Increasing the detection speed of an all-electronic real-time biosensor

Reduction of Biosensor False Responses and Time Delay Using Dynamic Response and Theory-Guided Machine Learning

Here, we provide a new methodology for reducing false results and time delay of biosensors, which are barriers to industrial, healthcare, military, and consumer applications. We show that integrating machine learning with domain knowledge in biosensing can complement and improve the biosensor accuracy and speed relative to the performance achieved by traditional regression analysis of a standard curve based on the biosensor steady-state response. The methodology was validated by rapid and accurate quantification of microRNA across the nanomolar to femtomolar range using the dynamic response of cantilever biosensors. Theory-guided feature engineering improved the performance and efficiency of several classification models relative to the performance achieved using traditional feature engineering methods (TSFRESH). In addition to the entire dynamic response, the technique enabled rapid and accurate quantification of the target analyte concentration and false-positive and false-negative results using the initial transient response, thereby reducing the required data acquisition time (i.e., time delay). We show that model explainability can be achieved by combining theory-guided feature engineering and feature importance analysis. The performance of multiple classifiers using both TSFRESH- and theory-based features from the biosensor's initial transient response was similar to that achieved using the entire dynamic response with data augmentation. We also show that the methodology can guide design of experiments for high-performance biosensing applications, specifically, the selection of data acquisition parameters (e.g., time) based on potential application-dependent performance thresholds. This work provides an example of the opportunities for improving biosensor performance, such as reducing biosensor false results and time delay, using explainable machine learning models supervised by domain knowledge in biosensing.

Surface Modification of Silicon Nanowire Based Field Effect Transistors with Stimuli Responsive Polymer Brushes for Biosensing Applications

We demonstrate the functionalization of silicon nanowire based field effect transistors (SiNW FETs) FETs with stimuli-responsive polymer brushes of poly(N-isopropylacrylamide) (PNIPAAM) and poly(acrylic acid) (PAA). Surface functionalization was confirmed by atomic force microscopy, contact angle measurements, and verified electrically using a silicon nanowire based field effect transistor sensor device. For thermo-responsive PNIPAAM, the physicochemical properties (i.e., a reversible phase transition, wettability) were induced by crossing the lower critical solution temperature (LCST) of about 32 °C. Taking advantage of this property, osteosarcomic SaoS-2 cells were cultured on PNIPAAM-modified sensors at temperatures above the LCST, and completely detached by simply cooling. Next, the weak polyelectrolyte PAA, that is sensitive towards alteration of pH and ionic strength, was used to cover the silicon nanowire based device. Here, the increase of pH will cause deprotonation of the present carboxylic (COOH) groups along the chains into negatively charged COO- moieties that repel each other and cause swelling of the polymer. Our experimental results suggest that this functionalization enhances the pH sensitivity of the SiNW FETs. Specific receptor (bio-)molecules can be added to the polymer brushes by simple click chemistry so that functionality of the brush layer can be tuned optionally. We demonstrate at the proof-of concept-level that osteosarcomic Saos-2 cells can adhere to PNIPAAM-modified FETs, and cell signals could be recorded electrically. This study presents an applicable route for the modification of highly sensitive, versatile FETs that can be applied for detection of a variety of biological analytes.

Electrically-receptive and thermally-responsive paper-based sensor chip for rapid detection of bacterial cells

Although significant technological advancements have been made in the development of analytical biosensor chips for detecting bacterial strains (E. coli, S. Mutans and B. Subtilis), critical requirements i.e. limit of detection (LOD), fast time of response, ultra-sensitivity with high reproducibility and good shelf-life with robust sensing capability have yet to be met within a single sensor chip. In order to achieve these criteria, we present an electrically-receptive thermally-responsive (ER-TR) sensor chip comprised of simple filter paper used as substrate coated with composite of poly(N-isopropylacrylamide) polymer (PNIPAm) - graphene nanoplatelet (GR) followed by evaporation of Au electrodes for capturing both Gram-positive (S. mutans and B. subtilis) and Gram-negative (E. coli) bacterial cells in real-time. Autoclave water, tap water, lake water and milk samples were tested with ER-TR chip with and without bacterial strains at varying concentration range 10¹-10⁵ cells/mL. The sensor was integrated with in-house built printed circuit board (PCB) to transmit/receive electrical signals. The interaction of E. coli, S. mutans and B. subtilis cells with fibers of PNIPAm-GR resulted in a change of electrical resistance and the readout was monitored wirelessly in real-time using MATLAB algorithm. Finally, prepared ER-TR chip exhibited the reproducibility of 85-97% with shelf-life of up to four weeks after testing with lake water sample.

Flocculation on a chip: a novel screening approach to determine floc growth rates and select flocculating agents

Flocculation is a key purification step in cell-based processes for the food and pharmaceutical industry where the removal of cells and cellular debris is aided by adding flocculating agents. However, finding the best suited flocculating agent and optimal conditions to achieve rapid and effective flocculation is a non-trivial task. In conventional analytical systems, turbulent mixing creates a dynamic equilibrium between floc growth and breakage, constraining the determination of floc formation rates. Furthermore, these systems typically rely on end-point measurements only. We have successfully developed for the first time a microfluidic system for the study of flocculation under well controlled conditions. In our microfluidic device (μFLOC), floc sizes and growth rates were monitored in real time using high-speed imaging and computational image analysis. The on-line and in situ detection allowed quantification of floc sizes and their growth kinetics. This eliminated the issues of sample handling, sample dispersion, and end-point measurements. We demonstrated the power of this approach by quantifying the growth rates of floc formation under forty different growth conditions by varying industrially relevant flocculating agents (pDADMAC, PEI, PEG), their concentration and dosage. Growth rates between 12.2 μm s^-1 for a strongly cationic flocculant (pDADMAC) and 0.6 μm s^-1 for a non-ionic flocculant (PEG) were observed, demonstrating the potential to rank flocculating conditions in a quantitative way. We have therefore created a screening tool to efficiently compare flocculating agents and rapidly find the best flocculating condition, which will significantly accelerate early bioprocess development.

A microfluidic electrochemical biosensor based on multiwall carbon nanotube/ferrocene for genomic DNA detection of Mycobacterium tuberculosis in clinical isolates

Herein we present a microfluidic-multiplexed platform that integrates electrochemical sensors based on carbon nanotubes associated with ferrocene as redox marker (carbon nanotube (CNT)/ferrocene) for direct detection of pathogenic viral DNA from Hepatitis C and genomic DNA from Mycobacterium tuberculosis in clinical isolates. By operating the fluidic device under high flow (150 μl/min), the formation of a very thin depletion layer at the sensor surface (δS = 230 nm) enhances the capture rate up to one DNA strand per second. By comparison, this capture rate is only 0.02 molecule/s in a static regime without flow. This fluidic protocol allows thus enhancing the limit of detection of the electrochemical biosensor from picomolar in bulk solution to femtomolar with a large dynamic range from 0.1 fM to 1 pM. Kinetics analysis also demonstrates an enhancement of the rate constant of electron transfer (kS) of the electrochemical process from 1 s(-1) up to 6 s(-1) thanks to the geometry of the miniaturized fluidic electrochemical cell. This microfluidic device working under high flow allows selective direct detection of a Mycobacterium tuberculosis (H37Rv) rpoB allele from clinical isolate extracted DNA. We envision that a microfluidic approach under high flow associated with a multiwall CNT/ferrocene sensor could find useful applications as the point-of-care for multi-target diagnostics of biomarkers in real samples.

Sequence specific detection of restriction enzymes at DNA-modified carbon nanotube field effect transistors

Protein-DNA interactions play a central role in many cellular processes, and their misregulation has been implicated in a number of human diseases. Thus, there is a pressing need for the development of analytical strategies for interrogating the binding of proteins to DNA. Herein, we report the electrical monitoring of a prototypical DNA-binding protein, the PvuII restriction enzyme, at microfluidic-encapsulated, DNA-modified carbon nanotube field effect transistors. Our integrated platform enables the sensitive, sequence specific detection of PvuII at concentrations as low as 0.5 pM in a volume of 0.025 μL (corresponding to ~7500 proteins). These figures of merit compare favorably to state of the art values reported for alternative fluorescent and electrical assays. The overall detection strategy represents a step toward the massively parallel electrical monitoring, identification, and quantification of protein-DNA interactions at arrayed nanoscale devices.

Self-assembled monolayer-based selective modification on polysilicon nanobelt devices

In this study, a self-assembled monolayer (SAM) of methoxy-poly (ethylene-glycol)-silane (mPEG-sil) was used to modify the silicon dioxide surface of polysilicon nanodevices (PNDs) to act as a passivation layer that inhibits nonspecific binding of proteins and reduces localized Joule heating power. Selective modifications of 3-aminopropyltrimethoxysilane (APTMS), NHS-biotin and dye-labeled Streptavidin on the removal regions were characterized. These PNDs, which consist of a two-level doping profile, were designed to confine heat in the low-level doping region during localized Joule heating. Localized Joule heating with pulse bias was examined in both vacuum and ambient, which indicated the removal region was longer in vacuum for the same pulse bias. Moreover, a comparison of selectively and nonselectively modified PNDs observed in time-lapsed fluorescence detection of dye-labeled Streptavidin showed a higher increasing rate in fluorescence intensity (∼2× enhancement) in the selectively modified PNDs. Finally, a COMSOL simulation was employed to evaluate the temperature distribution in the PNDs, with results showing that heat confinement was observed in the low-level doping region and a temperature very close to 673 K was achieved while applying a pulse voltage (40 V, 5 μs) to remove mPEG-sil.

Scalable graphene field-effect sensors for specific protein detection

We demonstrate that micron-scale graphene field-effect transistor biosensors can be fabricated in a scalable fashion from large-area chemical vapor deposition derived graphene. We electrically detect the real-time binding and unbinding of a protein biomarker, thrombin, to and from aptamer-coated graphene surfaces. Our sensors have low background noise and high transconductance, comparable to exfoliated graphene devices. The devices are reusable and have a shelf-life greater than one week.