Mechanisms of Electron-Induced Chemistry in Molecular Ices

Abstract

Electron-induced chemistry is relevant to many processes that occur when ionizing radiation interacts with matter. This includes radiation damage, curing of polymers, and nanofabrication processes but also the formation of complex molecules in molecular ices grown on dust particles in space. High-energy radiation liberates from such materials an abundance of secondary electrons of which most have energies below 20 eV. These electrons efficiently trigger reactions when they attach to molecules or induce electronic excitation and further ionization. This review focuses on the present state of insight regarding the mechanisms of reactions induced by electrons with energies between 0 and 20 eV that lead to formation of larger products in binary ice layers consisting of small molecules (H₂O, CO, CH₃OH, NH₃, CH₄, C₂H₄, CH₃CN, C₂H₆) or some derivatives thereof (C₂H₅NH₂ and (C₂H₅)₂NH, CH₂=CHCH₃). It summarizes our approach to identify products and quantify their amounts based on thermal desorption spectrometry (TDS) and electron-stimulated desorption (ESD) experiments performed in ultrahigh vacuum (UHV). The overview of the results demonstrates that, although the initial electron-molecule interaction is a non-thermal process, product formation from the resulting reactive species is often governed by subsequent reactions that follow well-known thermal and radical-driven mechanisms of organic chemistry.

1. Introduction

The interaction of free electrons with molecules can trigger chemical reactions that are relevant to a wide range of phenomena. This includes, for instance, radiation damage of living tissue [1,2], the curing of polymeric materials [3,4], nanofabrication processes such as focused electron beam induced deposition (FEBID) or extreme ultraviolet lithography (EUVL) [1,5,6,7] and also the chemistry in molecular ices that form on dust particles in space as studied in the field of astrochemistry [1,8,9,10,11]. In these processes, ionizing particulate or electromagnetic radiation incident on the material induces the chemical reactions. Common to all kinds of ionizing radiation is that it releases low-energy electrons when interacting with matter. High energy electrons or ions produce an abundance of low-energy secondary electrons (LESEs) with energies ranging from near 0 eV up to 50 eV [1,5,10] while electromagnetic radiation with energy above the ionization threshold of the irradiated material releases photoelectrons (PEs) that, if energetic enough, again produce LESEs [12]. As an example, it is now widely accepted that the chemistry inherent in FEBID is in fact driven by LESEs [1,13,14]. In EUVL, PEs and LESEs can have a similar efficiency with respect to the chemical conversion of a resist as the initial EUV photon [15]. In the context of astrochemistry, it is also nowadays proposed that LESEs released under the effect of ionizing cosmic radiation contribute to the chemistry that occurs in interstellar molecular ices [8,10,11].

Considering only the initial electron-molecule encounter, electron-induced chemistry may appear predominantly destructive in nature. In fact, interaction with an electron often leads to dissociative states of a molecule and consequent bond rupture, processes that are particularly efficient within the typical energy range of LESEs [1,16,17]. More specifically, after electron attachment (EA) to a molecule, the resulting radical anion can decay to an anion and at least one neutral fragment. This process is called dissociative electron attachment (DEA). Neutral dissociation (ND) refers to the fragmentation of a dissociative state formed by electronic excitation (EE) resulting in neutral fragments. As a third pathway, dissociative ionization (DI) leads to fragmentation of a radical cation formed by electron ionization (EI) yielding a cation and again at least one neutral product. Such fragmentation is, for instance, desired in FEBID which produces even freestanding 3D nanostructures by scanning a surface with a keV electron beam to control the fragmentation of adsorbed metal organic precursor molecules and thus the formation of metal deposits [5,6]. However, the reactive fragments produced upon electron-molecule interaction can also cause unwanted effects. As an example, radiation-driven chemistry of adsorbates is relevant to materials used in space. Even the most thoroughly cleaned instruments for trace-analysis [18] have water and other organics adsorbed on their surfaces so that desorption and possibly out-gassing are still measurable years after leaving Earth’s atmosphere [19]. Radiation-induced chemical reactions of the adsorbate or ice layer can then lead to corrosion of the underlying surface [20] and thus degradation of materials [21].

In contrast to the destructive effects mentioned so far, radiation processing of ice layers that accrete over time on cold interstellar dust grains is considered as the most likely scenario for formation of complex molecules in space [8,9]. This is based on the observation that electron interactions with small molecules can lead to the synthesis of larger and structurally more complex products. Such electron-induced syntheses have been summarized before [8,10,16,22] and we refer the reader to these earlier reviews for an overview of the developments in the field. Herein, we aim at providing an overarching perspective of work on the chemistry driven by low energy electrons in binary mixed ices that has been performed in our laboratory for more than ten years [16,23,24,25,26,27,28,29,30,31,32]. In particular, the goal of this review is to demonstrate that after the initial non-thermal electron-molecule interaction, the fate of the thus produced reactive species is not simply governed by random encounters with other molecules. In fact, these species often rather involve in thermal reactions which can be understood based on established reaction mechanisms of synthetic chemistry.

The reactants that were included in the studies reviewed herein comprise the majority of those molecules that are considered to be most abundant in interstellar icy grain mantles [33], namely water (H₂O), carbon monoxide (CO), methanol (CH₃OH), ammonia (NH₃), and methane (CH₄). Ethylene (C₂H₄) is also an important reactant because it is particularly involved in reaction mechanisms that lead to the synthesis of larger products and can be considered as the simplest model compound for unsaturated hydrocarbon species. In these studies, the ice constituents were systematically varied as summarized in

Table 1. Products that can be formed in reactions of C₂H₄ and CO with NH₃, H₂O, and CH₃OH when electron interaction leads to addition of one reactant to the other so that all atoms of the initial reactants become incorporated in the product. Such reactions are also referred to as atom-efficient. The products shown here are those that are formed with the smallest possible number of bonds being broken and reformed as compared to the reactants.

	NH3	H2O	CH3OH
C2H4	Ethylamine [29]	Ethanol [28]	Ethyl methyl ether [32]
CO	Formamide [30]	Formic acid [24]	Methyl formate [26]

which shows target products that incorporate all atoms of the reactants via a minimal amount of bond reorganization. Such systematic screening of ice systems is an important strategy of our approach. By exchanging one of the two constituents of the binary ice, it is possible to identify which of the reactants yields a key intermediate that leads to a particular synthesis. Thereby, deeper insight into the reaction mechanisms is obtained as will be discussed herein. Because of the large number of different products that are formed upon electron irradiation [8,11,34], CH₃OH stands out among the reactants of the binary mixed ices summarized in Table 1 and has therefore also been investigated as pure ice [31]. In addition, earlier work has studied electron-induced reactions in CD₃CN/C₂H₆ mixed ices [23] as well as in pure CH₃CN [25]. Furthermore, the electron-induced reactions of different amines (C₂H₅NH₂ and (C₂H₅)₂NH) with C₂H₄ and of NH₃ and propene (CH₂=CHCH₃) were also investigated [29]. Based on these combined results, this review:

(i)

provides a summary of our experimental approach which relies on a sufficient resolution in electron energy to identify resonant enhancements and energetic thresholds in the energy dependence of product yields and thus the electron-molecule interactions that initiate product formation,

(ii)

shows how information on the mechanisms of electron-driven chemistry in ices can be obtained by investigating the energy dependence of several products in the same sample or by comparing different ice systems, and

(iii)

discusses that the chemistry following the initial electron-molecule encounter can be understood based on established reaction mechanisms known from organic chemistry combined with quantum chemical calculations.

In line with the aims outlined above, this review is organized as follows. The experimental approach and methods are presented in Section 2. This is followed by a summary of the fundamental electron-molecule interactions that have been observed to initiate further reactions in the systems discussed herein (Section 3). Section 4 then discusses the principles of chemical reactions that are initiated by electron-molecule encounters and that lead to the synthesis of new stable products. Section 5 addresses some open questions and gives an outlook on potential topics for further study. Thereby, this review will hopefully serve future studies as reference to the chemical reaction mechanisms that underly the electron-induced chemistry in ices and multilayer molecular adsorbates.

2. Experimental Approach

The general approach of the work reviewed herein is to unravel the mechanisms of electron-induced chemical reactions in molecular ices by post-irradiation thermal desorption spectrometry (TDS) that can be combined, if appropriate, with experiments that directly monitor the electron-stimulated desorption (ESD) of products during irradiation. Both techniques rely on an ultrahigh vacuum (UHV) setup equipped with a variable temperature metal foil onto which ice layers are condensed, a defocused electron gun for irradiation, and a mass spectrometer for the detection of neutral desorbing species. In a TDS experiment, the surface with the ice layer is warmed up at a constant heating rate (typically 1 K/s) and the desorbing species are detected as function of temperature. ESD is studied at constant temperature. In both cases, mass scans can be performed to assess the range of products. In experiments that monitor the desorption rate of specific products as function of time, a maximum of four selected characteristic m/z values is measured in parallel. Based on this approach described in detail in our previous original publications [23,24,25,26,27,28,29,30,31,32], the following sequence of experiments is performed:

(i)

Products that are formed after electron irradiation are identified from TDS data for characteristic m/z values. In the ideal case, the desorption of a particular species can be seen in TDS curves for several m/z values and the intensity of the desorption peaks for these different m/z reflects the intensities in the mass spectra (MS) of this compound recorded on the same instrument when leaking the pure vapor of the product into the UHV chamber.

(ii)

Product formation is quantified from integrated desorption signals for a characteristic m/z ratio. The absolute product yields can be derived by comparing these integrated desorption signals to those obtained from ices that have been prepared with defined composition and surface coverage, as can be deduced from TDS, so that the amount of the respective product is known [31,35,36]. Relative amounts of different products can also be obtained without such a standard sample by using partial ionization cross sections for the characteristic m/z ratios used to monitor the products. The partial ionization cross sections ${\sigma }_p$ for a given m/z ratio with intensity $I_p$ of a specific product is defined by ${\sigma }_p=\frac{I_p}{\sum I_i}·{\sigma }_T$, where $I_i$ are the intensities of all m/z ratios of that product and ${\sigma }_T$ is its total ionization cross section. These cross sections, that refer to the electron energy applied for ionization in the mass spectrometer, define the intensities with which specific m/z ratios appear in the mass spectrum. If known, they can be used as correction factor that is applied to compare the intensities for specific m/z ratios and thus also the desorption peak areas of two different products and to convert them to a relative yield [24,29,31].

(iii)

The thus obtained product yields are determined as function of electron exposure, which is defined as the transmitted charge per surface area as measured on the metal foil, to establish over which range they increase linearly with exposure. In this linear regime, the decay of the initial reactants is still negligible and the product is not yet consumed by electron irradiation [24,26,29,30,31,32,35,36]. Under this condition, the rate with which a product is formed is directly proportional to the applied electron exposure. This is the basis on which the rates of formation can be compared between different products.

(iv)

Product yields are then measured as function of electron energy (E₀) for an exposure that is within the linear regime for all investigated electron energies. The energy dependence of the product yields gives information on the electron-molecule interactions that initiate product formation. This includes thresholds for electronic excitation or ionization as well as maxima within specific energy ranges that point to a contribution of electron attachment. These processes are summarized in Section 3. When monitoring product formation in mixed ices, it is not obvious from the energy dependence alone which constituent undergoes the initiating electron-molecule interaction if both show similar DI or ND thresholds or energy ranges for DEA. In these cases, more detailed mechanistic information can be derived by comparing the energy dependent yields of several products [24,26] as will be exemplified in Section 4. In particular cases, a unique assignment can be obtained by exchanging one component of the ice for another and checking whether product formation still occurs [37].

Figure 1 illustrates the individual steps of the data analysis performed to obtain electron energy dependent product yields from TDS experiments using the example of CO/NH₃ mixed ices. In the example presented here, m/z ratios of 43 and 45 have been monitored to reveal the formation of the stable neutral products formamide (H₂NCHO) and isocyanic acid (HNCO). In this particular case, the desorption peaks of the two compounds overlap (Figure 1a). While m/z 45 is representative of formamide alone, m/z 43 represents both the molecular ion of isocyanic acid and a cationic fragment obtained from formamide, both being produced in the ion source of the mass spectrometer. The desorption signal obtained for m/z 43 must therefore be deconvoluted into contributions of the two products. This is achieved by scaling the m/z 45 desorption signal of formamide with the relative intensity of the m/z 43 fragment which was deduced from the mass spectrum of formamide (Figure 1b). Subsequently, the area of this scaled peak is subtracted from the integrated m/z 43 desorption signal leaving only the contributions of HNCO to the desorption peak (Figure 1c). These integrated desorption signals are then converted to relative product yields using partial ionization cross sections (Figure 1d). An appropriate electron exposure within the linear regime is then selected based on the dependence of relative yields on exposure (Figure 1e) after which the energy dependence is obtained by measuring the relative product yields at this selected electron exposure for a range of different electron energies (Figure 1f).

For reference, we also summarize the technical details of the experiments. Briefly, the experiment is housed in a UHV chamber (Figure 2) with a base pressure of 10⁻¹⁰ mbar maintained by turbomolecular pumps. The UHV chamber is connected to a gas handling manifold which consists of two storage and one mixing reservoir (see Figure 3). Reactants can be leaked from the storage reservoirs into the mixing reservoir by precision leak valves. The increase of partial pressures in the mixing reservoir which is typically chosen as a few tens of mTorr is monitored with a MKS Baratron type 622B capacitance manometer which allows one to establish a defined gas composition. In between two consecutive experiments, the mixing reservoir is evacuated by a turbomolecular pump.

Molecular ices were deposited by leaking gases or vapors from the mixing reservoir into the main chamber via a stainless-steel tube whose end is located in front of the metal foil substrate (Au [23,25,27,28,29,30,35,36] or Ta [24,26,31,32]) held at 30–35 K by a closed-cycle helium cryostat (Leybold RGD210 or C/P 2-10). During this deposition step, the pressure in the main chamber typically increases by 1–2 orders of magnitude. This ensures that there is sufficient partial pressure of the gas from the manifold in the chamber to form a film. The only major background gas under UHV conditions is H₂, which does not co-condense. Prior to each deposition, the substrate is resistively heated. This ensures that adsorbates which may have accumulated on the surface are removed.

The thickness of the investigated ice layers is estimated from TDS data for individual compounds by observing at which pressure drop the transition from monolayer to multilayer desorption signal occurs [23]. TDS data are acquired by resistively heating the substrate with heating rates of ~1 K/s while monitoring characteristic m/z ratios of the investigated neutral desorbing species using a quadrupole mass spectrometer (QMS, Stanford Research Systems RGA200 or RGA300) residual gas analyzer with electron-impact ion source operating at an electron energy of 70 eV. Typically, ice layers are prepared with mean thickness in the range of 10 to 20 monolayers. This minimizes the contribution of the interface between the ice and the underlying substrate to the chemistry that is monitored. On the other hand, these layers are thin enough for the electrons to efficiently penetrate the ice. In the case of new results, the actual layer thickness is stated in the figure caption. We refer to the original publications for more detailed information on data from previous work.

A targeted mixing ratio of the two ice components of, for instance, 1:1 can be established in the simplest case by leaking a gas mixture with a molar fraction of 1:1 onto the substrate [28,29,32]. The actual stochiometric ratio of the two reactants in the ice can be deduced from the area under the TDS curves and considering the partial cross sections for electron impact ionization of the characteristic fragments that are monitored in the QMS [32]. The thus determined ice composition can deviate from the mixing ratio of the gases if one of the compounds adsorbs more strongly than the other. In this case, the gas mixture can be adjusted to produce a 1:1 composition of the ice as deduced from TDS [32]. Alternatively, the two compounds can be leaked sequentially. This is advantageous when one of them tends to form a porous ice such as H₂O. The pores can then accommodate the molecules of the second gas and thus enhance its sticking probability. Better control over layer thickness is achieved, when the two gases are leaked sequentially.

Commercial non-monochromatized electron sources (Specs FG15/40 or STAIB NEK-150-1) with energies between 1 eV and 500 eV with a typical energy resolution of 0.5–1 eV are used for electron irradiation [24,25,26,28,29,31,32]. The electron exposure was determined by integrating the transmitted current over time and dividing by the sample area. Before and after irradiation, the substrate is biased with a negative potential which exceeds the range of investigated electron energies, while the electron source is left on. This allows for precise control over start and end of the irradiation, while at the same time preventing any electrons from reaching the sample during deposition or TDS.

Data analysis and visualization is performed with Python [38] using the Numpy [39], SciPy [40], and Matplotlib [41] packages. Baseline correction of the ESD and TDS data was performed, in the simplest case, by subtracting a constant value from the raw data. This value can be calculated by averaging the background signal. Alternatively, the baseline is fitted by a line or polynomial or, in the case of ESD experiments, with an exponential function to correct for any temporal drifts of the background signal. In the case of low signal intensities, the noise level is reduced by applying a second-order Savitzky-Golay filter [42,43] with a window length of 9 points. In order to compare signals from different experiments quantitatively, the area under the desorption signal is numerically integrated by the trapezoidal method. An experimental error of the integrated desorption signals is established by repeating selected experiments 3–5 times to calculate a standard deviation.

3. How Do Reactive Species Form under Electron Irradiation?

3.1. Basics of Electron Molecule Interactions

Upon interaction with a free electron, a molecule can be promoted to different states that are often more reactive than the molecular ground state. In electron ionization (EI), the impinging electron liberates another electron from the target to leave a radical cation behind. Electronic excitation (EE) leads to a neutral excited state while electron attachment (EA) to the molecule yields a radical anion:

M + e⁻ → M^•+ + 2 e⁻     electron ionization (EI)

M + e⁻ → M* + e⁻      electronic excitation (EE)

M + e⁻ → M^•−      electron attachment (EA)

The dependence of the three processes on electron energy is different, with EE and EI showing a threshold-type behavior. When the energy of the impinging electron exceeds the threshold for electronic excitation (EE) or the ionization threshold (EI), the cross section for interaction steadily increases until it reaches a plateau between 30 to 70 eV (for EE) and 50 to 100 eV (for EI) [16]. After this plateau, the cross section decreases due to ever shorter interaction times owing to the increasing speed of the electron. In contrast, electrons can attach to a molecule only if there is an anion state available for the energy of the incoming electron. This leads to EA in well-defined energy ranges called resonances. At energies typically below 5 eV, this happens in so-called single particle resonances, where the incoming electron occupies a previously unoccupied orbital. At higher energies, electron attachment is usually accompanied by electronic excitation leading to core-excited resonances.

The three states that a molecule M can occupy after the encounter with the electron (radical cation, radical anion, excited) can be more chemically reactive than the ground state molecule M. If these reactive species are in contact with other molecules or atoms, further chemical synthesis can be triggered.

The described processes assume that the molecule itself stays intact. Yet, molecules do in fact often dissociate after interaction with an electron, if the energy of the electron is sufficiently high to overcome the dissociation energy of a chemical bond in the molecule. This energy can be offset to some degree by the formation of new chemical bonds [16]. There are thus a number of further processes that can lead to the formation of reactive species:

M + e⁻ → A⁺ + B^• + 2 e⁻     dissociative ionization (DI)

M + e⁻ → A^• + B^• + e⁻     neutral dissociation (ND)

M + e⁻ → A⁻ + B^•     dissociative electron attachment (DEA)

For DI, the thresholds are the same as for EI increased by the required bond dissociation energy. The same can hold for ND as compared to EE. Often, however, the presence of dissociation is not governed by additional energy but by whether the primary excited state M* is dissociative or not. This also applies to anion states M^•− formed by EA but here dissociation competes with autodetachment of the electron [44]. Furthermore, anion states M^•− are preferentially formed at near-thermal electron energies where excess energy can be removed from a particular bond by redistribution of energy to different internal degrees of freedom of the molecule [1]. There is also an additional dissociative process following electronic excitation, namely dipolar dissociation (DD) of the excited state into a cation and an anion. However, DD is usually considered to only have significant cross sections at energies higher than the ones discussed in this review [17,45,46,47]. We note that ND can also result from EA to a high-lying anion state which subsequently decays via autodetachment of the electron to produce neutral fragments [48]. This type of reaction mechanism has also been brought forward as bond breaking by catalytic electrons (BBCE) mechanism [49,50,51].

In the small organic molecules treated herein, the dissociative processes usually lead to separation of radical site and the ion into separate species. The radical species that are produced in this way drive most of the ensuing chemistry that occurs in condensed phase. Section 3.2, Section 3.3 and Section 3.4 present an overview of the reactive species (ionic and/or radical) that were identified as being relevant to the electron-induced chemistry discussed in Section 4. Section 3.5 then summarizes effects of the condensed phase that modify the fundamental electron-molecule interactions.

3.2. Reactive Intermediates Formed by EI/DI

In the studies reviewed here, the energy of the impinging electrons typically ranges up to around 20 eV. This energy range spans the lower part of the typical energy distribution of LESEs and includes their maximum yield for diverse materials [1,52,53]. The reactive species discussed here thus do not include ions and radicals with an appearance energy above ~20 eV.

Evidence for EI/DI as initiating step of electron-induced reactions in mixed ices has been obtained for the prototypical reactants C₂H₄, CO, NH₃, H₂O, CH₃OH and CH₄. The reactive species relevant to our studies that can be formed by EI/DI are summarized in

Table 2. Reactive species formed by EI and DI as relevant to the electron-induced reactions in molecular ices discussed in Section 4. Note that the listed species were not directly detected but their contribution to the reactions is deduced indirectly from the identified final reaction products. Other electron-molecule interactions may contribute as well as summarized in Section 3.3 and Section 3.4.

Source Molecule	Reactive Species	Ionization Threshold (Gas Phase) 1	Threshold for Product Formation (Ice)	References (Ice)
C2H4	C2H4•+	10.5 eV	8–10 eV	[28,29,32]
CO	CO•+	14.014 eV	13–14 eV	[30]
NH3	NH3•+	10.070 eV	≤12 eV/10 eV	[27,30]
H2O	H2O•+	12.621 eV	11 eV	[28]
CH4	CH4•+	12.61 eV	12 eV	this work
CH3OH	CH3OH•+ CH3O•+	10.84 eV 11.67 eV	9.8 eV ≤11 eV	[31,32] [32]

¹ Values for gas phase ionization energy are taken from NIST Webbook database [54].

. We omit for simplicity propene and amines because the reactions identified in this case were analogous to those of the shorter analogues C₂H₄ and NH₃ [29]. With the exception of CH₃OH, where significant fragmentation in the gas phase sets in almost at the ionization threshold, we did not find evidence that dissociation following EI contributes to the electron-induced reactions described in Section 4. While EI fragments with appearance energies below 20 eV are known [54], apparently their formation cross sections are too low to play a significant role in the studies reviewed here. We thus refer to the conditions present in the experiments as “soft ionization”.

3.3. Reactive Intermediates from ND

In many of the studied systems, product formation was observed starting from energies well below the ionization threshold [24,25,26,31]. When the energy dependence reveals threshold-type processes and not resonances, the only primary interaction available in this energy range (usually between 6 and 10 eV) is ND. Contrary to EI thresholds, the exact thresholds for ND are often unknown, even in the gas phase [45]. This is in part due to the fact that ND as a source for reactive species under electron irradiation is difficult to assess due to the experimental difficulties in monitoring reactive neutral radical species [17]. The only recourse for the identification of ND processes in electron-driven chemistry is thus by exclusion of EI/DI processes on grounds of the lower observed threshold energy and exclusion of EA/DEA processes by their non-resonant energy dependence. If an observed process starts near the minimum energy for electronic excitation, it is attributed to ND. We have observed product formation triggered by ND in systems containing NH₃, H₂O, CH₄, and especially CH₃OH, which produces a number of very important radical species (see

Table 3. Reactive species formed by ND as relevant to the electron-induced reactions in molecular ices discussed in Section 4. Note that the listed species were not directly detected but their contribution to the reactions is deduced indirectly from the identified final reaction products. Other electron-molecule interactions may contribute as well as summarized in Section 3.2 and Section 3.4.

Source Molecule	Reactive Species	Threshold for Product Formation (Ice)	References (Ice)
NH3	H• + NH2•	~7 eV	this work
H2O	H2 + O H• + HO•	6–7 eV 6–7 eV	[24] [24,28]
CH4	H• + CH3•	~10 eV	this work
CH3OH	CH3O• + H• H• + •CH2OH CH3• + HO•	~6 eV ~6 eV ~6 eV	[31] [31] [31]

). Note that in the case of CO and C₂H₄, the formation of products could be rationalized without having to resort to ND. While we did observe the formation of butadiene in the latter case, this was at energies that also allow DI processes [28,29].

3.4. Reactive Intermediates from (D)EA

Resonant structures in the energy dependence pointing to the formation of products via EA/DEA processes was observed for ices containing C₂H₄, CO, H₂O, CH₃OH and CH₃CN. The relevant reactive species are listed in

Table 4. Reactive species formed by (D)EA as relevant to the electron-induced reactions in molecular ices discussed in Section 4. Note that the listed species were not directly detected but their contribution to the reactions is deduced indirectly from the identified final reaction products. Other electron-molecule interactions may contribute as well as summarized in Section 3.2 and Section 3.3.

Source Molecule	Reactive Species	Resonance in Product Formation (Ice)	References (Ice)
C2H4	C2H4•−	<5 eV	[28]
CO	CO•− C + O•−	~4 eV 11 eV/16 eV	[24] [37]
H2O	H• + HO− O•− + H2	6.5 eV 10 eV	[24] [24]
CH3OH	CH3O• + H− CH3O− + H•	5.5 eV 5.5 eV	[32] [26,31]
CH3CN	CH3• + CN−	4 eV/10 eV	[25]

. In contrast, no true evidence for product formation that starts with EA to NH₃ and CH₄ was found in our studies. While we have invoked DEA to NH₃ as explanation for the resonant formation of formamide from NH₃ and CO at around 10 eV [30], recent preliminary results from our lab have led us to question this assignment, as we now see it as much more likely that the process is instead triggered by (D)EA to CO, as observed in the formation of other compounds from CO at that energy [24,26,37]. For reference, an overview of gas phase EA/DEA processes in the investigated reactants is given in Figure 4.

3.5. Effect of Condensed Phase

Most studies on the fundamental electron-molecule interactions are conducted in the gas phase. These crossed beam experiments typically have an energy resolution in the meV range and very low detection limits of charged particles [1,44]. For the interpretation of condensed phase experiments, these data are very helpful, but need to be treated with a bit of caution [63]. In particular, the threshold energies needed to form a charged particle are often lowered by 1 to 2 eV in the condensed phase, because charged particles are stabilized by polarization of the matrix [10,64]. For the same reason, desorption of particles is often hindered and occurs at higher energies than expected from the gas phase, or is suppressed entirely [25,65].

On top of the changes in energy, additional effects come into play once an electron-molecule interaction occurs in a molecular matrix. The caging effect can suppress fragmentation, as particles quickly lose energy to the matrix and can then recombine. This was studied in detail for methanol, where it was shown that in clusters, DI fragmentation is almost entirely absent, compared to the gas phase [66,67]. This can also lead to DEA resonances disappearing in the condensed phase (see for instance Ref. [68]). EA resonances can also initiate fast reactions with the matrix [69]. The radical anions formed from EA to CO or C₂H₄, CO^•− and C₂H₄^•−, are very strong bases and can quickly react with acidic protons from matrix molecules like H₂O, yielding HCO^• and ethyl radicals C₂H₅^• respectively [24,28].

In a similar vein, the radical cations produced from EI can undergo acid-base reactions with neutral molecules [70] leading to proton transfer reactions [71] between H₂O^•+ and H₂O to yield H₃O^•+ and HO^• radicals, or between NH₃^•+ and NH₃ to yield NH₄⁺ and NH₂^• radicals. Such acid-base chemistry has also been observed for larger molecules like acetaldehyde [36,72], primary and secondary amines [29] and alcohols [31]. It is also important for H⁻ that is produced by DEA to numerous molecules (see, for instance, Figure 4). As a strong base, H⁻ abstracts a proton from many reactants. This yields species such as HO⁻ or CH₃O⁻ when reacting with H₂O or CH₃OH [73].

While radical recombination reactions are typically barrierless, the reaction between a radical and a closed-shell species usually comes with a (small) energy barrier [74]. In cryogenic ices, this energy might not be available, which would lead to the stabilization of radicals. Observation has, however, shown that in addition to anionic and cationic species, also neutral radicals can undergo further reactions in the condensed phase. For instance, the hydroxyl radical HO^•, will react with CO from the matrix, yielding the radical species HOCO^• [24]. This is mostly due to the fact that dissociative electron interactions usually impart some excess energy to fragments, which makes them hotter than their surroundings, enabling rapid diffusion as well as radical molecule interactions. When attempting to understand reaction mechanisms, not only HOCO^• [24], but also other secondary radical species, such as HCO^• [24,26,30], C₂H₅^• [28] or HOC₂H₄^• [28] need to be considered.

4. How Do Reactive Species Undergo Chemical Transformation in Condensed Phase?

Electron irradiation is not only destructive but can also lead to molecular synthesis, i.e., the formation of more complex molecules from smaller entities [16]. In many cases, the underlying mechanisms are remarkably similar to those known from organic chemistry including nucleophilic displacement reactions [75], acid-base reactions [28], addition reactions [27,28,29], and oligomerization [76,77]. This suggests that electron-induced synthesis is in fact not a random re-assembly of atomic and molecular fragments but that it follows certain reaction principles. This chapter discusses the reactions that the initially formed reactive species can undergo by outlining these principles. In Section 4.1 (Radical recombination) we summarize evidence that concurrent reactions may occur so that the products are not simply explained by stochastic encounters of radicals. In Section 4.2 (Transfer hydrogenation). we present syntheses of new molecules that are initiated by formation of molecular radical anions via low-lying EA processes and subsequent protonation of the latter. Section 4.3 (Reduction by hydrogen radicals) deals with reactions that are caused by hydrogen radicals released upon dissociative electron-molecule interactions. Section 4.4 (Hydrogenation by methoxy radicals) then presents reactions where the reducing H^⦁ is instead supplied by transfer from a methoxy radical. In Section 4.5 (Addition reactions) we summarize reactions in which a radical or ion adds to multiple bonds of a reaction partner leading to atom-efficient synthesis of larger molecules. And finally, Section 4.6 (Oxidation of CO to CO₂) discusses the mechanism by which CO is oxidized in electron-induced chemistry.

4.1. Radical Recombination

Current astrochemical models assume that radical recombination is the dominant mechanism by which molecular synthesis occurs in interstellar ices [34,78,79,80]. However, neutral radicals which are produced in ices by the interaction of molecules with electrons or UV photons are typically surrounded by closed-shell molecules only. Radical–molecule reactions thus require no diffusion through the ice but are limited by activation barriers [74] which might not be accessible at cryogenic temperatures. Diffusion, on the other hand, requires less energy allowing the radicals to travel until they encounter a second radical where a typically barrierless recombination reaction can occur. On this basis, recombination would be governed by the stochastic encounter of different radical species as might also be suggested, for instance, from the products observed upon electron irradiation of condensed CF₂Cl₂ [10,81]. As has been discussed in Section 3.5, this paradigm does not necessarily hold true because radicals released during dissociation of the parent molecule might have considerable excess energy which can be sufficiently high to overcome the activation barriers of radical-molecule reactions even at low temperatures. In fact, theoretical studies suggest that transfer of a hydrogen atom between two radicals, also referred to as disproportionation, might dominate over recombination [82,83,84,85]. An example of such a disproportionation reaction would be the reaction between two CH₃O^• radicals yielding formaldehyde (H₂CO) and CH₃OH [31]. A similar case has been observed in matrix isolation experiments for the reaction between two HCO^• radicals which disproportionate to H₂CO and CO rather than recombine to glyoxal (HC(O)CH(O)) [82].

As an example of a recombination reaction, the formation of N₂ in ices containing NH₃ (see also Section 4.1) relates to the encounter of NH₂^• radicals. This is supported by the detection of hydrazine (N₂H₄) as the immediate recombination product [29,86]. N₂H₄ was identified, together with diimide (N₂H₂), upon electron irradiation of pure NH₃ layers [86]. The intermediate chemisorbed NH₂^• radicals were observed earlier upon electron-irradiation of a bilayer of NH₃ at E₀ = 50 eV as deduced from high-resolution electron energy loss spectroscopy (HREELS) and X-ray photoelectron spectroscopy (XPS) [87]. Loss of hydrogen from NH₃ may result from DEA. However, the present results (Figure 5) reveal a lack of resonances in the energy dependent yield of both N₂H₄ and N₂ from NH₃ ice (Figure 5) which argues against a DEA-driven mechanism as dominant source of NH₂^• radicals.

Among the molecules reviewed in this work, radical recombination is particularly important in the case of CH₃OH containing ices. In CH₃OH, three different bonds can be cleaved yielding CH₃O^•, ^•CH₂OH, CH₃^•, HO^•, and H^• radicals. Following electron irradiation of methanol ice, the recombination products methoxymethanol (CH₃OCH₂OH), ethanol (C₂H₅OH), ethylene glycol (HOC₂H₄OH), and dimethyl ether (CH₃OCH₃) were identified [31]. The similar energy dependencies for these products (Figure 6) suggest that the different radicals are produced by the same electron-molecule interactions. Assuming that radical recombination is in fact the dominant reaction, the relative product yields would reflect branching ratios with which the different radicals recombine. This approach was recently used in a photodissociation study of CH₃OH [79]. The authors deduced branching ratios into CH₃O^•, ^•CH₂OH, and CH₃^• radicals from the relative yields of ethanol, dimethyl ether, and ethylene glycol [79]. However, considering recent theoretical predictions that disproportionation dominates over radical recombination [82,83,84,85], it is obviously not possible to deduce the branching ratios by which CH₃O^•, ^•CH₂OH, and CH₃^• radicals are formed because, unlike for recombination products, the products of such hydrogen transfer are not linked to specific primary radicals. For example, it is not possible to distinguish between H₂CO that is formed by hydrogen transfer between two CH₃O^• radicals, two ^•CH₂OH radicals, or between a CH₃O^• and a ^•CH₂OH radical.

Considering the possible combinations of radical fragments of CH₃OH listed above, it is obvious that more compounds would be expected upon recombination governed by stochastics such as dimethyl peroxide (CH₃OOCH₃) and ethane (C₂H₆). These, however, were not detected which calls for an explanation. Dimethyl peroxide, which is formed by recombination of two CH₃O^• radicals, could only be detected by the Kaiser group by use of a very sensitive photoionization reflectron time-of-flight mass spectrometer [88]. This suggests that dimethyl peroxide is produced in much smaller yields than other recombination products so that it is below the detection limit of our mass spectrometer. The competition of radical recombination with disproportionation provides a reasonable explanation why only certain recombination products are observed in high yields. Here, a higher efficiency of the disproportionation reaction between two CH₃O^• radicals yielding CH₃OH and H₂CO as compared to recombination can rationalize the low yield of dimethyl peroxide.

There are no reports on the formation of C₂H₆ under irradiation of CH₃OH ice which is expected to form by coupling of two CH₃^• moieties, while its formation from the electron irradiation of CH₄ (Ref. [89] and this work) and CH₃CN [25] was indeed observed (Figure 7). In the case of CH₄ ice (Figure 7, top), the threshold-type behavior of the yield of C₂H₆ points to formation through a reaction driven by ND or EI. In contrast, pronounced resonances are observed in the energy dependence of C₂H₆ yields upon irradiation of CH₃CN [25]. They reflect known DEA processes that lead to release of CN⁻ which is presumably accompanied by loss of CH₃^• radicals [62,90]. Öberg proposed that the lack of C₂H₆ formation under irradiation of CH₃OH results from the higher abundances of CH₃O^• and ^•CH₂OH radicals which favors reactions of CH₃^• with either CH₃O^• or ^•CH₂OH over recombination of two CH₃^• radicals [79]. In addition, CH₃^⦁ radicals have a higher mobility at 35 K than CH₃O^⦁ and ^⦁CH₂OH [79,91] which might enable quick trapping of CH₃^⦁ radicals before the majority of CH₃O^⦁ and ^⦁CH₂OH radicals are consumed by other reactions. Disproportionation of two CH₃^• radicals would yield methylene (CH₂) and CH₄ which is endothermic by 0.21 eV as has been deduced from the standard enthalpies of formation [92] and should thus not be very efficient. However, CH₃^• radicals could react with CH₃OH by abstracting a H^• from the latter, which yields CH₄ as also observed in the irradiation of methanol [31] and either CH₃O^• or ^•CH₂OH. Thus, hydrogen radical abstraction, which competes with radical recombination, might be a crucial factor that prevents formation of C₂H₆, as will be discussed in the Section 4.2 in detail.

Recombination of radicals is also observed in CO/CH₃OH [26] and C₂H₄/CH₃OH mixed ices [32], although the yields may be lower compared to pure CH₃OH ice samples with the same film thickness as is the case for the product ethylene glycol (HOC₂H₄OH) (Figure 8, shaded in grey. On one hand, this is because the amount of CH₃OH is smaller when part of it has been replaced by a second reactant. On the other hand, radical recombination competes with concurrent reactions such as hydrogenation of CO and C₂H₄ which proceeds via different mechanisms (see Section 4.2, Section 4.3 and Section 4.4) and addition of radicals to CO or C₂H₄ (see Section 4.5). Because of these concurrent reactions, radical recombination is neither a purely stochastic process nor the only possible reaction by which radicals form stable products.

4.2. Transfer Hydrogenation

Transfer hydrogenation is an important reaction class in organic chemistry which subsumes all hydrogenation reactions where the reducing agent is not gaseous H₂ [93]. Many of those reactions rely on the proton transfer between a radical anion and a proton donor, often an alcohol. A prominent example of such reactions is the Birch reduction [94,95] which can be used to reduce aromatic compounds. The reaction employs solvated electrons as reducing agents which are provided by dissolving either Na or Li in liquid NH₃ [94,96]. Subsequently, these electrons add to an aromatic molecule yielding a radical anion. In a final reaction step, the radical anion is protonated by an alcohol, leading to hydrogenation. Depending on the reaction conditions, the molecule can undergo this reaction sequence multiple times leading to further hydrogenation. Several variations of the Birch reduction have been developed including electrochemical [97] and photochemical [98] all of which, however, rely on the initial formation of a radical anion.

Similar to the mechanism of the Birch reduction, EA to CO was observed at E₀ = 4 eV in CO/H₂O mixed ices. This was inferred from the formation of formaldehyde (H₂CO) at this energy (Figure 9). EA yields an intact CO^⦁− radical anion which can undergo proton transfer with H₂O yielding HCO^⦁ and HO⁻. Subsequently, HCO^⦁ reacts with a nearby H₂O molecule yielding H₂CO and HO^⦁ [24]. Recall that the resonance at 4 eV can only be due to EA to CO as DEA to H₂O occurs at considerably higher energies (Figure 4). As noted in Section 3.5, an analogous reaction leading to the formation of ethanol and its side products ethane (C₂H₆) and butane (C₄H₁₀) was observed after EA to C₂H₄ in C₂H₄/H₂O mixed ices at electron energies below 5–6 eV [28]. In this case, EA produces the C₂H₄^⦁− radical anion which undergoes proton transfer with H₂O to yield C₂H₅^⦁ and HO⁻ (Scheme 1). Subsequently, the C₂H₅^⦁ radical can attack the double bond of another C₂H₄ moiety to form a dimer C₄H₉^⦁ which can react further in a chain reaction. If either of the alkyl radical intermediates reacts with H₂O, however, the chain propagation is terminated yielding C₄H₁₀ and a HO^⦁ radical. The HO^⦁ radical, in turn, can attack a nearby C₂H₄ moiety forming HOC₂H₄^⦁. Finally, reaction with a nearby H₂O molecule again yields ethanol and a further HO^• (Scheme 1) potentially resulting in another chain reaction. Consideration of side products such as the different alkanes can help to identify the relevant intermediates and thus leads to a more comprehensive view of the reaction. We note, however, that such a reaction was not observed in C₂H₄/NH₃ mixed ices [29] suggesting that a sufficiently high acidity of the proton donor is essential for the reaction to occur.

This type of reaction has also neither been observed in C₂H₄/CH₃OH nor in CO/CH₃OH mixed ices despite the fact that the gas phase acidity of CH₃OH is in fact higher than that of H₂O [99]. One reason for this lack of proton transfer from CH₃OH could be the effect of the molecular environment. For example, the order of acidities of H₂O and CH₃OH is reversed in aqueous solution as HO⁻ is more strongly stabilized than CH₃O⁻ [100,101]. Thus, depending on the stability of the CH₃O⁻ anion in a CH₃OH matrix, the acidity of CH₃OH might be too low to enable proton transfer. Another factor might be the effect of different molecular environments on the stability of the radical anion as has been reported for acetonylacetone [102]. A water matrix might be able to stabilize the CO^•− and C₂H₄^•− radical anions so that they have a sufficient lifetime to undergo subsequent chemistry whereas CH₃OH is not. However, the finding that H₂CO is formed via EA to CO in CO/H₂O mixed ices [24] and the formation of ethanol in C₂H₄/H₂O mixed ices via EA to C₂H₄ [28] demonstrate the relevance of transfer hydrogenation reactions, analogous to the Birch reduction, for electron-induced chemistry of molecular ices.

4.3. Reduction by Hydrogen Radicals

ESD and crossed beam experiments on DEA to H₂O and CH₃OH revealed the production of HO⁻ [103] and CH₃O⁻ [61,104,105] (see Figure 4), respectively. This must be accompanied by the release of H^•. In addition, there is evidence that hydrogen radicals are produced by ND of H₂O [106]. Following these and possibly other initial electron-molecule interactions (see Section 3.3 and Section 3.4), the released H^•, which are particularly mobile, can react with unsaturated compounds such as CO or C₂H₄ enabling reaction pathways that are neither accessible in pure H₂O nor in CH₃OH. Hydrogenation of CO has been reported to occur during electron irradiation of CO/H₂O and CO/CH₃OH mixed ices yielding H₂CO, CH₃OH, and intermediate radicals (Reaction 1) [78,80,107,108,109].

$$\mathrm{CO}\stackrel{{\mathrm{H}}^{•}}{\to }{\mathrm{HCO}}^{•}\stackrel{{\mathrm{H}}^{•}}{\to }{\mathrm{H}}_2\mathrm{CO}\stackrel{{\mathrm{H}}^{•}}{\to }{ }^{•}\mathrm{C}{\mathrm{H}}_2\mathrm{OH}/{\mathrm{CH}}_3{\mathrm{O}}^{•}\stackrel{{\mathrm{H}}^{•}}{\to }{\mathrm{CH}}_3\mathrm{OH}$$

(1)

The hydrogenation of CO by H₂O and CH₃OH was also investigated by studying the energy dependence of H₂CO formation [24,26]. In the case of CO/H₂O mixed ices [24], the threshold around 7 eV (Figure 10) suggests that Reaction 1 is mainly triggered by ND of H₂O into H^• and HO^•. In fact, the yield of H₂CO does not exhibit resonances in the energy range from 6–13 eV suggesting that DEA of H₂O into HO⁻ and H^• (Figure 4) does not significantly contribute to the hydrogenation of CO. Note that the resonant formation of H₂CO at 4 eV results from EA to CO and subsequent protonation of the intermediate CO^•− radical anion by H₂O (Section 4.2).

In CO/CH₃OH mixed ices [26], a more sophisticated approach is required to study Reaction 1 as H₂CO is not only formed by hydrogenation of CO but also by fragmentation of CH₃OH [26,31]. Other studies have differentiated between these two possibilities by using ¹³C and ¹⁸O labelled CH₃OH and CO [78,80]. However, there is an alternative approach to isotopic labeling. Due to contributions of Reaction 1 to the overall yield, the energy dependence of H₂CO formation in CO/CH₃OH mixed ices differs from that in pure CH₃OH, where it is dominated by a high-lying resonant process [31] (Figure 11a). The energy dependence of this additional formation channel shows an onset at ~7 eV, in good agreement with the electronic excitation energy of CH₃OH which is about 6.4 eV in the gas phase [110] and 6.7 eV in the condensed phase [111]. The contribution of this channel to the H₂CO yield increases steadily with electron energy (Figure 11a), indicative of a non-resonant process. This agrees well with the hypothesis that ND of CH₃OH yields hydrogen radicals which then reduce CO according to Reaction 1. There is no further threshold near the ionization energy of CH₃OH suggesting that EI does not produce significant amounts of H^• which could contribute to hydrogenation of CO.

A similar result was observed at lower electron energies (Figure 11b). Here, the two DEA channels of CH₃OH at 5–6 eV yielding CH₃O^– and H^•, and CH₃O^• and H⁻ (Figure 4) are responsible for the formation of H₂CO [26]. In pure CH₃OH ice, the H₂CO yield increased linearly with electron energy. This linear increase is ascribed to an overlap of the DEA channel yielding CH₃O^• and H⁻ (Figure 4) [105,112] with the onset of ND. [110,111] Subsequently, two CH₃O^• radicals can undergo disproportionation to yield H₂CO and CH₃OH. The DEA channel yielding CH₃O⁻ and H^• (Figure 4) does not contribute to H₂CO formation in pure CH₃OH. In the presence of CO, the released H^• can react with CO (Reaction 1), enabling formation of additional H₂CO. This additional reaction channel is reflected in the energy dependence of H₂CO where the linear increase of product yield is overlaid with a more intense DEA resonance at ~6 eV (Figure 11b).

In the CO/CH₃OH mixed ice, the DEA channel at 5.5 eV yielding CH₃O⁻ and H^• (Figure 4) does not only lead to the formation of H₂CO but also produces methoxymethanol (CH₃OCH₂OH) which is not observed after electron irradiation of pure CH₃OH at E₀ = 5.5 eV (Figure 12). This is because this DEA channel only yields CH₃O^⦁ and H⁻ but no ^⦁CH₂OH radicals [112]. The recombination reaction between CH₃O^⦁ and ^⦁CH₂OH radicals needed for formation of methoxymethanol in pure CH₃OH thus does not occur at this low electron energy. In CO/CH₃OH mixed ices, however, ^⦁CH₂OH radicals are formed by successive hydrogenation of CO (Reaction 1) which produces both types of radicals and thus enables the formation of methoxymethanol by recombination of CH₃O^⦁ and ^⦁CH₂OH radicals.

The DEA resonance of CH₃OH at E₀ = 5.5 eV is also observed in C₂H₄/CH₃OH mixed ices where it causes hydrogenation of C₂H₄ yielding ethane (C₂H₆) (Figure 13a, shaded in grey) [32]. Note that no significant amounts of C₂H₆ are formed after irradiation of pure CH₃OH (Section 4.1) or C₂H₄ at E₀ = 5.5 eV (Figure 13a) suggesting that formation of C₂H₆ mainly occurs by a reaction between CH₃OH and C₂H₄. Two reaction mechanisms contribute to the formation of C₂H₆ which are based on the two competitive DEA channels of CH₃OH yielding CH₃O⁻ and H^•, and CH₃O^• and H⁻ (Figure 4) [104,112]. Following the first of these two DEA channels, the released hydrogen radicals can add to C₂H₄ producing an intermediate C₂H₅^• radical. Subsequently, the C₂H₅^• radical can abstract a H^• from a nearby CH₃OH molecule to yield C₂H₆ and CH₃O^•. The proposed mechanism is supported by the concomitant appearance of butane (C₄H₁₀) (Figure 13b) which is formed when the intermediate C₂H₅^• radical reacts with a second C₂H₄ molecule to yield C₄H₉^• before it abstracts a H^• from a nearby CH₃OH molecule. The reactions of the CH₃O^• radical which can also act as a reducing agent and the consequent outcome of the second DEA channel are discussed in Section 4.4. The relevance of reduction by H^• has also been noted in the case of the electron-induced reactions of C₂H₄ with NH₃ where electron-induced dissociation of NH₃ releases H^• [29] (Section 4.1 and Section 4.5). We note that related reactions of H^• have also been observed in the cases of cisplatin, which is a complex carrying a NH₃ ligand [113], and of η³-allyl ruthenium tricarbonyl chloride in the presence of NH₃ [114] which underlines the wider perspective of the chemical concepts described herein.

4.4. Hydrogenation by Methoxy Radicals

Unsaturated molecules cannot only be reduced by free hydrogen radicals as discussed in Section 4.3, but also by CH₃O^• radicals. These are produced by DEA to CH₃OH around E₀ = 5.5 eV along with H^– (Figure 4), by ND yielding CH₃O^• and H^•, and by EI yielding a CH₃OH^•+ radical cation which can protonate a nearby CH₃OH molecule to form CH₃OH₂⁺ and CH₃O^•. The CH₃O^• radical can transfer a hydrogen radical to C₂H₄ which produces H₂CO and C₂H₅^⦁ (Reaction 2). This is supported by the increase of the H₂CO yield upon electron irradiation at E₀ = 5.5 eV when CH₃OH is mixed with C₂H₄ (see blue shaded signal in Figure 13a). Analogous to the reactions following addition of free hydrogen radicals to C₂H₄ (Section 4.3), the so formed C₂H₅^• radical can react with a CH₃OH molecule to yield C₂H₆ and regenerate the CH₃O^• radical (Reaction 3). In a concurrent process, the intermediate C₂H₅^⦁ radical can add to a second C₂H₄ molecule before it reacts with CH₃OH to finally yield butane (C₄H₁₀) (Reaction 4).

$${\mathrm{CH}}_3{\mathrm{O}}^{•}+{\mathrm{C}}_2{\mathrm{H}}_4\to {\mathrm{H}}_2\mathrm{CO}+{\mathrm{C}}_2{\mathrm{H}}_5^{•}$$

(2)

$${\mathrm{C}}_2{\mathrm{H}}_5^{•}+{\mathrm{CH}}_3\mathrm{OH}\to {\mathrm{C}}_2{\mathrm{H}}_6+{\mathrm{CH}}_3{\mathrm{O}}^{•}$$

(3)

$${\mathrm{C}}_2{\mathrm{H}}_5^{•}+{\mathrm{C}}_2{\mathrm{H}}_4+{\mathrm{CH}}_3\mathrm{OH}\to {\mathrm{C}}_4{\mathrm{H}}_9^{•}+{\mathrm{CH}}_3\mathrm{OH}\to {\mathrm{C}}_4{\mathrm{H}}_{10}+{\mathrm{CH}}_3{\mathrm{O}}^{•}$$

(4)

Note that the CH₃O^• radical is recovered at the end of Reactions 3 and 4 so that the net chemical equations simplify to:

$${\mathrm{CH}}_3\mathrm{OH}+{\mathrm{C}}_2{\mathrm{H}}_4\to {\mathrm{H}}_2\mathrm{CO}+{\mathrm{C}}_2{\mathrm{H}}_6$$

(5)

$${\mathrm{CH}}_3\mathrm{OH}+2{\mathrm{C}}_2{\mathrm{H}}_4\to {\mathrm{H}}_2\mathrm{CO}+{\mathrm{C}}_4{\mathrm{H}}_{10}$$

(6)

The overall contribution of hydrogen transfer from CH₃O^• radicals to the hydrogenation of C₂H₄ can be estimated by comparing the relative yields of H₂CO, C₂H₆, and butane (C₄H₁₀). Considering the stoichiometry of Reactions 5 and 6, the amount of H₂CO should be equal to the total amount of C₂H₆ and butane (C₄H₁₀). A reanalysis of the data obtained after irradiation at 5.5 eV with 8000 µC/cm² [32] revealed the yields of H₂CO, C₂H₆, and butane (C₄H₁₀) to be 1.5%, 2.6%, and 1.3% with respect to the amount of CH₃OH in a non-irradiated sample. Consequently, the total amount of C₂H₆ and butane was 2.6 times higher than the amount of H₂CO. Thus, hydrogen transfer from CH₃O^• to C₂H₄ accounts for about 40% of the formed C₂H₆ and butane (C₄H₁₀), whereas the addition of free H^• radicals to C₂H₄ (see Section 4.3) accounts for about 60% of these products. Reaction 2 has been studied in particular after DEA to CH₃OH at 5.5 eV [32] but may also play a role for the reduction of C₂H₄ following CH₃O^⦁ formation after ND or EI of CH₃OH. Recall that the amount of H₂CO after irradiation of pure CH₃OH at E₀ = 5.5 eV was significantly lower than in C₂H₄/CH₃OH mixed ices (Figure 13a) suggesting that unimolecular dissociation of CH₃O^⦁ into H₂CO and H^⦁ (Reaction 7) is not efficient under the experimental conditions of cryogenic ices.

$${\mathrm{CH}}_3{\mathrm{O}}^{•}\nrightarrow {\mathrm{H}}_2\mathrm{CO}+{\mathrm{H}}^{•}$$

(7)

Notably and in contrast to C₂H₄, CO is not reduced by CH₃O^⦁ radicals (Reaction 8) [26].

$${\mathrm{CH}}_3{\mathrm{O}}^{•}+\mathrm{CO}\nrightarrow {\mathrm{H}}_2\mathrm{CO}+{\mathrm{HCO}}^{•}$$

(8)

This has been inferred by comparing the energy dependence of H₂CO with that of methyl formate (CH₃OCHO). As the production of methyl formate relies on the addition of CH₃O^• radicals to CO (see Section 4.5), it can be used as a proxy for CH₃O^• formation. Notably, only the energy dependence of methyl formate but not that of H₂CO shows a resonance at 10 eV (Figure 14). At this energy, CH₃O^• is produced through an ion-molecule reaction of CH₃OH with O^•− that is released by DEA to CO [115]. The lack of resonant enhancement in the yield of H₂CO suggests that hydrogen transfer from CH₃O^• to CO is not efficient.

The different reactivity of CH₃O^⦁ with regard to hydrogen transfer to CO and C₂H₄ can be explained by considering the energy profiles of these reactions. To this end, quantum chemical calculations were performed (Figure 15). The calculated reaction energies and barriers for the decomposition of CH₃O^⦁ into H₂CO and H^⦁, and the hydrogen transfer from CH₃O^⦁ to CO to yield H₂CO and HCO^⦁ agree well with published and validated data (Figure 15a,b) [116,117]. The values of our present calculation for the hydrogen transfer from CH₃O^⦁ to C₂H₄, which have not been reported so far, should thus have a comparable accuracy as the results for the other two cases (Figure 15c).

The computations reveal that the unimolecular decomposition of CH₃O^⦁ into H₂CO and H^⦁ is endothermic by 0.89 eV and has a reaction barrier of 1.07 eV (Figure 15a). The presence of CO, which acts as a hydrogen acceptor, significantly lowers the reaction energy although the reaction is still endothermic by 0.28 eV. The reaction barrier, however, is not significantly lowered compared to dissociation of CH₃O^⦁ into H₂CO and H^⦁ (Figure 15b). In the case of CH₃O^⦁ and C₂H₄, the reaction is exothermic by −0.64 eV, and the reaction barrier is lowered to a value of 0.76 eV (Figure 15c). Thus, the trend seen in the calculated reaction barriers agrees with the experimental observation that only hydrogen transfer from CH₃O^• to C₂H₄ occurs. Note, however, that even in the case of a reaction with C₂H₄ the CH₃O^• radical must have sufficiently high excess energy to overcome the reaction barrier of 0.76 eV. The thermodynamic threshold for the formation of CH₃O^• and H⁻ is 3.75 eV [112]. Under the assumption that the excess energy is only released as kinetic energy of the fragments, the kinetic energy of the CH₃O^• fragment can be calculated by [103] (1–31/32) × (E₀ –3.75 eV) which yields only 0.055 eV at E₀ = 5.5 eV, if vibrational excitation is neglected. This kinetic energy of CH₃O^• is not sufficient to overcome the reaction barrier suggesting that a considerable amount of the released excess energy must be stored in the rovibrational modes of CH₃O^•. Hydrogen transfer from CH₃O^• to C₂H₄ is thus relevant to the electron-induced chemistry in molecular ices under the condition that CH₃O^• is formed with sufficient excess energy to overcome the reaction barrier.

4.5. Addition Reactions Involving Radical Cations or Neutral Radicals

The formation of larger products from small ice components is most conceivable if the reactants do not dissociate upon interaction with the incident electron. This is the case when the electron energy is above the ionization threshold of a reactant but the excess energy is not sufficient to overcome any bond dissociation energy. The resulting intact radical cation of a reactant can then react with an adjacent molecule in the ice. If the two reactant molecules combine in an addition reaction to yield a larger product which contains all atoms of the reactants, the reaction is called atom-efficient.

An atom-efficient electron-induced reaction has first been identified in binary ice layers consisting of C₂H₄ and NH₃. Formation of ethylamine (C₂H₅NH₂) [27,29], i.e., an electron-induced hydroamination, was observed for electron energies above the ionization thresholds of C₂H₄ and NH₃ [27]. However, the electron energies were still below the onset of DI at least for NH₃ and not much above the threshold for C₂H₄ [54]. The proposed mechanism is shown in Scheme 2. Ionization of either C₂H₄ or NH₃ leads to an attractive Coulomb force between the reactants which allows them to circumvent the reaction barrier that prevents a spontaneous reaction between the neutral molecules [27]. Reaction between C₂H₄ and NH₃ after EI yields a cationic adduct (structure I in Scheme 2) as an intermediate. Intramolecular rearrangement of I into the ethylamine radical cation (structure II in Scheme 2) and subsequent neutralization by a thermalized electron eventually yields ethylamine. A theoretical study confirmed that bond formation between the reactants in the cation state is in fact barrierless in contrast to the neutral ground state where an activation energy of more than 2.2 eV must be overcome [120]. The mechanism in Scheme 2 has also been adopted to explain the formation of ethanol from C₂H₄/H₂O mixed ices [28], the formation of formamide from CO/NH₃ mixed ices [30], and the formation of ethyl methyl ether in C₂H₄/CH₃OH mixed ices [32].

The hydroamination mechanism presented in Scheme 2 has been examined more comprehensively in a subsequent study [29] by investigating the dependence of product yield on electron energy and sample composition, as well as the formation of side products. Experiments were also performed using NH₃ and propene (CH₂=CHCH₃), ethylamine and C₂H₄, and diethylamine ((C₂H₅)₂NH) and C₂H₄ in order to probe possible steric effects. Hydroamination products were observed in all cases and the product yields exhibited a threshold-type behavior with onset around 8 eV (Figure 16). This is lower than the gas phase ionization energies of most of the investigated reactants (10.1 eV for NH₃, 10.5 eV for C₂H₄, 9.7 eV for propene, values from 8.7 eV to 9.5 eV for ethylamine, and 7.9 eV to 8.6 eV for diethylamine [54]). The lower onset of product formation as compared to the gas phase ionization energies can be ascribed to the effect of the condensed phase (Section 3.5). This is illustrated by calculations performed on the B3LYP/6-311 + G(d) level of theory (Figure 17). In the first step, an isolated NH₃ molecule was ionized resulting in a vertical ionization energy of 10.8 eV, which is in reasonable agreement with the experimental adiabatic gas phase value (Table 2). The NH₃^⦁+ was then placed near a C₂H₄ molecule, embedded in a cluster of further six NH₃ molecules, and allowed to relax (MIN0 in Figure 17) resulting in a vertical ionization energy of 7.3 eV (MIN0^⦁+ in Figure 17) which coincides roughly with the observed onset of ethylamine formation (Figure 16). As discussed in Section 3.5, further relaxation of this structure proceeds via proton transfer from the NH₃^⦁+ to an adjacent NH₃ molecule producing NH₄⁺ and a NH₂^⦁ radical and results in a major energy gain of 5.9 eV as compared to an isolated NH₃^⦁+ (MIN1^⦁+ in Figure 17). From the relaxed state MIN1^⦁+, a shallow barrier with a height of only 0.3 eV where the NH₂^⦁ radical approaches the second reactant C₂H₄ (TS1^⦁+ in Figure 17) leads to another minimum (MIN2^⦁+ in Figure 17). Here, a C-N bond has formed and another neutral NH₃ interacts via one of its H atoms with the other C atom. Adding an electron to this cluster to simulate recombination with a thermalized electron in the condensed layer (MIN2 in Figure 17) then leads to bond flipping through the cluster which recovers the neutral NH₃ molecules and forms a new C-H bond relaxing to the final product ethylamine (MIN3 in Figure 17). The previous theoretical study, that included only one NH₃, placed that transition state on the cation potential energy surface only 1.38 eV below the isolated reactants NH₃^⦁+ and C₂H₄ [120]. The presence of more NH₃ molecules at the reaction site stabilizes the transition state and thus helps to drive the reaction towards the product.

Considering the wealth of different electron-molecule interaction that yield reactive species (see Section 3), it is not surprising that other products were also observed. For example, ethane (C₂H₆) and butane (C₄H₁₀) were identified together with ethylamine in C₂H₄/NH₃ mixed ices [29]. Although these products were also formed upon electron irradiation of pure C₂H₄, the product yields were considerably enhanced in the presence of NH₃ suggesting that NH₃ acts as a reducing agent. These reactions are governed by radical formation and recombination (Section 4.1), and are further driven by the action of free hydrogen radicals (Section 4.3).

It is also conceivable that the addition of a neutral radical to C₂H₄ is an alternative to the EI-driven hydroamination in the case of C₂H₄/NH₃ mixed ices. Here, NH₂^⦁ radicals, that can for instance be formed by ND (Section 3.3), would add to the double bond of C₂H₄. This would yield an intermediate H₂N–C₂H₄^⦁ radical which may then abstract a H^• from a nearby NH₃ to convert to ethylamine (Scheme 3). This type of reaction is also closely related to the mechanism presented in Figure 17 where the NH₂^⦁ radical results from hydrogen transfer from a neutral NH₃ to an ionized NH₃^⦁+ radical cation. Whether ND contributes to the yield of ethylamine can likely be unraveled by photon irradiation experiments on C₂H₄/NH₃ mixed ices analogous to those reported previously for pure ND₃ ice [122].

A similar reaction could occur whenever a radical intermediate is formed in a mixed ice that contains a species with multiple bonds, such as CO or C₂H₄. As discussed in Section 4.1, radical intermediates are particularly prominent when CH₃OH is present as reactant in an ice layer but can also be formed from H₂O (see Section 3). Therefore, a reaction mechanism involving radical addition is proposed for the formation of formic acid from CO and H₂O, and methyl formate and ethyl methyl ether from C₂H₄ or CO and CH₃OH, respectively. It is based on addition of a RO^• radical (R = H, CH₃) to CO or C₂H₄, respectively (Scheme 4). The energy dependences of product formation can then be traced back to the reaction channels that yield a RO^• radical (R = H, CH₃).

The energy dependences of methyl formate in CO/CH₃OH mixed ices [26] and formic acid in CO/H₂O mixed ices [24] reveal resonant product formation at E₀ = 10 eV (Figure 18). Assignment of this resonance to a specific DEA process in a particular molecule is not directly possible because CO, H₂O, C₂H₄ and CH₃OH do all evince DEA resonances at around 10–11 eV in both gas phase (Figure 4) [56,112] and condensed phase [105,123,124,125,126,127] experiments. No resonances are observed at E₀ = 10 eV for the formation of ethyl methyl ether in C₂H₄/CH₃OH mixed ices (Figure 18b) and for the formation of ethanol in C₂H₄/H₂O mixed ices (Figure 18a), suggesting that the 10 eV resonance is due to DEA to CO yielding C and O^⦁−. The latter can react with a nearby H₂O or CH₃OH molecule to yield HO^• or CH₃O^•, respectively (Reactions 9 and 10) [115].

$${\mathrm{O}}^{•-}+{\mathrm{H}}_2\mathrm{O}\to {\mathrm{HO}}^{-}+{\mathrm{HO}}^{•}$$

(9)

$${\mathrm{O}}^{•-}+{\mathrm{CH}}_3\mathrm{OH}\to {\mathrm{HO}}^{-}+{\mathrm{CH}}_3{\mathrm{O}}^{•}$$

(10)

DEA to CO thus clearly dominates over DEA to H₂O and CH₃OH at 10 eV, at least in the mixed ices investigated in the experiments summarized herein. This is in contrast to the gas phase where the cross sections for the formation of O^•− by DEA at 10 eV are 2 × 10⁻¹⁹ cm² for CO [45], 5.76 × 10⁻¹⁹ cm² for H₂O [60], and 4.5 × 10⁻²⁰ cm² for CH₃OH respectively [128].This demonstrates that the relative DEA cross sections in the mixed ices differ significantly from those in the gas phase.

In the case of CO/CH₃OH and C₂H₄/CH₃OH mixed ices, a second resonant process was identified when exposures of 12,000 and 8000 µC/cm², respectively, were applied (Figure 19). Here, DEA to CH₃OH at E₀ = 5.5 eV produces CH₃O^• and H⁻ (Figure 4) which leads to the production of methyl formate and ethyl methyl ether, respectively. The complementary DEA channel that yields CH₃O⁻ and H^• (Figure 4) also contributes to the production of methyl formate by hydrogenation of CO to CH₃O^• (Reaction 1, Section 4.3). DEA to C₂H₄ cannot be responsible for the observed resonance as the known DEA channels in the gas phase are located at considerably higher energies (Figure 4) [58,59,129], while the EA resonance of C₂H₄ which contributes to the formation of ethanol in C₂H₄/CH₃OH mixed ices occurs at lower energy (Figure 18a) (see Section 4.2).

In addition to the 10 eV and 5.5 eV resonances, there are non-resonant contributions to the formation of formic acid, methyl formate, and ethyl methyl ether, which are due to ND and/or EI of H₂O and CH₃OH. In the case of CH₃OH containing ices, EI clearly dominates over ND as can be seen by the onset for the formation of ethyl methyl ether at ~10 eV (Figure 18b). This value agrees nicely with the ionization threshold of CH₃OH which is at 10.84 eV in the gas phase [54], and at 9.8 eV in the condensed phase [64]. In the case of methyl formate and formic acid, the non-resonant contributions to product formation cannot be unambiguously identified to be dominated by ND or EI as the onset of this second formation channel is masked by the DEA resonance at 10 eV (Figure 18). However, both ND and EI of H₂O and CH₃OH can produce RO^• (R = H, CH₃) radicals and thus contribute to some extent to the formation of formic acid and methyl formate.

After EI of H₂O or CH₃OH, the respective radical cation can transfer a proton to a nearby H₂O or CH₃OH molecule to yield a HO^• or a CH₃O^• radical, respectively, along with a protonated H₂O or CH₃OH molecule (Reactions 11 and 12).

$${\mathrm{H}}_2{\mathrm{O}}^{•+}+{\mathrm{H}}_2\mathrm{O}\to {\mathrm{HO}}^{•}+{\mathrm{H}}_3{\mathrm{O}}^{+}$$

(11)

$${\mathrm{CH}}_3{\mathrm{OH}}^{•+}+{\mathrm{CH}}_3\mathrm{OH}\to {\mathrm{CH}}_3{\mathrm{O}}^{•}+{\mathrm{CH}}_3{\mathrm{OH}}_2^{+}$$

(12)

Furthermore, ND of H₂O or CH₃OH yields HO^• or CH₃O^•, and a H^• radical (Reactions 13 and 14) once the electron energy is higher than the electronic excitation energy.

$${\mathrm{H}}_2\mathrm{O}\to {\mathrm{HO}}^{•}+{\mathrm{H}}^{•}$$

(13)

$${\mathrm{CH}}_3\mathrm{OH}\to {\mathrm{CH}}_3{\mathrm{O}}^{•}+{\mathrm{H}}^{•}$$

(14)

In addition, DEA to CH₃OH yields CH₃O^• and H⁻ (Figure 4). Note that in CO/CH₃OH mixed ices, additional CH₃O^• radicals are also produced by hydrogenation of CO (Reaction 1, Section 4.3) following DEA of CH₃OH into CH₃O⁻ and H^• (Figure 4) and by ND of CH₃OH.

Regardless of how the HO^• or CH₃O^• radicals are formed and at which electron energy (Reactions 9–14), they can subsequently react with CO to yield an intermediate HOCO^• or CH₃OCO^• radical which can then abstract a H^• from a nearby H₂O or CH₃OH molecule to yield formic acid or methyl formate, respectively (Scheme 4). Similarly, they can add to C₂H₄ to yield an intermediate CH₃OC₂H₄^• or HOC₂H₄^• radical which can abstract a H^• from a nearby CH₃OH or H₂O molecule to yield ethyl methyl ether or ethanol (Scheme 4). At the end of the reaction, the RO^• radical is recovered, potentially inducing a chain reaction.

4.6. Oxidation of CO to CO₂

In contrast to the reduction by transfer of hydrogen as discussed in Section 4.2, Section 4.3 and Section 4.4, electron irradiation can also initiate oxidation reactions. For instance, CO is readily oxidized to CO₂ upon electron irradiation of CO/H₂O [24] and CO/CH₃OH [26] mixed ices. This CO₂ formation has usually been attributed to either of two mechanisms: Formation and decay of an intermediate ROCO^⦁ radical (R = H, CH₃), or reaction of an oxygen atom with CO [107,108,130,131,132]. Considering the first mechanism, reaction of HO^⦁ or CH₃O^⦁ with CO yields the intermediate HOCO^⦁ or CH₃OCO^⦁ radicals, respectively. These intermediates can either be stabilized by the condensed phase which favors their subsequent conversion to formic acid (HCOOH) or methyl formate (CH₃OCHO), respectively, or they can decay into CO₂, and H^⦁ or CH₃^⦁.

Results on electron-induced reactions in mixed CO/H₂O mixed ices [24] suggest that HOCO^⦁ is indeed sufficiently stabilized by the H₂O matrix to enable its further reaction to formic acid (see Section 4.5). In contrast, decay of HOCO^⦁ does not play a significant role for the formation of CO₂. This has been inferred by comparing the energy dependences of CO₂ with those of H₂CO and formic acid (Figure 20). As discussed in Section 4.2, H₂CO can only be formed via an intermediate HCO^• radical. This process is enhanced around 4 eV (Figure 20), where formic acid is not observed. Therefore, the resonant formation of formic acid at 10 eV cannot be ascribed to HCO^• but must proceed via the HOCO^• radical. Conversely, it can be excluded that a significant amount of CO₂ is formed by decay of HOCO^• as the resonance at 10 eV is not reflected in the energy dependence of CO₂ (Figure 20). It was thus proposed that CO₂ is mainly formed by the competitive ND channel of H₂O into H₂ and O and subsequent reaction of the latter with CO [24].

In contrast to the result for HOCO^•, the CH₃OCO^• radical which is formed upon irradiation of CO/CH₃OH mixed ices [26] readily dissociates into CO₂ and CH₃^•. This has been inferred from the formation of dimethyl ether (CH₃OCH₃) and methyl acetate (CH₃C(O)OCH₃) which are specific side products at E₀ = 5.5 eV formed by recombination of the CH₃^• radical with CH₃O^• and CH₃OCO^•, respectively (Figure 21). At this energy, CH₃O^• is produced by DEA to CH₃OH (Section 3.4) and CH₃OCO^• results from addition of the latter to CO (see also Section 4.5). Notably, dimethyl ether is not formed upon irradiation of pure CH₃OH at E₀ = 5.5 eV (Figure 21b) where CH₃O^• is resonantly produced together with H^- (Figure 4) so that direct production of CH₃^• radicals by dissociation of CH₃OH can be ruled out at this low energy [26,31]. Equally, methyl acetate cannot be formed in pure CH₃OH (Figure 21a) because this requires not only CH₃^• radicals but also CO which is known to be produced by electron irradiation of CH₃OH only at higher energies [48]. ND of CH₃OH into CH₄ and O may also contribute to CO₂ formation to some extent as has been suggested previously but not directly observed [130,133].

Studies applying molecular mechanics suggest that in water ice, HOCO^• loses all of its excess energy within a few picoseconds [134] which makes it impossible for HOCO^• to overcome the barriers toward CO₂ and H^•. Thus, such a scenario provides a reasonable explanation why CO₂ is not formed by dissociation of HOCO^•. Instead, CO₂ formation progresses by ND of H₂O into H₂ and O and subsequent addition of the latter to CO. In the case of CH₃OCO^•, stabilization in a CH₃OH matrix is much less pronounced as the capacity to form hydrogen-bonds in CH₃OH is lower compared to H₂O, where molecules can act as donors of two as compared to one hydrogen bond in CH₃OH. Furthermore, CH₃OCO^• can only act as an acceptor but not as a donor of hydrogen bonds. We propose that this reduced stabilization of the CH₃OCO^• radical enables it to readily dissociate and form CO₂, which HOCO^• cannot do.

5. Open Questions and Perspectives for Future Studies

As a conclusion to this review of our studies on electron-induced chemistry in ices, we discuss some cases where the reaction mechanisms remain to be resolved. This includes (a) the resonant formation of formamide from CO/NH₃ mixed ices [30], (b) the formation of isocyanic acid (HNCO) from the same mixed ices, and (c) methylation reactions that start from the neutral dissociation of methane.

In analogy to the hydroamination reaction, formation of formamide in CO/NH₃ mixed ices at electron energies above the ionization threshold was ascribed to an EI-driven reaction (see Section 4.5). However, the yield of formamide in CO/NH₃ mixed ices is resonantly enhanced around 9–10 eV [30]. This reaction channel has been assigned to DEA to NH₃ yielding H⁻ and NH₂^⦁, or NH₂⁻ and H^⦁ [135]. Following DEA, either NH₂^⦁ or H^⦁ can attack a nearby CO molecule to yield H₂NCO^⦁ or HCO^⦁. Finally, reaction with NH₃ yields formamide and either a NH₂^⦁ or H^⦁ radical, which potentially results in a chain reaction. However, a closer comparison of the formation of ethylamine in C₂H₄/NH₃ mixed ices [29] and of formamide in CO/NH₃ mixed ices [30] reveals that the proposed mechanisms may not be fully consistent, calling for a more comprehensive reevaluation. In fact, it could be expected that DEA to NH₃ at 9–10 eV should also contribute to the formation of ethylamine and/or C₂H₆. However, in contrast to the energy dependence of formamide [30], the energy dependences of ethylamine and C₂H₆ do not evince a resonance at 9–10 eV [29]. There might be several reasons for this discrepancy. Firstly, the resonance may have been overseen in the data for ethylamine and C₂H₆ production as the electron exposure of 800 µC/cm² used to study their formation was already above the regime where product formation depends linearly on the electron dose for E₀ > 10 eV [29]. Secondly, ionization of C₂H₄ may already start to contribute to the formation of ethylamine at 10 eV if it is considered that the ionization energy of C₂H₄, which is about 10.5 eV in the gas phase [54], is very likely shifted to lower values in the condensed phase [44,136]. In contrast, the ionization energy of CO is 14.0 eV in the gas phase [54] and thus at considerably higher energies. Thus, resonant contributions to the overall product yields in the C₂H₄/NH₃ mixed ices might be masked by higher contributions of the EI-driven mechanism.

As an alternative mechanism, the resonance observed in the formamide yield from CO/NH₃ mixed ices may also stem from DEA to CO rather than NH₃ [56,123,137]. Dissociation of CO into C and O^⦁– could provide an alternative route to the formation of NH₂^⦁ via

$${\mathrm{O}}^{•-}+{\mathrm{NH}}_3\to {\mathrm{HO}}^{-}+{{\mathrm{NH}}_2}^{•}$$

(15)

Similar to Reaction 9, but with NH₃ instead of H₂O. From here, formation of formamide would again proceed by reaction of NH₂^⦁ with CO. However, all other products that form from NH₂^⦁ radicals should be resonantly enhanced when CO is present, but in CO/NH₃ mixed ices, no such enhancement in the production of hydrazine (N₂H₄) is observed.

Non-dissociative EA to CO could lead to formamide via the reaction of the base CO^⦁- with an adjacent NH₃ to yield HCO^⦁ and NH₂⁻. HCO^⦁ could then react with intact NH₃, or possibly with the NH₂⁻ to eventually form formamide. However, so far, non-dissociative EA to CO at electron energies around 10 eV has not been reported, neither in gas phase nor in the condensed phase. A more conceivable scenario is that EI to NH₃ creates a secondary electron with almost no kinetic energy, which could then attach to CO, triggering the low-lying non-dissociative resonance. At the same time, the NH₃^⦁+ radical cation could react with another intact NH₃ to yield NH₄⁺ and NH₂^⦁, as described in Section 3.5. This would provide both, a good acid to protonate CO^⦁− and an NH₂^⦁ to complete the reaction to formamide. In conclusion, the interpretation of the experimental data with respect to the mechanism by which formamide is produced is still ambiguous which calls for further investigation.

The interpretation of the mechanism for formation of formamide from CO and NH₃ is further confounded by the fact that formation of formamide is always accompanied by the formation of isocyanic acid (HNCO). Our preliminary experimental data show a very consistent ratio of about 2:1 moieties of HNCO to formamide (see Figure 1). Observation of extra-terrestrial occurrences of formamide have also always co-detected isocyanic acid [138], though ratios between the two vary. There is, to date, no good explanation for the formation of HNCO. We can only exclude that it is formed by radiolysis of formamide. Any substantial progress in the mechanistic study of the formation of formamide could thus be very helpful in understanding the origin of isocyanic acid as well.

The final suggestion for further study is a whole class of reactions, namely the methylation of compounds starting from ND of methane. Figure 7 shows the formation of ethane from methane, which could be considered the simplest methylation reaction, as it is the recombination of two methyl radicals. The energetic onset at around 10 eV hints at an ND process, rather than impact ionization, as the ionization threshold of methane in the gas phase lies at 12.6 eV and neutral dissociation of methane is known to have higher cross sections than EI for energies up to 30 eV, at least in the gas phase [139,140]. In the irradiation of CO/CH₄ mixed ices, our preliminary results show that acetaldehyde (CH₃CHO) is ormed. The dependence on electron energy of its formation is fairly similar to the formation of ethane from methane, which could be interpreted that acetaldehyde is formed by the methylation of CO with subsequent addition of a hydrogen radical. The methylation of small molecules is certainly a class of reactions that needs to be investigated in more detail.

These examples show that there is indeed plenty of room for future studies that aim at unravelling the mechanisms of electron-driven reactions in molecular ices. The dimension of this problem will certainly increase when more than two compounds are presents in the ice. However, the insight obtained so far demonstrates that the methodology presented in this review is a valuable approach to tackle these problems.