Silicon nanowire field-effect transistors for sensing applications

Sensing chemical and biological species is essential in many areas like diagnostics of diseases, screening and development of new drugs or environmental monitoring. All these fields experience a strong demand in transducers that convert specific (bio-)chemical processes into measurable signals. Ion-sensitive field-effect transistors (ISFETs) were invented in the 1970s and offer a promising set of features for integrating sensor and readout electronics on the same chip, providing a portable and cost-effective solution. Such devices convert chemical reactions into electrical signals via electrostatic gating of the FET by adsorbed chemical or biological species.
During the past decade, the ISFET concept has been applied to nanoscale devices, in particular silicon nanowire field-effect transistors (Si NW FET). The reduced size offers the possibility of a dense integration of differently functionalized nanowires, which is useful for the multiplexed detection of analytes. In this dissertation, we investigate the sensing properties of Si NW and establish a reliable and versatile sensing platform. The basic physical concepts as well as device fabrication are introduced in Chapter 1.
One of the key aspects is the stable operation of Si NW in electrolyte environment. This is achieved by a drastic reduction of leakage currents using atomic layer deposition (ALD) of high-quality oxides, such as aluminum oxide Al2O3 or hafnium oxide HfO2. Bare oxide surfaces display active hydroxyl (OH) groups that can interact with protons in solution. In Chapter 2, we show that Si NWs coated by ALD oxides are excellent pH sensors with nearly ideal Nernstian response of 58.2 mV/dec at 20°C. This is attributed to the high number of available surface OH groups.
Chapter 3 focuses on the low-frequency noise measurements and explores the ultimate detection limits set by the noise of the transistor. The deduced gate-referred threshold voltage noise ?Vth is the true figure of merit in such transistor-based sensors. We demonstrate that ?Vth is constant over the full operation range of the transistor, as long as the intrinsic nanowire resistance dominates and the contact resistance can be neglected. Both linear and subthreshold regime can be used for equally sensitive pH measurements with a resolution of ~200 ppm of a full Nernstian pH shift (1 Hz bandwidth at 10 Hz, 1 µm wide NW).
Even though great progress has been achieved in pH sensing, it has turned out to be much harder to realize a true reference electrode, which – while sensing the electrostatic potential – does not respond to the proton concentration. In Chapter 4, we demonstrate a highly effective reference sensor, whose proton sensitivity is suppressed by as much as two orders of magnitude. To do so, the Al2O3 surface of a nanowire FET is passivated with a self-assembled monolayer of silanes with a long alkyl chain (C 18). We find that a full passivation can be achieved only after several days of self-assembly. In addition, we determine the number of proton-binding surface sites as a function of silanization time by quantitatively comparing the experimental data with the theoretical site-binding model. Furthermore, we find that a partially passivated surface can sense small changes in the number of active sites caused by the adsorption of uncharged species. A detection limit for this type of application is estimated. This result might extend the application field of ion-sensitive FETs to the detection of neutral species.
Understanding the electrolyte background is a basic prerequisite for any further biochemical sensing. In Chapter 5, we study the sensor response to changes in KCl concentration at several constant pH values. A significant signal change is only observed at high ionic strengths > 10 mM. Our measurements suggest that anions (Cl-) rather than cations (K+) adsorb on the surface, independent of the pH value. By comparing the data to three well-established models, we find that none of those can completely explain the results. We propose a new adsorption model instead which gives excellent agreement with the data. According to our model, the chloride ions directly interact with the hydroxyl surface groups and replace previously attached protons.
In Chapter 6 we report our efforts to selectively sense targets other than protons. Using ionophores embedded in PVC membranes, we demonstrate selective detection of potassium ions in a differential set-up. We also show that a different functionalization scheme, based on covalent silane chemistry, can be equally effective. Moreover, first successful and reproducible biosensing measurements of a specific sugar-lectin interaction are presented. Together with the ongoing work towards the integration of the sensor and the read-out electronics on one chip, the mentioned functionalization methods may enable a highly efficient multi-analyte detection system with a commercial potential, for example in the area of water quality monitoring.