Several currently available techniques to measure gaseous formaldehyde require the vapor to be adsorbed onto a filter or into a liquid solution, which is then further analyzed using methods such as electrochemical detection, ion chromatography, high performance liquid chromatography [9], voltammetry [10,11], or photometric or fluorometric determination. These methods are not suited for real-time monitoring, and they often require large, expensive laboratory equipment.

Bioelectronic sniffers for gaseous formaldehyde have been developed which incorporate immobilized enzymes such as aldehyde dehydrogenase (ALDH) or formaldehyde dehydrogenase (FALDH) [12,13]. The enzyme electrodes are located in a liquid compartment which is separated from the gas compartment by a diaphragm. In this way, formaldehyde in the gas phase can be detected amperometrically [14]. Formaldehyde dehydrogenase requires the presence of a co-factor, β-nicotinamide adenine dinucleotide, which enables formaldehyde conversion. Storage stability of enzyme-based sensors may be an issue. Kataky et al. reported a disposable FALDH-based sensor with screen-printed electrodes [15]. The enzyme and co-factor were placed behind a polyurethane membrane, and this device had a 50% loss in sensitivity after two weeks when stored at 4 °C, but <10% decrease in response after two weeks when stored at room temperature. Achmann et al. used a Teflon membrane to separate the liquid phase from the gas phase and phenothiazine (PT) as a mediator to detect the formed NADH [13,16]. They reported a linear response in the tested range (1–15 ppm) with a sensitivity of 1.9 μA/ppm and a detection limit of about 130 ppb. The sensor showed no significant response to other tested chemical gases. The group later modified the sensor and increased the sensitivity to 76 ppb [17]. Korpan et al. used the enzyme alcohol oxidase (AOX) to develop a potentiometric formaldehyde sensor using a field-effect transistor which had stable response for more than 60 days when stored at 4 °C [7].

Gaseous formaldehyde can also be directly measured using cataluminescence [18], chemiluminescence [19], colorimetry [20,21], infrared absorption [22], amperometric detection [23-25], or using sensors incorporating semiconductor metal oxide thin films [26], metal oxide films [27,28], or metal oxide nanoparticles [29]. Portable units have detection ranges from 0–10 ppm, but at low concentrations (<0.1 ppm) there is significant cross-sensitivity to alcohols and other interferents. Although some metal oxide sensors operate at temperatures as low as 95 °C [26], the more typical operating range is from 300 to 400°C, thus requiring high power consumption [30].

To date, the majority of microscale formaldehyde sensors have relied on metal oxides as the sensing material. In metal oxide sensors, electron donors or acceptors in the gas phase adsorb onto the metal oxide. At high temperatures (>200 °C), the adsorbed species can exchange electrons with the metal oxide: an acceptor molecule will take electrons and reduce its conductivity, while an electron donor will increase the conductivity. These surface interactions are thermally activated. The absorption probability of the gas onto the surface can be predicted by a Langmuir isobar which limits the sensor operation at the high-temperature side, and the surface combustion of the adsorbed analyte limits the sensor operation at the low temperature side. The spontaneous desorption of the gas from the surface also has a minimum thermal energy [30]. These effects combine to determine the temperature window of operation for a given material. Commonly used materials include SnO₂, TiO₂, and WO₃, which are metal oxide semiconductors. These materials are typically thick films in commercially available sensors. Many authors report as well an enhanced response magnitude when doping the metal oxides with metal nanoparticles [31-33]. The underlying interaction between the gas and the solid involves the catalytic oxidation of the target gas over the surface. In most cases adsorbed oxygen is utilized for the oxidation reaction and the consumption of surface oxygen reduces the surface space charge which in turn is manifested in a change in electrical conductivity. If the bare oxide is not active enough the presence of metal particles such as Au, Pd, Pt, Al enhances the adsorption/desorption properties of oxygen on the metal oxide surface and therefore as well the catalytic oxidation of gases.

Sensors based on metal oxide semiconductors are inexpensive and relatively easy to use: they convert a chemical concentration directly into an electrical output by simply measuring the resistivity of an exposed thin film [30]. A key issue when using metal oxides is the requirement of elevated temperature of operation. The most simple approach is using an external heat source [27,34]. The substrate under the thin sensing film is placed on a heated substrate, or hotplate (Figure 1a). Although this setup is not suitable for direct miniaturization, it remains very useful for characterizing thin films.

Other work was done by depositing the sensing film onto a ceramic tube [35,36] with electrodes and a heating wire (Figure 1b). This ceramic tube structure can be welded onto a standard electronic packaging base element. The temperature is controlled by the current flowing through the heating wire. Xu et al. used this configuration to detect formaldehyde gas with a mixed oxide of ZnO/ZnSnO₃ doped with Au [32]. They found that exposure to 50 ppm formaldehyde induced a 34.5 times change in the conductivity; the sensor also had high selectivity against ammonia, benzene, toluene, and other interfering gases. Huang et al. [35] used a sol-gel process to dope rare earth oxides LaFeO₃ with Zn. The obtained powder was dispersed in a polyvinyle acetate (PVA) solution and then coated onto the ceramic tube (Al₂O₃). To improve stability and repeatability the tube was calcinated at 400 °C for 2 h. Zhang et al. [36] used the same ceramic tube and LaFeO₃ powder but doped it with Pb [36]. Chen et al. fabricated CdO-In₂O₃ powder through calcinations on ceramic tubes at different temperatures [26,37]. They report the best sensing properties when calcinated at 650 °C of about an 80 fold increase in electric resistivity when exposed to 10 ppm of formaldehyde. They later doped the CdO-In₂O₃ film with SnO₂ to further increase the sensitivity [37]. The improved sensing layer showed a sensitivity defined as the ratio of the electrical resistance in air to that in gas of 559 when exposed to 300 ppm formaldehyde.

Wang et al. used the ceramic tube structure to characterize the sensing properties of SnO₂ doped with hydroxyl-functionalized multi-wall carbon nanotubes (MWCNTs) [8]. They report a much higher response to formaldehyde than that of undoped SnO₂. The lowest concentration of formaldehyde detected by a 5 wt% MWCNT doped SnO₂ sensing layer was 0.03 ppm. They explain the increase in sensitivity with the high adsoption capacity of MWCNTs. The material showed a response when exposed to acetone, methanol, toluene, benzene and ammonia. However the response was twofold stronger towards formaldehyde.

The wide-scale deployment of metal oxide sensors operating at high temperature in large area sensor networks may in future be limited due to their high power consumption. Some gas sensors have also been found to have gas sensitivity at room temperature under ultraviolet (UV) illumination [30], and the mechanism for photoinduced reactivity of semiconductors such as TiO₂ has been widely studied [46,47].

Metal oxides such as TiO₂, ZnO, and WO₃ exhibit a small and slow low-temperature gas sensitivity that is enhanced with UV light. Under UV illumination, the photons have sufficient energy to excite valence band electrons to the conduction band, which then migrate to the surface of the solid. These electrons participate in reactions with molecules adsorbed at the solid surface. TiO₂ is one of the most commonly used photocatalysts, with anatase (E_g = 3.23 eV) and rutile (E_g = 3.02 eV) crystal forms commonly used for this application. It can be activated by near UV illumination (300–370 nm) as well as UV light (254 nm), and it can degrade a broad range of molecules for use in indoor air purification [48], including alkanes, ketones, alcohols, phenols, and aromatic compounds [49]. The photodegradation of formaldehyde results in carbon dioxide [50]: TiO 2 ⟶ h ν e - + h + H 2 O → H + + OH - h + + OH - → OH • e - + O 2 → O 2 - HCHO + OH • → HCO • + H 2 O HCO • + OH • → HCOOH HCOOH + 2 h + → CO 2 + 2 H +

This reaction requires the presence of oxygen as well as water adsorbed at the metal oxide surface. The oxygen captures TiO₂ electrons and assists in charge separation [51], preventing electron-hole recombination. The adsorbed water is oxidized to produce hydroxyl radicals OH͘ which are the primary oxidant [52]. The adsorbed water also enhances oxygen photoadsorption by trapping photogenerated holes at OH^- sites [53]. The production of stable reaction intermediates can stop the reaction by adsorbing to the catalyst and blocking reaction sites on the TiO₂. In the case of formaldehyde, the formic acid intermediate is converted to CO₂ after further UV illumination.

Obee and Brown [54] found that increasing the humidity level increased the TiO₂ oxidation rate of formaldehyde up to a maximum, then decreased with increasing humidity for concentrations in the range of 10 ppm and above. For low formaldehyde concentration, 1.5 ppm and below, the oxidation rate had a weak dependence on humidity. The oxidation rate also depended on the formaldehyde concentration, increasing to a maximum at ∼100 ppm. The formaldehyde photo-oxidation rate decreased with increasing temperature.

The presence of other interferents in the air will lead to competitive interaction at the TiO₂ surface adsorption site. Noguchi et al. [55] found that oxidation rate of formaldehyde was higher than that of acetaldehyde because TiO₂ has higher adsorption capacity for formaldehyde; Obee and Brown [54] found that formaldehyde is more strongly adsorbed than toluene. Ao et al. [50] worked with VOC concentrations in the 20–200 ppb range, which better reflect levels commonly found in the indoor environment. They found that the photodegradation rate of formaldehyde was reduced in the presence of other VOCs such as benzene, toluene, ethylbenzene and o-xylene, because the photodegradation intermediates of these molecules blocked the active surface sites on the TiO₂.

Thick film TiO₂ sensors were developed which used the photocatalytic response to different organic molecules in the gaseous phase [56]. The cermet sensors operated at room temperature and the adsorption of different gases to the surface gave different resistivity signatures. The sensors were regenerated by exposing to ambient air for 24 hours and did not require heat.

Peng et al. have used ZnO nanorods to detect formaldehyde under UV illumination [57]. The nanorods were deposited onto ITO electrodes on a glass substrate. They measured increased conductivity in the nanorod due to photocatalytic formaldehyde oxidation. In comparison to nanoparticles, which can also be made into composite nanofibers, nanorods have increased delocalization of charge carriers, increasing the time before photogenerated charge carriers (electrons and holes) will recombine. The nanorods were thus investigated for increased photocatalytic reaction efficiency.

Shie et al. investigated the use of an ultraviolet light emitting diode (UVLED) as the light source to activate photodegradation of formaldehyde on TiO₂ and Ag/TiO₂ photocatalysts [58]. UVLEDs are available from several manufacturers with different wavelengths and intensities, and they have been investigated as alternative light sources for photolithography [59]. In the work of Shie et al., each UVLED had a maximum luminous intensity at λ = 383 nm and 20 mW power consumption. They found that compared to both traditional 254 nm and 365 nm UV lamps, the UVLED had the highest energy effectiveness E_e, defined as E_e = [decomposition mass of formaldehyde (mg)]/[input power (kWh)], while all of them had high decomposition efficiency η. The low cost and long lifetime of UVLEDs make them an attractive option for the development of low-power sensors for wide-area arrays.

Microfabricated cantilevers have been used as sensors for detecting the presence of proteins, bacteria and viruses [60-62]. In the dynamic mode, the cantilevers can be used as sensors when molecules or cells adsorb onto the cantilever, increasing its mass and thus its resonant frequency. The cantilevers can also be used in static mode, when the adsorption of molecules onto the cantilever cause bending and result in surface stress (Figure 5a). The bending of the cantilever can either be detected using optical means to measure deflection of a beam, or using piezoresistors, which undergo a change in resistance [63] (Figure 5b). These cantilevers can be fabricated using the same process as for atomic force microscopy (AFM) cantilevers [64] but are usually functionalized with specific recognition (“probe”) molecules which will capture the “target” molecules from a mixture. The cantilevers are often coated with gold so that the probe molecules can be attached to the cantilever using thiol chemistry [65]. In the case of piezoresistive detection, the cantilevers can be integrated with microelectronics to produce a microscale sensor package.

Seo et al. used piezoresistive cantilevers as a formaldehyde sensor [66]. They found that cantilevers coated with 3-mercaptophenol had the largest downward deflection compared to 2-mercaptophenol, 4-mercaptophenol, and 1-mercapto-6-hexanol. The lowest detection limit was in the range of 0.027 ppm. While reaction of formaldehyde with the 3-mercaptophenol induced compressive stress in the cantilevers, the adsorption of benzene, toluene, and xylene produced tensile stress and upward bending of the cantilevers. However, Seo et al. found that the sensor curvature did not recover its original value after formaldehyde exposure and subsequent rinsing in air, indicating that the formaldehyde may have chemically reacted with the 3-mercaptophenol. Thus, regeneration of this sensor still requires further work.

The most widely used sensor configuration is the chemiresistor (Figure 6). A chemiresistor consists of one or several pairs of electrodes and the electrical resistance change is measured at the output. A simple ohmmeter suffices to collect the data. Several models have been developed to correlate the normalized gas concentration γ in the ambient to the concentration inside the conducting film [79]. Other often used configurations include organic thin film transistors OTFTs [80,81], or diode type sensors [82], which require more fabrication steps.

Recently a new class of polymers known as intrinsically conducting polymers or electroactive conjugated organic polymers has been increasingly applied in sensing materials based on the unique electrical and optical properties [83]. The charge transfer mechanisms differ fundamentally from inorganic crystalline semiconductors such as silicon, in that they are molecular in nature. A polymer becomes intrinsically conductive if the molecular orbitals overlap, allowing a delocalized wave function. The free movement of the charge carriers throughout the lattice is maximized when the orbitals are only partially filled. To describe the electronic phenomena the concept of solitons, polarons and bipolarons have been introduced [84]. The most widely used intrinsically conductive polymers for sensor applications are polypyrrole (PPy), polyaniline (PANi), polythiophene (PTh) and their derivatives. Although the pure polymers may have relatively low conductivities (<10^–5 S/cm^–1[69]) they can be reversibly doped through oxidation or protonation. Conductivities in the range of semiconductors or conductors (∼10⁰–10⁵ S/cm^–1) can be achieved in the doped polymers. The detection limits for conducting polymer sensors can be <1 ppm for acid-base active analytes, and on the order of several ppm for inert organic analytes, with response times on the order of seconds for ultra thin film sensors [69].

Hosono et al. synthesized a highly conducting PPy thin film by plasma polymerization [85]. They then doped the polymer using 4-ethylbenzenesulfonic acid (EBSA) as reported earlier [86] resulting in three orders of magnitude larger conductivity than iodine-doped PPy films. They define the sensitivity as (R_air-R_gas)/R_air × 100(%), in which R_air and R_gas were the electrical resistances when exposed to air and gas respectively. In Figure 7(a) the sensitivity to different gases is reported. The group suggests that the conductivity change is mainly caused by the increased charge carrier concentration in the polymer backbone. They reported as well sensitivity changes as a function of analyte concentration Figure 7(b) and humidity.

The sensitivity and time response of the sensing film is in general proportional to the surface area of the film per unit mass. Changing the morphology at the nanoscale offers therefore the possibility to increase the performance of the sensor. Nanofiber-based sensing films have one to two orders of magnitude larger sensing area per unit mass than solid films [87] and therefore offer faster response times and increased sensitivity. Polymer nanofibers have been prepared using interface polymerization [88], surfactant methods [89], as well as electrospinning [87]. Of these nanofiber production methods, electrospinning offers easily controllable fibre diameter and high uniformity using a relatively simple fabrication system. An electric field is applied between the polymer solution reservoir and a collection ground plate. The stretching forces induced on the jet by the electric field and the rapid evaporation of the solvent causes the formation of solid nanofibers during the passage from the tip to the collection screen. Polymer nanofibers have been interfaced with microfabricated structures [90] and the characteristics of single fibers are investigated. More than one hundred different polymers, polymer blends, and polymer-nanoparticle composites have already been electrospun into nanofibres, summarized in a recent summary of gas sensors based on electrospun nanofibers [87].

Alternatively, conductive polymers can be used as the organic fraction of organic/inorganic hybrids. Zheng et al. [91] used a PANi-TiO₂ nanocomposite as the active layer on a QCM sensor. They dispensed an aqueous solution containing PANi and TiO₂ nanoparticles directly onto the electrodes and allowed it to dry. They reported response times of less than 300 s and they showed the response to formaldehyde and several other VOCs (Δf ≈ 100 Hz for 150 ppm of formaldehyde). They report as well that the addition of TiO₂ improved the thermal stability of the polyaniline sensor. Ma et al. [92] showed that the size and shape of the oxide particles, degree of dispersion, kind of interaction between the organic and inorganic phase, as well the material of the electrode and the concentration of the analyte itself are all factors which affect the response of the sensor.

Another class of organic/inorganic hybrids is synthesized by an intercalation reaction. Matsubara et al. interleaved PPy chains with molybdenum oxide (MoO₃) layers through ion exchange [93]. The structure of pure MoO₃ consists of layered, covalently bonded double-octahedral oxide sheets held together by Van der Waals forces. A variety of large organic compounds can be intercalated between the layers [94]. The conductive nature of the PPy chains is responsible for the drastic resistivity decrease from 2 × 10¹⁰ Ω-cm of pure MoO₃ to 9.6 Ω-cm of the PPy doped MoO₃ layers. The charge carrier transport properties are governed by the host layer (MoO₃) but the magnitude of resistivity is dependent on the corresponding polymer. A detailed study on the sensing mechanism is still in progress. Matsubara et al. suggest that the analyte has two effects on the resistivity of the hybrid material. The first is the partial electron transfer between the analyte and the polymer causing a change in charge carrier density in the polymer backbone. The second is a physical effect due a change in the interlayer distance affecting the charge transfer between the two different materials. The response to a formaldehyde exposure of 500 ppm was up to 6% with response times from 90–120 s. The group further studied other organic/MoO₃ hybrid systems and improved their response to formaldehyde [95]. They found that the PPy/MoO₃ and PANi/MoO₃ exhibited 4-8% change in conductivity to an exposure of 50 ppm of formaldehyde. They concluded that the sensitivity is limited by large insoluble PANI moieties adsorbed onto the hybrid surface. PANI shows an opposite response compared to the hybrid. They modified the intercalation process accordingly by eliminating the insoluble polymer residues [96,97]. With their new sensing layers they achieved a 4% change in conductivity when exposed to 200 ppb.