2.1. Graphene Properties
Graphene, or single atomic thick carbon, is the first purely two-dimensional (2D) material to be obtained [
18]. Graphene is made up of carbon atoms which are bound to three others with a 120° bond angle, resulting in a hexagonal lattice arrangement of
sp2-hybrised carbon [
19]. The 2D nature and hexagonal carbon arrangement is the basis of graphene’s high specific surface area (2630 m
2/g), a trait which is particularly advantageous in biosensing applications [
20]. Graphene is considered attractive for electronic applications due to its intrinsically exceptional ballistic charge transport [
18]. Experimentally, carrier mobilities have been reported to be about 2 orders of magnitude larger than the “gold-standard” semiconductor, silicon. Carrier mobilities have been known to exceed 10
7 cm
2·V
−1·s
−1 in graphene that has been decoupled from bulk graphite, to be as high as 10
5 cm
2·V
−1·s
−1 in suspended graphene devices [
21], and about 4 × 10
3 cm
2·V
−1·s
−1 for CVD graphene on a SiO
2 substrate. Moreover, graphene material can be manufactured in large quantities and relatively cheaply, therefore making it a suitable substrate for large-scale electronic device manufacturing [
22].
Graphene consists of two energy bands, the valence band (VB) and the conductance band (CB), which hold holes and electrons, respectively [
23]. The arrangement of the carbon atoms of graphene in a honeycomb lattice creates a completely full VB and an empty CB, as depicted in
Figure 1 [
19]. The two bands intersect at a point called a Dirac point, or the
K and
K’ points in the Brillouin zone. At the point where they meet, depicted by the Dirac voltage (V
D) in V
g–I
DS measurements, the Fermi level passes across. This Fermi level can be tuned and adapted because of doping by external influences, such as electron deficient (p-doping) or electron rich (n-doping) molecules [
18], therefore essentially causing a shift in the V
D to a more positive voltage (p-doping) or to a more negative voltage (n-doping). The V
D can therefore be monitored and utilized as a means of sensing biological molecules. The electronic properties, such as the V
D, carrier mobility, and resistance, can be influenced by many external sources, these include: applying an electrical field, charged moieties near the graphene’s surface, or by chemically modifying the surface, such as chemical binding to the graphene both covalently and non-covalently [
18].
2.2. G-FET Development
Graphene FETs are generally fabricated using micro fabrication techniques, such as photolithography coupled with metal evaporation or physical vapor deposition (PVD), to pattern and develop the device contacts. The graphene is either then transferred from a copper substrate used for its growth (CVD graphene) or from exfoliated graphene on to a patterned device [
24]. Alternatively, a bulk graphene layer (CVD graphene on SiO
2/Si or epitaxial graphene) is plasma etched away to form a channel [
25]. Many G-FETs produced in this manner are highlighted in
Table 1.
The channel is then modified to detect target biomarkers by immobilizing bioreceptors onto the graphene channel. This can be done directly (adsorption) or through a linker molecule. The immobilization of a highly specific bioreceptor (a process termed biofunctionalization) to the graphene surface induces chemical specificity towards the target biomarker. Such receptors may include amino acids, enzymes, antibodies, aptamers, or indeed any selective and specific molecule [
26]. However, if a linker molecule is required, the graphene channel must first be chemically functionalized to enable the immobilization of the bioreceptor. The chemical functionalization of graphene can be also be used to tailor the electronic properties of graphene via doping and band-gap engineering effects, produced by chemical modification or adsorption of molecules on to the graphene [
18].
The functionalization of graphene with a linker molecule can be performed through covalent binding to the carbon atoms of the hexagonal matrix or by non-covalent binding to the graphene by electrostatic and/or weak Van der Waals forces [
18]. A wide range of potential functionalization chemistries, such as halogenation, hydroxylation, epoxidation, carboxylation, amination, alkylation, and azidation, have been developed for graphene [
27]. The presence of
sp2 carbon atoms makes the graphene surface a potential candidate for covalent bonding [
28]. Covalent chemistries used to make graphene functional include fluorination [
29] and hydrogenation [
30] by plasma treatments. Also utilized is free-radical addition to the carbon atoms of the hexagonal matrix [
31], such as diazotization [
32]. Other covalent methods include the covalent attachment of polymers such as PEG [
31] and silanization by 3-aminopropyltriethoxysilane (APTES) [
33]. Tehrani et al. demonstrated the development of a G-FET for cancer risk biomarker (8-OHdG) with a limit of detection of 0.1 ng·mL
−1 using the diazonium functionalization chemistry [
34]. Teixeira and co-workers reported the detection of human chorionic gonadotropin (hCG) at 0.62 ng·mL
−1 using an epitaxial G-FET functionalized using the APTES method [
33]. Although covalent chemistry has proven to be successful, it also creates undesirable disruption to the
sp2 nature of the carbon atoms. As a result, the
sp2 hybridization will be converted to
sp3 hybridization [
28], which disrupts the electron structure of graphene, and therefore diminishes the excellent and desirable electronic properties of graphene. Therefore, other avenues of graphene functionalization have been explored [
18].
Non-covalent functionalization is dominated by the physisorption of molecules to the graphene through weak Van de Waals forces [
18]. More specifically, this non-covalent functionalization often occurs through an interaction between the π-electron cloud of the graphene and the functional molecule, otherwise known as π–π stacking. Graphite (bulk graphene) is an example of π–π interaction. Graphite is multiple layers of graphene sheets stacked upon one another through an interaction between their respective π-electron clouds [
31]. Since this non-covalent functionalization of graphene occurs in this way, the
sp2 nature of the carbon atoms is not affected. Therefore, the electronic and structural properties are not severely disrupted [
18], making this a desirable method of functionalization for G-FET biosensor development. Often, the molecule used for functionalization has a polyaromatic hydrocarbon base, such as benzene, naphthalene, or pyrene, with pyrene exhibiting a strong affinity towards graphene through π-stacking [
35]. Chen et al. demonstrated the effect of some of these electron withdrawing and electron donating molecules on the graphene’s electronic properties. It was reported that functionalization with tetrafulvalene (TTF), an electron donor, acts to p-dope the graphene, whilst an electron acceptor, hexaazatriphenylene-hexacarbonitrile (HATCN), acts to n-dope the graphene. However, both remained non-destructive to the graphene’s electronic and structural properties [
18]. Furthermore, functionalizing the graphene surface using a pyrenebutanoic acid succinimidyl ester (PBASE) through π-stacking is attractive, as the pyrene base of this molecule exhibits a strong affinity to the graphene sheet, whilst the succinimidyl ester provides a binding site for amines of various biomolecules, including antibodies, enzymes, bacteria, and nucleic acid probes [
25,
36,
37,
38,
39,
40]. Moreover, several non-covalent functionalization techniques have been developed to decorate the graphene surface using metal nanoparticles, such as gold [
41], platinum [
42], palladium [
43], and zinc oxide [
44]. Metal nanoparticles can be deposited onto the graphene channels by immersing the channel into the metal salt solution, electrochemical deposition, or by a chemical reduction process. Gutes et al. reported that the nature of the metal dictates the size and densities of the as-prepared metal nanoparticles, despite the same experimental conditions. For example, platinum metal appeared to form smaller particles with lower density when compared to gold and palladium [
43]. Cai et al. utilized gold nanoparticles on a G-FET to create a binding site for a sulphur-terminated biorecognition molecule. Moreover, Cai et al. reported the presence of nanoparticles to increase the active surface area of the G-FET, which in turn improved the sensitivity by providing more binding sites for biomolecule immobilization [
41].