Adsorption dynamics of bifunctional molecules: Allyl methyl ether on Si(001)

Adsorption of organic molecules on the Si(001) surface typically proceeds via an intermediate state that is characterized by a relatively weak, non-covalent bond between the functional group of the adsorbing molecules and the surface. The type of the functional group thus determines the nature of the binding in the intermediate state,^1–4 which, in turn, governs the adsorption dynamics^5–9 (Fig. 1). The adsorption dynamics of alcohols and ethers on Si(001)^6,9 may serve as a good example: in both cases, the adsorbing molecules form a dative bond with the surface in the intermediate state^10–18 (lower right sketch in Fig. 1), and despite major differences in their structure, mass, and the energetics of the further reaction into the final state, they show almost identical adsorption dynamics as determined by the dependence of the initial sticking probability s₀ on the kinetic energy E_kin of the incoming molecules.⁹ On the other hand, molecules with different functional groups such as alkenes feature a considerably stronger dependence of s₀ on E_kin and thus a much lower reactivity at high kinetic energy of the incident molecules⁵ (Fig. 1). In the latter case, the intermediate state involves a weaker three-center bonding between the surface and the carbon–carbon double bond of the alkene molecules^19–23 (upper right sketch in Fig. 1).

In the case of molecules with two or more functional groups, the reaction pathway was determined for a couple of systems,³ both theoretically and experimentally. To the best of our knowledge, however, the effect of different functional groups on the adsorption dynamics of such bifunctional molecules has not been investigated so far.

In this paper, we study the adsorption dynamics of the bifunctional molecule allyl methyl ether (AME) on Si(001). Using molecular beam techniques for the adsorption experiments, the kinetic energy of the molecules impinging on the surface can be controlled, in contrast to adsorption from thermal gas. Whereas the latter experiments measure the kinetics of a reaction, e.g., at different surface temperatures, and give valuable information on the energetics of the reaction,^24–26 the molecular beam experiments exploit, in addition, the dynamics of the reaction in terms of the influence of the kinetic energy of the incoming molecules on the reactivity.²⁷ 

The two functional groups in the AME molecule, i.e., the ether group with its oxygen heteroatom and the C=C double bond, react via two very different intermediate states that exhibit different adsorption energetics and dynamics. We compare the results of AME to the adsorption dynamics of diethyl ether (Et₂O) and ethylene (C₂H₄) on Si(001) in order to disentangle the contribution of each functional group to the overall reaction dynamics. Despite the high reactivity of the C=C double bond on Si(001),^24,28–32 we show here that the adsorption dynamics is mainly determined by the ether group. The C=C double bond then influences the further reaction from the intermediate into the final state as deduced from the dependence of s₀ on surface temperature T_s.

The experiment was described in detail in Refs. 27 and 33. In short, the experimental setup consists of a four stage molecular beam apparatus, and the beam is generated via supersonic expansion through a nozzle of 100 μm in diameter. The sample is located 30 cm downstream of the nozzle. The kinetic energy of the molecules in the beam was varied by heating the nozzle, seeding the beam in noble gas of different atomic mass (i.e., helium), or both seeding and heating of the nozzle. The mean kinetic energy and its distribution were determined by means of time-of-flight measurements. Typical values of the ratio between the mean velocity of the molecules and the width of the velocity distribution, i.e., the speed ratio S, range between S = 2.5 for the unseeded beam to S = 11 for the seeded beam of E_kin = 550 meV.

The sample was prepared by resistive heating up to 1300 K and subsequent cooling with cooling rates of ≈1 K/s. We verified the 2 × 1 reconstruction of the Si(001) surface by means of low energy electron diffraction. During the experiment, the sample was heated up to 1100 K between two successive adsorption measurements in order to desorb the adsorbates of the former measurement from the surface. The temperature of the sample was measured using a thermocouple glued to the rear side of the sample.

The absolute sticking coefficient s was determined by means of the King-and-Wells method:³⁴ when the molecular beam enters the sample chamber and an inert shutter is placed in the front of the sample, the partial pressure related to the investigated molecules is increased due to the molecules being reflected from the inert shutter. The partial pressure is monitored by using a quadrupole mass spectrometer (QMS), which is positioned out of the line of sight of the molecular beam. The QMS was operated at m/z = 45, the most intense signal in the mass spectrum of AME. When the shutter is opened, a drop of the QMS signal is observed due to adsorption on the reactive sample surface. This drop of the partial pressure is proportional to the initial sticking coefficient s₀ = s(t = 0) as long as the beam diameter (≈8 mm) is smaller than the sample surface, which is 5 × 1 cm² in size. From the raw data, the background partial pressure is subtracted. The latter is caused by thermal gas that enters the main chamber effusively via the aperture and can be measured by blocking the beam in the first differential pumping stage; it typically accounts for less than 5% of the total signal. After background subtraction, the absolute value of s₀ is determined from the initial signal drop when the data are normalized to the maximum value of the signal.

In Fig. 2, the adsorption measurements, which are represented by the partial pressure measured as a function of time, are shown for three different beam energies E_kin at a given surface temperature T_s = 540 K. The initial signal drop after opening the shutter at t = 0 s as indicated by the black arrows decreases with increasing E_kin. This implies a reduced initial sticking coefficient s₀ at higher E_kin. The kinetic energy was varied between 90 meV (unseeded beam) and 550 meV by seeding the beam and heating the nozzle to T_N = 1000 K.

The dependence of the initial sticking coefficient on kinetic energy is shown in Fig. 3 for two different surface temperatures. At low kinetic energy, the sticking coefficients exhibit the highest values, indicating a high sticking probability for the adsorption of thermal gas, in qualitative agreement with thermal gas adsorption measurements. With increasing E_kin, the initial sticking coefficient decreases continuously for a given surface temperature. For comparison, we additionally measured s₀(E_kin) for diethyl ether and further compare to the data for ethylene on Si(001) taken from Ref. 5. For Et₂O/Si(001) (blue lines, data shown in Fig. S1 in the supplementary material), the decrease in s₀ with increasing E_kin is very similar to AME/Si(001). For C₂H₄/Si(001) (ocher dashed-dotted line, from Ref. 5), s₀ drops much faster with increasing E_kin when compared to AME and Et₂O.

The dependence of the initial sticking coefficient on surface temperature was determined at a constant kinetic energy of E_kin = 90 meV (room temperature nozzle, unseeded beam). The results are shown in Fig. 4 (red data): s₀ is constant for T_s < 400 K and drops continuously at surface temperatures T_s > 400 K. This dependence indicates that the adsorption proceeds via an intermediate state: Once the impinging molecule is trapped in the intermediate state, it can either react into the final state or it desorbs back into the gas phase. For a non-activated reaction, the desorption barrier ϵ_d is higher than the conversion barrier ϵ_a; thus, at increased surface temperature, desorption is favored over conversion, leading to the drop of s₀. With ν_d and ν_a being the prefactors for the desorption and conversion rate, respectively, and s_0,max being the probability for adsorption in the intermediate state, the temperature dependence of s₀ can be expressed according to the Kisliuk model³⁵ as

Fitting Eq. (1) to the data as shown in the inset of Fig. 4, the difference between the two energy barriers ϵ_d − ϵ_a can be deduced. For AME/Si(001), we obtain $(ϵd−ϵa)AME=0.31±0.04$ eV. Despite the similarity of the dependence of s₀ on E_kin for AME and diethyl ether on Si(001), the s₀(T_s) data of Et₂O/Si(001), which are shown in Fig. 4 as well (blue data, taken from Ref. 36), significantly deviate from the AME/Si(001) system. For Et₂O/Si(001), $(ϵd−ϵa)Et2O=0.24±0.04$ eV was reported.³⁶ 

We first discuss the continuous decrease in the initial sticking coefficient s₀ with increasing beam energy E_kin , as shown in Fig. 3. Such a dependence is typical for a non-activated adsorption channel. The higher the kinetic energy of the incoming molecules, the less likely it is that the molecules dissipate enough energy that they are trapped in the intermediate state. When comparing the results for AME and diethyl ether, it is obvious that both datasets follow the same functional form, i.e., a linear dependence of s₀ on E_kin. Moreover, the slopes are almost identical for comparable values of s₀. In contrast, the data for ethylene on Si(001) show a much stronger dependence of s₀ on E_kin. In a previous study, we have shown that the nature of the intermediate determines the adsorption dynamics.⁸ As we observe almost identical adsorption dynamics for AME and diethyl ether, but very different adsorption dynamics in case of ethylene on Si(001), we conclude that in the case of the bifunctional molecule AME, the dynamics are largely determined by the ether group and the corresponding intermediate. The latter comprises a dative bond between the oxygen atom of the ether group and the unfilled D_down state of a silicon dimer. This bond is relatively strong,³⁶ in particular, when compared to the three-center π complex, which is formed in the intermediate between the C=C double bond of ethylene and the unfilled D_down state of the silicon surface.^19,23 From our results, we conclude that the C=C double bond as a second functional group in the AME molecule does not significantly affect the adsorption dynamics into the intermediate state, which are governed by the ether group.

In our discussion, we made use of the fact that the nature of the intermediate determines the adsorption dynamics,^8,9 which enabled us to identify the reaction channel that governs the reaction of AME into the intermediate state. At this point, we would like to note that one can further exploit this observation to determine the reaction channels including the involved intermediate state also for molecules which do not give easy access to the intermediate state, e.g., in case it is too short-lived for the experimental techniques applied.

Although the C=C double bond in the AME molecule does not influence the adsorption dynamics into the intermediate, it has a non-negligible influence on the further reaction from the intermediate into the final state as deduced from the temperature dependence of the sticking probability: The energy difference ϵ_d − ϵ_a = 0.31 eV and the prefactor ratio $νdνa=3.8×102$ for AME/Si(001) deviate significantly from the values measured for Et₂O/Si(001) (ϵ_d − ϵ_a = 0.24 eV, $νdνa=7×102$³⁶). This difference is attributed to the presence of the C=C double bond in the AME molecule, which is the major difference between the two molecules.

The further reaction of the ether group into the final state proceeds via the cleavage of the O—C bond (ϵ_a = 0.4 eV in the case Et₂O³⁶), leading to a Si—OCH₂CH₃ and a Si—CH₂CH₃ entity. The presence of the C=C functional group in the AME molecule may provide an additional reaction channel for the conversion from the intermediate into the final state via the [2 + 2] cycloaddition of C=C on silicon, which proceeds with a reduced barrier.²⁰ Indeed, the higher value of ϵ_d − ϵ_a indicates a lower energy of the transition state between the intermediate and final state, which translates into a lower conversion barrier if we assume the same binding energy in the intermediate state for both molecules. Both the decreased transition state energy and the increased value for the ratio of the prefactors, $νaνd$⁠, are thus in line with an additional reaction channel for the further reaction into the final state, which might be a combination of the final states of the ether and the C=C entities.

The adsorption dynamics of the bifunctional molecule AME was found to be mainly determined by the ether group with a negligible influence of the C=C double bond. On the other hand, the further reaction into the final state deviates from the kinetics deduced for molecules with an ether functionality only, indicating a substantial influence of the C=C double bond in the further course of the reaction. The results thus demonstrate that the initial adsorption dynamics can be controlled by one functional group only even when the further reaction comprises both functional groups of a bifunctional molecule.

The supplementary material includes the data on the initial sticking probability for diethyl ether on Si(001) measured as a function of the kinetic energy of the incoming molecules.

We gratefully acknowledge funding by Deutsche Forschungsgemeinschaft through Grant Nos. DU 1157/4-1, GRK 1782, and SFB 1083 (Project No. 223848855).

The data that support the findings of this study are available from the corresponding author upon reasonable request.