In general, the sticking coefficient strongly depends on the coverage, therefore, an exact measu-rement of the fraction of adsorbed molecules versus the pulse number is necessary to evaluate the heat release per unit of molecules and the coverage.
For many large organic molecules on cold surfaces, however, the sticking coefficient is one and independent of the coverage. In this case, the number of adsorbed molecules can be determined without performing a sticking coefficient measurement, if the molecular flux has been determi-ned. To calculate the coverage, the number of adsorbed molecules in a filled monolayer has to be known. If this number is not known or adsorption in multilayer occurs, I will refer to the number of adsorbed molecules or number of adsorbed atoms instead of the coverage.
In cases, where the sticking coefficient is non unity and coverage dependent, the King-Wells sticking coefficient is evaluated as discussed in section 3.3. Therefore, the QMS signal taken for this evaluation is filtered with a median filter, which replaces each data point with the median of 5 circumjacent data points and a frequency filter, that removes frequencies which are higher than 4 Hz. This significantly reduces the noise and facilitates the evaluation of the signal. Figure 5.4 shows the original signal (gray) along with the filtered QMS signal (black). Not only the noise is removed by using the data processing procedure, but also the signal shape is influenced.
Therefore, several tests with different data sets were performed to confirm, that the accuracy is not influenced by the filtering procedure.
A small amount of gas exits the molecular beam in case the chopper is closed but the gas inlet to the molecular beam is open. Although the flux on the sample is only 1-2 % compared to the flux in case of an open chopper, a noticable rise in the background pressure can be detected in this case. This background is substracted from the QMS signal before the analysis.
Fig. 5.5 shows a schematic representation of QMS signals, which are processed to determine the King-Wells sticking coefficients. In principle, the sticking coefficient can be obtained by
Abbildung 5.4: QMS signal during pulsing on a Pd(111) crystal surface with a chopper opeing time of 266 ms compared with the filtered signal used for the evaluation
Abbildung 5.5: Schematic representation of the QMS signal during a pulsed molecular beam experiment on the gold flag/gold reference (a) and on the gold flag/reference (b)
calculating the difference QMS signals from sample and gold flag. However, the slightly diffe-rent position of sample and gold flag lead to diffediffe-rent scattering conditions for the two cases.
Therefore, a systematic error would be introduced when using this procedure. To take this into account, a separate measurement with a gold reference is performed [194]. The sticking proba-bility S versus the pulse number i can be calculated as follows:
Si=
Ri·t0+tend
i·t0 (Igold f l−Isample) Fcorr
Rtend 0 Igold f l
(5.2) Igold f lis the average intensity of all pulses on the gold flag andIsampleis the signal intensity on the sample. The correction factorFcorrincludes the above mentioned correction due to the diffe-rent scattering geometries during measurement with the gold flag and with the gold reference.t0 is the time between two pulses (typically 2000 ms),tend is the time until which the integration is performed. Accordingly, the sticking coefficient can be calculated versus the pulse number with Equ. 5.2 after integration over the QMS signals from sample, gold flag and gold reference.
For the choice oftend, two cases have to be considered. If the pulse shape from the sample and the gold flag are identical, as depicted in Figure 5.7a (a),tendcan be chosen arbitrarily as long as t0≥tend>0.
Figure 5.7b (b) shows a case, where the QMS signal shapes from sample and gold reference are Abbildung 5.6: Comparison of the QMS signal shape during a pulsed molecular beam experi-ment on the sample and on the gold reference: (a) O2 adsorption on Pd(111) at 300 K (b) CO adsorption on 1.5 Å Pd/Fe3O4at 110 K
(a) (b)
significantly different. Such a situation can arise, when molecules desorb in between the pulses from the sample. In the present work, such a behavior has only be observed for CO adsorption at low temperatures (≈110 K) and close to the saturation of the adsorption sites with CO.
Campbell et al. described a procedure in which they used two different procedures for the evalua-tion of the QMS data in an SCAC measurement in case transient adsorpevalua-tion/desorpevalua-tion occurs in the timescale of the pulse period. For the evaluation of the surface coverage, the sticking proba-bility for a given pulse is defined as the fraction of the molecules, which have not desorbed from
the surface before the next pulse starts to hit the surface. Accordingly, tend =t0. This sticking probability has been defined as the long-time sticking probability [58].
For the determination of the adsorption heat per mole adsorbed, it has to be considered that molecules initially adsorb on the surface and deposit heat. However, some of these molecules desorb later while removing heat from the sample. These desorbing molecules only influence the peak height if they desorb in between the beginning of the pulse and the time when the pulse maximum is reached, as only this time frame in the SCAC signal is used to determine the ad-sorption heat. Accordingly, integration of the QMS signal is performed from the beginning of the pulse until the maximum of the pulse,
tend=t0+pt (5.3)
where pt is the time until the pulse maximum. The sticking probability, calculated by this pro-cedure has been defined by Campbell et al. to be the short-time sticking probability [58]. As the effect of molecules, which desorb in between the pulses close to saturation cannot be fully ta-ken into account by determining the long-time sticking probability. Accordingly, an error in the determination of the number of adsorbed molecules is introduced. In that case, the determined number of adsorbed molecules is not accurate at high surface coverages where desorption of molecules between the pulses is significant.
The long-time and the short-time sticking coefficients coincide in the low coverage regime. At high coverages, however, the long-time sticking coefficient is lower compared to the short-time sticking coefficient due to desorption in between the pulses. An example is shown in Fig. 5.7, where the long-time and the short-time sticking probability for CO on 1.5 ÅPd/Fe3O4, measured at 110 K is plotted as a function of number of adsorbed molecules.
It should be mentioned, that Campbell et al. made an additional correction for the sticking pro-bability in case the line shape of the heat signal changes due to transient adsorption/desorption [58]. This sticking probability has been termed the weighted short-time sticking probability. As a change in the line-shape of the SCAC signal has never been observed in our experiments, this correction has not been applied in the current work.
Abbildung 5.7: Short-time and long-time sticking probabilities for CO adsorption on 1.5 Å Pd/Fe3O4 at 110 K as a function of the number of adsorbed CO molecules