Predicting Nano and the things with it...
Descriptors and quantitative structure property/activity relationships (QSARs, QSPRs) are powerful approaches widely used in drug discovery to predict physical, chemical and biological properties of chemical compounds. Surprisingly however, these strategies have not yet reached the nanotechnology arena even though nanotechnology is used in '000's of products in the market and it's use increases. Without attempting to be comprehensive, the inventory of publicly available products currently using nanomaterials includes electronics, batteries, automotive products, paints and coatings, clothing, cosmetics, sunscreens, construction materials as well as medical products. Looking at this list you may wonder about the reason behind not having proper descriptors developed for nanomaterials. In accordance with the well-known QSAR paradigm, if these molecular parameters were either experimentally or theoretically calculated for a group of structures, then using well-defined mathematical models would enable materials scientists to forecast experimentally not yet available phys-chem or biological properties for compounds belonging to that group. This is based on a simple empirical hypothesis used in QSAR, namely that similar compounds act similarly.
Nanomaterials, for a number of reasons, are a special case though, which makes modeling them exceedingly complicated. At first, they are nothing like a well-defined organic molecule. They are extremely diverse and complex systems, and most of the time they’re described with unknown exact composition and with sizes up to 100s or 1000s of atoms, which limits the possibility of using quantum chemical ab initio calculations. Secondly, there still are lots of unrecognized barriers preventing the reliable and reproducible commercial synthesis of nanoparticles. And last but not least, the literature currently lacks systematic characterization standards that could apply to all the classes of nanomaterials, which is neigh on impossible to have, and this brings up the question of developing models for each known nanoparticle class (metals, metal-oxides, CNTs, QDots, etc.). The recently proposed list of descriptors for this purpose includes size, size distribution, surface morphology/topology, surface chemistry (capping agents, lipophilicity), composition, charge, distribution of atoms throughout the whole particle, “solubility” and the parameters of agglomeration. However, this list is probably far from being complete.
Although developing descriptors for nanomaterials is in a very early stage, things are already moving forward. Research groups are putting tremendous efforts into coming up with reliable nano-QSAR (QNAR) models. However, these are almost exquisitely designed for toxicity predictions. Of course biological effects on human health as well as on the environment are very important, this much is clear. You would want to know if the sunscreen you are using is more dangerous to put on than the UV radiation it is supposed to block, right? Although the results of the currently existing models are quite promising (e.g. Puzyn and his group developed a model that can successfully describe the cytotoxicity of oxide NPs to E. coli; Tropsha et al. showed that the cellular effects of 51 tested nanoparticles depend on the nature of the metal core rather than anything else)1,2, they are only built around relatively small sample sizes. Nevertheless, predicting nanotoxicity even for a small series of nanoparticles is a good first step and may invite attention from chemical agencies worldwide to complete their regulations, such as REACH (Registration, Evaluation, Authorization and Restriction of Chemical substances).3
Another attractive and regularly seen test case is the one of carbon nanotubes’ (CNT's). Essentially CNTs are rolled up graphene sheets that are built up from sp2 hybridized carbons atoms, and show extraordinary mechanical, electrical properties. Besides being one of the most exciting materials of recent years, they also possess all the above mentioned uncertainties when it comes to their synthesis, systematic characterization in literature and prediction of their important phys-chem properties. Imagine for a moment that you have a tool to predict the mechanical properties of the composite material you are about to make from your nanotubes just using one or two simple descriptors from experimental results or theoretical calculations (e.g. chiral angle or defect concentration). How much easier that could make the job of materials engineers! Things are happening in that direction as well.4
Not having enough proper data available for being able to build models, which is the major issue here, leaves a lot of room for further improvements, and this can only be done through effective collaboration and knowledge sharing. I just came across a research portal from The University of Arizona that is to bring experts together in order to solve problems associated with semiconductor manufacturing.5 Some more existing nanotechnology hubs can be also found in the Nanoinformatics 2020 Roadmap published by the National Nanomanufacturing Network.6 If you think we need more of these kinds of initiations to overcome the shortage of existing and useful experimental data around all classes of nanomaterials, or you already aware of such collaborations please share your thoughts with us.References used
- Tomasz Puzyn et al., Nature Nanotechnology, 6, 2011, 175.
- Alexander Tropsha et al., Combinatorial Chemistry & High Throughput Screening, 14, 2011, 217.
- http://ec.europa.eu/environment/chemicals/reach/reach_intro.htm Tammie L. Borders et al., Journal of Chemical Information and Modeling, dx.doi.org/10.1021/ci300482n
- http://www.internano.org/; doi: 10.4053/rp001-110413