Research on metallic nanomaterials attracts more and more attention since the last decade. Amongst them Reactive Multilayer Foils (RMFL) are of particular interest. RMFL are obtained by the superposition of very thin A-B films, where A and B are pure metals. The layer thickness typically varies between a few (3-4) to 100 nanometers and the number of layers may be extremely large, about 5000. Such multilayer nanofilms are now easily produced by a magnetron sputtering technique or high-vacuum electron beam deposition (for a relatively small number of layers). During the alloying of the elements and the formation of intermetallic compounds, a very high density of energy is release by RMFL. Once ignited, the solid-solid or solid-liquid reaction releases a large amount of energy in a short period of time (microseconds).
The highly reactive properties of RMFL give them many technological applications (joining, ignition, heating, hyperthermia for medical applications, …). As an example, such systems might be used for efficient joining within the micro and macro scale. This advanced technology can be used for welding metal glasses or soldering microelectronic components without thermal damage of the bonded components. Another example is the nanoenergetics-on-a-chip (NOC) technology which allows one to incorporate energetic nanomaterial into electronic devices. The recent technological advances have been accompanied by increasing production of RMFL and metallic nano-powders.
The control of exothermic properties of these systems requires a better understanding of their specific behavior. Both from experimental and theoretical point of view, a great effort has been made to explore their specificity. Molecular dynamics simulations and related approaches appear as a valuable tool for the nanoscale investigation of these materials. Even if MD stands out as a way of studying phase transformations, reactive and transport processes, it is nevertheless important to define the contours of its applicablity.
A puzzling question is to ask what makes the specificity of metallic multilayers compared to metallic nanoparticles. Metallic nanoparticles have also very interesting physical properties that often vary from the bulk material. Some of these properties, especially the increased reactivity, are due to the high surface area to volume ratio of nanoparticles. Fundamental aspects need to be addressed in the framework of thermodynamics of nanosystems and investigated by computational methods.
RMLF are particularly good candidates for computational Materials Science for the following reasons. One is that the time scale of the reactions and the length scales of the multilayer period match the capabilities of MD quite well. The second is that we now have experimental techniques such as DTEM that can probe the reactions in situ at these time (10ns) and lengths scales (10nm). The other is the fact that multilayers provide a model material for studying phase transformations under steep concentration gradients and very rapid heating rates. This includes both solid state and liquid state reactions.
The process of reaction propagation in RMFL has been described for the first time in 1996 by Barbee and Weihs . The reaction wave can be initiated by means of short-time local heating; after initiation, this reaction wave propagates over the entire film and does not need external sources of heat. The self-sustained character of the process, a drastic increase in temperature in the front, and almost instantaneous formation of solid products are characteristics of this phenomenon. The parameters of propagation of the reactive front for RMFL significantly differ from those of systems with a similar composition obtained by mixing of powders. For instance, in the case of NiAl system, the burning rate is as large as 10 m/s. Moreover the reaction can start at relatively low temperature (i.e. 600 K).
If we look at the layered structure of nanofoils, two situations can happen. The system is chemically stable if A and B are nonmiscible. Otherwise, A and B will first mix into a solid solution and then react to give intermetallics compounds. Since the material is made of alternating very thin layers, it may become highly reactive even at low temperature and the reaction proceeds quite rapidly. It is this property which makes RMFL well suited for welding and soldering heat-sensitive materials in a few seconds without damage. The performance and the efficiency of RMFL are directly related to the promotion or the inhibition of reactions such as compound formation, crystallization, or grain growth.
Both because of their practical and fundamental interest, the study of RMFL attracts more and more attention (for a recent review, see [2,3]).
To study the dynamic processes such as defect motion, nucleation and growth, and phase transitions, very powerful experimental techniques are now available. High-resolution transmission electron microscopy (HRTEM) can spatially resolve the nanoscale phenomena. But advanced techniques coupling "snapshot" diffraction and dynamic TEM (DTEM) allows one to image the transient structures with a 15 ns time-resolution [4-6]. Intermetallic formation reactions can be studied under rapid heating using time-resolved x-ray diffraction (TRXRD) using synchrotron radiation [7,8]. Differential scanning calorimetry (DSC)  and nanocalorimetry  provide valuable thermodynamic and kinetic data on phase transformations  at both low (1 K/s) and high (10^5 K/s) heating rates.
Molecular dynamics (MD) is a promising tool to study RMFL since the typical length scales (a few nanometers) correspond precisely to the ones accessible by the simulation. Moreover, a great effort has been made to develop interatomic potentials for binary metallic systems and their alloys [12,13]. However, the transferability of potentials to various local environments that can be encountered during atomistic simulations have to be enhanced to accurately describe the reaction to form intermetallics compounds during solid-solid or solid-liquid-reaction. Furthermore, only very few reactive potentials (e.g., comb, reaxFF) have the capacity to simulate the thermite reaction between a metal and an oxide [14,15].
Molecular dynamics has been used successfully to study the interface between crystalline copper and liquid aluminium , the demixing phenomena in NiAl nanosized particles , the reaction pathways in an Al-coated Ni nanoparticle , the melting and crystal growth in Al50Ni50 system  the sintering process and the alloying between Al and Ni nanoparticles  the exothermic alloying of Ni/Al multilayers induced by shock loading [21,22] or to determine the pressure-dependent melting temperature of Al and Ni .
This method has also been used to investigate the interfacial mixing behaviour in transition metal (Fe,Co,Ni)-Al multilayer systems during deposition  or to study the atomic scale structure of sputtered metal multilayers . Adsorption and penetration of Al and Ni atoms are investigated by first principles calculations, to determine the driving forces impacting Ni/Al interfaces produced during multilayer deposition .
Using an embedded-atom method type potential, the mechanisms of energy release in a reacting NiAl multilayer has been studied [27,28]. These last works open new perspectives in the understanding of basic atomistic steps associated with alloying reactions. It is now possible to link MD simulations and novel experimental techniques to studying irreversible formation reactions at unique length and time scales. The effects of steep concentration gradients and rapid heating can be simulated and characterized in these nanolayered films.