AFSID project has ended on July 31st, 2011.

Upload the public final report here

Executive Summary

The FP7 STREP Project AFSiD “Atomic functionalities on Silicon Devices” has established a fruitful collaboration between a CMOS foundry (CEA-LETI-MINATEC) and six academic partners to design, fabricate and characterize state-of-the-art  silicon CMOS devices featuring radically new functionalities: single atom transistors (SATs) and metal-oxide-semiconductor single electron transistors (MOS-SETs).

The CMOS facility offered state-of-the-art top-down fabrication processes to fabricate several tens of thousands of samples with dimensions as low as 20×20nm². For comparison, the Bohr orbit of a donor electron in silicon is about 2 nm which is a substantial fraction of the smallest device size.  A wide range of designs were realized, from basic three-terminal transistors up to complex multi-gate and multi-channel structures. The whole CMOS platform capability was used for designing samples sensitive enough to manipulate just a few electrons, while staying very close to pre-industrial Fully-Depleted Silicon-On-Insulator (SOI) device processes .

At the scale reached the active region of the devices, made of ultra-thin 10 nm SOI, includes only few carriers and few implanted dopants.  The radically new regime where carriers can be controlled one-by-one or where the source-drain current is mediated through only one centered donor becomes then accessible.

For CMOS scaled down beyond the 22nm node, which will appear in the market within the next few years, statistical variability in the atomic-level device structure means the end of the standard design and function of field effect transistors.  But the AFSiD project has shown how to take advantage of devices at these severe limits to establish the proof-of-concept functions based on the electrostatic control of a single donor orbital and the associated discrete quantum energy spectrum.

A major contribution of the project is that, for the first time, a clear link between the transport spectroscopy performed at low temperature and the room temperature characteristics of silicon devices has been established. A striking correlation between the atomic position of shallow donors in the channel and the threshold variability of field effect transistors (FET) has been unambiguously established.  A room temperature selection of a Single Atom Transistor (SAT) based on simple automated current versus gate voltage tests has been done with very good correlation with the low temperature characteristics.

Furthermore, deterministic implantation by monitoring drain current modulation from ion impacts in custom designed devices has been demonstrated.  Associated with this has been the extensive study of implanted dopant activation and implant damage repair by laser annealing by a process compatible with the CMOS process flow.

Having firmly established both the SAT and MOS-SET concepts on an unprecedented large series of samples, the AFSiD partners have obtained several remarkable results:

The control of surface channels in silicon ultra-thin gated nanowires, the hybridization of these channels with donors in the body of the silicon and the control of two donors with 3 gates have all been demonstrated.

The control of single electron transistors with charging energy larger than 10meV down to a typical size of 20 nm has been achieved. These devices are robust, compact and can be further downscaled.  The constant progresses in downscaling CMOS devices can be readily integrated to implement even better MOS-SETs.

The MOS-SET concept has been extended to the realization of intrinsic or doped single electron transistors and to the realization of 2 or 3 coupled SETs in parallel or in series opening the way for large-scale integration.

The SETs have been used as charge detector to access the ionization of single donors and for several functions including proof–of-concept for an electron pump functioning as a latch switch for a full adder.  The SETs have also been co-integrated on chip with a FET leading to new read-out and control functions.  Isolated double quantum dot for charge quantum bits have been also investigated.

In summary the AFSiD project is a vivid example of a very beneficial collaboration between academic partners and the CMOS world.  New disruptive functions such as the large scale quantum computer proposed by B. Kane in 1998 have never been as close as in our good old CMOS switches.