Imaging the way molecules desorb from catalysts

Unlike surface catalytic reactions, desorption has been thought as a relatively simple process consisting of a series of statistically independent events randomly and uniformly occurring over the surface: in this picture adsorbates take off as soon as their thermal energy exceeds the binding energy. Following Irving Langmuir’s mean-field treatment, the rate of recombinative desorption of adsorbed particles is thought to be proportional to the square of the coverage. However, several measurements contradicted this model. As a matter of fact, temperature-programmed desorption (TPD), the standard method for determining desorption kinetics, hardly ever shows pure Langmuir behavior. The exponent m in the desorption rate law, referred to as desorption order, is often a fractional number rather than an integer (m = 1 or 2); further, the desorption peaks in TPD exhibit unexpected widths or symmetries or are split into multiple peaks. In such cases, the desorption mechanism cannot be uniquely understood from TPD data. A detailed characterization of the complex microscopic mechanisms occurring into the adlayer is instead needed. In order to shed light on the microscopic origin of desorption, we have investigated oxygen on Ag(111) combining structure sensitive electron microscopy with TPD into the same experimental set-up, the low-energy electron microscope (LEEM) at the Nanospectroscopy beamline of Elettra.

Our experiments confirmed the well-known features of the oxygen (4x4) TPD spectra, exhibiting a double peak structure which becomes more and more pronounced upon consecutive NO2 adsorption - O2 desorption cycles. LEEM allowed us to correlate the TPD spectrum to the evolution of the O-adlayer morphology, which was imaged at 10 nm lateral resolution. As can be seen in the images in the right hand column of Figure 1, two different types of islands are visible, the darker ones being considerably larger than the bright ones, which exhibit a well-defined triangular shape. Actually, both islands exhibit the same (4x4) LEED pattern. The low and high temperature peaks in the TPD data, labeled “A” and “C” in the top panel of Figure 1, are associated to such island types. Interestingly, the larger islands appear only upon consecutive NO2 adsorption – oxygen desorption cycles and never break up into smaller domains. We  suggest that their fragmentation is hampered by small amounts of oxygen which are trapped in the subsurface region upon subsequent reconstruction/deconstruction cycles. By diffusing towards the surface, such oxygen heals  possible vacancies in the (4x4) phase. As a result, the perimeter to area ratio of the large islands remains low leading to a reduced oxygen desorption rate when comparing with the one belonging to the small islands. Further desorption of subsurface oxygen gives origin to the high temperature tail of the TPD spectrum, which increases with the number of preparation cycles.
The analysis of the LEEM data allowed us to determine the total area and perimeter of the (4x4) islands. As can be seen in the second panel of Figure 1, left column, the (4x4) island area determines the oxygen coverage; note that the temperature derivative of the coverage of the (4x4) phase, expected to be proportional to the desorption rate measured in TPD, also shows a double peak. 

Figure 1.  Top panel: TPD spectra of oxygen, exhibiting the double peak pattern (A-C). The corresponding island morphology, simultaneously measured in LEEM, is shown in the right hand column of the figure. The correlation TPD – LEEM is evident. In all images, field-of-view is 15 μm, bright field operation at start voltage of 19 eV. Second panel from top: the O-(4x4) coverage, obtained from area analysis of LEEM images acquired during TPD; third panel from top: derivative of the O-(4x4) coverage; bottom panel; variation of the island perimeter versus increasing temperature.

Similarly, the variation of the island perimeter exhibits a double peak with a maximum and a shoulder at exactly the same temperature as observed in the TPD spectra. By assuming that the oxygen molecules desorb from the edges of both types of islands, we could formulate a rate law to model the TPD spectra. Such desorption is most probably due to the reduced coordination of the O atoms at the edge sites, combined with the high recombination probability of the O atoms, which are on neighboring sites in the (4x4)-O structure. The calculated TPD spectra were obtained after numerical integration of the equations for the evolution of the oxygen coverage for the two types of islands with a desorption order of m=0.5 for small triangular islands (red) and m= 0.3 for large islands (orange). The resulting rate vs temperature curves were fitted to the experimental TPD data (see Figure 1, top panel), obtaining almost perfect agreement. Our model was thus able to quantitatively reproduce the observed desorption kinetics.
That desorption kinetics of adsorbed particles is affected by island morphology and that the simplistic mean-field treatment is incorrect has been a long-standing postulate. So far, it was not possible to prove this, because of the limited information provided by macroscopic kinetics. Here, by imaging the surface morphology using LEEM, we could provide direct evidence on how the desorption kinetics is impacted by the formation and annihilation of islands. 

Further info

This research was conducted by the following research team:

  • S. Günther, Chemie Department, Technische Universität München, Garching, Germany
  • T.O. Menteş, M.A. Niño, A. Locatelli, Elettra - Sincrotrone Trieste S.C.p.A., Trieste, Italy
  • S. Böcklein, Department Chemie, Ludwig-Maximilians-Universität München, Munich, Germany
  • J. WintterlinDepartment Chemie, Ludwig-Maximilians-Universität München, Munich, and Center for NanoScience, Munich, Germany.

Contact persons:

Sebastian Günther:  ,
Joost Wintterlin: 



 S. Günther, T.O. Menteş, M.A. Niño, A. Locatelli, S. Böcklein, and J. Wintterlin, “Desorption kinetics from a surface derived from direct imaging of the adsorbate layer”,  Nat. Comm. 5, 3853 (2014); doi: 10.1038/ncomms4853.
Last Updated on Thursday, 14 August 2014 13:29