Several metal-oxide semiconductor materials have been used as active elements in FETs ^[33][34][41,42]. The most common ones, along with their respective electrical properties and examples of biosensing applications. Several criteria should be considered when selecting a metal-oxide semiconductor for chemical/biosensor FET-based PoCT. First, the specific applications should be considered. For instance, target concentration levels and diagnostically relevant cut-offs, sample nature (e.g., blood, saliva, urine, sweat), singleplex or multiplexed measurement, and intended implementation settings (e.g., remote/low resource area, emergency department). These considerations influence the selection of the most suitable materials and designs. Second, metal-oxide semiconductor materials’ commercial availability and their electrical/mechanical stability in the measurement environment. For example, in the case of gas/organic molecules sensing for a breath analyzer, where a high temperature is usually required, both the semiconductor and the metallization material must be able to withstand that temperature. On the other hand, the material must be sufficiently stable in the intended solution for wet-chemical and biological sensing.

2.2. Operation of Metal-Oxide FET-Based Biosensor

Several types of analytes/biomarkers (e.g., antigen, nucleic acid, virus and virus protein/capsid, bacteria) have been successfully implemented in metal-oxide FET sensors. A FET-based biosensor typically relies on the integration of an ISFET and bioreceptors with suitable binding affinity and specificity to the target analyte. The nature of the target analyte influences the design of the overall assay and detection mechanisms. As with other FET sensors, different classes of bioreceptors can be used, the most common being antibodies and antibody-fragments, enzymes, nucleic acid-based probes, and aptamers. Bioreceptors are typically immobilized on the semiconductor material (the sensing channel) and display biding sites to capture the target analyte(s). The surface potential of the FET sensors and, therefore, the channel conductance changes when these bioreceptors bind with the targets. The channel conductance variation resulting from the presence of the target can be correlated to a sensitivity index by measuring the changes in the drain current. The presence of the target on the sensing channel is typically detected either directly (label-free operation) or via a secondary amplification step. In addition, competitive and displacement affinity assays can also be used, with or without amplification. Label-free assays rely on the intrinsic charges present on the target at the measurement condition ^[7][35][15,74] or conformational changes induced by its binding and are conceptually easier to design and implement than two-step sandwich assays. However, the issue of the sensor non-specific fouling by biomolecules present in the test matrix typically imposes a limit on the analytical performance. In addition, it is worth mentioning that the Debye length, which governs the extent to which the analytes’ charges affect the FET channel electrical surface characteristics, should also be considered, as it severely limits the direct detection of the analytes in physiological solutions. Various approaches have been proposed to overcome the Debye length effect. For example, an oligonucleotide stem-loop bioreceptor was successfully used for adaptive target recognition in ultrathin In₂O₃ FETs ^[36][55]. Another approach modulated the Debye screening length without affecting the FET’s channel surface by fabricating a Meta-Nano-Channel Bio-FET to tune the double-layer shielding electrostatically ^[37][75]. Signal amplification is typically achieved via an enzymatic reaction, either directly if the bioreceptor is an enzyme or indirectly via a secondary probe conjugated with an enzyme. Direct enzymatic assays are easier to implement than two-step sandwich assays; however, only a limited number of analyte/bioreceptor pairings are available. Biochemical sensing based-FETs can be carried out either in quasi-real time or in steady-state manners. In the former case, a short pulse of gate voltage is applied to enable high-frequency transient measurements while the drain current is continuously monitored, with the sensing channels being immersed in the measuring solution ^[10][14][38][18,22,49]. Real-time measurements are beneficial for accessing binding kinetic information. However, measuring binding kinetic data ideally requires a fluidic system to deliver the sample and buffer solutions ^[39][45] and is, therefore, more complex to implement. In the steady measurement, the affinity FET sensors are exposed to the sample containing the target molecules to allow for the binding reaction to take place and eventually reach binding equilibrium or saturation state. Drain currents before and after completion of the assay are acquired, and these measurements are used to extrapolate the concentrations of the target analytes based on a calibration curve ^[6][40][14,79].

3. Development and Integration of Metal-Oxide Nano-FET Biosensor for PoCT

3.1. Fabrication of Nanoscale Metal-Oxide Semiconductors: Vapor-Based Approaches

Scaling down metal-oxide semiconductors into the nanoscale increases their analytical performance due to the comparable size of metal-oxide nanostructures to the targeted molecules ^[41][82]. Metal-oxide nanostructures, including nanowires (NWs), nanoribbons (NRs), nanobelts, nanorods, and nano-thin films, can be grown using vapor-phase fabrication techniques, which include chemical vapor deposition (CVD) and physical vapor deposition (PVD). Vapor-phase-grown metal-oxide nanostructures. Metal-oxide NWs and nanobelts can be synthesized using vapor processes such as CVD and PVD. As detailed previously, two mechanisms typically govern these fabrication methods: vapor-liquid-solid and vapor-solid mechanisms ^[42][83]. CVD synthesizes nanostructures (NWs, nanobelts, or nanorods) through chemical reactions in the vapor phase with the assistance of a noble metal catalyst (e.g., Pt, Au). Different metal-oxide NWs have been synthesized through laser ablation CVD procedure ^{[42][43][44][45]}[83,84,85,86]. Here, NWs are grown in a pre-coated substrate (usually Si/SiO₂ substrate) with a catalyst (either a thin film or nanoparticles). During the reaction of the laser beam with the targeted material, clusters or droplets of the targeted material are generated and form the NW backbone based on the catalyst size in the pre-coated substrate. A drawback of using a thin film catalyst is that this produces significantly different NW diameters, so it lacks reproducibility ^[7][15]. On the other hand, monodisperse metal clusters catalysts allow more control over the NW diameters as the nanocluster catalyst’s size guides the formation of the metal-oxide NWs ^[46][47][87,88]. Since the synthesis reaction takes place at a high temperature (>770 °C), NWs grown with CVD are crystalline as reported for SnO₂ ^{[48][49][50][51]}[89,90,91,92], In₂O₃ ^{[45][52][53][54]}[53,86,93,94], and ZnO ^[10][55][56][18,44,46]. On the other hand, PVD produces nanostructures by either thermal evaporation or gaseous plasma. In the thermal evaporation process, a high temperature evaporates the material powder under a vacuum. To prepare a functional FET device from metal oxide nanostructures grown in the vapor phase, a patterning/transfer technique is typically required, which aims to transfer the as-synthesized nanostructures to a secondary (receiver) substrate. The patterned semiconductor nanostructures can then be integrated with electrical contacts and isolated using a passivation layer. Shadow masks and conventional photolithography with an etching or lift-off process are mainly used in the patterning step ^{[15][16][57][58][59]}[23,24,99,100,101]. Conversely, dip-coating is the most used method for transferring metal-oxide FETs to the patterned substrate ^[60][98]. In this methodology, the as-grown semiconductor materials are transferred from the growth substrate into an organic solvent (usually an alcoholic solution) using ultrasonication. The resulting suspension is then dispensed drop-by-drop onto the secondary substrate. Although there are many reports of successful transfer of CVD/PVD-fabricated metal-oxide nanostructures using dip-coating (e.g., In₂O₃ ^[60][61][62][51,98,102], ZnO ^[10][55][18,44], and SnO₂ ^[49][51][63][90,92,103]), this procedure is limited by the substantial damages occurring in the metal-oxide nanostructures during the ultrasonication step. Alternatively, CVD/PVD-fabricated metal-oxide semiconductors can be transferred with direct contact printing ^[42][51][83,92]. Vapor-phase thin film deposition techniques. When combined with a patterning process, metal-oxide FET can also be fabricated by top-down vapor-phase thin film deposition approaches. The most commonly used vapor-phase thin film deposition techniques for this purpose include sputtering, atomic layer deposition (ALD), and pulsed laser deposition (PLD). Sputtering is a widely used thin film deposition technique ^{[15][16][57][58]}[23,24,99,100]. Metal-oxide FET fabrication based on sputtering has several merits, namely (1) relatively low-temperature processing (from room temperature to a few hundred degrees), which makes it readily compatible with a wide range of substrates, for example, flexible plastic substrates ^[64][65][104,105], silicon/glass substrate, (2) efficient control of the thickness and morphology of the metal thin films by modulating the sputtering conditions (e.g., sputtering power, time, and gas flow rate), and 3) control of the composition of the deposited film by co-sputtering different material targets ^[65][105]. Metal-oxide thin films can be deposited either from the respective metal-oxide targets in an inert atmosphere or from a pure metal target within an oxidative gas environment, which is typically referred to as reactive sputtering ^[66][106]. Sputtering of oxide targets in an inert atmosphere is superior due to the simplicity and superior reproducibility of the process, while reactive sputtering is more sensitive to contaminants and process parameters.

3.2. Fabrication of Nanoscale Metal-Oxide Semiconductors: Solution-Based Approaches

Most metal-oxide nanostructures/thin films can be synthesized through solution-based routes (e.g., sol-gel ^[67][68][69][113,114,115], and wet-chemical synthesis ^[43][68][70][84,114,116]). In this paradigm, the nanoscale metal-oxide elements are subsequently deposited on substrates, for example, using spin coating, spray coating, bar coating, and printing ^[65][67][71][105,113,117]. When combined with the patterning technique, the solution-based route is cost-effective and more compatible with large-area deposition than vapor-based techniques ^[67][68][72][113,114,118]. In a typical sol-gel process, a dissolved metal salt precursor is spin-coated or printed directly on a substrate at room temperature. A high-temperature annealing (200–350 °C) step is then used to convert the precursor framework into the desired metal-oxide framework by decomposing and desorbing by-products of the synthesis reaction ^[45][71][73][86,117,119]. This step determines the metal-oxide FET’s electrical properties. In₂O₃ thin films ^[17][73][74][25,119,120], In₂O₃ nanoribbons ^[30][72][38,118], and ZnO thin films ^[69][115] have been successfully fabricated by a sol-gel process with electron mobility μ_FE > 10 cm² V⁻¹ s⁻¹, current on/off ratios from 10⁴ to 10⁷, and SS values from 81 mV/decade to 600 mV/decade ^{[17][72][73][74]}[25,118,119,120]. High-temperature annealing requirement is a drawback of the sol-gel approach. It restricts the substrate material choice as many polymeric substrates cannot tolerate the required high temperature. Therefore, new strategies to lower the annealing temperature are needed to combine this approach with flexible substrates, for example, using novel precursors and/or developing innovative annealing methods ^[69][71][115,117].

3.3. Surface Functionalization of Metal-Oxide-Based FET Sensors

A range of biorecognition approaches, including enzyme/substrate, antibody/antigen interactions, and nucleic acid hybridization, has been exploited to impart selectivity to biosensors for a specific molecular target ^[75][124]. These bio-affinity recognition methods have been successfully implemented to realize experimental FET biosensors for testing many diseases with high prevalence, such as cancers (e.g., protein biomarkers) ^[13][67][21,113], cardiovascular diseases (e.g., protein biomarkers) ^[76][77][125,126], infectious diseases (e.g., nucleotide biomarkers) ^[29][37], and diabetes (e.g., protein and enzyme biomarkers) ^[16][24]. An essential step in preparing a nano-FET biosensing device is introducing a molecular bioreceptor with high and specific binding affinities to the target of interest on the surface of the FET. Metal-oxide-based FETs are, therefore, typically functionalized first with a chemical agent to enable covalent immobilization of the specific bioreceptors on their surface. In this regard, organosilanes and phosphonic acids are widely used. Organosilanes covalently bind to many metal-oxides such as In₂O₃ ^{[30][58][78][79]}[38,78,100,127], SnO₂ ^[31][39], ZnO ^[55][44], SiO₂ ^[80][81][128,129], Fe₃O₄ ^[82][130], and β-Ga₂O₃ ^[83][131]. Presently, 3-aminopropyltriethoxysilane (APTES), 3-mercaptopropyltrimethoxysilane, and 3-(trimethoxysilyl) propyl aldehyde are the most used amongst the organosilanes. Silanization is carried out either in vapor or liquid phases using a mixture of ethanol/water 95%/5% (v/v), or toluene ^[80][84][85][128,132,133]. Prior to surface functionalization, metal-oxide surfaces are often activated using UV-ozone or oxygen plasma. This activation step generates –OH groups on the surface, facilitating the reaction with the organosilanes. The mechanisms and characteristics of metal-oxide functionalization with organosilanes have been reported in detail elsewhere ^{[80][82][86][87]}[128,130,134,135]. In the case of APTES, the amine functional groups can be subsequently reacted with a mono or hetero bifunctional linker, such as glutaraldehyde, to introduce reactive groups able to covalently conjugate bioreceptors including DNA probes ^[88][136], antibodies ^[13][89][21,137], proteins ^[89][137], or enzymes ^[10][18]. After immobilization, unreacted groups (e.g., CHO) are usually blocked, and the surface is passivated toward reducing non-specific adsorption events ^[84][90][91][132,138,139].

4. Tailoring Metal-Oxide Nano-FETs toward Point-of-Care Testing Applications

PoCT as a diagnostic device must be fast time-to-result (ideally in less than an hour), cost-effective, portable, instrument- and technician-free, robust enough in the implementation environments (i.e., weather and shelf-time), and sensitive. Recent advances in metal-oxide nano-FET make it possible to meet the PoCT requirements satisfactorily.

4.1. Sample Processing Integration

Besides a few notable exceptions, FET biosensors remain to date mostly in the research and development realm, with R&D focussed mainly on sensor development ^[92][93][94][145,146,147] and bioassay/detection elements ^[95][96][97][144,148,149]. However, important underlying issues associated with sample processing typically required prior to/during PoC testing have received far less attention, contributing to the limited real-life deployment of FET-based PoC platforms. The need for sample processing in FET biosensors lies within the complexity of biofluids typically used for PoCT, including blood, saliva, urine, and to a lesser extent sweat. Such biofluids vary substantially in their composition (e.g., pH, protein concentration, ionic strength) and typically interfere with biological assays and/or analytical performance of the sensors. Various biofluid processing approaches are commonly used with FET biosensors, including blood filtration/desalting ^{[40][84][91][98]}[79,132,139,150], centrifugation/washing ^[14][22], chemical pre-treatment, microfluidic biomarker pre-enrichment ^[99][151], and novel sensing methodology ^[100][152]. These approaches typically rely on the intervention of trained staff and the use of external analytical equipment (e.g., centrifuge, micropipette, pump), which is often challenging in PoC settings. In terms of simplifying PoCT and eliminating analytical errors typically associated with sample manipulation by operators, integrated approaches have been actively explored. This includes operation by benchtop instrument through supply energy or timed triggering mechanism ^{[101][102][103][104]}[153,154,155,156], internalized vacuum/chemical reactions, and capillary pump ^{[105][106][107][108]}[157,158,159,160]. Various strategies have been investigated to integrate sample processing features within FET biosensing platforms for enabling PoCT. The sample ionic strength is especially significant in the context of the FET platform as the Debye screening length (λ_D) is a key parameter dictating the sensitivity to a given analyte ^{[29][109][110][111][112]}[37,161,162,163,164]. Briefly, the presence of counter ions in the measurement solution effectively screens the charge of the analyte that can be sensed at the FET surface, with the screening length in physiological solutions being below 1 nm ^[113][114][165,166]. To circumvent this issue, molecular probes with smaller dimensions have been used, including cleaved antibodies and aptamers. Alternatively, polymeric biointerfaces have been shown to extend the λ_D of the sensor and consequently enable FET measurement at high ionic strength. For example, Gao et al. utilized polyethylene glycol (PEG), small-molecule spacer, and aptamer on Graphene FET biosensors to directly measure prostate-specific antigen (PSA) in physiological conditions ^[115][167]. 

4.2. Analytical Validation and Regulatory Approval of Point-of-Care Testing Devices

As with all medical devices, PoCT devices should be thoroughly validated and subjected to regulatory approval prior to being implemented and commercialized. A critical aspect here is to demonstrate that the device’s real-world performance is acceptable and that it complies with regulatory standards and requirements of a given jurisdiction (e.g., United States-United States Food and Drug Administration, FDA; European Union-European Conformite, CE; Australia-Australia Therapeutic Goods Administration, TGA; China-National Medical Products Administration (NMPA); Japan-Japan Pharmaceuticals and Medical Devices Agency, PMDA). The bench-to-bedside journey for PoCT devices is complex, expensive, and undoubtedly associated with a “valley of death”. Cross-disciplinary collaboration with all key stakeholders is, in most cases, essential to successfully navigating the various regulatory processes required for developing and commercializing PoCT devices. Validation of the analytical performance of PoCT platforms. Validating the analytical performance is a critical step in the development of any IVD medical device, including PoCT biosensors. A consideration specifically relevant to the PoCT platform is that it must be designed to achieve the required level of performance, taking into account the skills and the means available to the intended users, for example, lay persons. Testing should therefore consider the variation that can reasonably be expected in the user’s technique and environment. The assessment of the analytical performance should include all aspects relevant to the intended use of the PoCT device, including analytical sensitivity (e.g., the limit of detection or limit of quantitation, inclusivity) and specificity (e.g., interference, cross-reactivity), accuracy (derived from trueness and inter/intra assay precision) and linearity (as applicable) ^{[116][117][118]}[52,169,170]. Regulation of PoCT biosensor. For most PoCT diagnostic devices, regulatory approval is mandatory prior to the product launching to the market. In addition, post-market surveillance is also necessary and provides additional insight into the patient population and the use of the device in the real world. The regulatory oversight is not only applied to the PoCT device itself but also to any associated software, consumables (regents, calibrator), and user instructions ^[117][169]. Medical PoCT biosensors are classified under low and medium risk to the user, and low complexity tests are usually streamlined in most of the regulatory landscapes, while higher risk tests are subjected to more stringent regulation.  It is worth noting that new PoCT biosensors where there are no approved/predicated technologies are automatically classified as Class III and must undergo a full pre-market approval process in the US FDA landscape. On the other hand, class III PoCT biosensors with existing predicated technologies can apply to be accessed as class II devices ^[119][173]. 

5. PoCT Adoption Barriers and Limitations

5.1. PoCT Performance Issues

The limited performance—real or perceived—of some PoC tests (i.e., specificity, sensitivity, and precision) compared to tests performed within centralized laboratory facilities may impede their adoption and utility. Poor PoCT device performance, particularly “false negative results”, can have serious consequences. Performance issues are also inherently linked to the issues associated with the quality control of PoCT devices.

5.2. Challenges and Considerations for Metal-Oxide FET-Based PoCT

Metal-oxide FETs are cost-effective devices derived from the availability of well-established fabrication protocols compatible with scaled-up manufacture. In addition, their performance is high enough to meet the requirements of most PoCT applications. They, therefore, provide an excellent compromise between ultra-high-performance solid-state sensors that typically suffer from limited manufacturability and low-sensitivity sensors that can be readily mass-produced. Despite these advantages, several issues should be considered. A critical challenge associated with the implementation of metal-oxide FET sensors is batch-to-batch performance uniformity. To mitigate this issue, it is essential to consider all factors that affect their performance. For example, it has become evident that the characteristics of metal-oxide semiconducting materials are controlled not only by their structure and geometry but also by the presence of functional defects and their crystal structure. Importantly, FET-based PoCT devices should have a small footprint to enable integration with signal processing systems. The inherently small size of FET sensors increases compatibility with wearable technologies, which is currently attracting a huge amount of research. In addition, impedance matching between the subsystem units should be considered, as it affects the signal capture within these units’ signal-to-noise ratio level ^[75][124]. Finally, an often-overlooked consideration is the issue of the packaging and shelf-life of FET-based sensors. The environmental stability of the metal-oxide FET itself and that of the biointerfacial layers containing the bioreceptors should be considered and optimized to ensure compatibility with PoC settings, where substantial variations in storage conditions are likely to occur.

5.3. Economic Considerations

The economic dimension is a prominent barrier to the adoption of PoCT. In general, the cost per test is higher for PoCT than that associated with conventional batch testing in centralized laboratories. Regarding the PoCT implementation cost, there are both direct and indirect costs associated with the implementation of PoCT. While direct costs are relatively straightforward to evaluate ^[120][185], indirect costs related to operational aspects, such as staff training, quality assurance, laboratory accreditation, etc., are often difficult to cost. But while accurate data is lacking, the implementation cost of PoCT is likely higher compared with testing conducted within centralized facilities.  Finally, it is worth noting that, in most cases, reimbursement schemes are not adjusted to cover the direct and indirect additional costs associated with PoCT ^[121][187]. For instance, the UK’s central fund for clinical pathology provides the same fee for PoCT as for centralized testing, regardless of the workload or the patient care pathway ^[122][188]. This is likely a further barrier to adoption, and reimbursement should consider the patient care pathway to overcoming it.