The ability to mold materials into arbitrary nano-structures is one of the cornerstones of our society. To this end, as a society we have developed various techniques that have broadly been classified as top-down or bottom-up. Top-down approaches are used regularly to manufacture billions of devices, some with features as small as 7nms, in the microelectronics industry. The origin of many of these techniques can be traced back to the 1960s during the birth of the modern microelectronics industry. Although, today these techniques are being regularly used for making optical, micro-mechanical as well as biomedical device. Typically, the top-down process flow involves patterning a polymer resist with an optical mask or electron beam followed by pattern transfer via etching or material growth (Vision (top)). In a parallel effort, various bottom-up techniques have enabled synthesis of complex, atomically precise, nano-structures like quantum dots, carbon nanotubes and metallic nanoparticles (Vision (bottom)). In my opinion, all biotechnology is also at its very core bottom-up nano-fabrication and there we have made phenomenal progress like developing technologies necessary for creating synthetic DNA and designer proteins.
Although these two approaches have independently matured and succeeded in meeting various technological demands, it is becoming increasingly clear that emerging devices require a unified, ``hybrid'', approach. Fields such as quantum-information processing, large area metasurfaces, quantitative biology, bio-electronics and neural-interfaces demand unique capabilities, like spatial organization of individual molecules, which cannot be achieved scalably with any existing methods. This need has been well recognized in the nano-fabrication community for the better part of last three decades and there has been several different methods that have been proposed with varying degree of success. These can be broadly classified into three categories: (1) fully stochastic assembly, (2) fully deterministic scanning-probe assembly, and (3) directed self-assembly. In (1), fully stochastic assembly, bottom-up structures are either deposited or grown at random locations on a substrate or array of top-down devices. When top-down devices are predefined, the yield of functional devices, in which a single component lands within the top-down device, is limited to 37% due to Poisson statistic, but yields are much lower since precise location within the device itself is uncontrolled. Alternatively, via `select and post-process', randomly arranged components can be mapped using atomic force microscopy (AFM), scanning electron microscopy (SEM), or some high resolution optical microscopy and a device is built top-down around it. Scanning probe assembly (2) involves pushing components into a device using the scanning probe tip, one at a time. Neither of fully stochastic nor scanning-probe approaches can be scaled up, i.e. billions of devices can't be made using this technique the same way computer chips are made. In contrast, directed self-assembly (3) uses top-down lithographically defined growth sites or binding sites to localize components to microstructures with high probability, and is thus inherently scalable. But this method is not generalizable, i.e. even if directed self-assemble can be used to organize carbon nanotube, quantum dots or a particular protein that same method need not work for a different bottom-up component. And, this is a stark contrast compared to top-down techniques which are incredibly modular, for instance the etcher or photolithography resists that is used to make devices on Gallium phosphide can also be used to make devices silicon nitride or silicon. This modularity in top-down approach has enabled rapid device design cycles and ultimately the success of many industries.
In my opinion, the ideal "hybrid" nano-fabrication technique need, above all else, be compatible with the majority of existing bottom-up nano-structures and should use existing top-down nano-fabrication infrastructure. I define these condition because if the technique plays well with only one particular class bottom-up nanostructure it is unlikely to be successful since it severely limits design flexibility. Also, if the technique requires fundamentally new top-down tools then it's unlikely to be adopted since capital cost for top-down nano-fabrication tools are extraordinarily high and thus unlikely to be widely used. To the best of my knowledge, DNA origami is the only tool currently available that satisfies these condition and thus it represents
DNA origami is a self-assembly technique to synthesize arbitrarily shaped 2D or 3D nanostructures by folding a long single stranded DNA(ssDNA) "scaffold" using ~200 distinct ssDNA "staple" strands. DNA origami is particularly attractive due to the ease with which its precise shape can be programmed, its high yield, geometric homogeneity and the possibility of inexpensive bio-synthesis (one gram of DNA origami can be produced for $200). Further, since every part of the origami can be addressed, at ~5nm resolution with a unique DNA sequence, it can be used as a scaffold for organizing other DNA labeled nano-structures. This last ability is particularly powerful given the large catalog of existing methods available to label almost any inorganic or organic molecules, like proteins or florescent dyes, with arbitrary DNA strand.
My vision for using DNA origami as a unifying tool is illustrated in Fig 'Vision' : DNA origami will be used as a modular adaptor to Organize or Integrate individual or well-defined groups of bottom-up (self-assembled) nanocomponents on planar surfaces, before or after, building top-down nanodevices. Further, DNA origami could also be used form bio-inspired patterning of surfaces with structures at an extremely low price point in comparison to regular top-down patterning. My efforts towards making the above vision a reality has two complementary aspects: (1) investigating the fundamental processes involved in using DNA origami simultaneously with top-down and bottom-up nanofabrication and (2) engineering transformative devices using the new capabilities. I will currently focusing on the following directions:
These different directions share several aspects in common and working towards these goals will help develop a rich knowledge base for using DNA origami as the tool that unifies the two existing approaches to nanofabrication. Along the way, I also expect the development of a number of transformative nano-devices with relevance to quantum technologies, nano-optics, material science and biomedical sciences.
The work I have been doing over the past few years has enabled us to demonstrate that it's possible to position and orient DNA origami on a surface (or with respect to devices) using standard top-down nano-patterning. However, I believe the merger of DNA origami with top-down nanofabrication is much more powerful. I envision an entire suite of techniques for create unique structures. The cartoon on right summarizes key aspects of this toolbox and the different subsection are,
1. DNA origami placement (DOP)
In its current version DNA origami placement works with fairly reasonable yield of ~98%, i.e if you have a binding site at a particular location then there is ~98% probability that only one origami will bind there. While this is good for a large number of applications this value is ridiculously low compared to traditional top-down nanofabrication in which the average error rates are close to 1 in a billion. So, how can we improve the errors rates in DOP from ~1 in 100 to 1 in a billion. One possible method involves decorating the ends of the origami with polymer (maybe PEG or even just ssDNA) to sterically occlude other origami from getting close. This approach is thematically similar to the trick I used in the absolute orientation in order to break the up/down symmetry of the origami which suggests that there is a good chance of success, although the precise chemical composition and length of the polymer needs to be explored.
A second "problem" with DOP is it's reliance of e-beam lithography which is extremely expensive and can be very challenging to develop from scratch in a lab. So, it would be worthwhile to develop an easier approach and the best way to do this would be to use nanosphere lithography. It might be possible to create a regular pattern of binding sites using the monolayer of polystyrene beads whose footprints can be regarded as a polymer resist. So, the general idea would be to create a monolayer of beads on an activated substrate (treated with plasma or acid to make hydroxyl groups) followed by HMDS for passivation and bead removal. At this stage the substrate should behave like a normal placement substrate ready for DNA origami to be organized. This method would most easily give circular binding sites and hexagonal lattice, though there might be a room for some tuning of the lattice. There might also be a bit of tuning necessary for the type of passivation layer most appropriate for this method.
2. Orthogonal placement:
The current version of DOP involves origami of a single shape interacting uniquely with it’s partner binding site. However, it might be possible engineer multiple DNA origami-binding site pairs such that each origami interacts specifically with its partner binding site with minimal cross-talk. While this vision of orthogonal placement seems ambitious, an analogous process exists in nature. Specifically, natural protein-protein interactions have exquisite specificity and are highly orthogonal: a typical protein interacts strongly only with its binding partner, and very weakly with a background of 100s of different proteins. And, an important part of this specificity is 3D shape complementarity, i.e. the geometric fit between the surfaces of two proteins. While the development of these shapes could be done analytically a much easier option might be use the insights we gained in the absolute orientation paper which also provides a qualitative simulation tool to access how a given origami shape would interact with a binding site.
3. DNA origami liftoff
One of the potential devices that can emerge from having the technology to position individual molecules (or ions) is electronic or optical gated for quantum computers. This is because both tese applications require electrical or optical manipulation of the internal quantum state of a single excitation which is most easily attained with a single excitable entity. While origami placement could be used to immobilize molecules (or ions) at desired location, it can’t however be used for such applications in it’s current form. This is because placement will result in immobilization of not only the molecule (or ion/atom) of interest at a specific location but also a large amount of DNA in it’s vicinity whose presence will adversely effect the properties of the molecule. A similar argument could actually be made for almost all applications involving DOP though in many instance the presence of DNA origami might not be detrimental. Ideally, for such applications, it would be desirable to immobilize only the molecule, or more generally bottom-up nanostructures, at a specific location without any DNA in the vicinity. This can be achieved using “origami liftoff” technique in which placement is initially done using an origami carrying a special cleavable linker. After the placement the linker is covalently attached to the surface (Origami toolbox(c)(1)) and cleaved in middle to reveal a functional group that is now bound to the surface (Origami toolbox(c)(2)) without any direct bond to the DNA. Finally, the the DNA origami is removed (Origami toolbox(c)(3)) revealing just a single functional group on the surface, that can then be used as a binding site for the nanocomponent of interest, serve as a nucleation site or functional element itself. While I have already demonstrated covalent coupling of origami as well as removal of origami from the surface much work remains in designing an appropriate linker. I think the best first step would be to using a validate photocleavable linker but I think in the long run it might be more prudent to develop custom linkers.
4. Origami-directed etching and mineralization
One of the most attractive feature of DNA origami is the homogeneity, resolution and complexity of the synthesized 2D or 3D nanostructures which is comparable to lithographically attainable features. However, since DNA does not possess any technologically relevant optical, electrical or mechanical properties the origami can’t be used directly as a functional unit. The origami nanostructures, in this case, is analogous to features formed in polymer resist immediately after photo or e-beam patterning. Just like the DNA origami, polymer features by itself isn’t useful until top-down processes like etching, ion implantation or metal deposition transfers the geometric pattern onto a more functional material. Thus, if techniques existed to faithfully transfer the origami shape into inorganic material then we would be able to create 2D/3D nanostructures with features at higher resolution or more complex than what is achievable by conventional top-down techniques as well as create well defined nanostructures on the surface without any top-down patterning whatsoever.
While there are a myriad of different techniques one can develop to attain the stated goal. I think the following three are the most exciting and potentially most useful.
a. An extension of the established metal-assisted chemical etching (MaCE) technique would be very interesting. In this case origami would be first immobilized on a silicon (<110>) substrates (randomly or using placement) followed by metal (silver or gold) reduction and exposure to HF/H2O2. In such a setting the metal structure, formed in the shape of the origami, will catalyze the etching of the silicon underneath leading to creation of an origami shape aperture in silicon. One of the biggest potential drawback of this approach is the extent to which the native oxide will effect the etching as well as potentially move the metal nanostructure, if not remove them completely. It might also not even require metal reduction since there is some evidence that suggest just metal ions on the DNA might be catalytically active.
b. Converting the 2nm thick origami into a polymer pillar using atom-transfer radical-polymerization (ATRP), which is a standard polymer growth technique that utilizes a transition metal catalyst to create polymers of well-defined length on alkyl-halides initiators. Operationally, we will be creating origami with ATRP initiator (2-bromoisobutyrate) at the end of each staple, this could be achieved by attaching an initiator with a NHS ester moiety onto an amine terminated DNA that inturn would be bound to a staple extension. Then polymer growth is initiated, after immobilization on a substrate, using appropriate monomer and heat. During the initial phase it might be best to create PMMA pillars on the origami since PMMA is a standard lithography resist. However, during later stages one could use the proven modularity of ATRP to grow semiconducting/florescent polymer like poly-paraphenylene vinylene from the origami to act like wires. A related line of research would also be to use a similar method to create biocompatible polymer like NiPAM on the DNA origami so as to form nanoparticles of well defined shapes that can find utility for therapeutic functions. Incidentally ATRP on origami has already been demonstrated and while the work did not attempt to have a very dense group of initiators the results are nonetheless promising.
NOTE: There is another method to creating polymer pillars that I have described in more details within the large-area nanotexturing section.
c. Using the DNA origami as a solid carbon source in metal catalyzed graphene formation. The approach would involve evaporating nickel/copper a dielectric substrate (like silicon dioxide) on which origami has been immobilized. This would be followed by annealing in Ar/H2 environment at 1000c. While I am fairly certain that this will yield graphene bilayer in the shape of origami, the graphene itself is going to be doped with phosphorous and nitrogen from the DNA. It might be possible to get around this problem if a metal catalyst were one in which phosphorous, nitrogen and oxygen might dissolve. Another option would be to transfer the origami pattern onto a self-assembled monolayer (SAM) and then use the SAM as the carbon source for graphene formation. It would be really fun to apply this technique with origami structures with 2nm features which when converted to graphen (even bilayer) would display very interesting optical properties. Especially interesting would be to use 2D chiral structures that have been designed to break up/down symmetry, since it would allow one to create electrically tunable chiral surfaces. These structures would definitely form electrically tunable antennas in the visible/NIR spectrum. Further, since bilayer graphene is semiconducting the structures could also be used for electronic devices.