We started studying the growth of semiconductor nanowires (NWs) in 2004. Our activity now extends from the in depth study of growth of basic mechanisms to the fabrication of NWs for applications (for which, see III-V nitride nanowiresand Nanowires for photovoltaics).
The growth mechanisms of semiconductor NWs constitute a rich domain of basic physical investigations, the understanding of which also conditions the controlled fabrication of NWs with predefined morphology, size, local composition and crystal structure and composition, as well as ordered arrangement, that are essential for applications. Rather uniquely, Elphyse and the C2N Materials department can combine growth proper, analyses, and modelling. Most of our studies deal with the Vapor-Liquid-Solid (VLS) growth of NWs of III-V materials. In this case, new monolayers (MLs) form one by one at the interface between the already grown solid NW stem and an apical liquid catalyst nanodroplet, using atoms transferred from the latter, itself fed by the vapor (for non-VLS growth, see III-V nitride nanowires).
A striking feature of the NWs of III-V compounds is that they frequently adopt the hexagonal wurtzite (WZ) crystal structure whereas, under any other form (bulk, 2D, 0D…), these materials (except the III-N family) form in the cubic zinc blende (ZB) structure. Early on, we proposed a kinetic explanation of WZ formation, based on the concept of nucleation at the VLS triple phase line, which has been widely adopted and is now confirmed experimentally. Equally important is that, depending on the nanodroplet contact angle, the apical NW-liquid interface may be either planar or present a peripheral truncation which oscillates in phase with the ML growth cycle. In 2024, we developed a kinetic description of this oscillation. NW polytypism remains a hot topic, the precise control of which will allow the routine fabrication of crystal phase heterostructures (associating two phases of the same material), which are intensely explored as building blocks of novel quantum devices. In situ experiments reveal an intimate link between polytypism and interface morphology.
To clarify and exploit this link, our current research combines experiments and theory. Using in-situ TEM, we investigate and control crystal phase switching via morphology by playing on growth conditions or applying an electic field. We work on transferring this knowledge to achieve the same control blindly in a standard epitaxy setup. Via thermodynamic and kinetic modelling, we aim at clarifying the subtle interplay between ML nucleation and growth and droplet and interface geometry, depending on growth conditions. Nucleation must be revisited when our applied electic field distorts the liquid droplet. Besides, we also study theoretically why polytypism is unfortunately so rare in VLS-grown Si or Ge NWs.
Beyond growth kinetics, we now concentrate on the statistical aspects of the growth of individual NWs. The control of composition and structure at the ultimate ML level cannot indeed neglect the stochastic nature of the nucleation events that trigger the formation of each ML. In this respect, NWs are a choice system for studying nucleation and growth from a nanosize supersaturated mother phase, which presents many specific features as compared with a bulk medium. Our initial post-growth studies of the chronology of single NWs are now largely superseded by in situ investigations at the sub-monolayer spatial and temporal levels.
For III-V NWs in particular, a key fact is that one NW component is present at very low concentration in the droplet. We now understand that, in a wide range of growth conditions and NW diameter, the amount of volatile group V atoms at nucleation is insufficient to form a full ML. This explains why the growth cycle of a WZ ML comprises three stages, (1) nucleation and growth of a fractional ML using all available atoms, (2) slow step flow (at the pace of vapor input) up to ML completion, (3) waiting time with droplet refill until next nucleation. We must assess how the fluctuations of the durations of these stages between MLs will affect the blind control of the number of grown MLs via growth time that we aim to implement.
To this end, we initilally calculated numerically and analytically the sub-Poissonian distributions of these durations in absence of group V desorption and assuming a perfect incomplete ML (IML) regime, whereby the group V amount at nucleation is insufficient for all MLs.Detailed comparison with precise in situ data yield likely values of key parameters governing nucletion and reveal some self-regulation, but why full agreement is lacking must be understood.
We started by lifting our initial restictions. Performing calculations at arbitrary temperature, we derived the domains of growth conditions and NW geometry where either the IML regime or the alternative mixed regime (with some MLs fully forming at nucleation) prevails. We found that a very broad range of growth conditions and NW diameter should produce a quasi-IML regime. An intrinsic self-regulation mechanism should then operate, whereby the ML cycle times are very narrowly distributed, despite the random fluctuations of the durations of the individual stages. Further work is needed to confront our theory to systematic in situ experiments and to identify and model other possible sources of randomness.
The investigation of the energetics and statistics of the ZB MLs is less advanced. A reason is that each ZB ML forms very quickly. We believe that, even if they have not yet been observed, the same three growth stages are present, but that stage 2 is made quasi-instantaneous by the formation of the truncation and transfer of its material to the ML. The corresponding kinetics have been described, but the statistics remain to be explored. Moreover, the quasi-perfect correlation WZ/planar and ZB/truncated that is frequently observed is not well understood. Alternative models of truncation are worth considering. Beyond the stable growth of the individual phases, the formation of crystal phase heterostructures requires studying the transition between the two, noting for instance that truncation then involves dissolving one phase to grow the other one.
NWs offer many prospects for growing material combinations inaccessible in other systems, such as joint axial and radial heterostructures or single-material crystal phase heterostructures. Despite the freedom granted by the NW geometry, the formation of high quality heterostructures in NWs raises issues commonly encountered in planar structures. In VLS-grown NWs, the sharpness of compositional heterostructures may be limited by the accumulation of the currently grown element in the liquid (reservoir effect). We have developed and compared thermodynamic and kinetic models of the formation of such heterostructures. This provides guidance to the choice of materials, catalysts and growth sequence and allows one ultimately to achieve interface widths of only a few MLs. The impact of the oscillating truncation on interface sharpness is an open question.
Strain relaxation in mismatched heterostructures is another issue. Our model predicting quantitatively the critical dimensions for axial heterostructures (below which no extended defect should form) has become standard and is widely used by growers. The literature about core-shell heterostructures being quite confusing, we carry out intermittent investigations of this case via numerical and analytical modelling.
People
Collaborations
