3-D Atomic Printers?
December 15, 2014, IEDM, San Francisco—Enrico Prati from Consiglio Nazionale delle Ricerche talked about atomic scale devices and the potential for 3-D atomic printers. The driver of Moore’s Law and the progressions of molecular-scale processing begs the question of the possibility for atomic-scale devices and the mechanisms to enable their creation.
Over time, normal functions seem to go from centralized to distributed like printing. Existing platforms using silicon and gallium arsenide seem to be on the path to finer granularity in processing down to the atomic level. The challenge is to work at a granularity not of a level, but at the atomic level. The challenges are that the forces and effects at the atomic level vary considerably from the macro-levels of atomic layers.
One issue is that at the short scale of atoms, electrons exhibit wave-like properties which violate the assumption that an atom is an intact whole. Obviously, the progressions make us consider a FINFET evolving to a single electron device. The problem is that we are starting to see quantizing effects in the FINFETs at 4° K. Early experiments are showing that silicon starts losing its bulk properties at the 3 nm range, and the electrons behave like waves.
The molecular-level behaviors start to disappear below 5 nm and regular shapes show energy levels with different spacing of the peaks and valleys of the sub-orbital states. The nature of the transport mechanisms changes and vary with the device dimensions. Quantum effects like spin and changes in the valley levels change the conductivity. As a result, the quantum effects only allow 3 electrons in two quantum dots or about 2.6 M qubits /cm2.
These conditions allow the pumping of single electrons, making transfers much like a CCD with the current proportional to frequency. It is possible to manufacture devices below 3 nm, but the devices will exhibit extremely high variability and will see extremely high parasitics in the wire transport layers.
Single atom devices will show atomic states that will approach a hydrogen spectrum. A single atom (ion) doping is possible with focused ion implants to achieve 30-80 nm precision, but the following thermal anneal process will cause a shift in position. A scanning-tunneling microscope has better spatial resolution but is very slow.
The quantum transport within a single atom will have to overcome coulomb and other blockages like valley blockage at the atomic scale as well as the issue of charge sensing of an individual donor. The quantum mechanical issues would allow a valley change-over switch to detect the “flavor” of an electron. The donor-band formation of 4 atoms at a temperature of over 8° K indicates a change from atomic to band-gap behaviors. One of the limits to scaling is the doping limit of a single atom.
The most viable alternative to a single atom device is directed self assembly. The advantage is that the processing can be done in parallel and still allow for sub-nanometer precision. This assembly is a 3-D process that allows for various control methods. The mono-layers allow for deterministic doping but also allow for oxygen and carbon impurities at 800-1000° C processing.
Supra-molecular chemistries in active and passive modes provide good control of the nano-crystals and wires. The nano-devices can be placed with phase-separated block copolymers by starting from a high resolution pattern. This technique is already used in colloidal quantum dots that are used in color displays.
The next generation of devices will bridge the lithography and self assembly to meet the sub-nanometer manufacturing requirements. The lithography will be used to establish the patterns and the active devices and connections will be self assembled. This process is probably as close as possible to the 3-D atomic-level printer.
