Please use this identifier to cite or link to this item:
Title: Detection of biomolecules using multivariant data analysis from silicon nanowire field-effect transistor arrays
Authors: Schwartz, Miriam
Advisors: WAGNER, Patrick
Ingebrandt, Sven
Issue Date: 2018
Abstract: The aim of this thesis was to establish different medically relevant bioassays on silicon nanowire field-effect transistor sensors (SiNW FET). In parallel to this doctoral thesis with biomedical and biosensing focus, another thesis work was done by a colleague, where new readout instruments and a circuit model description of the SiNW FETs and the amplifiers were elaborated. The SiNW FETs used in both thesis works were fabricated in the framework of a previous project. The nanoelectronic SiNW FET platform has unique features, like a label-free detection of biomolecules, an ultra-sensitive and highly specific response as well as a fast detection of the biomolecular binding reaction within some minutes. These beneficial attributes make this platform a promising technology for future healthcare monitoring and, in particular, for the usage in diagnostic applications, where a fast, sensitive, portable and early detection is very important and can save lives. In initial experiments, the working principle and the functionality of the SiNW FETs were verified by pH and conductivity measurements as well as by detecting the layer-by-layer deposition of polyelectrolyte multilayers. Afterwards, two types of medically relevant bioassays were established on the nanowire platform and on microsized, ion-sensitive field-effect transistors (ISFETs) as control sensors to compare micro- to nanoscale devices. In the first bioassay, DNA hybridization of short oligonucleotides on the sensor surfaces was detected. In initial tests, a synthetic 20 base pair (bp) DNA sequence was used to establish a robust assay. Afterwards, a 20 bp DNA sequence specific to Human Leukocyte Antigen-B27 (HLA-B27) was used towards biomedically relevant experiments. Carriers of the HLA-B27 allele have a higher risk to suffer from auto immune diseases, such as Morbus Bechterew, Morbus Reiter and other inflammatory disorders. In contrast to standard DNA-microarrays, which typically utilize fluorescence of previously labeled target analytes, the detection with the SiNW FET biosensors is direct and label-free. Therefore, electronic or electrochemical assays reduce costs and time. Secondly, protein bioassays were performed, in which the target protein binding to its specific antibody was measured. With electrochemical biosensors such as SiNW FETs, the detection of larger, globular biomolecules is more difficult than the detection of short DNA sequences. In the bioassay, the brain-derived neurotrophic factor (BDNF), which is reduced in Alzheimer, Parkinson’s and Huntington’s disease, was detected. In the human body, BDNF is present in serum, tears, saliva and liquor. The detection of BDNF was done in differently concentrated phosphate buffers and in Hank’s balanced salt solution. It was possible to measure clinically relevant concentrations with the SiNW FET biosensor. Moreover, the detected BDNF concentrations were lower than the limit-of-detection of a commercially available ELISA kit. Besides its higher sensitivity, the SiNW FET method presented in this thesis has further advantages compared to an ELISA assay: It is label-free so that no secondary antibody is necessary, fully-electronic, which means it could be miniaturized and integrated into a battery powered handheld device, and the results are available within some minutes depending on the affinity of the binding partners. The results obtained in this thesis indicate that the SiNW FET platform has a tremendous potential as a future point-of-care platform technology for biomedical applications. In a biomedical company, further research was done in highly concentrated buffer, synthetic and human serum on the nanowire-based platform. The obtained results were promising and some of them were published in a patent. For the practical experiments presented in this work, the SiNW FETs were fabricated and encapsulated such that they can electronically record the biomolecule binding at the solid-liquid interface in different liquid matrices. For the establishment of a bioassay on the sensors, the sensor surface had to be functionalized with a siloxane layer and modified with specific capture molecules as a bioreceptor layer. To avoid false-positive signals, free areas of the sensor surface were blocked with a suitable blocking agent. To verify that the experimental procedures of the assays were stable and reliable, optical controls were firstly established on bare glass or silicon surfaces and on non-functionalized nanowire chips. The electronic recordings were done in two different modes: Firstly, potentiometric DC detection was used by measuring the transfer characteristic of the SiNW FETs before and after the target binding. This detection scheme is used by most of the research groups in the silicon nanowire field. Upon binding of biomolecules to the SiNW FET surface, the typical shift of the transfer characteristic was observed, which was already reported before for ISFET and other silicon nanowire devices. The results obtained in this thesis work indicated that the SiNW FETs function similarly to long-channel field-effect transistor devices. Furthermore, the dimensions of the nanowires have a strong influence on the shape of the transfer characteristic curves and, with this, also on their sensitivity. This was not reported before in the research field. In DNA experiments, it was shown that in thinner wires the curve was steeper than in wider wires. The second detection method was an impedimetric AC mode, where the transistor transfer function, i.e. the frequency bandwidth, of the SiNW FETs was measured. The theory of the AC recording is still under discussion and latest experimental findings are presented. The transistor transfer function in general is the mathematical representation of the relation between the input and the output signals of a frequency-dependent system. Biomolecular interactions at the sensor surface lead to a change in their input impedance and, hence, to a change in the transistor transfer function spectrum. However, the experimental results demonstrate that various parameters have an influence on this recording mode. In case of the microscale ISFET devices, the transistor transfer function decreases after each biomolecule layer. This is in agreement with previously published observations. For the SiNW FET sensors, the results were not fully reliable, wherefore it was concluded that, besides the attachment of the biomolecules, also the charge type, the charge density and the conformation of the biomolecules might influence this readout mode. To derive further information and conclusions from the results of the protein bioassay, a multivariant data analysis was developed with the help of a custom-made MATLAB program. In doing so, the two previously mentioned readout principles were combined, since they can be measured simultaneously. Seven different parameters of the DC and AC readouts were evaluated and plotted against each other in a radar plot to allow a comparison of all variants. A classification model consisting of various decision trees was implemented. The purpose was to differentiate between the different experimental steps as well as between the different concentrations. Finally, the accuracy of the classification model was evaluated. So far, it is only feasible to verify an analyte by functionalizing the sensor surface with specific capture molecules. However, no further information of the target molecules, such as size, surface activity or conformation can be derived. Therefore, the ultimate goal of this multivariant data analysis is to create a library of various parameters in order to identify an unknown analyte and the corresponding characteristics by means of its structure, concentration, total charge, and charge distribution in future works. In conclusion, this thesis work demonstrates that it is possible to detect ultra-low concentrations of biomolecules (down to the fM regime) with the SiNW FET platform, which make it an interesting tool in different applications areas, in which the detection of very tiny concentrations is crucial. Because it was possible to detect BDNF as a target molecule, which is involved in different diseases, the SiNW FETs are a promising technology for the detection of other analytes related to medical disorders. This could be realized by changing the surface modification protocol of the sensors and by applying other, specific capture molecules. In future, more experimental studies need to be performed in human samples, such as serum, saliva or tears. It is assumed that the sensitivity of the SiNW FETs might be reduced in body fluids due to the various components of these liquids. However, by optimizing the surface modification and functionalization, it is expected that the SiNW FET platform can compete with commercially available ELISA tests or fluorescence microarrays. The differential readout of SiNW FETs in an array format offers the possibility to identify different target molecules, simultaneously. This property would be beneficial to detect multiple biomarkers at the same time for panel assays using multivariant recording with different parameters. It is well known in the biosensor field that many novel bioassays on electronic platforms are highly prone to false recordings caused by side influences from parameters such as temperature, pH and ionic strength variations in the detection matrix. The here developed method for multivariant data analysis of multiple, partly independent transducer principles might also be expanded to include these parameters on future multisensoric SiNW FET platforms. By this, the multivariant method offers many promising features for future point-of-care instruments.
Keywords: silicon nanowire field-effect transistors; multivariant data analysis; DNA; BDNF; supervised machine learning; transistor transfer function; transfer characteristic
Document URI:
Category: T1
Type: Theses and Dissertations
Appears in Collections:PhD theses
Research publications

Files in This Item:
File Description SizeFormat 
PhD thesis_Schwartz Miriam.pdf
  Until 2023-04-11
PhD Thesis6.38 MBAdobe PDFView/Open    Request a copy
Show full item record

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.