**2. Field‐effect transistor‐based biosensors**

*1.1.2. Electrochemical transducers*

168 Different Types of Field-Effect Transistors - Theory and Applications

biosensor market [10, 19, 20].

*1.1.3. Electronic transducers*

Electrochemical sensors originally called "enzyme electrodes"emerged as highly sensitive, easy‐to‐use, portable, and user‐friendly sensing devices. Electrochemical biosensors are now‐

Electrochemical sensors measure electrochemical processes occurring in an electrode function‐ alized with enzyme bioreceptors immersed in an electrolyte solution containing the analytes. Electrochemical processes include the measurement of tiny changes of voltage (potentiom‐ etry), current (amperometry), or resistance/conductance (conductometry) specific to the presence of an analyte. Electrochemical sensors usually involve two electrodes, the working electrode and the counter electrode under which a voltage is applied. A third reference elec‐ trode can be employed to set and monitor the potentials vs an absolute reference value.

In the absence of the reference electrode, it is somewhat difficult to measure a small current change in a reproducible way, thus the reproducibility of electrochemical sensors is generally less accurate than those employing optical transducers [19]. However, because electrochemi‐ cal sensors can be miniaturized and manufactured inexpensively, they actually dominate the

Electrochemical glucose sensors alike their analogue glucose optical sensors employ glucose oxi‐ dase (GOx) enzyme to bind the glucose molecule and facilitate its oxidation process. Oxygen and GOx oxidize glucose into gluconic acid and generate hydrogen peroxide as by‐product. During this process, GOx is reduced and can be regenerated (oxidized) by adding ferricyanide which in turn is reduced into ferrocyanide. A metal electrode can regenerate ferrocyanide (reduced form) into ferricyanide. The reduction‐oxidation cycle of glucose generates electrons at the metal elec‐ trode and can induce a spontaneous electric current (with no voltage applied) proportional to the glucose concentration. To obtain a quick glucose measurement, a change in the electrical current is measured as a function of a voltage applied between the electrolyte and the metal electrode.

Another popular redox system to detect glucose is based on enzyme glucose dehydrogenase as catalyzer and nicotinamide adenine dinucleotide, replacing oxygen to oxidize glucose [21, 22].

After the discovery of the enzyme electrode, ion‐sensitive field‐effect transistors (ISFETs), **Figure 2** (right), in which the gate electrode in a conventional metal‐organic field‐effect tran‐ sistor (MOSFET), **Figure 2** (left), is replaced by an aqueous solution and a reference electrode, emerged to measure ionic species in electrochemical and biological environments. Double lay‐ ers formed at the oxide electrolyte interface result in a different conductive state at the transistor channel proportional to the electrolyte ion concentration [23]. The gate oxide can be sensitive to specific ions similar to a glass electrode [24] or can be modified with a selective membrane or molecular receptors to filter specific ions [25, 26]. ISFETs are not strictly biosensors because they do not employ a biomolecular receptor as an active component to sense ions but they laid

Electronic transducers include the field‐effect transistor and the organic electrochemical tran‐

the foundation toward field‐effect transistor biosensors (bio‐FETs).

sistor, reviewed in greater detail in the following sections.

adays the most widespread transducers for glucose sensing [18].

ISFETs can be considered a close version of field‐effect transistor‐based biosensors (bio‐ FETs). Bio‐FETs have the transistor semiconductor channel directly coupled with molecular bioreceptors sensitive to specific analyte molecules in the electrolyte without an insulating layer and can be directly gated through an electrolyte medium or through a back gate/gate‐ insulator as shown in **Figure 3**. The molecular bioreceptors act as a filter that allows only one type of analyte to selectively interact with the semiconductor channel. The analyte of interest can bind covalently to a specific molecular bioreceptor and change the semiconductor doping state. The conductive state of the functionalized semiconductor channel can be tracked by measuring the transistor drain‐source current as a function of the electrolyte analyte concen‐ tration and the voltages applied.

**Figure 3.** Schematic structure of field‐effect transistor‐based biosensors (bio‐FETs).

Sensing mechanism may differ significantly when applied to different analyte molecules despite the common architectures among bio‐FETs. As a thought experiment, a substance impermeable to ions that prevents the ionic doping of the semiconductor can be placed on top of the transistor channel. If the permeability of this substance changes following an inter‐ action with a specific analyte allowing access of ions from the electrolyte into the semicon‐ ductor, the increased permeability of the barrier can be revealed by the modulation of the semiconductor conductive state as a function of a gate voltage [28]. Ions in proximity with the semiconductor can induce a doping/dedoping process by the field‐effect or tune the effective energy barrier height required to inject charge carriers from the drain‐source metal electrodes into the semiconductor. Because the chemical structure of the semiconductor remains unmod‐ ified, desorption of the analyte or ionic species from the semiconductor surface is possible for real‐time reversible sensor detection.

Sensing DNA hybridization with FET devices is paving the way toward virus sensing and DNA disease prevention [29]. The working mechanism of DNA field‐effect transistor sensors deserves attention. Carbon nanotube FETs typically operate as unconventional Schottky bar‐ rier transistors in which current modulation occurs primarily by tuning the contact resistance rather than the channel conductance [30].

Synthetic DNA hybridization consisting of random generated sequences and different oligo lengths (15 and 30 mer) was detected with bio‐FETs made of gold drain‐source electrodes functionalized with MercaptoHexanol and single‐walled carbon nanotube channel material, **Figure 4** (left). Selective response to addition of complementary DNA was observed and almost no change occurred upon addition of phosphate‐buffered saline (PBS) solution or mismatched DNA, **Figure 4** (right). MercaptoHexanol self‐assembled monolayer provides a nice passivation on gold electrodes against nonspecific binding of mismatched DNA and provides ideal conditions for efficient hybridization with nearly 100% binding efficiency of analytes carrying complementary sequences. The formation of double stranded DNA on gold electrodes lowered the effective work function of gold facilitating charge carrier injection.

**Figure 4.** Schematic illustration of DNA FETs sensing device in operation (left) and real‐time normalized conductance monitoring of 30 mer DNA hybridization in phosphate‐buffered saline solution, pH 7.4. Two other devices were used to simultaneous test complementary (CM30) and mismatched (MM30) DNA buffer solutions. Selective response to addition of complementary DNA is observed and almost no change upon addition of phosphate‐buffered saline solution or mismatched DNA [29]. Copyright © 2006 American Chemical Society.

Real‐time DNA biosensors to detect cystic fibrosis genes have been demonstrated with sensitivities up to the femtomolar range [31]. These bio‐FETs hinge on silicon nanowire (SiNWs) channels functionalized with peptide nucleic acid receptors that are complemen‐ tary to the wild type of cystic fibrosis genes. Tailoring SiNWs is relatively easy due to the well‐known modification of silicon oxide surfaces as opposed to graphene or carbon complex surfaces. Although a nonspecific interaction of negatively charged oligonucle‐ otides with NW sensors was observed, specific detection was possible by analyzing the magnitude of the conductance change following introduction of wild vs mutant DNA sample solutions. After detection, the conductance was reversible to its original state upon addition of DNA‐free solution.

Sensing mechanism may differ significantly when applied to different analyte molecules despite the common architectures among bio‐FETs. As a thought experiment, a substance impermeable to ions that prevents the ionic doping of the semiconductor can be placed on top of the transistor channel. If the permeability of this substance changes following an inter‐ action with a specific analyte allowing access of ions from the electrolyte into the semicon‐ ductor, the increased permeability of the barrier can be revealed by the modulation of the semiconductor conductive state as a function of a gate voltage [28]. Ions in proximity with the semiconductor can induce a doping/dedoping process by the field‐effect or tune the effective energy barrier height required to inject charge carriers from the drain‐source metal electrodes into the semiconductor. Because the chemical structure of the semiconductor remains unmod‐ ified, desorption of the analyte or ionic species from the semiconductor surface is possible for

Sensing DNA hybridization with FET devices is paving the way toward virus sensing and DNA disease prevention [29]. The working mechanism of DNA field‐effect transistor sensors deserves attention. Carbon nanotube FETs typically operate as unconventional Schottky bar‐ rier transistors in which current modulation occurs primarily by tuning the contact resistance

Synthetic DNA hybridization consisting of random generated sequences and different oligo lengths (15 and 30 mer) was detected with bio‐FETs made of gold drain‐source electrodes functionalized with MercaptoHexanol and single‐walled carbon nanotube channel material, **Figure 4** (left). Selective response to addition of complementary DNA was observed and almost no change occurred upon addition of phosphate‐buffered saline (PBS) solution or mismatched DNA, **Figure 4** (right). MercaptoHexanol self‐assembled monolayer provides a nice passivation on gold electrodes against nonspecific binding of mismatched DNA and provides ideal conditions for efficient hybridization with nearly 100% binding efficiency of analytes carrying complementary sequences. The formation of double stranded DNA on gold electrodes lowered the effective work function of gold facilitating charge carrier injection.

**Figure 4.** Schematic illustration of DNA FETs sensing device in operation (left) and real‐time normalized conductance monitoring of 30 mer DNA hybridization in phosphate‐buffered saline solution, pH 7.4. Two other devices were used to simultaneous test complementary (CM30) and mismatched (MM30) DNA buffer solutions. Selective response to addition of complementary DNA is observed and almost no change upon addition of phosphate‐buffered saline solution

or mismatched DNA [29]. Copyright © 2006 American Chemical Society.

real‐time reversible sensor detection.

170 Different Types of Field-Effect Transistors - Theory and Applications

rather than the channel conductance [30].

Highly selective and sensitive real‐time protein detection was reported with SiNW bio‐ FETs [25]. Nanotube/nanowire semiconductor channels offer higher surface area than pla‐ nar devices and increase sensitivity to the point that single‐molecule detection is possible. Sensing of calcium ions (Ca2+) was possible by immobilizing calmodulin onto SiNW surfaces. A drop in conductance upon addition of 25 μM Ca2+ solution was observed with a subse‐ quent recovery in conductance when the device was rinsed with a Ca2+ free solution, **Figure 5**. Control experiments with unmodified SiNWs did not show a change in conductance when Ca2+ ions were added.

SiNW arrays modified with antibodies for influenza A or adenovirus displayed selective conductance change corresponding to single‐virus binding/unbinding of influenza A and paramyxovirus [32]. Many bio‐FETs can be built in parallel for multiplexed biodetection.

**Figure 5.** Real‐time detection of calcium ions. Plot of conductance vs time of a bio‐FET channel of calmodulin‐terminated silicon nanowire where regions 1 and 3 correspond to the operation under pure buffer solution and region 2 corresponds the operation under 25 μM Ca2+ solution [25]. Copyright © 2001 by the American Association for the Advancement of Science.

No purification of virus samples was required in these measurements. **Figure 6** shows the conductance vs time recorded simultaneously for influenza A (nanowire 1) and adenovi‐ rus group III (nanowire 2) sensors set in proximity with a microfluidic system that allows the sequential flow of 1–4 adenovirus, influenza A, pure solution, and 1:1 mix solution of adenovirus and influenza A, respectively. Adenovirus is negatively charged in the solu‐ tion resulting in a positive conductance change of nanowire 2 with an on‐time duration of ca 16 s. Negative conductance change of similar duration was observed when influenza A was introduced and binded to nanowire 1. Bottom arrows in **Figure 6** indicate singulari‐ ties in which the proximity of adenovirus in device 1 resulted in a short‐lived positive change of conductance ca 0.4 s, and similarly, the proximity of influenza A resulted in a short‐lived negative change of conductance in device 2. The excellent binding selectivity, single viral particle sensitivity and selective multiplexed detection enables rapid identi‐ fication of viral samples as required for robust medical solutions, fundamental virology, and drug discovery.

Label‐free amplified biodetection has been reported with floating‐gate transistor archi‐ tectures, **Figure 7** (left), in which the gate voltage is indirectly applied via a secondary electrolyte [33]. In such structures, the semiconductor does not need to be modified with selective bioreceptors and is not in direct contact with the analytes therefore the semicon‐ ductor can be selected on its electronic performance and doping mechanism (field‐effect vs electrochemical) and not on the ease of its chemical modification or robustness in elec‐ trolytes, thereby reducing fabrication complexity. Bioreceptors are set on one part of the floating gate coupled with a secondary electrolyte compartment and entirely separated

**Figure 6.** Conductance vs time recorded simultaneously from two silicon nanowire sensors, nanowire 1 was modified with influenza A antibody (top) and nanowire 2 was modified with adenovirus group III antibody (bottom). Arrows 1–4 correspond to the introduction of (1) adenovirus, (2) influenza A, (3) pure buffer, and (4) 1:1 mixture solution of adenovirus and influenza A. Bottom arrows highlight short‐duration conductance changes corresponding to nonspecific diffusion of viral particles. Solutions made by 40 viral particles per μl in phosphate buffer 10 μM, pH 6.0 [32]. Copyright © 2014 National Academy of Sciences.

Transistors as an Emerging Platform for Portable Amplified Biodetection in Preventive Personalized... http://dx.doi.org/10.5772/67794 173

**Figure 7.** Device architecture of the floating‐gate transistor structure (left) and transistor transfer characteristics (right), drain‐source current (Ids) vs gate voltage (Vgs), of control device, curve 1 and floating‐gate functionalized with alkylthiol derivatives right‐shifted 225 mV from the control device, curve 2 [33]. Copyright © 2015 American Chemical Society.

from the primary electrolyte [34–36]. The bioreceptor‐free gate electrode is coupled with the primary electrolyte. Well established self‐assembled monolayer chemistry on metal electrodes can be employed to tune the effective potential of the floating gate for bioelec‐ tric signal amplification [37]. Geometric considerations of the gate electrode play a crucial role in the device operation [38, 39]. Self‐assembled monolayers of alkylthiol derivatives on the floating gate had a measurable voltage shift (ca. 200 mV) on the transistor electric characteristics, **Figure 7** (right).

Label‐free DNA hybridization has been sensed by a floating‐gate transistor based on poly(3‐ hexylthiophene) and an ion gel electrolytes [36]. The DNA is hybridized at the floating gate resulting in a shift in the threshold voltage of the transistor with a relative magnitude pro‐ portional to the DNA mismatch. The response of DNA with three mismatched base pairs was undistinguishable from a fully random DNA sequence.
