Surface Science - from static to dynamic surface systems

In this Surface Science educational series in two parts, you will be introduced to how it all began and  guided through key events and important aspects in the evolution from the early days to the mature area.

The series cover topics such as the bridging of surface physics and surface chemistry, the paradigm shift from the study of static surface systems to the exploration of dynamic ones, as well as the importance of the advances in theory without which the evolution of the field could not have happened. It will also guide you through the different analytical methods that are used in the surface science field, and dig deeper into some of the areas that have, such as biomaterials and nanotechnology.

The Evolution of Surface Science, part 2

This post about the dynamics of atom/molecule interactions with surfaces, is a natural extension of the previous post, part 1, which addressed static properties of surface systems. Static systems refer to systems not changing in time, while dynamic systems refer to systems where temporal changes are an inherent characteristic. In contrast to kinetics, which are descriptions using phenomenological rate constants, dynamics aim for a detailed atomistic description of all steps of the interaction. The relation between kinetics and dynamics is briefly articulated below and in more detail in a later post.

The presentation describes central concepts in molecule surface dynamics such as surface scattering, potential energy surfaces(PESs), energy exchange and transfer between a surface and an incident/desorbing molecule. It talks about surface corrugation, charge transfer and adiabatic versus non-adiabatic descriptions of the interactions. Specific surface-molecule systems are not covered, except for some examples that illustrate general concepts and types of model systems. Experimental methods are only briefly described and will be covered in later posts.

The early days of surface science focused on static properties

In the early development of surface science and particularly in surface physics, the focus was on the “static properties” of simple surfaces. It was challenging enough to describe the surface, with or without adsorbates, in a “frozen” situation, without involving temporal changes and dynamics.

For many years, the determination of structural aspects of adsorbed atoms and molecules, both as single adsorbates and as dense monolayers, and the associated electron structure, was a primary field. Bond positions of atoms, bond lengths, and vibrational properties were in focus together with the associated electron structure aspects, such as bond orbitals, surface band structure and charge distributions. This development was described in the first post in this series, The evolution of surface science, part 1, and more will follow in a later post,  where theoretical aspects will be described.  

More complex static systems

Today the position(s), orientation(s) and electron structure of simple adsorbates, like, e.g., CO molecules, on single crystal metal surface like Ni, Cu, Pt can be described in atomistic detail and with chemical accuracy in the energies. Also “simple” compound surfaces, like metal oxides, are fairly well described. Challenges of static systems still remain as the systems addressed become more complex, like molecules with many atoms or more complex compound surfaces. Oxides, carbides, polymers and “soft” surfaces, and biological surfaces like lipid bilayers are now in focus, as are larger, e.g., organic or even biological molecules. More about this in later posts, e.g., when I will talk about biomaterial surfaces and biointerfaces.

Dynamics is the studies of temporal changes

While the static approach addresses the frozen state, the dynamics concern the time evolution of reversible or irreversible changes of the surface system. Examples are adsorption processes and diffusion, reaction, and desorption of adsorbates. Theoretically, static systems can be described by solving the time-independent Schrödinger equation, while dynamic systems require the even more challenging task to solve the time-dependent equation.

Historical remark

Historically gas surface dynamics was “born” from two “parents”; molecular collision dynamics in the gas phase, an active field already during the first part of the 20th century, and the surface community, until then focusing on static surface systems (Part 1 in this blog post series). Both communities were in the 2nd half of the 20th century looking for new areas, and challenges for their studies and they merged their knowledge, techniques, and concepts in gas surface dynamics.  The molecule-molecule collision community had performed their studies by using one or two molecular beams to allow gas molecules to collide under well-controlled collisions in a vacuum, with prescribed translational and internal (electronic, vibrational and rotational) energies. This community was the main contributor of the molecular beam technique and post-scattering analytical tools, while the surface community’s main contribution was the surface knowledge, e.g., how to prepare and analyze surfaces and construct potential energy surfaces for molecule-surface systems.

Drivers towards studies of dynamic systems

One strong driver towards dynamics studies was the rich, purely basic research questions in this field; to just ask and learn how a molecule scatters against a surface, Fig. 1, and is caught or reflected by it. However, a major motivation was also the relevance for technically important surface processes and reactions like heterogeneous catalysis, dry etching, CVD (chemical vapor deposition) and oxidation. Knowledge about the static systems certainly provides significant insight also about dynamic systems, but it is insufficient to develop detailed and well-founded models and scenarios for systems developing in time. It is like comparing a few photographic snapshots (the static properties) versus a movie (dynamics) to describe a system developing in time.

Figure 1. The principles of a scattering experiment. The incident particles are prepared and characterized with respect to their energies (translational and internal) and angles of incidence. Molecular beam techniques and optical excitations are major preparative tools. The scattered particles are analyzed with respect to their energies (translational and internal) and angular distributions. Mass spectrometry and optical (laser) techniques are major analytical tools.

Moving on to dynamic interactions

The overall ambition in dynamics studies is thus to understand, at an atomic level, how things happen and evolve at the surface. Typical questions asked are: “How can a molecule that bounces towards a surface lose enough energy to stick to it?", "Which are the detailed steps involved in desorption of atoms and molecules from surfaces?”, and “What causes or prevents dissociation of simple molecules like H₂, CO, O₂ on different surfaces?”. “How do two different atoms/molecules react on a surface to create a new molecule, as in a catalytic reaction?” Some of the elementary processes in focus here are shown in Fig. 2, and for a specific system in Fig. 3. Adsorption, desorption and diffusion events of simple atoms and molecules, like noble gas atoms and H_2, CO, O₂, NO, became popular model systems for studies of the collision dynamics and associated energy and charge transfer events.

Figure 2. Examples of gas-surface interaction events referred to in the text.

How do molecules in the gas phase interact with a surface?

Qualitatively what we want to understand in gas surface dynamics are (Fig. 2) how molecules or atoms from the gas phase interact with surfaces. For example, how do they land on the surface (sticking, adsorption, chemisorption), exchange energy with it, and then (sometimes) leave it?

As an illustration of what dynamics address and contain, we imagine the space-time path of a CO molecule approaching a partially oxygen-covered catalyst surface like Pt (Fig. 3). First, it lands on the surface (sticking) by getting rid of some of its excess energy, then it diffuses over the surface, i.e., jumps between different adsorption sites, and eventually reacts with an adsorbed oxygen atom and thereby forms a CO₂ molecule. Along this path, we can identify many of the basic processes studied in gas surface dynamics; some displayed in Fig. 2. The first step is adsorption/sticking of the molecule (described below). It then most likely diffuses on the surface until it finds a pre-adsorbed oxygen atom (which was formed from a dissociated O₂ molecule) with which it forms a new bond, the CO₂-bond. At ambient temperature, CO₂ does not (in contrast to O atoms or CO molecules) bind strongly to the surface. It, therefore, maybe after a few diffusion jumps, desorbs (see also below) from the surface.

Figure 3. Illustration of the system CO + O₂ → CO₂ + O

Potential energy surfaces (PESs)

A useful concept and quantity in the discussion of dynamics is the potential energy surface (PES), Fig. 4. A point on a PES, e.g., for an atom on a real surface, corresponds to the total energy of the whole system at that point. In real space, the point is defined through all the relevant coordinates in the system and the total energy of the system at that point. For an atom, the coordinates are the atom’s distance from the surface and the x-y coordinates parallel to the surface. This gives a four-dimensional space for the atom-surface PES (xyz and energy). The x-y coordinates are necessary to define the atom's position in relation to the surface atoms in the unit cell. For a molecule, one must, in addition, define the positions of all atoms in the molecule. This means that the PES is multidimensional. For example, if a diatomic molecule rotates without changing distance from the surface, it still moves to another point on the PES. For a uniquely defined point one must specify both the azimuthal and polar coordinates of the molecular axis, and the intramolecular distance, yielding a six-dimensional PES space.

Figure 4. A one-dimensional cut of the 6-dimensional PES for a system like O₂ on Pt, where there exists a molecular adsorption well (minimum) in the entrance channel for the molecule to the surface, and then an activation barrier for dissociation of the O₂ molecule and final formation of two chemisorbed O atoms. This way of drawing the 1 D cut of the 6 D PES means that the coordinate r is rather complex – it is not just the vertical distance to the surface, z, but runs through xyz coordinates of the 6 D PES such that the local minimum O₂ (ad), the local maximum E^a and the absolute minimum 2O(ad) are visited. An O₂ molecule from the gas phase can undergo the following events as alternatives: (i) bounce back to the gas phase against the barrier E^a (if too little energy is dissipated in the region of the O₂ (ad) well, (ii) adsorb in the O₂ well and then (iii) either be thermally desorbed back to the gas phase or (iv) be thermally activated to pass the barrier E^a to form two adsorbed O atoms. (v) If the kinetic energy of the incident molecule is higher than E^a it can also go directly into the O(ad) adsorption well. Note; if the molecule is trapped in the molecular adsorption well O₂ (ad), and stays there for some time (length of time determined by temperature), it can perform a diffusive motion to many identical adsorption sites, before either desorbing or dissociating to 2 O(ad) (see Fig. 3).

From ground state PESs for static systems to excited state PESs for dynamics.

When addressing dynamic events on surfaces, the static ground state properties, i.e., the lowest energy state, is insufficient as a platform for the description. The static ground state is a (adiabatic) multidimensional PES, where the ground state is represented by minima. In reality, in dynamics, the system often evolves along PESs that represent excited states, rather than the ground state, for example, because of too high activation barriers between the excited PESs’ local minima and the absolute minima of the ground state PESs’. For practical time scales, the system under study may thus often be “locked” or semi-frozen in a metastable state. An example of the latter is the formation of thin, self-passivating oxide mono- or multilayers on Si and many metals like Al or Ti. Thermodynamically the whole metal piece should be oxidized, but the thin oxide scale of one or a few nm stops growing due to a too high activation barrier for atom transport and/or O2 dissociation.

Timescales

The timescale of a simple molecule scattering event, with no trapping or sticking, but just a single collision, where the molecule bounces back into a vacuum, is typically 10^-12 – 10^-13 seconds. This is the same as one vibrational period of a molecule in free space, and can be estimated by dividing the molecule’s surface interaction range, typically of order 0.1 nm, and a typical velocity in the thermal regime, 100 – 1000 m/s.  If the scattering instead involves temporal trapping on the surface, and then desorption, the time scale can be extended from many picoseconds to nanoseconds. Depending on the details of the scattering event, it can also be longer, and even go to infinity when the molecule is stuck permanently on the surface.

If a molecule has been thermally equilibrated with the surface, it can desorb due to thermal fluctuations, if the combination of temperature and binding energy to the surface are in the right range. The timescale for the desorption event, or surface "lifetime" of the molecule, can be estimated using the Arrhenius approximation (see below), where the time constant, τ^d, for desorption is the inverse of the desorption rate constant, τ^d = 1/k^d = (ν exp-E^d/k_BT)^-1 and where ν is the so-called preexponential factor, in simple cases of order 10¹³. E^d is the desorption energy, which is equal to the binding energy to the surface, if there is no activation energy for adsorption. For example, when the temperature is around room temperature, binding energy of a little less than 0.8 eV gives a lifetime of order 1 s. At 100 K higher or lower T, the lifetime is of order ms and essentially infinite, respectively.

Classification of gas-surface interaction events

For classification of gas surface interaction/scattering events one frequently uses the three classes I) elastic, II) inelastic and III) reactive scattering. In the former two, all chemical bonds are preserved, and new ones are not formed, while bond-breaking and bond-formation characterize class iii). The term “scattering” is used because in the more detailed studies one always uses molecular beams, where the beam atoms or molecules scatter against a single crystal surface (Fig. 1).

I) Elastic scattering

A “cleanest” example of elastic scattering is when He atoms scatter against a single crystal surface. A unidirectional and monoenergetic beam of He atoms scatters predominantly without energy deposition, or energy pick-up, to/from the surface. One branch of studies even focuses on diffraction of He atoms to extract structural information about the scattering surface (diffraction is inherently an elastic event). As a curiosity, even H₂ molecules may be diffracted by some crystalline surfaces. The strong elastic component for He and hydrogen scattering derives from the small mass ratio between the scattered atom and the surface atoms, making inelastic phonon dissipation ineffective, and from the weak electronic interaction (holds only on some surfaces for H₂) – it is like throwing a ping-pong ball towards a bowling ball. When the mass ratio is reversed, it leads to stronger inelastic scattering involving phonons (see below).

II) Inelastic scattering

In inelastic scattering, energy is deposited into the surface and/or picked up from it. Possible energy conversion events are shown in Fig. 5. For atoms, a typical example is heavier noble atoms than He, e.g., Ar, and especially Xe, where kinetic (translational, T) energy of the incident atoms can be transferred to, or picked up from, the phonon bath (vibrations, V) of the surface atoms. If the incident particle is a molecule, additional energy conversion/dissipation events are possible like rotational (R) and vibrational (V) and even electronic (E) excitations. In the latter cases, the molecules in the molecular beam are prepared with specific vibrational, rotational or electronic excitations.

On the detection side, after scattering, one measures the changes in these vibrational-rotational-electronic quantum numbers by laser spectroscopy and translational energy changes by pulsed/chopped beam techniques. Large sets of data are obtained by varying the initial vibrational-rotational quantum numbers, translational energy and angles of incidence. With theoretical (molecular dynamics) tools, one then tries to make models of the surface PES and of how all the scattering events (elastic and inelastic) take place.

Figure 5.  Possible energy conversion events in molecule-surface scattering events.
T: translational energy of incident or scattered/desorbed atoms or molecules.
V: Vibrational energy of incident or scattered molecules or vibrational excitations in the surface phonons.
R: Rotational energy of incident or scattered molecules.
E: Electronic excitations in the incident or scattered atoms/molecules and/or in the surface region (e.g., electron-hole pairs).

Surface corrugation

Especially in connection with elastic and inelastic scattering the concept “surface corrugation” is often used. This concept describes how ‘corrugated’ the energy landscape of the PES under study is. As we learned above, a point on the PES is the total energy of an atom/molecule at a specific point in real space. Corrugation refers to how much this energy varies as we move over the PES, or more correctly how much that energy varies over the surface unit cell. An extreme case with low corrugation is He atoms on a metal surface which is very ‘smooth’ with little energy variation as the He atoms explore the surface unit cell. He atoms mainly interact (repulsively) with the tails of the metal valence electron wave functions spilling out into vacuum (although there is an attractive van der Waals component): The repulsive “wall” has only small variations in charge density along the surface. More polarizable noble gas atoms and especially reactive atoms/molecules can in contrast exhibit very large corrugation.

III) Reactive scattering

In reactive events chemical bonds are broken and/or new ones are formed. One of the most widely explored sub-class here is catalytic reactions of the formal type A + B → C where A, B are the original reactants and C the reaction product. The catalyst surface enhances the reaction dramatically, without being changed itself. But there are also other types of reactive events, e.g., reactive etching where A species from the gas phase may react with a surface atom, B, to form a gas phase product C, thereby removing surface material (often chemically and/or spatially selective).

Experimentally reactive scattering can be performed with one or two molecular beams (MBs). In the former case one typically first deposits some predetermined coverage of atoms/molecules, say reactant A, on the surface. Then one directs molecular beam molecules of B to the surface and detects the product molecules AB for varying incident MB parameters and for different surface temperatures. One may also detect the velocity and angular distribution of product molecules AB.

The experiment may also be reversed, i.e., by exchanging A for B. With two molecular beams, there are more opportunities for variability in the experiment. One example is that a steady state can be established with two beams, where the surface coverage stays constant, but the reaction goes on continuously. This is often made at different ratios of the A/B fluxes. One may furthermore catch reaction events involving transient states of reactants on the surface, like molecules in precursor motion.

Dry etching

Dry etching reactions constitute a sub-class of reactive scattering. Dry etching is a method used extensively in semiconductor surface processing in combination with photolithography for microelectronics. For example, a silicon surface may be dry etched with fluorine-containing molecules. A typical example is XeF₂ etching of the Si surface by forming SiF₄ which desorbs from the surface. The net result is a removal of one Si atom from the surface from two incident XeF₂ molecules. In practice, the etching is performed in a plasma etcher, but MB experiments can help elucidate the reaction mechanisms.

Additional aspects of dynamics and special cases 

Above I have outlined the essence of gas-surface dynamics. There are some additional concepts that are common in dynamics and which I briefly define and comment on below. It is also appropriate to articulate a few special cases like ‘sticking’ and thermal desorption. Let’s begin with ‘sticking’.

Sticking

The ‘sticking’ of molecules and atoms on surfaces is a central concept in surface reactions. Sticking is essentially synonymous with “adsorption” as a process and leads to adsorbed/chemisorbed molecules.  It connects kinetics and dynamics via the adsorption rate constant k^a. The kinetic equation (no desorption) for first-order adsorption (commonly seen for adsorbing atoms) is

dθ/dt = k^a (1-θ(t))                                     (1)

where θ(t) is the surface coverage. k^a is the product of the gas impingement rate (gas pressure) and the sticking probability, s₀, on the clean surface. k^a contains via s₀ (in an embedded form) all the atomistic details of the scattering event when a molecule hits the surface.

The simple solution to the Eq. 1 is

θ(t) = 1- e^-t/τ                        (2)

and τ=1/k^a

The solution describes how an initially clean surface, exposed to a constant flux of gas species, fills up to saturation in the same functional way as a capacitor is charged with a constant voltage. Higher gas flux and higher values of s₀ shorten the time scale τ to fill up the surface with adsorbates. Typically, the time constant for filling the surface is of order a few seconds at 10^-6 torr pressure and s₀=1. First order adsorption is often called monoatomic adsorption but can also apply for simple molecular adsorption. If the type of adsorption is instead second order, common for dissociative adsorption of a diatomic molecule, we have dθ/dt = k^a (1-θ(t))², with a somewhat slower filling up of the surface compared to 1^st order adsorption.

Sticking coefficient measurements have been subject to many studies over the years. Methods range from surface spectroscopic or desorption measurements of the coverage θ(t) for different gas doses to the surface, to sophisticated MB methods. Depending on the studied system the sticking coefficient can vary from unity, meaning that every molecule/atom striking the surface sticks to it, to many orders of magnitude lower values.

Numbers lower than unity are indicative of the existence of an activation barrier for adsorption. In such cases it may be elucidating to measure s₀ at different incident energies of the molecule; an activation barrier may prevent slow molecules from sticking but not sufficiently fast ones. An activation barrier may also be overcome by internal vibrational or rotational energy of an incident molecule, due to concerted adsorption and energy conversion events (Fig. 5).

A smaller than unity sticking coefficient can, however, also be seen without an activation barrier due to inefficient energy transfer. This happens when the molecule/atom does not dissipate energy sufficiently fast in one single scattering event, to be trapped by the surface and therefore returns to the gas phase.

Thermal desorption

A sub-class of experiments in this area, not requiring molecular beams, but just a gas inlet to adsorb molecules on a surface, focuses on the desorption of molecules or atoms. The experiment starts with a surface partially or fully covered with an adsorbate that stays on the surface indefinitely, unless some external perturbation causes them to desorb. The most common perturbation is an externally applied temperature increase, leading to what is called thermal desorption, TDS or TPD (thermal desorption spectroscopy and temperature programmed desorption, respectively). Typically, in TDS, the surface is subject to a linear increase in temperature and the flux of desorbing molecules are recorded in real time by mass spectrometry until the surface is empty. By analyzing the recorded thermal desorption spectrum with some kinetic model, e.g., the Langmuir model, it is possible to determine the adsorption energy of the initially adsorbed molecules. The kinetic equation for 1^st order adsorption reads

dθ/dt = – k^d θ(T(t))             (3)

where k^d is the desorption rate constant, and

 k^d=ν exp(-E^d/kT)                  (4)

E^d is the desorption energy, which for non-activated adsorption is essentially equal to the molecule’s binding energy to the surface. ν is the so-called preexponential factor, which in simple absolute rate theory has a value of order 10¹³.The applied temperature rise is an experimental parameter and enters as T(t) = ct, where c is the heating rate, and where k^d (T)  as we see contains the binding energy to the surface.

For diatomic molecules that dissociate upon adsorption, like O₂ on many metals, the desorption is frequently found to be 2^nd order in the coverage, i.e., dθ/dt = – k^d θ²(T(t)). The 2^nd order dependence indicates that two neighboring atoms are required to form a diatomic molecule that desorbs.

More about energy dissipation and energy exchange

A molecule outside a surface, to which it eventually binds, constitutes an excited state of the system. When a molecule-surface bond is formed, a lower energy state is formed, and in the process, the initial higher energy of the system must in some way be dissipated.

The candidates for energy dissipation are many (Fig. 5). A strong candidate is the phonons (lattice vibrations) of the solid surface. As, e.g., a heavy and energetic Xe atom hits a surface, a major energy dissipation channel is phonons. Since the energies involved are typically much larger than that of single phonons, the dissipation often involves cascades of phonons. 

Another surface candidate is electronic excitations, i.e., electron-hole pairs. It is less common, but increasingly more important, as the mass ratio of incident particles and surface atoms get smaller, making phonon excitations less effective, e.g., hydrogen molecules on transition metals.

Yet another example (reactive interaction) is when electronegative molecules, like Cl₂, hit an electropositive (low work function) surface. Then, electron transfer can occur from the surface to the (electron affinity level of an) approaching Cl₂ molecule and lead to electron-hole pair creation as an energy dissipation channel (see more about this case below). The lifetime of electron-hole pair excitations is extremely short for metals, 10^-14 – 10^-15 s. Phonon lifetimes are one to two orders of magnitude longer.

But the surface excitations (phonons and electron-hole pairs) are only one set of energy dissipation channels (Fig. 5) associated with the surface. For molecules, there are also the internal molecular excitations that can pick up or give away energy, namely rotations and vibrations. In reality, a full scattering/sticking process may involve many of these elementary excitations and energy conversion events (the arrows in Fig. 5), simultaneously or sequentially. The lifetime of molecular vibrations at surfaces are in the range from tens of picoseconds to fractions of picoseconds -  orders of magnitude shorter than in the gas phase, due to the efficient channeling of energy into phonons or electron-hole pairs.

It is also possible that a molecule with excess kinetic, i.e., translational energy, collides with the surface and converts translational energy to internal, rotational or vibrational energy of the same molecule, which then bounces off the surface after a translational energy-to-internal energy conversion. Fast incident molecules can thus scatter on the surface back into a vacuum or gas phase, with much higher internal vibrational-rotational excitation than in the incident molecules (and lower translational energy).

The discussion above can also be applied to the reverse process to adsorption, i.e., desorption. This step, when a molecule or atom leaves the surface, requires energy pick up from the surface (from its heat-bath of phonons, electron-hole  pairs...) to convert the lower energy state of a bound molecule to a free molecule in the gas phase.

Adiabatic vs. non-adiabatic processes

Another aspect of the scattering and energy dissipation, that has generated much interest and work, is whether the energy dissipation in e.g., an adsorption event is adiabatic or non-adiabatic, which in turn affects on which PES the adsorbing particle moves.

If energy dissipation happens gradually, like in a friction process, and the energy is continuously dissipated as very low energy quanta, e.g., as phonons, i. e. heat, we call the energy dissipation and adsorption process adiabatic. However, in many cases, the system does not dissipate energy adiabatically but in larger, sudden (quantum) energy release events, like a photon, or an electron or a cascade of phonons. Then we call the energy release non-adiabatic. In PES language the latter event corresponds to first moving on an excited state PES and then making a sudden transition to a new, lower-lying (in energy) PES with a large change in energy. If the system adiabatically dissipates excess energy, the trajectory takes place on the ground state PES. In rare cases, non-adiabatic events can be detected experimentally as excited electrons or photons (electron-hole pairs, exoelectrons and surface chemiluminescence) emitted from the surface. 

Gas surface dynamics and scattering event analysis

The scattering processes mentioned above are preferably studied with molecular beams. This allows well-defined translational and internal energies of the molecules in the beam. The outcome of such collision and energy conversion/dissipation events are studied by analyzing the scattered molecules with respect to angular distributions, translational energy and internal energy, and also by measuring the fraction (if there is one) of molecules stuck on the surface. Of course, also the parameters of the incident molecules (angle of incidence, translational and internal energies, etc.), are varied.  The collective name of this area became gas-surface dynamics.

The scattering events take place in a multi-coordinate space. For example, to uniquely define the scattering of a two-atom molecule against a single crystal surface several aspects must be defined; (i) the direction of the molecule i. e. the azimuthal and polar angle of its trajectory to the surface, (ii) where does the molecule hit the surface in the surface unit cell (e.g., on top of a surface atom, or in the hollow between three or four surface atoms), and (iii) what is the rotational orientation of the molecule, and what is the vibrational phase. To develop a theory for the scattering event, the calculations need to include all these coordinates. Since all of them cannot be measured and some are averaged over all possible values in the experiment like the rotational orientation and vibrational phase, a challenge is to use all experimentally accessible information about the incident beam and the scattered molecules to learn about the scattering details and participating the PESs.

Charge transfer

When a strongly electronegative molecule like Cl₂ approaches a metal surface with low work-function like potassium, the electron affinity level of Cl₂ shifts downwards in energy due to so-called image interaction. As a result, an electron can jump or tunnel from the potassium surface to the Cl₂ molecule, forming a transient Cl₂^-   molecule (the affinity level is the lowest unoccupied electron level in the Cl₂ molecule, i.e., where the weakest bound electron of a Cl₂^-   sits.) As a result, the Cl₂ molecule gets stuck to the surface and dissociates to form two KCl bonds on the surface.  The timescale for the involved processes are in the range tens of fs to ps.

If the incident particle is instead very electropositive and the surface has a high work function, like K or Na incident on Pt or W, the reverse process occurs.  Electrons tunnel from the alkali atom’s valence orbital into the valence band of the metal, and leads to a process called surface ionization, where Na⁺ or K⁺ ions leave the surface. These examples are extreme cases of charge transfer. Similar charge transfer can occur more smoothly involving fractional charge transfer and orbital mixing (surface and molecular orbitals) for other systems, e.g., for oxygen molecules on metal surfaces. Fractional charge transfer means that an electron orbital is partially occupied, partially empty.

Precursors

In the context of adsorption/sticking, precursors and precursor motion are frequently appearing concepts. This usually refers to a motion of a molecule/atom just after it has struck the surface, but before it has dissipated its excess energy and come to rest and equilibrated with the temperature of the surface (Fig. 2c). The motion on the surface is driven by the adsorption energy and occurs before that energy has been fully dissipated.

There are several possible mechanisms for precursor motion. The molecule may in the first encounter convert some of its momentum normal to the surface into parallel momentum (elastically) or just dissipate that part of its energy (inelastically), resulting in a ‘hot’ precursor moving along the surface with higher than thermal energy, but trapped by the surface interaction. During the parallel motion, which occurs on the corresponding PES, the excess energy may be successively dissipated until the molecule is trapped in some local or absolute minimum on the PES and just becomes an adsorbed and thermally equilibrated molecule. The motion is similar to diffusion, but the latter is purely thermal in nature. A hot precursor may visit parts of the PESs not accessible via thermal diffusion.

An alternative outcome of the “hot” precursor motion on the surface is that the precursor regains energy and/or vertical momentum and scatters back to the surface. This is sometimes referred to as a trapping-desorption channel (Fig. 2f).

Another type of precursor, is when the adsorbate partially covers a surface under study and an incident molecule of the same kind lands on an area already occupied by the adsorbate. The incident adsorbate may then perform a similar parallel motion on the surface on top of adsorbates, as described above, until the molecule finds an empty site and adsorbs there (or desorbs before that). This type of precursor is sometimes referred to as an extrinsic precursor, while the type of precursor discussed above may be called “intrinsic” precursors..

Photon and electron induced dynamics

When a surface with adsorbed atoms/molecules is irradiated with electrons of a few or a few tens of eV, i.e., a little more than the surface chemical bond strengths, it may lead to a special type of desorption, called photo-induced (PID) or electron stimulated (ESD) desorption. The mechanisms involved in such events are almost always electronic excitations, either in the adsorbed molecule itself or, more common, in the surface or in the molecule-surface bond.

A common case is excitation of an electron-hole pair in the valence band of the surface. Typically, the initially excited hot electron moves to the adsorbate and into an antibonding orbital of the adsorbate – surface complex, which in turn causes the adsorbate to be repelled from the surface into a vacuum. Alternatively, it is instead the excited hole state that moves to the adsorbate and empties a bonding orbital, thereby causing desorption.  Similar events may lead to dissociation of adsorbed molecules, which in turn may cause a surface reaction, if there are other adsorbates on the surface (see below for the CO + O₂ case on platinum). PID studies are conceptually important also because they elucidate processes occurring in photocatalysis, of interest e.g. in photocatalytic decomposition of water to H₂.

A frequently studied model system

One of the most studied model reactions in the reactive scattering category is CO reacting with an oxygen atom, the latter derived from a dissociated O₂ molecule (Fig. 3) or NO₂,. A major reason for the interest in this system is that it involves key reactions in automotive emission cleaning catalysis. Key reaction steps are reactions with CO and NO or NO₂, in the presence of O₂. Let's, therefore, return to the CO + O₂ catalytic reaction on platinum (Fig. 3), which we briefly touched upon several times above.

At typical catalytic operation temperatures, say 150 – 700 C, the reaction proceeds by simultaneous adsorption of CO molecules and O₂ molecules, where the latter dissociate. CO molecules stay intact on the surface with stay-times of a tenth of a second (at the lowest T) to only around a ns at the higher temperatures, before they desorb again. During this time, they diffuse rapidly on the surface, faster the higher the temperature is. The O₂ molecules dissociate and form adsorbed O atoms on the surface, via the molecular intermediate state shown in Fig. 4. These O atoms are found by the diffusing CO molecules, and the two react to form CO₂ molecules (the mobility of O atoms is much smaller than that of CO). CO₂ molecules are by comparison very weakly bound to the surface and desorb almost instantaneously.

The CO+ O₂ reaction at low temperature

This simple scenario (previous paragraph) hides some essential details of the reaction, which are discovered if one runs the experiment at a much lower temperature and in a slightly different way. When CO and O₂ are co-adsorbed on Pt at sufficiently low temperature, typically well below 150 K, both molecules stay intact and non-dissociated on the surface, and they do not react. If the surface temperature is then raised monotonically with co-adsorbed CO and O₂, one suddenly sees CO₂ molecules leaving the surface already around 150 K, indicating reactions between the absorbed CO and oxygen. However, only a fraction of the CO molecules reacts at this temperature, and a new desorption flux of CO₂ is seen when the temperature reaches around 330 K. The latter is caused by the same reaction as was just discussed above between adsorbed CO and O atoms.

The actual events that occur are that the low-temperature formation of CO₂ is caused transiently by dissociating O₂ molecules (thermally activated dissociation). Some of the O atom dissociation products react with neighboring CO molecules and some just form adsorbed oxygen atoms as in Fig. 3. The latter cannot react with remaining CO molecules on the surface, because of a rather high activation barrier, until the temperature reaches around 300-350 K. The latter is the ‘normal’ CO oxidation reaction on Pt, with which this paragraph began, and which has been studied extensively since the days of Langmuir. The former reaction occurring at around 150 K is in a way more exotic, but mechanistically interesting, and reveals that there is a stable adsorbed molecular O₂ state that is permanently occupied at sufficiently low T and transiently at high T. This state has been spectroscopically verified.

Photoinduced CO + O₂ reaction

If in the latter experiment, with co-adsorbed O₂ and CO molecules, at < 150 K, where they don’t react, the temperature is not raised, but instead a flux of near UV photons is sent to the surface, CO₂ molecules are also formed, without any T increase. The mechanism of this reaction is first a photoexcitation of electron-hole pairs in the Pt surface, followed by (hot) electron transfer to an antibonding orbital of the adsorbed O₂ molecule, forming a transient O₂^-, thereby inducing its dissociation to O atoms. The latter helps to get over the activation barrier in Fig. 4. Some of these O atoms bounce into and react with neighboring CO molecules, and form CO₂, in much the same way as in the TPR experiment.

In PES language, this experiment shows that there exists a ground state PES for O atoms with a deep minimum corresponding to chemisorbed O atoms, and an excited state PES for O₂, with a shallow minimum for O₂. It also shows that there is a local minimum for an adsorbed O₂ molecule with an activation barrier towards dissociation

Concluding remarks

My ambition with this post has been to present an overview of the field called gas surface dynamics, its general concepts and phenomena and what this field is all about. I have covered scientific questions and types of studies embraced by this area. Detailed descriptions of specific model systems are beyond the scope of this post. For example, the descriptions of adatom or admolecule motion to, and on, a surface, which is dealt with in gas – surface dynamics, is also highly relevant for thin film growth by physical vapor deposition (PVD) or chemical vapor deposition (CVD), but these areas have not been covered.

As also mentioned in post 1 on static systems, I want to emphasize the important role played by theory, both for interpretation of experiments, for developing concepts and understanding and for planning experiments. The theory aspect will be discussed in a later post, as will the relevant and extensive topic of experimental methods be.

Finally, as was indicated in the introductory part of this post, dynamics of molecule-surface interactions are very far from atomistic detail for more complex systems. Let’s take as an example protein-surface interactions. Even if we choose the simplest possible, yet relevant surface, like graphite or gold, and the simplest possible protein, we are still very far from an atomistic description. The latter would require describing a very complex and multidimensional PES on a time scale from picoseconds to seconds or even minutes, keeping track of all the 100ds or more atoms in the protein and how they (re)orient, bind to the surface, break or form bonds etc., creating a full molecular-dynamics ‘movie’ of how the protein binds, orients, unfolds, (partially) denatures etc. Although we are very far from such descriptions it is the direction of future studies.