Letter pubs.acs.org/JPCL
Binding Energy and Dissociation Barrier: Experimental Determination of the Key Parameters of the Potential Energy Curve of Diethyl Ether on Si(001) Marcel Reutzel,† Marcus Lipponer,† Michael Dürr,*,†,‡ and Ulrich Höfer† †
Fachbereich Physik und Zentrum für Materialwissenschaften, Philipps-Universität, D-35032 Marburg, Germany Institut für Angewandte Physik, Justus-Liebig-Universität Giessen, D-35392 Giessen, Germany
‡
S Supporting Information *
ABSTRACT: The key parameters of the potential energy curve of organic molecules on semiconductor surfaces, binding energy of the intermediate state and dissociation barrier, were experimentally investigated for the model system of diethyl ether (Et2O) on Si(001). Et2O adsorbs via a datively bonded intermediate from which it converts via ether cleavage into a covalently attached final state. This thermally activated conversion into the final state was followed in real-time by means of optical second-harmonic generation (SHG) at different temperatures and the associated energy barrier ϵa = 0.38 ± 0.05 eV and pre-exponential factor νa = 104±1 s−1 were determined. From molecular beam experiments on the initial sticking probability, the difference between the desorption energy ϵd and ϵa was extracted and thus the binding energy of the intermediate state was determined (0.62 ± 0.08 eV). The results are discussed in terms of general chemical trends as well as with respect to a wider applicability on adsorbate reactions on semiconductor surfaces.
T
fragments of the former diethyl ether molecule on two neighbored dimer rows thus quenching two dangling bonds (inset Figure 1).10 We use this reaction as a prototype system for determining the energy barrier ϵa = 0.38 ± 0.05 eV and preexponential factor νa = 104±1 s−1 by means of SHG. We combine this measurement with a molecular beam experiment in order to also get access to the binding energy of the intermediate state. In the case of a barrierless adsorption channel from the gas phase into the intermediate state,11 the binding energy of the intermediate state is equal to the desorption barrier ϵd for desorption into the gas phase. By measuring the initial sticking coefficient as a function of surface temperature, the energy barrier difference ϵd − ϵa = 0.24 ± 0.03 eV and thus ϵd = 0.62 ± 0.08 eV were determined. With νd/νa = (7 ± 3) × 102, the respective pre-exponential factor νd = 107±1.3 s−1 was deduced. Both νa and νd were found to be considerably lower than the typically assumed values of 1012 to 1013 s−1; thus, the temperature-dependent rate measurements are necessary for an absolute determination of the energy barriers. Figure 1 shows the results of a typical SHG experiment following the ether cleavage reaction in real-time. The
he functionalization of semiconductor surfaces with organic molecules has attracted much interest,1−5 especially with respect to the challenges arising from the miniaturization in semiconductor device physics.6 On Si(001), adsorption of most organic molecules is controlled by an intermediate state; for various systems, this intermediate was observed by means of spectroscopic methods or scanning tunneling microscopy.5,7 However, experimental access to the main parameters of the underlying potential energy curve, especially to the barrier for conversion from the intermediate to the final state, is challenging: the experiment both has to be able to differentiate between the intermediate and final state as well as it has to be fast enough to follow the conversion between the two states in real-time. In this Letter, we show that optical SHG can be employed for the determination of this conversion rate ka and, thus, the energy barrier ϵa. For crystal structures with inversion symmetry, such as silicon, SHG is symmetry forbidden in the bulk and, thus, provides a background free, surface sensitive probe. The key for the present application is the sensitivity of SHG on the number of reacted Si dangling bond states,8,9 as this number typically changes from the intermediate to the final state. For example, the overall nonactivated adsorption of diethyl ether on Si(001)10 includes an intermediate with a dative bond7 between the molecule’s oxygen atom and the DDown state of the lower Si atom of the silicon dimer. Ether cleavage converts this intermediate into two covalently attached © 2015 American Chemical Society
Received: July 27, 2015 Accepted: September 15, 2015 Published: September 21, 2015 3971
DOI: 10.1021/acs.jpclett.5b01510 J. Phys. Chem. Lett. 2015, 6, 3971−3975
Letter
The Journal of Physical Chemistry Letters
Figure 1. Typical measurement of the nonlinear susceptibility χ(2) s , which is proportional to the square root of the measured SHG signal, for real-time monitoring of the thermally activated ether cleavage is observed due to the dosage of Et2O reaction. A change of χ(2) s (t = −155 s), the rising surface temperature (t = −22 s) and the surface reaction (t > 0 s). In the inset, scanning tunneling microscopy (STM) images (positive sample bias) and schemes of the datively bonded intermediate state (80 K) and the covalently attached final state (300 K) are shown (taken from ref 10).
Figure 2. Isothermal SHG experiments monitoring the ether dissociation at different surface temperatures by means of Δχ(2) s (t). The inset shows an Arrhenius plot of the deduced dissociation rate ka; the error bars for statistical errors lie within the diameter of the data points. Activation energy and pre-exponential factor were deduced from a least-square fit to the data (straight line).
of Et2O.13 The inset in Figure 2 shows the decay rates ka plotted as a function of inverse surface temperature. The data points are well described by the Arrhenius law ka(Ts) = νaexp(−ϵa/kBTs) with activation energy ϵa = 0.38 ± 0.05 eV and pre-exponential factor νa = 104±1 s−1. In addition to random measurement errors, we take into account a possible systematic underestimation of Ts by up to 10 K due to nonuniform laserheating; in the given temperature regime this leads to the relatively high error bars. Complementary information was obtained when measuring the initial sticking probabilities s0 as a function of surface temperature Ts, which gives access to the difference between desorption and conversion barrier, ϵd − ϵa. In the left inset of Figure 3, two typical measurements following the King and Wells method14 are shown; in the case of the 171 K measurement, the molecular beam enters the main chamber at t = −22 s and is blocked by a nonreactive shutter, thus the background pressure pQMS(t) as measured by means of a
nonlinear optical response of the surface is characterized by the nonlinear susceptibility χ(2) s , which is proportional to the square root of the measured SHG signal. For small surface coverage, χ(2) s depends linearly on the number of dangling bonds on the 8,12 surface.8 In addition, χ(2) s changes as a function temperature. In Figure 1, χ(2) is plotted as a function of time; prior to s adsorption of Et2O, the measured SHG signal and, thus, χ(2) s is constant. At t = −155 s, χ(2) s drops as Et2O was adsorbed in the datively bonded intermediate state at Ts = 80 K and quenches one dangling bond per adsorbed molecule; coverage was approximately 0.1 ML (1 ML equals one Et2O molecule per silicon dimer). At t = −22 s, the surface temperature was increased to Ts = 293 K. Up to t = 0 s, the concomitant decrease of χ(2) s is dominated by the temperature dependence of the SHG signal. At t = 0 s, the final temperature is reached and for t > 0 s, the change in χ(2) s can be exclusively attributed to the conversion from the intermediate into the final state and the concomitant quenching of two dangling bond states. In order to compare different isothermal measurements, χs(2) is (2) (2) normalized to Δχ(2) s,max = χs (t = 0) − χs,∞ and plotted in the (2) (2) (2) (2) form Δχs (t) = [χs (t) − χs,∞]/Δχs,max, where χ(2) s (t) is the (2) time dependent nonlinear susceptibility and χs,∞ is the nonlinear susceptibility when all adsorbates are converted into the final state. χ(2) s,∞ is obtained by heating the surface to temperatures slightly above 300 K (at t = 670 s in Figure 1) and again measuring the SHG signal at the temperature at which the reaction was followed (293 K at t = 800 s in Figure 1). In order to qualitatively compare the overall change in χ(2) s , in Figure 1 the sample is further cooled down to 80 K at t = 920 s. As χ(2) s of a surface with still datively bonded molecules would return to the base level, the observed difference in χ(2) s can be directly attributed to the conversion from the intermediate into the final state. For a quantitative analysis, the isothermal measurements are plotted on a logarithmic scale (Figure 2) and fitted with the (2) exponential decay law Δχ(2) s (t) = Δχs (t = 0)exp(−kat) of a first order chemical reaction, with ka being the dissociation rate
Figure 3. Main panel: Initial sticking coefficient s0 of Et2O on Si(001) as a function of surface temperature Ts (Ekin = 0.1 eV). Left inset: King and Wells measurement of sticking probability s for two different surface temperatures, the less pronounced drop of pQMS at 371 K indicates a lower initial sticking probability at higher temperatures. Right inset: Kisliuk plot for the determination of ϵd − ϵa and νd/νa. For details see main text and Supporting Information. 3972
DOI: 10.1021/acs.jpclett.5b01510 J. Phys. Chem. Lett. 2015, 6, 3971−3975
Letter
The Journal of Physical Chemistry Letters
Beside the comparison of different heteroatoms, it seems interesting to extend the discussion on molecules with different substitutes on datively bonded oxygen atoms like, for example, water (H2O), alcohol (R1OH), and ether (R1OR2). For water molecules, the energy difference for molecular desorption and O−H dissociation could be extracted experimentally to be ϵd − ϵa = 0.29 eV.21 As no molecular water adsorption was spectroscopically observed down to 80 K,22,23 ϵa is expected to be small, and thus, ϵd ≈ 0.3 eV. Compared to water, the bond strength of datively bonded Et2O is significantly higher, which can be explained considering the positive inductive effect of the alkyl chains, thus stabilizing the dative bond. For alcohols, to the best of our knowledge no experimental studies are reported in literature for the binding energy of the datively bonded intermediate; within the dicussed trend, its bond strength is expected to be 0.3 eV < ϵd < 0.6 eV. Comparison with theory is difficult, as the calculated absolute values of ϵd span a wide range.24−26 Another type of coordinatively bonded intermediate is observed in cycloaddition reactions of unsaturated carbon hydrates.5,27 For example, ethylene and acetylene adsorb via an intermediate state that consists of a three-membered ring with the π-electrons donating electron density into the DDown state of the silicon dimer.28,29 For ethylene, the energy barrier for conversion into the final state, ϵa = 0.13 eV,30 and the energy barrier ϵd = 0.33 eV for desorption30,31 are both lower compared to Et2O, indicating that the interaction of the DDown state with the lone pair electrons of the oxygen atom is stronger when compared to the π electrons of C2H4. With respect to the kinetics, we observe a first order chemical reaction (Figure 2). We recently proposed some kind of SN2type adsorption pathway at which the Dup state (nucleophile) of the neighboring dimer row donates electron density into an antibonding O−C orbital of Et2O.10,32 As the available dangling bonds are given by the surface reconstruction, the reaction rate is only dependent on the number of datively bonded molecules and the proposed reaction mechanism is in good agreement with the observed kinetics. On a first glance, both pre-exponential factors, νa = 104 s−1 and νd = 107 s−1, seem to be very low compared to the typically assumed attempt frequencies of 1012−1013 s−1. On the other hand, for the conversion of ethylene on Si(001) from the weakly bound intermediate into the final adsorption state, νa ≈ 102 s−1 was determined,30 even lower than in our experiment. Along with the argumentation in ref 30, the low pre-exponential factors can be rationalized by the high number of degrees of freedom of the intact molecules in the datively bonded intermediate state and the corresponding large partition function ZI. In the framework of transition state theory, the ratio between the partition function in the transition state, Z⧧, and ZI is proportional to the pre-exponential factor νa; a large value of ZI and in consequence a small ratio of Z⧧/ZI, thus, results in a low value of νa. In the gas phase, the number of degrees of freedom is only moderately increased when compared to the datively bonded molecule;33 thus, also, νd is expected to be lower than for desorption of a more rigidly bound molecule. As temperature-dependent rate measurements are necessary for an absolute determination of the energy barriers and prefactors, we want to discuss the more general applicability of SHG for studying the kinetics of such conversion processes on semiconductor surfaces. In principle, these reactions are also accessible by spectroscopic methods such as, for example, X-ray
quadrupole mass spectrometer (QMS) rises and saturates at equilibrium pressure peq. At t = 0 s, the nonreactive shutter is retracted and the silicon surface is exposed to the molecular beam. The pressure drop peq − pQMS(t) is proportional to the sticking coefficient s(t) = [peq − pQMS(t)]/peq. When the surface reaches full coverage, the background pressure rises again to peq. In the main panel of Figure 3, the initial sticking coefficient s0 as obtained from the initial pressure drop is plotted as a function of Ts: with increasing surface temperature, s0 stays close to unity up to approximately 300 K; for higher temperatures, s0 drops to values below 0.1. This temperature dependence is well described within the Kisliuk model15,16 taking into account an adsorption process via an intermediate state. Once trapped in the intermediate state, the molecules can either convert into the final state via ϵa or desorb back into the gas phase via ϵd (Figure 4). In the right inset of Figure 3, s0(Ts) is analyzed
Figure 4. Potential energy curve of Et2O on Si(001). The datively bonded intermediate state (I) is kinetically stabilized against the ether cleavage reaction by ϵa = 0.38 eV and is bonded with ϵd = 0.62 eV relative to the molecule in the gas phase.
within a Kisliuk plot (compare Supporting Information) from which ϵd − ϵa = 0.24 ± 0.03 eV and νd/νa = (7 ± 3) × 102 are deduced. The experimental results are summarized in Figure 4. With ϵa = 0.38 ± 0.05 eV (νa = 104±1 s−1) and ϵd − ϵa = 0.24 ± 0.03 eV, the binding energy of the intermediate state can be deduced to ϵd = 0.62 ± 0.08 eV (νd = 107±1.3 s−1) and thus the combination of both techniques determines the main parameters of the potential energy curve of the datively bonded Et2O on Si(001). In the following, we will first discuss the binding energy of Et2O with respect to other coordinatively bound organic molecules. Second, we will be focusing on the unusually low preexponential factors as well as having a short look on the kinetics of the ether cleavage reaction and its implications on the reaction mechanism. Finally, as we see SHG to be more generally applicable to the measurement of the reaction kinetics of organic molecules on semiconductor surfaces, we will review the method in more detail at the end of this section. In a simple model, the dative bond strength ϵd of heteroatoms on the Si(001) surface is expected to decrease with increasing electronegativity of the heteroatom, as also observed in DFT calculations.7,17 The presented results confirm this trend experimentally as Et2O is more weakly bound on Si(001) than amines. For example, at low coverage trimethyl amine (TMA) desorbs intact from the surface as N−C cleavage is activated with respect to the molecule in the gas phase.18,19 In that case, the binding energy of the datively bonded TMA was directly determined by desorption experiments to be ϵd = 1.1 eV,20 substantially higher than in the case of Et2O/Si(001). 3973
DOI: 10.1021/acs.jpclett.5b01510 J. Phys. Chem. Lett. 2015, 6, 3971−3975
Letter
The Journal of Physical Chemistry Letters
sample low (
