Electron-induced reactions in condensed films of acetonitrile and ethane

Abstract

Reactions in pure and mixed films of C₂H₆ and CD₃CN deposited on a Au surface at 35 K have been induced by low-energy electrons and investigated by Thermal Desorption Spectrometry (TDS). The incident electron energy (E₀) was varied between 5 and 16 eV and a number of different products were identified. Beside the main products, CD₄, CD₃H, and C₂D₆, molecules resulting from atom scrambling during radical chain reactions (C₂H₅D) and recombination products (CD₃CD₂CN and C₂H₅CD₃) were identified while others were characteristically absent. The quantity of the different products varied with E₀. The observed electron-driven processes are in accord with previous findings from gas phase experiments on dissociative electron attachment and electron impact ionization. On this basis, reaction mechanisms leading to the formation of the observed products are suggested for different ranges of E₀.

---

1. Introduction

Functionalized surfaces are important because they allow to tailor the properties of a material or provide chemical groups to which more complex molecular architectures can be attached. This calls for the development of processes to modify the chemical composition of surfaces in a controlled way. It has been shown that exposure to low-energy electrons can be used to functionalize semiconductor surfaces.^1,2 This approach is motivated by the finding that specific chemical reactions can be initiated by proper tuning of the energy of electrons incident onto a molecule or surface. A very fine example is provided by recent in depth investigations of electron-induced reactions in gaseous thymine.³ Therefore, it should be possible to achieve a better control of surface modifications than in the case of high-energy radiation such as X-rays or keV-electrons, which abundantly produce secondary electrons with energy below 20 eV but with a wide energy distribution.⁴

Processes for surface modification are not only of interest for inorganic but also for organic materials. While simple hydrocarbon materials like polymers and alkanethiols or alkanesilanes for the formation of self-assembled monolayers are readily available, more complex molecules, which carry the desired functional groups already before being deposited or self-assembled on a surface, usually require a sophisticated synthetic strategy and are therefore costly. Thus, it is desirable to establish processes that allow to couple functional groups to hydrocarbon materials after their deposition on a surface.

This work pursues the idea to functionalize a hydrocarbon by producing, through an electron-induced reaction, species that would both activate the hydrocarbon and bind to the resulting radical site. While examples of intermolecular reactions have been studied both in gaseous clusters⁵ and condensed samples,^6,7 attachment of specific molecular groups to a hydrocarbon could not be demonstrated so far. As a first effort to demonstrate functionalization of hydrocarbons, an investigation of the reactions induced by low-energy electrons in pure and mixed films of ethane (C₂H₆) and deuterated acetonitrile (CD₃CN) and monitored by use of Thermal Desorption Spectrometry (TDS) is presented here. This system served as a model for materials consisting of a hydrocarbon surface (represented by C₂H₆) and a reactive agent (CD₃CN) that is expected to functionalize the hydrocarbon material upon electron exposure. CD₃CN was chosen as example because electron-induced processes in the non-deuterated molecule have been well characterized in the gas phase.^8–10 The deuteration was not expected to fundamentally modify these processes but helped to unravel the reaction mechanisms. Multilayer films of CD₃CN and C₂H₆ condensed onto a Au surface cooled to 35 K in ultra-high vacuum have been exposed to low-energy electrons. Desorbing products were detected by mass spectrometry in the TDS experiments. The results of these experiments show that intermolecular reactions are, in fact, initiated. The underlying mechanism are investigated by variation of the incident electron energy.

2. Experiment

The experiments were performed in an UltraHigh Vacuum (UHV) chamber pumped by a turbomolecular pump to a pressure of about 10⁻¹⁰ Torr, with the residual gas consisting mainly of hydrogen. If necessary, additional pumping can be provided by a titanium sublimator. Low-energy electron-induced reactions in both mixed (with a ratio of 1 : 1) and pure films of C₂H₆ and CD₃CN condensed onto a polycrystalline Au substrate were monitored by TDS using a Quadrupole Mass Spectrometer (QMS) residual gas analyser (Stanford, 200 amu). The substrate is held at 35 K by means of a closed-cycle He cryostat. The temperature was measured using a thermocouple type E. A commercial fload gun delivering currents of the order of a few μA cm⁻² as measured at the sample was used for electron exposure. The energy (E₀) of the non-monochromatized electron beam with an estimated resolution of the order of 0.5–1 eV was varied between 5 and 16 eV. Typical exposure times yielding 5000 μC ranged from 5 min at 15 eV to 1 h at 5 eV. At the relatively high incident current densities used here, charging of the irradiated molecular films is likely. While this modifies the exact value of E₀, we have at present no established procedure to determine exactly the amount of this effect. This is due to the divergence of the non-focussed electron beam that increases upon decrease of E₀. This prohibits the measurements of the current onset that could be used to calibrate E₀. Therefore, all stated E₀ are merely nominal energies as deduced from the difference in potential between the filament of the electron gun and the substrate.

For each experiment, thin films of C₂H₆ (Messer-Griesheim, purity 99.95%) and/or CD₃CN (Acros Organics, NMR-grade, nominal purity 99 atom % D) were deposited onto the Au substrate by introducing gases or vapours via a gas-handling manifold. The manifold consists of precision leak valves and a small calibrated volume where the absolute pressure is measured with a capacitance manometer. For each film deposition, a calibrated amount of gas or vapour, measured as a pressure drop, is leaked via a stainless steel capillary whose end is located just in front of the substrate. Prior to each deposition, the substrate is cleaned by resistive heating to 270 K. The thickness of the films was estimated from TDS measurements performed on pure and non-exposed CH₃CN (Riedel-de Haen, 99.9%) for increasing amounts of deposited vapour as described in Section 3.1.

After an irradiation time corresponding to a specific electron exposure, stated as an accumulated charge, thermal desorption was induced by increasing the substrate temperature from 35 K up to 250 K at a rate of 1 K s⁻¹ (exposure experiment). This was achieved by resistive heating of two thin Ta ribbons spotwelded to the thicker Au substrate. The temperature was controlled using a standard power supply, which is computer-controlled by software developed in our lab. In each experiment, TDS curves were recorded for four different molecular masses, among which in all cases the masses of the parent positive ions of the investigated molecules formed after 70 eV electron impact at the entrance of the QMS. The other chosen signals were characteristic of specific product molecules and also correspond to the parent positive ions. As reference, a TDS measurement was performed using a non-exposed film of the same composition and thickness prior to each exposure experiment (denoted as 0 μC). TDS measurements at 44 amu (parent ion of CD₃CN, see section 3.2) and 43 amu (parent ion of CD₂HCN, not shown) recorded on pure and non-exposed films of CD₃CN both revealed a single desorption peak at 140 K with very similar shape and an intensity ratio of 100 : 5. Assuming CD₃CN and CD₂HCN to have the same ionization efficiency, this suggests that some isotope exchange has occurred prior to the experiments. This needs to be taken into account in the discussion of the results.

3. Results and discussion

3.1. Thickness calibration

TDS measurements of films of CH₃CN produced from increasing amounts of gas were performed to calibrate the monolayer coverage. The obtained asymmetric desorption peaks were fitted using two Gaussians with maxima at 135 K and 148 K. An example is shown in Fig. 1a. The integral signal corresponding to the monolayer should increase linearly with the amount of leaked vapour, up to a point where the monolayer is complete, and remain constant at higher doses. This behaviour is observed for the higher temperature desorption signal (Fig. 1c). In contrast, the lower temperature peak only starts to grow upon completion of the monolayer and increases linearly with the dose thereafter as a characteristic of the formation of a multilayer. A dose of 0.14 ± 0.03 mTorr is required for monolayer formation as deduced from linear least square fits, depicted as a dashed and a compact line in Fig. 1b and c, to TDS curves from measurements for amounts of leaked vapor ranging from 0.02 to 1.2 mTorr. From this calibration, the thickness of the CD₃CN films as used in most of the electron exposure experiments was estimated to be of the order of 15 monolayers (MLs). Some experiments aiming at larger product molecules were performed at a thickness of 30 ML to achieve a larger signal. At these thicknesses, effects from the Au surface during electron exposure can be largely excluded. The metal may nonetheless contribute to the thermal desorption signal, as will be discussed later. The thickness of the mixed layers is estimated to be of similar magnitude due to the roughly comparable size of CD₃CN and C₂H₆.

Fig. 1 (a) TDS measurement at 41 amu for a film of CH₃CN produced by depositing an amount of vapor corresponding to a pressure drop of 0.4 mTorr in the manifold. The desorption signal is fitted by two Gaussian with maxima at 148 K and 135 K, ascribed to the monolayer and the multilayer signal. Integrated multilayer (b) and monolayer (c) signal as a function of the amount of vapor leaked into the chamber.

3.2. Reactions in pure CD₃CN films

Each specific molecule can be identified in a TDS curve corresponding to a characteristic molecular mass by the temperature at which desorption from the surface occurs. Under the conditions of the present experiments, the partial pressure of CD₃CN typically starts to increase at 120 K and peaks at or near 140 K. Assuming that overlap with signals due to other molecules with fragments of identical mass does not occur, the intensity of the desorption peak is proportional to the amount of its neutral precursor molecule on the surface. The magnitude of this signal thus reveals the variation of the amount of CD₃CN on the substrate under electron exposure.

Fig. 2a shows representative TDS curves at 44 amu (parent ion of CD₃CN) without exposure and after exposure at three different E₀. Fig. 2b summarizes the peak intensities of the 44 amu desorption signal after electron exposure of 5000 μC, normalized to the same signal in the corresponding reference experiment, i.e., prior to exposure. The results indicate that the rate of loss of CD₃CN increases with increasing E₀, to become noticeable around 10 eV and to reach an order of 50–60% after exposure to 5000 μC at and above 14 eV. Assuming first order kinetics, a cross section σ for the reaction can be calculated from

where n and n₀ are the maxima of the desorption signals, the product of incident current and exposure time, I₀·t, is given by the electron exposure, and S₀ is the irradiated area. We note that this is just a crude estimate because only two data points are used at each energy. From the electron exposure of 5000 μC, the corresponding drop of the 44 amu signal (Fig. 2b) and the substrate surface of 2.8 cm², we calculate cross sections for the loss of CD₃CN as shown in Fig. 2c.

Fig. 2 (a) TDS measurements at 44 amu for 15-layer films of CD₃CN before and after electron exposure of 5000 μC at the stated incident energies. The magnification is the same for all curves. (b) Peak intensities of the desorption signal after exposure, each normalized to the same intensity for a non-exposed film recorded immediately before. (c) Depletion cross sections for CD₃CN under electron exposure estimated from the data in (b).

The mass spectra of CD₃CN also show a small signal at 28 amu corresponding to a fragment ion DCN⁺. The intensity of this signal amounts to about 6% of the 44 amu peak. Fig. 3a shows representative TDS measurements for 28 amu, while the intensity of this signal relative to the 44 amu data from the same experiment (Fig. 2) is plotted in Fig. 3b. Above a threshold of roughly 10 eV, the relative intensity of the 28 amu signal is found to increase. At the same time the desorption peak broadens and, at 15 eV, the maximum shifts to 130 K. This gives evidence of the production of another species with a signal at 28 amu that desorbs at a similar temperature as CD₃CN. As a first guess, the literature was searched for data on DCN. CD₃CN has a reported vaporization enthalpy in the 30–35 kJ mol⁻¹ range, while 36–38 kJ mol⁻¹ are reported for DCN.¹¹ In a pure film, DCN should thus desorb at a higher temperature than CD₃CN. If DCN is only present as a small amount in CD₃CN, hydrogen bonding must be less pronounced. Thus, we suggest that the 28 amu signal actually gives evidence of the production of a certain amount of DCN under electron exposure starting around E₀ = 10 eV.

Fig. 3 (a) TDS measurements at 28 amu for 15-layer films of CD₃CN before and after electron exposure of 5000 μC at the stated incident energies. The magnification is the same for all curves. (b) Peak intensity of the desorption signal at 28 amu before and after exposure, each normalized to the intensity at 44 amu in the same experiment, as function of incident electron energy.

Production of CD₄ in CD₃CN exposed to electrons is deduced from a desorption peak around 55–60 K in the TDS scan for 20 amu, as shown in Fig. 4 for representative E₀. This signal remains small through the 6–9 eV range, with a somewhat higher intensity at 6 and 7 eV, and starts to increase around 10 eV to reach a new constant level at 12 eV. The observation of CD₄ suggests that cleavage of the C–C bond in CD₃CN occurs and the resulting CD₃ species reacts with hydrogen from its environment. In order to trace the origin of this hydrogen, the signal at 19 amu corresponding to CD₃H was also recorded. The fact that this signal was very small or even absent at all E₀ shows that CD₃CN, and not H₂ from the residual gas of the vacuum chamber, is the source of hydrogen. It must be noted that both the 19 amu and the 20 amu signal yield a flat baseline in most experiments when the TDS measurement is performed without electron exposure. A signal above 150 K, appearing upon electron exposure and becoming stronger at higher E₀, may stem from gas accumulated on other parts of the sample holder than the substrate or from fragments that attach to the metal surface and are released in a reactive process at higher temperature. The position and shape of this signal changed somewhat between the single measurements. The same applies to the signal in the 130–135 K range seen at higher E₀. While this signal is absent from all measurements without electron exposure, its position shifted between different measurements. The fact that the Au surface itself was not freshly prepared prior to each measurement may explain these changes if one assumes that the higher temperature signal stems from molecules or fragments in immediate contact with the metal. Chemisorption on polycrystalline or stepped Au surfaces has been reported previously for CO.¹² In the absence of sputter cleaning, such chemisorption processes may change the state of the metal surface from one experiment to the next, thus accounting for the variations in the TDS curves. The detailed investigation of this effect was beyond the scope of the present work and therefore not pursued.

Fig. 4 TDS measurements at 20 amu for 15-layer films of CD₃CN before and after electron exposure of 5000 μC at the stated incident energies. For comparison, measurements at 19 amu for both a fresh film and a film after the same exposure are included. The magnification is the same for all curves.

As a third product from CD₃CN under electron exposure, this time performed at 6 eV, 10 eV, and 15 eV, C₂D₆ could be observed as concluded from a 36 amu desorption signal in the 75–80 K range (Fig. 5). All reference measurements, i.e., experiments without electron exposure, show a flat baseline throughout the investigated temperature range. Similar to the results for CD₄, the intensity of the 36 amu signal is higher at 15 eV than at lower E₀ but, in contrast, it displays a minimal intensity at 10 eV. The energy dependence of C₂D₆ formation will be investigated more closely in the case of mixed films, as described in the next section.

Fig. 5 TDS measurements at 36 amu for 15-layer films of CD₃CN before and after electron exposure of 5000 μC at the stated incident energies. The magnification is the same for all curves.

In addition, as will be discussed in detail in Section 4, it was anticipated that larger products like CD₃CD₂CN and NCCD₂CD₂CN might be formed. This possibility was investigated at E₀ = 7 eV. As mass spectra from the NIST database¹¹ usually resemble those acquired on the present apparatus, these data were used as reference for the identification of the two compounds. For CD₃CD₂CN, the parent cation (60 amu) and fragments 58 amu and 56 amu are thus expected with relative intensities of approximately 15 : 85 : 9, while the strongest signal for NCCD₂CD₂CN should be 56 amu. Fig. 6a shows the TDS curves for 44 amu (CD₃CN) as well as 56 amu, 58 amu, and 60 amu after an exposure of 10 000 μC onto a 30 ML film of pure CD₃CN. While measurements for the latter three masses exhibit a flat baseline prior to exposure (not shown), desorption signals are observed at 140 K and around 180 K for 58 amu. A very weak increase of the signal is observed around 180 K for 60 amu. The relative intensities of the signals at 180 K corresponds roughly to that expected for CD₃CD₂CN, according to the database.¹¹ We thus ascribe the observed desorption signals to this compound. Assuming similar ionization efficiencies for CD₃CD₂CN and CD₃CN, we estimate, from a comparison of the peak intensities for 44 amu and 58 amu, that of the order of 0.1% of CD₃CD₂CN was formed. The very close agreement of the lower temperature peak with the desorption signal of CD₃CN can be rationalized by the similar vaporization enthalpies reported for the two compounds.¹¹ Also, it is possible that some CD₃CD₂CN could be driven to the vacuum by evaporation of the matrix material, CD₃CN. The higher temperature desorption signal may tentatively be assigned to recombinative desorption of fragments that are initially trapped at the metal surface. The lack of signal for 56 amu indicates that formation of NCCD₂CD₂CN cannot be observed within the sensitivity of the present experiment. It must be noted that at the larger film thickness and exposure time, while the intensity of 44 amu (CD₃CN) has roughly doubled, the intensities of CD₄ and C₂D₆ have now increased by factors of 4 and 8 (Fig. 6b). This can be traced back not only to the larger amount of material and longer exposure but also to the somewhat higher E₀, as compared to the experiments described earlier (Fig. 4 and 5). Fig. 6b also demonstrates that while the amount of CD₃CD₂CN appears to be smaller than that of CD₄ and C₂D₆, as far as can be judged from the mere comparison of the mass spectral signals, it is nonetheless not negligible.

Fig. 6 (a) TDS measurements at 44 amu as well as 56 amu, 58 amu, and 60 amu for 30-layer films of CD₃CN after electron exposure of 10 000 μC at E₀ = 7 eV. The magnification is the same for the lower three curves. (b) Direct comparison of desorption signals at 20 amu, 36 amu, and 58 amu recorded under the same conditions. The magnification is the same for all curves.

3.3. Reactions in mixed CD₃CN/C₂H₆ (1 : 1) films

Mixed films of CD₃CN and C₂H₆ were produced by leaking into the UHV chamber a mixture of the vapors containing equal amounts of both gases. Again, TDS curves prior to and after electron exposure of 5000 μC at different E₀ were recorded. Fig. 7a summarizes the peak intensities of the 44 amu desorption signal after electron exposure of 5000 μC normalized to the same signal without exposure. Again, the rate of loss of CD₃CN increases with increasing E₀ to become noticeable above 10 eV and reach an order of 60–70% after exposure to 5000 μC at 15 eV. C₂H₆ exhibits a similar behavior to reach a loss of 50–60% at 15 eV (Fig. 7b). 30 amu was used to quantify the amount of C₂H₆, despite the fact that the 28 amu signal is stronger for this compound, in order to avoid possible overlap with the DCN⁺ signal from CD₃CN. Again, some formation of DCN under electron exposure is deduced from an increase of the 28 amu signal around 140 K after exposure at 15 eV (Fig. 8). At the same time, the signal near 80 K ascribed to a fragment of C₂H₆ decreases under exposure. While the 28 amu signal at 140 K might also be due to a higher temperature channel for production of C₂D₆, the decrease of the 30 amu signal at the same temperature gives evidence against this interpretation.

Fig. 7 (a) Peak intensities of the desorption signal at 44 amu for 15-layer films of CD₃CN/C₂H₆ mixtures (1 : 1) after exposure, each normalized to the same intensity for a non-exposed film recorded immediately before. (b) Peak intensities of the desorption signal at 30 amu for the same films after exposure, each normalized to the same intensity for a non-exposed film recorded immediately before.

Fig. 8 TDS measurements at 30 amu and 28 amu for 15-layer films of a CD₃CN/C₂H₆ mixture (1 : 1) before and after electron exposure of 5000 μC at E₀ = 15 eV. The curves for 0 μC are offset for clarity.

As in pure CD₃CN, production of methane in the mixed films under exposure of the sample to 5000 μC is obvious from desorption peaks between 50 and 55 K, this time for both 19 and 20 amu (Fig. 9). This result indicates that both CD₄ and CD₃H are formed under exposure to the electron beam. At 15 eV, production of CD₃H is now even more important than CD₄. Again, experiments performed without exposure of the film (upper frame of Fig. 9) do not show any signal and CD₄ and CD₃H desorption signals at higher temperatures are not discussed further here. It must be noted that the desorption of CD₄ and CD₃H occurs at a slightly lower temperature than in the pure CD₃CN films, most probably due to the weaker polarization forces resulting from the smaller average amount of CD₃CN present in the environment of a single product molecule.

Fig. 9 TDS measurements at 19 amu and 20 amu for 15-layer films of CD₃CN/C₂H₆ mixtures (1 : 1) before and after electron exposure of 5000 μC at the stated incident energies. The magnification is the same for all curves.

Formation of C₂D₆ under exposure to electrons is again observed in the mixed films. In this case, a larger number of E₀ was investigated (Fig. 10). This time, the signal reaches its maximum at E₀ = 7 eV before dropping to a roughly constant lower level at higher E₀. Desorption takes place in the 70–75 K range. This is again, and in agreement with the result for CD₄ and CD₃H, slightly lower than in the pure CD₃CN film.

Fig. 10 TDS measurements at 36 amu for 15-layer films of CD₃CN/C₂H₆ mixtures (1 : 1) before and after electron exposure of 5000 μC at the stated incident energies. The magnification is the same for all curves.

Assuming that reactions between fragments from CD₃CN and C₂H₆ can occur in mixed films, TDS measurements aiming at a number of different products were performed. First, we searched for some possible adducts between fragments from CD₃CN and C₂H₆. Assuming that CD₃ could abstract H from C₂H₆ and one of the fragments from CD₃CN could then react with the resulting radical, we investigated whether C₂H₅CN or C₂H₅CD₃ are formed for E₀ ranging from 6 to 16 eV. Evidence for production of C₂H₅CD₃ was found above E₀ = 11 eV (Fig. 11), while at lower E₀ a flat baseline was obtained. Evidence for the former could not be found at any of these energies.

Fig. 11 TDS measurements at 47 amu for 15-layer films of CD₃CN/C₂H₆ mixtures (1 : 1) after electron exposure of 5000 μC at the stated incident energies. The magnification is the same for all curves.

Given that the strongest production of C₂D₆ is observed at E₀ = 7 eV (Fig. 10) and formation of C₂H₅CD₃ takes place at higher E₀, mixed films were investigated specifically after exposure at 7 eV and 15 eV to find evidence for other products. In analogy to the pure CD₃CN film, 30 ML films after exposure to 10 000 μC at E₀ = 7 eV were searched for traces of CD₃CD₂CN, NCCD₂CD₂CN, and, in addition, of n-C₄H₁₀. Desorption around 180 K was again observed in the 58 amu curve (not shown) but with smaller intensity, while evidence for NCCD₂CD₂CN was missing. On the other hand, additional information is now needed to distinguish CD₃CD₂CN from n-C₄H₁₀. From literature data,¹¹ we expect for n-C₄H₁₀ a signal at 43 amu that is about an order of magnitude stronger than the parent cation (58 amu). Evidence for such a strong signal at 180 K was not obtained from TDS curves recorded at 43 amu, but a strong desorption peak occurred at 140 K. This signal, on the other hand, can be traced back to an isotopic impurity of about 5% CD₂HCN in CD₃CN. As recording TDS curves for reference mixtures containing all potential products was beyond the scope of the present work, possible contributions of n-C₄H₁₀ to the strong signal at 140 K can not be excluded a priori. On the other hand, a desorption temperature of 110 K has been reported recently for n-C₄H₁₀.¹³ We thus conclude that n-C₄H₁₀ was not formed in the present experiment.

As will be discussed in Section 4, scrambling of fragments might lead to products such as CD₂HCN or C₂H₅D. While the former was already present as impurity in the sample of CD₃CN, and thus prohibited the observation of a minor product of the same mass, formation of the latter is clearly seen already at E₀ = 7 eV (Fig. 12). While the desorption signal for 30 amu decreases slightly upon exposure, the signal for 31 amu increases. Again, this mass was already present as a small isotopic impurity (¹³C) in C₂H₆ but the increase of the signal upon exposure must be ascribed to formation of C₂H₅D. Interestingly, the loss of CD₃CN is much more pronounced here than for the pure film at the same thickness and E₀, where it amounts to approximately 10%. Formation of C₂H₅D was also deduced to occur at E₀ = 15 eV (not shown). While the desorption signals at 75 K decrease strongly for both 30 amu and 31 amu in a 30 ML pure film of C₂H₆, and also for 30 amu in a mixed film of CD₃CN and C₂H₆, the same peak in the 31 amu curve of the mixed film remains roughly constant, thus hinting towards formation of C₂H₅D.

Fig. 12 TDS measurements at 30 amu, 31 amu and 44 amu for 30-layer films of a CD₃CN/C₂H₆ mixture (1 : 1) before and after electron exposure of 10 000 μC at E₀ = 7 eV. The curves for 0 μC are offset for clarity.

Finally, the films after exposure at E₀ = 15 eV were searched for traces of C₂H₅CD₂H as well as CD₃CD₂CN and CD₂HCD₂CN. Formation of the former might occur together with C₂H₅CD₃ if more than two fragments are involved in the reaction. Unfortunately, mass 46 (parent cation of C₂H₅CD₂H) is also present as fragment M-H in C₂H₅CD₃. On the other hand, signals at 46 amu and 47 amu should have similar intensity in C₂H₅CD₃ according to the database.¹¹ TDS curves of mixed 30 ML films after 10 000 mC exposure at 15 eV did reveal roughly equal intensities for these two masses. Evidence for formation of C₂H₅CD₂H was thus not obtained.

While the formation of CD₃CD₂CN was already deduced, a reaction mechanism involving more than two fragments could, again, also lead to formation of species like CD₂HCD₂CN. As the most intense fragment for CD₃CD₂CN has mass 58 amu, 57 amu should be most intense for CD₂HCD₂CN. Such a signal was also not detected within the sensitivity of the present experiment.

4. Discussion

Electron-induced chemical modifications in the condensed phase can involve intermolecular reactions of reactive fragments formed during the initial electron–molecule interaction. In order to unravel the reaction sequences in CD₃CN or mixed CD₃CN/C₂H₆ films leading to the observed products, we first review the available information on the products of Dissociative Electron Attachment (DEA) and Electron Impact (EI) ionization studies in the gas phase.

A recent study on electron attachment cross sections for CH₃CN in the energy range from near 0 up to 10 eV has revealed the production of the anionic fragments CH₂CN⁻, CHCN⁻, CCN⁻, CN⁻ and CH₃⁻ ,⁸ presumably accompanied by the radicals as follows in reactions (1a–e)

It was concluded that electron attachment in the gas phase is thus a purely dissociative process. The different fragments were observed at specific E₀. The most abundant product, CH₂CN⁻, appears solely around 3.2 eV, with a cross section of 4 × 10⁻¹⁹ cm² ,while CHCN⁻, which was observed at the same E₀ but two orders of magnitude less frequent, is also produced above 8 eV together with a smaller amount of CCN⁻. CN⁻ is the second most important fragment, with a maximum cross section of 1 × 10⁻²⁰ cm² in the 6–8.5 eV range and a smaller signal near E₀ = 2 eV. Finally, a very small amount of CH₃⁻ was observed between 6 and 8 eV. These signals essentially coincide with earlier results covering a broader energy range but showing no further important resonance structures above 10 eV.^9,10 Only formation of CN⁻ was observed above 10 eV but with considerably smaller intensity than in the 6–8.5 eV range. Our interest, and thus also the discussion, focuses mainly on fragments that are produced most efficiently as suggested by the gas phase studies. In addition, Electron-Stimulated Desorption (ESD) of D⁻ from solid films of CD₃CN has been observed above 7 eV, with maxima near 9 and 11.5 eV.¹⁴

Electron attachment to C₂H₆ has only been studied in thin condensed films.¹⁵ A characteristic H⁻ signal due to ESD was observed around 10 eV in this case. Other fragments appearing in the same energy range were CH₃⁻, CH₂⁻, and, with a smaller intensity, CH⁻.

Positive ion formation above the ionization threshold¹¹ can also contribute to reactions taking place under electron exposure. The thresholds for gaseous CH₃CN, CD₃CN, and C₂H₆ have been reported at 12.2–12.4 eV, 12.2–12.3 eV, and 11.4–11.8 eV. Fragmentation of CH₃CN was observed with thresholds around 14 eV (CH₂CN⁺) and 15–16 eV (CHCN⁺, CH₂⁺). C₂H₆ produces C₂H₄⁺ already near the ionization threshold and C₂H₅⁺ near 12 eV, while C₂H₃⁺, C₂H₂⁺, and CH₃⁺ appear above a threshold between 13.5–15 eV. Dipolar dissociation to CH₃⁺ and CH₃⁻ also takes place in gaseous C₂H₆ starting at 13.6 eV¹¹ but does not noticeably contribute to ESD below E₀ = 18 eV.¹⁵

Assuming essentially the same behavior for CH₃CN and CD₃CN, the gas phase electron attachment channel around 7–8 eV producing CN⁻ can explain the formation of CD₄ and CD₃H at 6–7 eV in the present study. The resonant process which presumably leads to a CD₃ radical as a second product (reaction 1d) is likely to be shifted towards lower E₀ in the condensed phase due to polarization of the molecular environment. Production of CD₄ and CD₃H at this energy requires additional abstraction of D or H atoms from adjacent molecules (see also Scheme 1):

Regarding only pure CD₃CN, it is not clear whether formation of CD₄ is due to abstraction of D from another molecule by the initially-formed CD₃ radical or to recombination with D atoms released upon formation of a fragment CDCN⁻:CHCN⁻ is produced in the gas phase by reaction (1b) with a rate that amounts to 50% that of reaction (1d) and peaks at 8.3 eV. From mere stoichiometric arguments, this process would provide an important source of D in the present experiments. Comparison with the mixed films, on the other hand, suggests that (2a–c) must dominate because dissociative processes in C₂H₆ leading to free H are not known at such low energies. Also, from energetic arguments, it is more likely that formation of CHCN⁻ or CDCN⁻ upon electron attachment is accompanied by production of H₂ or D₂.⁸ Processes (2a–c), as the dominant production channels of CD₄ and CD₃H in the condensed phase at E₀ between 6 and 7 eV, are thus consistent with the observed approximate ratio of 1 : 1 for the two products in the mixed films produced from equal amounts of CD₃CN and C₂H₆.

Scheme 1

It is possible that DEA according to process (1d) initiates a radical chain reaction. This might lead formally to migration of a CD₂CN radical as depicted in Scheme 1. If this species does not undergo further fragmentation, it could be expected to eventually recombine with CD₃ produced by a second DEA event (1d) at another site to form CD₃CD₂CN or with another migrating CD₂CN radical producing NCCD₂CD₂CN. Among those, the latter should be most probable due to the prediction (Scheme 1) that CD₂CN should recur throughout the chain reaction. In contrast, based on the statistical argument derived from Scheme 1, the probability for recombination of two CD₃ radicals should be considerably smaller as they are assumed to be consumed upon formation of CD₄. On the other hand, C₂D₆ is, in fact, observed with an intensity comparable to CD₄ (Fig. 4 and 5) and CD₃CD₂CN is formed but not NCCD₂CD₂CN. This suggests that part of the CD₃ radicals can not overcome the activation barrier to D abstraction¹⁶ at low temperature, and thus remains immobilized within the film so that recombination with other CD₃ or with CD₂CN are efficient concurrent reactions to abstraction of a D from an adjacent molecule when the temperature increases. This is reasonable if we assume radical recombination to be barrier-free.¹⁷ On the other hand, recombination requires sufficient mobility within the film and CD₂CN is probably not mobile enough to recombine with a second CD₂CN at a noticable rate. This rationalizes the fact that NCCD₂CD₂CN is not observed. The results so far suggest that the mobility of fragments is a rate-determining factor in the observed reactions.

In analogy to Scheme 1, Scheme 2 presents a hypothetical reaction sequence induced in the condensed mixed films of CD₃CN and C₂H₆ by DEA process (1d). Here, both of reactions (2b and c) take place as deduced from the simultaneous production of CD₄ and CD₃H. Again, we start by assuming radical chain reactions following the initial DEA event. In this case, propagation of the chain could produce new products, namely, CD₂HCN and C₂H₅D. Evidence obtained for formation of C₂H₅D at E₀ = 7 eV demonstrates that such chain propagation in fact takes place. In contrast to pure CD₃CN, where chain propagation does not lead to new products, the mixed films provide the opportunity to detect such reaction steps.

Scheme 2

Based on the finding that chain reactions must take place, a larger variety of recombination processes is expected in the mixed films. Recombination of two CD₃ radicals formed as immediate products of DEA explains the observed formation of C₂D₆, while recombination of the other intermediate radicals would produce CD₃CD₂CN, CD₃C₂H₅, C₂H₅CD₂CN, NCCD₂CD₂CN, or n-C₄H₁₀. Among those, CD₃C₂H₅ that would be formed according to

has been searched for but not detected at 7 eV, as obvious from Fig. 11. Also, NCCD₂CD₂CN and n-C₄H₁₀ could both not be detected, while evidence for CD₃CD₂CN, at least via the higher temperature process presumably involving trapping of radicals at the metal surface, was obtained. A search for C₂H₅CD₂CN was not performed here but given the lack of detectable signal for NCCD₂CD₂CN and n-C₄H₁₀ we expect a negative result for this products as well.

One possible explanation for the fact that CD₃CD₂CN, but not CD₃C₂H₅, is observed at E₀ = 7 eV invokes a different reactivity for CD₂CN and C₂H₅ radicals. While the former can be stabilized by a mesomeric effect, the latter is more reactive and would quickly attack adjacent molecules to abstract a D or H atom. CD₂CN can thus survive until recombination with other radicals such as CD₃ occurs. This becomes possible upon temperature increase, while C₂H₅ would react immediately with any neighbor molecule, thus preventing eventual recombination with CD₃. On the other hand, if C₂H₅ is in fact as reactive as assumed here, the same should apply for CD₃. This contradicts the interpretation given for the results on pure CD₃CN films because such a high reactivity of CD₃ should favor the formation of CD₄ and CD₃H over recombination products, such as CD₃CD₂CN, which is in contrast to Fig. 6b.

In the light of these contradictions, another explanation needs to be envisaged. The detection of C₂H₅D in the mixed films clearly demonstrates that chain reactions take place. In the case of pure CD₃CN films, we may assume that these reactions lead to accumulation of CD₂CN both within the bulk of the film and at the metal surface. After some exposure, newly formed CD₃ may encounter not only adjacent CD₃CN but also CD₂CN frozen in the film. The latter situation could lead to immediate recombination without increase of the temperature. CD₃CD₂CN formed in this way could freely evaporate upon temperature increase. We attribute the 140 K peak in the 58 amu curve (Fig. 6) to this process. As with CD₂CN, CD₃ may also get trapped at the metal surface. Trapped CD₃ and CD₂CN, on the other hand, may recombine through an activated process only at higher temperature, and can thus be held responsible for the 180 K desorption signal.

Trapping, i.e. chemisorption of molecules, at Au surfaces is not only known for CO.¹² Dissociative adsorption on Au(111) at low temperature has been reported for SiH₄.¹⁸ Also, electron-induced processes can induce chemisorption as reported for O₂ at temperatures below 50 K.¹⁹ The scenario suggested above for CD₃CN is consistent with such a process, even if the present polycrystalline Au surface is not as well defined as a single crystal surface.

In the mixed films, the situation is more complex. Here, the chain reactions lead to atom scrambling, as witnessed by the formation of C₂H₅D. This would lead to a larger variety of products. It is thus possible that the signal for each single product may drop below the level of sensitivity of the present experiment. The only remaining signal is the 180 K desorption of CD₃CD₂CN that was suggested above to involve radicals trapped at the metal surface. This signal is smaller than in pure films of CD₃CN, which can be ascribed to both a lower amount of CD₃CN but also to scrambling via the chain reactions that consume some of this compound. The fact that the 140 K signal is completely suppressed would thus reflect the lower probability that a CD₃ radical is produced in the vicinity of an already present CD₂CN.

A further question concerns the absence of a signal due to CD₃C₂H₅ at 7 eV. According to Scheme 2 and the detected amounts of CD₄ and CD₃H, both CD₂CN and C₂H₅ should be equally present within the mixed films. On the other hand, in the situation where trapping of the radical at the metal surface is important, a different reactivity might again be invoked. CD₂CN, due to the possible mesomeric structures, might offer a binding site to the metal even if it is immobilized, and the D abstraction occurs on the far side with respect to the metal. CD₃ is equally small and mobile enough to rearrange near the metal so that it may become trapped. C₂H₅, on the other hand, may not easily rotate to bring the radical site near the metal. This may decrease its lifetime to the point that recombination reactions become less likely than for CD₂CN. We must note that CD₃CD₂CN has been detected here in 30 ML films, while the search for CD₃C₂H₅ has only been performed with 15 ML films. Also, so far, we are not aware of experimental results supporting the suggested different binding efficiency for CD₂CN and C₂H₅ radicals. A final decision on the reason for the absence of detectable amounts of CD₃C₂H₅ can thus not be given here.

It must be noted that production of DCN according to

is not observed (Fig. 3 and 8) in the range of the resonant processes (1d–e) in good agreement with the low intensity of CD₃⁻ (1e/5a) in the gas phase. Similarly, C₂H₅CN has not been observed in the mixed films, giving evidence that recombination of a C₂H₅ radical produced adjacent to the site of an initial DEA event with CN⁻ does not take place, as expected from the stability of the latter anion in its ground state.

The rate of production of CD₄, CD₃H, C₂D₆ in pure CD₃CN, and C₂H₅CD₃ clearly increases above 11 eV. This must be ascribed to the onset of ionisation, which opens cationic reaction pathways. Interestingly, all four products could be most easily explained if CD₃ fragments were formed in the initial reaction step. To the best of our knowledge, formation of CD₃ from CD₃CN or the analogous CH₃ from CH₃CN does not appear to be important between 11 and 16 eV.¹¹ Formation of CN⁻ would imply also that CD₃ is produced, but in the gas phase this process is much less intense at higher E₀ than at 8 eV⁹ and can thus not explain the strong increase in CD₃-containing products at higher E₀. An intense production of CD₃ in this energy range, on the other hand, could stem from reaction (1d) if induced by low energy secondary electrons released upon ionisation according to

Note that, as mentioned above, in comparison with gas phase results processes involving ions are likely to be shifted by 1 eV or more towards lower E₀ in the condensed phase. Since the production of CN⁻ and CD₃ (1d) in the gas phase has a broad resonance (1 eV FWHM) with a peak below 2 eV, this reaction may be effective in the condensed phase with secondary electrons of energies even below 1 eV. Considering that the energy resolution of our experiment is of the order of 0.5 to 1 eV, the measured onset of all reactions driven by the initial electrons is expected to be additionally shifted to lower E₀ by almost 0.5 eV, thus, an increased production of CD₃via reaction (1d) following ionization (6a/b) is energetically possible above 11 eV in the mixed film, and at a slightly higher E₀ in pure CD₃CN. Somewhat more energetic secondary electrons (roughly 2 eV²) can induce the most efficient reaction (1a) of CD₃CN, thus opening a reaction channel for production of free D radicals. These, and the D as well as H radicals that are formed as by-products of higher energy ionizations, can recombine with CD₃ to form CD₄ and CD₃H. This opens additional channels for formation of these latter products as compared to lower E₀, where abstraction of D from CD₃CN or H from C₂H₆ by CD₃ can lead to production of CD₄ or CD₃H, thus rationalizing their higher rate of formation at higher E₀. Given that the DEA process (1a) has a much higher cross section in the gas phase than the lower energy resonance yielding CD₃ and CN⁻, one could expect a much higher production rate of CD₄ than C₂D₆ in the pure film of CD₃CN. In fact, the present results demonstrate a similar order of magnitude for the two products, which can be explained by escape of D or D₂ from the molecular film.

The production of C₂D₆ can again be explained by recombination of CD₃ radicals upon increase of the sample temperature. On the other hand, the fact that less C₂D₆ is produced in the mixed film than in pure CD₃CN suggests the presence of concurrent recombination channels, for example yielding C₂H₅CD₃ (Fig. 11).

As summarized above, ionization processes also lead to a number of positively charged molecular fragments. Under exposure to the electron beam, these are likely to be neutralized by electrons that are thermalized through inelastic collisions within the film, thus again yielding reactive radicals. C₂H₅⁺ formed upon ionization can recombine with CD₃ radicals produced by secondary electrons to form C₂H₅CD₃. If this is the dominant reaction pathway, other isotopic isomers are not expected. This product could also be formed by recombination of three fragments, namely CD₂⁺ and C₂H₅⁺ neutralized under the electron beam and a free D radical. In this case, not only C₂H₅CD₃ but also C₂H₅CD₂H is expected to form. The missing evidence for the latter suggests that, in fact, deceleration of incident electrons and the release of slow secondary electrons dominates the reaction mechanism.

Recombination processes like the one just described could lead to production of a number of other species. For example, CD₂CN⁺ and CD₃, preferably in pure CD₃CN, could form CD₃CD₂CN, while C₂H₅⁺ and D would again produce C₂H₅D that was already predicted to form via the radical chain reactions of Scheme 2 for the mixed films at low E₀. Recombination of three species—CD₂CN⁺, CD₂⁺, and H—could possibly form CD₂HCD₂CN. Again, missing evidence for the latter suggests that such ternary reactions are of minor importance.

While the present results give qualitative information on the reaction mechanism under electron exposure and suggest that secondary electrons are involved at E₀ above the ionization threshold, a more quantitative study of this latter effect would require a comparison with experiments performed at lower E₀. This way, the assumption that reaction (1a) is also the dominant one in the condensed phase could be verified. Some support that this reaction indeed plays a role has been obtained in a previous study that aimed at the functionalization of hydrogen-terminated diamond.² At present, we hope to improve our TDS setup in order to achieve electron exposure at lower E₀ within acceptable time scales.

Finally, we want to emphasize that there is no clear enhancement of any production rate around E₀ = 10 eV. Some production of CD₄, CD₃H, C₂D₆, and C₂H₅CD₃ is observed within the range of E₀ between 9 and 11 eV, which coincides with the resonant channel producing H⁻ in C₂H₆,¹⁵ taking into account not only the inherent energy width of this process but also the low energy resolution of the present experiments. On the other hand, the production rates of these products increase above 11 eV thus ruling out a dominant role of the H⁻ producing resonance in C₂H₆ in the formation of the products observed so far. For the same reason, we conclude that DEA to CD₃CN, producing D⁻ as observed by ESD above 7 eV with maxima at 9 and 11.5 eV,¹⁴ is less efficient than the higher energy reaction channels described before. Also, the fact that a product C₂H₅CN is not observed at any of the investigated E₀ demonstrates that CN⁻ produced by DEA becomes stabilized within the film and is not available for reactions with hydrocarbon radicals.

5. Conclusions

Thermal desorption experiments following exposure to low-energy electrons on pure films of CD₃CN and mixed films of CD₃CN and C₂H₆ were performed in order to obtain a detailed picture of the reactions that are initiated by electron impact. It was found that, while the initial events can be explained in line with previous gas phase results on dissociative electron attachment and electron impact ionization, complex intermolecular reactions take place.

The electron-driven dissociative process initiates radical chain reactions that lead to atomic scrambling. Products are formed via both abstraction of hydrogen by radicals such as CD₃ and also by recombination of radicals. In these processes, the mobility of the participating species is relevant, as witnessed by the prevalence of products resulting from recombination of CD₃ with other radicals. Evidence also points towards an influence of intermediate trapping of radicals at the metal. This process increases the lifetime of reactive fragments that are released upon increase of the temperature during the TDS scan.

At energies above the ionization threshold, the observed products point towards a dominant contribution of secondary electrons to the reactions. Again, these low-energy electrons initiate fragmentations that lead to release of CD₃, and possibly also D, that can again initiate radical chain reactions but this time with a higher efficiency due to the increased number of electrons.

The results show that intermolecular reactions and formation of larger molecules take place in mixed films of CD₃CN and C₂H₆. The insight obtained from this study will guide future work aimed at the development of new processes for the functionalization of hydrocarbon materials.

Acknowledgements

I. I. acknowlegdes a travel grant from ESF program EIPAM.

References

1. W. Di, P. Rowntree and L. Sanche, Phys. Rev. B: Condens. Matter Mater. Phys., 1995, 52, 16618.
2. A. Lafosse, M. Bertin, D. Caceres, C. Jäggle, P. Swiderek, D. Pliszka and R. Azria, Eur. Phys. J. D, 2005, 35, 363.
3. (a) H. Abdoul-Carime, S. Gohlke and E. Illenberger, Phys. Rev. Lett., 2004, 92, 168103; (b) S. Ptasińska, S. Denifl, V. Grill, T. D. Märk, P. Scheier, S. Gohlke, M. A. Huels and E. Illenberger, Angew. Chem., Int. Ed., 2005, 44, 1647; (c) S. Ptasińska, S. Denifl, V. Grill, T. D. Märk, E. Illenberger and P. Scheier, Phys. Rev. Lett., 2005, 95, 093201; (d) S. Ptasińska, S. Denifl, P. Scheier, E. Illenberger and T. D. Märk, Angew. Chem., Int. Ed., 2005, 44, 6941.
4. V. Cobut, Y. Frongillo, J. P. Patau, T. Goulet, M.-J. Fraser and J.-P. Jay-Gerin, Radiat. Phys. Chem., 1998, 51, 229.
5. I. Martin, T. Skalicky, J. Langer, H. Abdoul-Carime, G. Karwasz, E. Illenberger, M. Stano and S. Matejcik, Phys. Chem. Chem. Phys., 2005, 7, 2212.
6. M. Imhoff, L. Parenteau, L. Sanche and M. A. Huels, Phys. Chem. Chem. Phys., 2005, 7, 3359.
7. P. Swiderek, M. C. Deschamps, M. Michaud and L. Sanche, J. Phys. Chem. B, 2003, 107, 563.
8. W. Sailer, A. Pelc, P. Limão-Vieira, N. J. Mason, J. Limtrakul, P. Scheier, M. Probst and T. D. Märk, Chem. Phys. Lett., 2003, 381, 216–222.
9. M. Heni and E. Illenberger, Int. J. Mass Spectrom. Ion Processes, 1986, 73, 127.
10. J. A. Stockdale, F. J. Davis, R. N. Compton and C. E. Klots, J. Chem. Phys., 1974, 60, 4279.
11. NIST Chemistry WebBook, NIST Standard Reference Database Number 69, eds. P. J. Linstrom and W. G. Mallard, June 2005, National Institute of Standards and Technology, Gaithersburg MD, 20899 (http://webbook.nist.gov).
12. J. M. Gottfried, K. J. Schmidt, S. L. M. Schroeder and K. Christmann, Surf. Sci., 2003, 536, 206.
13. S. Funk, B. Hokkanen, J. Wang, U. Burghaus, G. Bozzolo and J. E. Garcés, Surf. Sci., 2006, 600, 583.
14. M. Lachgar, Y. Le Coat, R. Azria, M. Tronc and E. Illenberger, Chem. Phys. Lett., 1999, 305, 408 . We note that according to a personal communication by R. Azria, Fig. 3 of this reference indeed shows results for pure CD₃CN in contrast to what was erroneously stated in the text of the same reference.
15. P. Rowntree, L. Parenteau and L. Sanche, J. Phys. Chem., 1991, 95, 4902.
16. B. Kerkeni and D. C. Clary, J. Chem. Phys., 2005, 123, 064305.
17. H. Tamura and M. S. Gordon, Chem. Phys. Lett., 2005, 406, 197.
18. M. J. S. Spencer and G. L. Nyberg, Surf. Sci., 2004, 573, 151.
19. J. M. Gottfried, K. J. Schmidt, S. L. M. Schroeder and K. Christmann, Surf. Sci., 2002, 511, 65.

---

Footnote

† Present adress: Airbus Deutschland GmbH, Hünefeldstr. 1–5, 28119 Bremen, Germany.

---