E. Law, P. R. Westmoreland, PCCP, 2005, 7, 806–813.

[46] P. Hemberger, B. Noller, M. Steinbauer, I. Fischer, C. Alcaraz, B. K. Cunha de Miranda, G. A. Garcia, H. Soldi-Lose, J. Phys.

Chem. A, 2010, 114, 11269–11276.

[47] S. J. Klippenstein, J. A. Miller, A. W. Jasper, J. Phys. Chem. A, 2015, 119, 7780–7791.

[48] H. P. Reisenauer, G. Maier, A. Riemann, R. W. Hoffmann, Angew.

Chem. Int. Ed. Engl., 1984, 23, 641.

[49] M. S. Schuurman, J. Giegerich, K. Pachner, D. Lang, B. Kiendl, R. J. MacDonell, A. Krueger, I. Fischer, Chem. Eur. J., 2015, 21, 14486–14495.

[50] E.E. Rennie, C.A.F. Johnson, J.E. Parker, D.M.P. Holland, D.A.

Shaw, M.A. Hayes, Chem. Phys., 1998, 229, 107–123.

[51] F. Holzmeier, I. Wagner, I. Fischer, A. Bodi, P. Hemberger, J.

Phys. Chem. A, 2016, 120, 4702–4710.

[52] P. Glarborg, A.D. Jensen, J.E. Johnsson, Prog. Energy Combust.

Sci., 2003, 29, 89–113.

[53] E. Ikeda, P. Nicholls, J. C. Mackie, Proc. Combust. Inst., 2000, 28, 1709–1716.

[54] Z. Tian, Y. Li, T. Zhang, A. Zhu, Z. Cui, F. Qi, Combust. Flame, 2007, 151, 347–365.

[55] A. V. Friderichsen, E.-J. Shin, R. J. Evans, M. R. Nimlos, D. C.

Dayton, G.B. Ellison, Fuel, 2001, 80, 1747–1755.

[56] A. M. Scheer, C. Mukarakate, D. J. Robichaud, G. B. Ellison, M.

R. Nimlos, J. Phys. Chem. A, 2010, 114, 9043–9056.

[57] J. C. Mackie, K. R. Doolan, P. F. Nelson, J. Phys. Chem., 1989, 93, 664–670.

[58] B. Sztáray, A. Bodi, T. Baer, J. Mass Spectrom., 2010, 45, 1233–1245.

[59] L. M.P.F. Amaral, Ribeiro da Silva, Manuel A.V., J. Chem. Ther- modyn., 2012, 48, 65–69.

[60] Chase, M. W. NIST-Janaf Thermochemical Tables, 4th ed., J.

Chem. Ref. Data, Monograph 9, 1998.

[61] B. Cronin, M. G. D. Nix, R. H. Qadiri, M. N. R. Ashfold, PCCP, 2004, 6, 5031–5041.

[62] J. C. Biordi, Prog. Energy Combust. Sci., 1977, 3, 151–173.

[63] D. Felsmann, K. Moshammer, J. Krüger, A. Lackner, A. Brockh- inke, T. Kasper, T. Bierkandt, E. Akyildiz, N. Hansen, A. Lucas- sen, P. Oßwald, M. Köhler, G. A. Garcia, L. Nahon, P. Hemberger, A. Bodi, T. Gerber, K. Kohse-Höinghaus, Proc. Combust. Inst., 2015, 35, 779–786.

[64] J. Krüger, G. A. Garcia, D. Felsmann, K. Moshammer, A. Lack- ner, A. Brockhinke, L. Nahon, K. Kohse-Höinghaus, PCCP, 2014, 16, 22791–22804.

[65] N. Hansen, T. A. Cool, P. R. Westmoreland, K. Kohse-Höinghaus, Prog. Energy Combust. Sci., 2009, 35, 168–191.

[66] T. Bierkandt, P. Hemberger, P. Oßwald, M. Köhler, T. Kasper, Proc. Combust. Inst., 2017, 36, 1223–1232.

[67] Y. Li, L. Zhang, T. Yuan, K. Zhang, J. Yang, B. Yang, F. Qi, C. K.

Law, Combust. Flame, 2010, 157, 143–154.

[68] Y. Di, C. S. Cheung, Z. Huang, Atmos. Environ., 2009, 43, 2721–2730.

[69] P. Hemberger, A. J. Trevitt, T. Gerber, E. Ross, G. da Silva, J.

Phys. Chem. A, 2014, 118, 3593–3604.

[70] P. Hemberger, A. J. Trevitt, E. Ross, G. da Silva, J. Phys. Chem.

Lett., 2013, 4, 2546–2550.

[71] J. H. Frank, A. Shavorskiy, H. Bluhm, B. Coriton, E. Huang, D. L.

Osborn, Appl. Phys. B, 2014, 117, 493–499.

[72] J.B.A. Mitchell, C. Rebrion-Rowe, J.-L. LeGarrec, G. Taupier, N.

Huby, M. Wulff, Combust. Flame, 2002, 131, 308–315.

[73] D. Schleier, P. Constantinidis, N. Faßheber, I. Fischer, G. Fried- richs, P. Hemberger, E. Reusch, B. Sztáray, K. Voronova, PCCP, 2018, 20, 10721–10731.

[74] L. Nahon, L. Nag, G. A. Garcia, I. Myrgorodska, U. Meierhenrich, S. Beaulieu, V. Wanie, V. Blanchet, R. Géneaux, I. Powis, PCCP, 2016, 18, 12696–12706.

[75] P. Hemberger, V. B. F. Custodis, A. Bodi, T. Gerber, J. A. van Bokhoven, Nat. Commun., 2017, 8, 15946.

[76] G. Zichittella, P. Hemberger, F. Holzmeier, A. Bodi, J. Pérez- Ramírez, J. Phys. Chem. Lett., 2020, 856–863.

[77] L. Barreau, K. Veyrinas, V. Gruson, S. J. Weber, T. Auguste, J.-F.

Hergott, F. Lepetit, B. Carré, J.-C. Houver, D. Dowek, P. Salières, Nat. Commun., 2018, 9, 4727.