Design and optimization of ultrathin silicon field effect transistor's for sensitive, electronic-based detection of biological analytes

Noncommunicable diseases (NCD) are currently the leading cause of death worldwide. Over 57 million deaths occur globally each year, with close to 36 million of them attributed to NCD’s, and 80% of those in low and middle income countries. Most of these were due to such chronic illnesses as cancer, cardiovascular disease, diabetes, and lung disease. Moreover, the prevalence of these diseases is rising fastest in low-income regions which have little resources to combat these large, yet avoidable costs. In particular, over 1.6 million cases of cancer are caused each year in the United States, with nearly 600,000 of these cases being fatal. Cancer is an uncontrolled growth and spread of abnormal cells in the body, and unfortunately, can exist in many different cell types. The complexity in the causes of cancer has made it tougher to diagnose since several factors may weight into its prevalence such as: genetic factors, lifestyle factors, certain types of infections, and different environmental exposures. As a result, the protocols for the most cost-effective intervention are available across four main approaches to cancer prevention and control: primary prevention, early detection, treatment, and palliative care. Early diagnosis based on awareness of early signs and symptoms and, if affordable, population-based screening improves survival, particularly for breast, cervical, colorectal, skin and oral cancers. If primary prevention of cancer fails, secondary prevention (early detection) may be the difference between irreversible spread of a malignant cancer, and the patient’s survival. Early detection commonly refers to the diagnosis of a disease before individuals show obvious signs or symptoms of illness. With cancer, RNA and protein biomarkers of cells are currently assayed to determine their serums level and if they have deviated from the normal ranges. However, these assays commonly require large centralized lab facilities, frequent monitoring during treatment, and expensive equipment and/or supplies. This has led to point-of-care diagnostics becoming a $16 billion global market, aimed at miniaturizing technology and making it cost-effective for individual patient testing and treatment without the use of centralized lab facilities. A main point-of-care testing platform being pursued utilizes Complementary Metal Oxide Semiconductor (CMOS) technology. CMOS-based products can enable clinical tests to be conducted in a fast, simple, safe, and reliable manner, with improved sensitivities. Moreover, CMOS products offer portability and low power consumption, in large part due to the explosion in the semiconductor and communications markets. Silicon nanowires are of great interest for point-of-care testing as they are a CMOS compatible structure, require the use of no labels, and are highly sensitive to the binding of molecules to their surfaces. This is due to the large surface area to volume ratio afforded to nanowires. Moreover, arrays of silicon nanowires have demonstrated multiplexed, label-free sensing of cancer markers from undiluted serum samples. However, the research going into CMOS for point-of-care is in its infancy compared to other optical (surface plasmon resonance, fluorescence) or electrochemical methods (glucose sensors), although the technology for CMOS has been around for decades. Thus, the protocols for optimization of the sensors and their bioconjugation have not matured to the point DNA microarrays and ELISA’s have. The protocols for creation of a dependable silicon nanowire biosensor revolve around three main aspects: semiconductor processing, device functionalization, and choice of analytes. In this dissertation, I discuss our efforts to create a stable, silicon nanowire based sensor using CMOS compatible techniques and optimization processes. Moreover, I talk about our efforts into creating a device functionalization protocol using monofunctional silanes which affords the best sensitivity and specify for an electronic based biosensor. Finally, I discuss our look towards the future in silicon nanowires by using high-k dielectrics in our fabrication process, as well as an alternative monolayer deposition method which utilizes sub-nanometer thickness poly-l-lysine monolayers, for sensing clinically relevant targets of microRNA. Using a special type of silane, called a monofunctional silane, and a vapor based deposition method, we were able to achieve sub-nanometer levels functional monolayers on thermally oxidized silicon surfaces. We employed a variety of characterization techniques (XPS, AFM, ellipsometry) to determine the densities of the monolayer, uniformity, topography, and their point of saturation. Furthermore, we demonstrate this method’s applicability to biosensors by using it to functionalize substrates for silicon nanowires, gold nanoparticles, and protein microarrays. In tandem with this work, we constructed a “top down” silicon nanowire processing protocol which yielded nanowires capable of long-term, stable measurements in aqueous solutions. The combination of anneals, dry etching, and final wet etching gave mV standard deviations in device threshold characteristics. This protocol combined with the monolayer protocol above allowed an in-depth characterization of the pH sensitivity of bare devices, ones with silanes, and ones conjugated with proteins to be determined. Similarly, different oxide thicknesses and their effect on device sensitivity for proteins were also explored. Using a bunch of different linker chemistries and characterizing their conjugation of antibodies through fluorescence and the device, allowed for a chemistry to be chosen which was used to sense mouse immunoglobulins in pg/mL levels with high specificity. Finally, we take the fabrication of nanowires to the next level by using high-k dielectrics (HfO2) as the gate insulator. We deposit HfO2 through ALD (atomic layer deposition) and optimize the anneals to provide nanowires with ~200mV subthreshold slopes, sub-mV threshold deviations, and sub nanoampere gate leakages. All these characteristics exceed the processes for thermal oxide gated silicon nanowires, some by an order of magnitude. Since HfO2 is a high-k material, reaction of silanes and its density were unknown, but high-k materials do form stable amide linkages. Thus, we optimized a wet deposition of small molecular weight poly-l-lysine to provide a sub-nm conjugation layer for proteins and nucleotides by using AFM, XPS, and ellipsometry to understand the process. Using these combined protocols, we were able to conjugate probe oligonucleotides to surfaces and detect target microRNA’s down to 100fM concentrations, with a dynamic range over 4 orders of magnitude. With these ranges well within the clinical levels (1pM-100pM), we believe silicon nanowires have the capability to become a well-established point-of-care diagnostic platform