Over the last decades, organic electronic materials and their use in semiconducting devices have attracted enormous attention.1–5 The interest in developing this class of materials is due to the many advantages this technology has to offer. Their unique features allow for flexible electronic devices that can be fabricated by low-cost manufacturing processes such as spin coating, spray coating, inkjet printing, and screen printing. These materials offer compatibility with biological systems, and in addition, their electronic properties can be tuned via chemical synthesis or doping. Therefore, organic electronic materials are an excellent candidate for the development of small, portable and inexpensive sensors ideally suited for point-of-care analysis.
Many organic electronic devices have been explored for sensing applications over the years. Among these, organic thin-film transistors (OTFTs) offer several advantages for use as transducers due to their inherent signal amplification that allows for small sample volumes and ease of incorporation into circuits.6 For these reasons, OTFTs have been used in sensor applications for gaseous analytes (e.g., H2O2 and O2,7 NO2,8 and O39,10), humidity sensing,11–13 acidity levels (pH),14–17 ion concentrations,18,19 and a variety of biologically important analytes [e.g., using enzymes,12,20–29 antibodies,30 and deoxyribonucleic acid (DNA)31].
An interesting class of OTFTs is organic electrochemical transistors (OECTs). These devices have been used widely in recent years due to their low operating voltages and their simplified structure. Furthermore, OECTs have the ability to operate in aqueous environments that are essential for biological applications. In addition, OECTs' low-voltage requirements significantly limit the risk for hydrolysis. The above key properties enable OECTs to be integrated with microfluidic devices,32 and ultimately move the technology closer to the “lab-on-a-chip” concept.
Electrochemical Transistors (ECTs)
Typically, an OECT consists of the channel, a thin film of semiconducting polymer [in its doped (conducting) state] deposited onto a supporting substrate; the source (S), drain (D) and gate (G) contacts, and finally, an electrolyte medium in contact with the channel and the gate (Fig. 1). In the case where the channel is composed of a highly conducting polymer, that same polymer can be used for the source and drain contacts as well. In such devices, the electrolyte medium can be either a liquid,12,13,33 a gel,13 or a solid13,34,35 (depending on the sensing application), and the OECTs operate in an electrochemical mode where the conductivity of the polymer film is modulated by oxidation or reduction. This doping and de-doping process is reversible and involves the migration of mobile ions out of and into the polymer film. As a result, the polymer switches between its redox states. Therefore, OECTs act as ion-to-electron converters6 providing a bridge between the biomolecular world and electronic devices for effective communication. This enhances our ability to extract information about the electrolyte, and will further assist in the development of sensor technology.
The operation of OECTs is based on the application of a gate voltage (Vg) through the electrolyte medium that causes a change in the channel current (Isd). More specifically, upon the application of a small gate voltage, ions from the electrolyte medium migrate into or out of the polymer film altering the doping level of the active area. This causes a change of the channel conductivity that can be detected by a simple current measurement. In order to reach maximum current modulation between the redox states of the active area, the area of the gate electrode must be considerably larger compared with the area of the channel. Hence, the device geometry plays a role in the response.
In general, upon the application of a gate voltage, an OECT can be turned “on” (high source–drain current) or “off” (negligible source–drain current) depending on the initial level of doping and the sign of the gate bias. The switching time between the “on” and “off” states (or different oxidation states of the polymer) is usually limited by the ion migration. Therefore, the choice of the ionic electrolyte is critical. In order to achieve high levels of current modulation in the channel, a high degree of ionic involvement is required. If the polymer utilized for the channel of the OECT has high electronic conductivity, then the kinetics that governs the ion migration can become the main hindrance to a fast response. Thus, electrolyte mediums comprising small ions that can migrate readily into and out of the polymer matrix are desirable. A detailed description of OECT modeling the ionic and electronic aspects of device operation is currently in press by Bernards and Malliaras.37
Poly(3,4-Ethylenedioxythiophene) Doped With Poly(Styrene Sulfonic Acid) (PEDOT:PSS) ECTs
Conducting polymers have long been used in OECTs, and have consequently made their way into sensing applications. In 1984, White et al.38 reported results of their work on the first OECT where polypyrrole was employed as the active material on a three-electrode configuration. This work paved the way for a series of demonstrations of chemical and biological sensing by several groups using different conducting polymers. The most commonly used conjugated polymers in this type of devices are polyaniline,11,16–18,21–28,39–41 polypyrrole,20,42 polycarbazole,19 polythiophene, and their derivatives,30,31 with applications covering a broad spectrum of chemical and biological sensing. Some examples include humidity sensors, pH, and metal ions, analyte sensing (e.g., glucose, urea, triglycerides, etc), DNA, and more. However, despite the widespread use of the above polymers, there are some limitations concerning their applications. For example, polypyrrole's conductivity gets irreversibly destroyed upon exposure to H2O25,43 limiting its use with enzymes producing H2O2, like glucose oxidase (GOx). Moreover, while polyaniline does not react with H2O2,21 at pH higher than 5, it loses its electrochemical activity and thereby, its ability to sense analytes.5,27 Provided that most of the biologically interesting reactions take place at neutral pH, this loss of activity significantly limits its use. There have been successful efforts to overcome these limitations by using modified polyanilines or composite films that extend operation at neutral pH.5 Nevertheless, there is a need for a more robust conducting polymer that will operate in a broad pH range.
One of the most successful polymers with these characteristics that has been widely used for many types of applications is poly(3,4-ethylenedioxythiophene) (PEDOT).44 This conducting polymer is electrochemically active, i.e., it can conduct ions and undergo electrochemical switching. In addition, it exhibits high environmental stability and can retain its activity in a broad pH range. All these properties make PEDOT a prime candidate for sensor applications and other devices like electrochromic displays.45–48 Several attempts to incorporate PEDOT into transistors for the development of biosensors have proven to be successful.30,31 When p-doped PEDOT can display conductivities in the range of 1–100 S cm−1 depending upon the film morphology and the choice of counter ion. The use of poly(styrene sulfonic acid) (PSS) provides enhanced conductivities of up to 10 S cm−1, but it offers other advantages as well. As PEDOT itself is insoluble, the use of PSS as the counter ion allows for a stable suspension and, moreover, increases its processability.
The chemical structure of PEDOT:PSS is shown in Figure 2. PEDOT doped with PSS is commercially available as an aqueous solution in a highly conducting state (p-doped state). This solution (Baytron P, H.C. Starck GmbH, Goslar, Germany) yields films that have excellent stability under ambient environment and can operate over a wide pH range. Further enhancement of the film conductivity can be achieved by the addition of an appropriate organic compound to the PEDOT:PSS solution, e.g., ethylene glycol, dodecylbenzenesulfonic acid, dimethyl sulfoxide, or N-N-dimethylformamide.49–51 It has been proposed that organic solvents with one or more polar groups change the film morphology and enhance the film conductivity.50 The conformational changes of the PEDOT chains are a result of the interaction between the dipoles in the polar groups of the solvent and the positively charged PEDOT.
In a PEDOT:PSS-based OECT, PEDOT:PSS is in its pristine state that is highly conductive and can serve as the source (S), drain (D) and gate (G) electrode material for the transistor. In a PEDOT:PSS film, the mobile holes in PEDOT are compensated by the negatively charged sulfonate groups in PSS. This transistor, at zero gate voltage (Vg = 0.0 V with respect to ground), is inherently in the “on” state at which high current flows through the channel (Isd). Upon the application of a small positive gate voltage, the transistor turns “off ” as cations from the electrolyte medium are driven into the channel. This migration causes the electrochemical de-doping of the PEDOT, hence the decrease of the source–drain current, Isd.37 The de-doping effect is not uniform throughout the channel. In particular, the side of the film closer to the source electrode will eventually become more reduced compared with the side closer to the drain electrode which is positively biased with respect to the ground (source electrode). This is indicated in Figure 1 by the gradual variation in the color of PEDOT:PSS channel closer to the electrodes.
The reduction of the highly conducting form of PEDOT+ (oxidized state) to the less conducting form of PEDOT0 (neutral state) is reversible and occurs according to the following electrochemical reaction:6
PEDOT+:PSS− + M+ + e−?↔?PEDOT0 + M+ : PSS−
where M+ is a cation in the electrolyte medium, and e− is an electron from the source electrode. The migration of M+ into the channel causes the reduction of the PEDOT:PSS film. PSS− ions are too large to migrate out of the polymer matrix. Upon the removal of the gate potential, M+ ions diffuse back into the electrolyte medium, and the original conductivity of the active layer is restored. This switching of the PEDOT:PSS channel between its oxidized and neutral states can change its bulk conductivity over several orders of magnitude. This change in conductivity can be achieved with the application of only a few volts.
Typical I-V curves, characteristic of the device behavior, are shown in Figure 3 for different applied gate voltages. In general, the relationship between Isd and Vsd is almost linear for low, either positive or negative, source–drain voltage and all applied Vg. Upon the application of higher positive Vsd (first quadrant), the current increases at a higher rate until over-oxidation of the PEDOT:PSS active area occurs. This is a nonreversible state where the transistor is no longer functioning properly. With the application of a “higher” negative Vsd (third quadrant), the current starts to saturate and eventually remains constant for higher source–drain voltages. In Figure 3, for zero gate voltage, the PEDOT:PSS ECT is inherently “on”; the source–drain current is high, whereas for applied gate voltages, there is a significant current decrease. This decrease in the source–drain current has been attributed to the following two mechanisms. As the cations migrate into the PEDOT:PSS channel, the hole tunneling is disrupted (ion-leverage mechanism), therefore the hole mobility decreases.34 At the same time, the cation migration causes a reduction of the PEDOT:PSS channel (electrochemical mechanism); hence, the hole density decreases.52 The observed current decrease is likely due to the combination of both mechanisms.
Driven by the need for small, portable, and inexpensive sensors for point-of-care analysis, there is a demand for low-cost fabrication techniques of organic semiconducting devices. PEDOT:PSS-based OECTs can be patterned by standard photolithography and dry etching.53 Thin layers of PEDOT:PSS can then be deposited onto a variety of solid substrates using processes such as spin coating. When a metal (Pt, Au, or Ag) is used for the gate, source, and drain electrodes, then additional deposition and patterning steps are added to the process. This approach provides high-quality devices but might be too expensive for mass fabrication of sensor devices. Therefore, simpler and more versatile patterning techniques are desirable.
Soluble polymers such as PEDOT:PSS offer more options for simpler patterning techniques due to their ease of processing. Some of the existing alternatives to photolithography are screen printing, soft embossing, microcontact printing, and inkjet printing (i.e., piezoelectric, thermal, electrostatic, or acoustic). Among these techniques, inkjet printing is rapidly becoming one of the most attractive technologies. Its advantages lie on the patterning capability, high speed and low cost of the process, efficient use of material, and most importantly, the ability to print thin films on flexible or nonflexible substrates. PEDOT:PSS films deposited by commercial inkjet printer and by spin coating process have shown no appreciable differences in the electrochemical characteristics and surface properties of the films.54 PEDOT:PSS devices deposited by thermal inkjet printing have been used for chemical and biochemical sensing applications.54–56 Currently, thermal inkjet printing can provide patterning of polymers with 10 µm feature size.57 For devices where their operation is based on charge injection or charge accumulation induced by field effects, a high resolution is essential. However, the operation of PEDOT:PSS-based OECTs does not require particularly small features; therefore, inkjet printing represents a promising fabrication method. OECTs offer a unique opportunity for the development of robust multifunctional sensors in a very economical way.
PEDOT:PSS-based OECT in its simplest form can have a lateral architecture. In this case, PEDOT:PSS is used for all the features which are printed out next to each other in a single step. This lateral configuration consists of the PEDOT:PSS features, and the electrolyte medium defines the ECT. Adding a recognition element will convert the OECT into a sensing device. It has been recently demonstrated that biorecognition elements such as enzymes can be incorporated in the PEDOT:PSS solution, and can even be printed by thermal inkjet printing with minimal loss of enzyme activity.54,56
Another commonly used OECT architecture is the vertical configuration, where the gate electrode and the transistor channel are facing each other and the electrolyte medium is in between. This configuration can give a relatively faster response compared with the lateral one, because the distance between the gate electrode and the transistor channel can be made very small. However, the lateral geometry might have the advantage of greater exposure of the channel to the target analyte, especially in gas-sensing applications, as it offers a more open and easily accessible structure.
Applications in Sensors
Organic electrochemical thin-film transistors have been explored in various sensor applications. Their low operational voltage requirements, easy fabrication, biocompatibility, as well as their ability to efficiently transduce ions to electrons render them prime candidates for the development of sensor arrays. Some novel work on PEDOT:PSS-based OECTs for various sensor applications is described in the sections below.
Nilsson et al.13 exploited the capabilities of OECTs as a humidity sensor. They delivered an inexpensive flexible sensor which was fabricated by printing the commercially available PEDOT:PSS on thin polyester foils and paper as shown in Figure 4. A lateral geometry was used, and the patterned PEDOT:PSS layer was printed in a single step. A sulfonated tetrafluorethylene copolymer, Nafion® (DuPont Wilmington, DE), was used as the solid-state electrolyte forming a thick film structure covering the transistor channel and the gate electrode. Nafion® is known to change its ionic conductivity upon exposure to different humidity levels. Therefore, the electrochemical switching of the transistor depends on the ambient humidity level which can result in a change of the electrolytes' conductivity by several orders of magnitude. This group reported on an exponential decrease of the source–drain current at room temperature, when the relative humidity increases from 40 to 80%. More specifically, a current drop of two orders of magnitude is reported within the above-mentioned humidity range. The results are shown in Figure 5.
Ion-selective membranes are extensively used in membrane-based ion selective electrodes (ISEs) for trace metal analysis in environmental samples or the determination of inorganic ions in clinical analysis. For example, in clinical analysis, the monitoring of physiological electrolyte concentrations such as Ca2+, K+, Na+, and Cl− in various body fluids is very important for all organisms because various metabolic processes are related to them.58 Abnormal electrolyte concentrations may be either the cause or the consequence of a variety of disorders. The importance of this type of sensors has attracted a lot of interest in the last decades,59 which resulted in remarkable improvement of the detection limit to the order of picomolar (10−12 M).60
Several types of ISEs have been developed over the years using conducting polymers,18,61–63 including PEDOT:PSS as an ion-to-electron transducer.64,65 A Ca2+ selective PEDOT:PSS-based OECT has been demonstrated by Berggren et al.36 [ion-sensitive organic electrochemical transistors (IS-OECT)]. In this work, an ionophore-based solvent polymeric membrane, which coats the top of the PEDOT:PSS channel, provides the desired selectivity. The polymeric membrane on top of the active channel is immersed in an electrolyte solution containing 0.1 M KCl and 0.5 M of CaCl2. The transistor used in this case had a vertical geometry with the gate electrode immersed in the electrolyte solution on top of the configuration. The polymeric membrane is composed of 2-nitrophenyl octyl ether, poly(vinyl chloride), potassium tetrakis(4-chlorophenyl)borate (KTpClPB), and N,N,N′,N′-tetracyclohexyl-3-oxapentanediamide (ETH 129). The latter component serves as the Ca2+ selective ionophore or, in other words, the Ca2+ carrier through the membrane. Upon the application of a small gate voltage (Vg = 0.15V), this PEDOT:PSS IS-OECT had different responses for different Ca2+ concentrations in the electrolyte solution. In summary, the coupling of the polymeric membrane with the ECT provided an ion-sensing mechanism with a simple current readout.
Bernards et al.33 employed PEDOT:PSS-based OECTs coupled with biological recognition elements that enabled them to distinguish between monovalent and divalent cations. The recognition element in this case was a bilayer lipid membrane (BLM) where gramicidin was incorporated. Gramicidin is a well-known linear polypeptide which forms ion channels within the bilayer membrane. The channels are selective for monovalent cations, but are not permeable to anions or polyvalent cations. Monovalent cations and water can move through a pore that is formed by the peptide backbone.66–69 When a BLM is formed between the active channel and the gate electrode, assuming a vertical geometry as shown in Figure 6, the gating effect is completely suppressed. BLMs are shown to block ion migration in the electrolyte medium even for gate voltages higher than 0.1 V. Thus, upon the application of a gate voltage, the conductivity of the PEDOT:PSS channel remains unchanged. Addition of gramicidin into the BLM renders the membrane permeable to monovalent cations, K+ in the case of KCl electrolyte solution. This restores the gating effect that is observed by the source–drain current modulation results (Fig. 7). When a CaCl2 electrolyte solution is used instead of KCl, a very small current modulation is observed. This confirms the preference of monovalent cations by gramicidin channels in comparison to divalent cations. The incorporation of other ionophore proteins into the BLMs can provide different kinds of selectivity and will open more possibilities for sensing applications.
Enzymes are highly specific and selective catalysts, frequently used for clinical and diagnostic purposes. These functional proteins enable delicate and complex biological reactions to occur under mild conditions and with accelerated rates. Oxidoreductases are a category of enzymes that facilitate electron transfer through binding a target analyte on their active sites; mediators can be used to assist this transduction. The integration of enzymes into electrochemical transducers can provide excellent sensing devices for clinical diagnostic analysis.
The most widely used enzyme in sensor applications is GOx. This is attributed to the demand for commercially available glucose biosensors that offer a quick way for easy monitoring in diabetic individuals. Diabetes mellitus is a group of metabolic disorders of carbohydrate metabolism, a disease in which glucose is underused, producing hyperglycemia (high levels of blood glucose), or overused, producing hypoglycemia (low levels of blood glucose).58 With an estimate of 21 million people in the United States alone suffering from diabetes,70 the development of inexpensive glucose sensors is important.
As an example of this work, Zhu et al.12 utilized a PEDOT:PSS ECT to demonstrate its capability of glucose sensing at neutral pH. They employed a vertical transistor configuration where the active channel, source, and drain electrodes were deposited by spin-coating PEDOT:PSS. A Pt wire was used as the top gate electrode of the transistor. When the electrolyte medium consisted solely of phosphate-buffered saline (PBS), the application of a gate voltage resulted in a small decrease of the source–drain current, Isd. A similar response was obtained upon addition of GOx in the electrolyte solution. However, upon the addition of glucose in the GOx containing solution, a dramatic decrease of the Isd current was observed. This response can be seen in Figure 8. The inset in the same figure shows the relative response of the transistor with different gate voltages: −ΔIsd/Isd, where −ΔIsd/Isd = [Isd(Vg = 0 V) − Isd(Vg)]/Isd (Vg = 0 V).
Systematic studies show that the current modulation, Isd, had a monotonic dependence on glucose concentration within the range of 0.1–1.0 mM upon the application of a gate voltage, Vg (results not shown here). GOx catalyzes the reaction of glucose in the presence of oxygen and produces hydrogen peroxide (H2O2) and gluconic acid. The H2O2 is oxidized at the Pt electrode, while PEDOT+ of the active channel gets reduced in order for the device to maintain charge balance. The production of gluconic acid can induce changes in the pH levels of the electrolyte solution which is what many existing glucose sensors detect.21,28 However, in this case, the current modulation was measured in standard buffer solutions of different pH and was found to be independent of pH within a range from 5 to 9. This indicates that the sensing mechanism is entirely of electrochemical nature. Furthermore, the use of free-floating enzyme in the electrolyte medium decouples the biorecognition element from the transducer enabling simultaneous optimization of both. This approach offers great potential for the detection of a variety of analytes by simple utilization of the appropriate enzymes. Finally, by not immobilizing the enzyme, the enzymatic activity remains intact, and in addition, the detection time is decreased since the analyte does not need to diffuse into the polymer matrix.
Currently, the determination of irregular levels of blood glucose is based on glucose sensors that utilize finger stick assays. However, the method of collecting blood samples is painful and can cause stress to the patient. Alternative noninvasive methods to monitor blood glucose levels have long been considered, e.g., biological screening using urine, sweat, and saliva. An effort for a urine glucose-screening test was rendered fruitless as monitoring of urine glucose concentration lacks the sensitivity and specificity needed, and moreover, it does not indicate hypoglycemia.58 Efforts on developing a method for glucose determination by using sweat have not produced reliable results.71 A promising alternative to blood testing is the use of saliva samples. The typical glucose level in human saliva is within the range of 0.008 and 0.21 mM,71 which is out of the detection limits of several existing sensors. OECTs could potentially lower the current limit of detection by offering an inexpensive, disposable saliva sensor that could help millions of diabetic patients.
As an example, the work of Macaya et al.72 and Bernards et al.73 might help move the technology a step closer to the direction of noninvasive glucose testing. This group has developed a simple glucose sensor with sensitivity well within the clinical range for saliva testing utilizing a PEDOT:PSS-based OECT. The transistor consists of PEDOT:PSS active channel, source, and drain electrodes, and a Pt wire as the gate electrode immersed in the electrolyte solution from the top. They showed that although the response of the device was small when operated as a chemiresistor (Vg = 0 V), the response of the device when operated as a transistor was considerably larger due to the inherent signal amplification of these devices (Fig. 9). In addition, they were able to show that the sensitivity can be tuned by changing the gate voltage to cover the range from 1 µM to 10 µM.
Conclusions and Future Outlook
Over the last decade, PEDOT:PSS ECT appeared to be one of the most promising transducers for chemical and biological sensor applications. This is due to the enormous potential it presents for facile processing of small, portable, and inexpensive sensors ideally suited for point-of-care analysis. PEDOT:PSS-based OECTs can be used in gaseous or aqueous environments to detect a wide range of target analytes for a variety of possible applications in fields such as health care, environmental monitoring, and food industry.
Furthermore, the possibility of using inkjet printing technology for transistor fabrication is rapidly becoming one of their most attractive aspects. This process offers efficient use of material, high-speed, and low-cost fabrication, and most importantly, the ability to print thin films on flexible substrates. This capability, in combination with the ability of the material to integrate chemical or biological recognition elements, make PEDOT:PSS-based OECTs an ideal platform for sensor applications. OECTs can translate chemical and biological signals into electronic signals very efficiently. Other advantages of these devices include low-voltage requirements that limit the risk of hydrolysis in aqueous environments, and their inherent signal amplification that enhances sensitivity and allows for small sample volumes. Most of the applications of PEDOT:PSS-based OECTs as sensors will probably involve disposable devices for which the long-term stability and lifetime of the sensor are not issues. Finally, their ease of integration with microfluidics and incorporation into circuits enable them to move the current technology closer to the “lab-on-a-chip” concept.
Although a lot of progress has been achieved with the PEDOT:PSS-based OECTs for chemical and biological sensor applications, there is still work that needs to be done toward the practical realization of these devices. Primarily, there are many more recognition elements that can be used with these devices for sensing a vast amount of target analytes. This will allow for the realization of the design of arrays of sensors. Moreover, further development and optimization will be required for analyzing real samples in order to avoid interferences from other compounds, and to ensure high selectivity even in complex samples.
The authors would like to acknowledge Dr. Nikolaos Chalkias, John DeFranco, Dr. Daniel A. Bernards, Dr. Róisín M. Owens, and Seiichi Takamatsu for useful comments and discussions regarding this work. This work is supported primarily by the Nanoscale Science and Engineering Initiative of the National Science Foundation under NSF Award # EEC-0117770, 0646547 and the New York State Office of Science, Technology and Academic Research under New York State Office of Science, Technology and Academic Research (NYSTAR) Contract # C020071.
Maria Nikolou is currently a postdoctoral research associate at Cornell University where her research is focused on the development of biological sensors based on organic thin-film transistors. She received a B.S. in Physics with a specialization in microelectronics from the University of Crete (Greece). Her interests at that time were focused on growth and optical characterization of inorganic semiconductors. In 2000, she attended the University of Florida, where she received her Ph.D. in Physics (2005). Her graduate work was on “in situ spectroscopic studies of single-walled carbon nanotubes and conjugated polymers in electrochromic devices.”
George Malliaras is the Lester B. Knight Director of the Cornell NanoScale Facility and an associate professor in the Department of Materials Science and Engineering at Cornell University. He received a B.S. in Physics from the Aristotle University (Greece) and a Ph.D., cum laude, in Mathematics and Physical Sciences from the University of Groningen (the Netherlands). Prof. Malliaras is the recipient of the NSF Young Investigator Award, the DuPont Young Professor Grant, and a Cornell College of Engineering Teaching Award. He serves on the board of directors of Infotonics, a microelectromechanical systems (MEMS) fabrication facility designed to provide a rapid pathway to commercialization, is the chairman of the editorial board of the Journal of Materials Chemistry, and serves on the editorial board of Sensors. His research interests span several aspects of organic electronics, including structure and morphology of organic thin films, their processing and patterning, charge transport and injection in organic semiconductors, device physics, and applications of organic devices in biosensors.