Nano field effect transistors as basic building blocks for sensing

SiNW FETs and CNT FETs have been studied as candidates for integrable
biosensors. Both types show sensing capability, CNT FETs even for
uncharged molecules. The “top–down” process developed for SiNW FET
fabrication gives much freedom in device designing and is fully compatible
with CMOS production schemes. Due to the fact that the SiNW fabrication
reported here uses the same structuring techniques as in CMOS
processing, it is very reliable. The fabrication of CNT FETs is based on
CVD. By this process CNT FETs with interesting device parameters can
be reached. Since CVD needs high temperatures it is not compatible with
CMOS techniques. CVD growth leads to a mixture of CNT types which
results in a low device yield. The contact regions between the nanoscale
CNTs and fabricated metal contacts are very critical. Hysteresis, device
break down due to metal etching or chemical disruption and instability
in liquid environment have been attributed to the Pd–CNT contacts. It
was shown that the degradation of CNT FETs can be recovered. With
this both device types show good long term stability. The concept of
combining the electrolyte–gate with the back–gate, presented in the example
of pH sensing on SiNW FETs, enables stable performance in liquid
environment.
Based on our current experience we propose the following improvements
regarding future device design:
SiNW FETs:
• Improvement in sensitivity could be reached by removing the top oxide
locally and by reducing the nanowire width. An optimal width
needs to be found for high sensitivity and low device to device fluctuations.
• A higher doping, up to ~ 1018 cm−3, would enhance the contact
properties and lower the threshold voltages. But as the screening
length LD is reduced, the nanowire width has to be reduced as well.
• Control of the nanowire surface by a thin oxide layer or covalent
functionalization could help for stabilization and would allow specific
sensing.
• Differential measurements between a sensing and a reference SiNW
FET would allow stable performance.
CNT FETs:
Especially with respect to stability and to reproducibility further improvements
need to be done. Some possible solutions are:
• Humidity free fabrication and annealing before the deposition of the
metal contact.
• Titanium as contact material. Annealing leads to titanium carbide
contacts [32] which are less transparent, but expected to be more
stable.
• Contact passivation has to be very robust and innert. Therefore
we suggest to cover the contact region with silicon oxide or silicon
nitride.
• Increase of the hydrophobicity of the SiO2 surface by a monolayer
of alkane ended molecules.
Due to the improved stability in fabrication and performance reported
here and with the further improvements suggested above, we can think
about possible applications of SiNW FETs and CNT FETs and give an
outlook on possible future developments.
The high intrinsic sensitivity of CNT FETs is of interest in real time monitoring
of minute amount of molecules — or ultimately single molecules
— without labelling. In contrast to conventional sensing techniques, such
as fluorescence microscopy, the devices can easily be integrated on a micro
chip which is of special interest for in vivo experiments where access
with optical instruments is limited. Need in current basic research of biophysics
can be met this way. However the limited reliability in fabrication
and performance of CNT FETs hinders their integration into complex,
highly integrated devices or their implementation in industrial products.
In contrast to this SiNWs are well suited in this respect. The fabrication
is entirely CMOS compatible and very flexible in design. Devices
proved to be robust under the conditions of all performed experiments.
Depending on the application of interest the sensitivity can be chosen by
adjusting the NW size. The SiNW FETs introduced here can be used
as transducers for various sensor types. Implementing on–chip micro pH
sensors would be the most direct way to an industrial product. For this
purpose, two different surface types are needed: a pH–sensitive surface
for the working FET and a pH–insensitive surface for a reference FET.
Going beyond this into the field of biosensing, specific receptors have to
be immobilized. For multiplex sensing, each SiNW in an array has to be
functionalized individually. This can be realized by techniques such as microcontact
printing [149], dip-pen- [150] or conventional lithography [151]
combined with UV assisted Si-H chemistry [152] or by electrical addressing
of individual NWs, for example by electrochemical [153] or local heat
assisted [154] Si chemistry. In a next step the SiNW sensor array has to be
combined with an electronic read out system — preferably implemented
on–chip. Further on such a multiplex sensor device can be integrated in a
micro fluidic system. Complex micro fluidic systems including actuators
such as pumps and valves, separation and mixing chambers, and liquid
multiplexing, all on one chip have been reported [155, 156, 157, 158]. Integrated
sensors that can be coupled to actuators via integrated electronic
control elements can enhance the functionality of such devices. As an example
we think of a ring mixer where the liquid is cycled repeatedly. An
integrated sensors could not only monitor the mixing, but the product of
chemical reactions in situ. Combining the sensors, logic on–chip electronics
and micro fluidic actuators would lead to complex, even programmable
Lab-On-Chip systems. Such devices could be used for fast and massively
parallel analyzing tools — maybe a key technology for systems biology
and an opening to the “post microarray area”.