α-Fe₂O₃ based nanomaterials as gas sensors

Abstract

Interest in detecting and determining concentrations of toxic and flammable gases has constantly been on the increase in recent years due to increase of modernization, industrialization and high standards of life. Detection of such gases is very important in many different fields such as industrial emission control, household and social security, vehicle emission control and environmental monitoring. Metal oxide gas sensors are among most important devices to detect a large variety of gases. α-Fe₂O₃, an environmental friendly semiconductor (E _g = 2.1 eV), is the most stable iron oxide under ambient atmosphere and because of its low cost, high stability, high resistance to corrosion, and its environmentally friendly properties is one of the most important metal oxides for gas sensing applications. This is the first review about gas sensing properties of α-Fe₂O₃ nanostructures. In this paper gas sensing properties of α-Fe₂O₃ are extensively reviewed. After a brief explanation about metal oxide gas sensors and α-Fe₂O₃, sensors based on α-Fe₂O₃ nanomaterials have been reviewed. Gas sensing section is divided into five subsections: pure α-Fe₂O₃ gas sensors, metal/α-Fe₂O₃ gas sensors, metal oxide/α-Fe₂O₃ gas sensors, polymer/α-Fe₂O₃ gas sensors and graphene/α-Fe₂O₃ gas sensors.

1 Metal oxide gas sensors

Air pollution caused by toxic, flammable and explosive gases, such as CO, H₂S, NH₃, CH₄, NO₂, H₂ and C₃H₈ is one of the critical factors that contribute to global warming, climate changes, and harm to human health. The human olfactory system can detect odorous gases, such as H₂S, NH₃, and most of volatile organic compounds (VOCs), but in spite of excellent human olfactory system, detection of some hazardous gases is impossible for human, because some gases like CO and H₂ are both odorless and tasteless as well as colorless [1–3]. Furthermore in some cases absolute gas concentrations is very low to be detected by human nose. Therefore development and fabrication of a device for early detection and/or alarm of certain flammable, explosive, and toxic gases are extremely necessary. For this purpose, different devices have been invented and developed toward tract detection and/or concentration monitoring of such pollution gases [4, 5]. Generally, the most common methods that have been developed for the purpose of detecting gases and VOCs are mainly based on high-performance liquid chromatography (HPLC) [6] and gas chromatography/mass spectroscopy (GC) [7]. However, these techniques are time consuming and quite expensive. Thus, simple and convenient methods for this purpose are in great demand. Consequently, the development of cheap and reliable devices for detection of gases is considered to be a significant goal in science. Gas sensors can be made from various materials depending on the purposes they serve. Regardless type of gas sensor, general requirements for a reliable gas sensor is high sensitivity, fast response, and good selectivity. Among different available gas sensors such as optical gas sensors [8], surface acoustic gas sensors [9], microwave gas sensors [10], thermal conductivity gas sensors [11] and etc., gas sensors based on metal oxides semiconductors (MOS) have been one of most investigated devices [4].

The idea of using semiconductors as gas sensitive devices leads back to 1952 when Bradeen and Barteen first reported gas sensitive effects on Germanium [12]. Ten years later, Seiyama was the first to apply this phenomenon to the detection of some gases like CO₂, toluene and propane gases using a ZnO thin film [13]. In the same year Taguchi patented and subsequently marketed a SnO₂ based gas sensor capable of detecting low concentrations of combustible and reducing gases using a simple electrical circuit [14]. Soon after first sensor devices based on tin oxide were commercialized, the research was expanded to other metal oxide candidates and metal oxide gas sensors aroused the attention from many researchers interested in gas sensors due to the low cost, ease of fabrication, simplicity of their use, and ability to detect many gases including toxic (e.g. H₂S, CO) and flammable (e.g. H₂, CH₄) gases. However despite many worthy advantages, metal oxide gas sensors generally require high operating temperature (typically 150–500 °C), moreover they have poor selectivity and low sensitivity to the very low concentration of gases [15]. Nowadays metal oxide semiconductors are the most utilized calss of inorganic materials for chemical gas sensing [16] and many companies such as Figaro, FIS, UST, MICS, CityTech, NewCosmos, Applied-Sensors, etc. provide this type of gas sensors [4].

Semiconductor based gas sensor can be fabricated into three types of devices, i.e., sintered pellet [17], thick film and thin film. However thin or thick film sensors are more promising detectors (Fig. 1) as compared with the pellet sensors, since the main processes in gas sensors take place actually at the surface and in a near-surface thin layer of the sensing material. The remaining part of semiconductor leads often to only an increase in ohmic losses and serves as a substrate for a sensor and a volume where a heater is placed [18].

Fig. 1

Comparative maximum response of NdFeO₃ pellet and film [19]

Thin or thick film gas sensors are fabricated by depositing a sensing layer over an insulating substrate provided with electrodes and a heater (Fig. 2). The electrodes are used for the readout of sensor resistance; the heater raises the temperature of the semiconductor gas sensor sufficiently high to allow for their fast and reproducible operation [20].

Fig. 2

Typical structure of resistive metal oxide gas sensors [21]

According to the literature, a reduction in particle size can significantly increase the sensitivity of sensors. A bulk solid material typically has <1 % of its atoms on the surface, while a nanoparticle can possess over 90 % of its atom on the surface, so the high surface-to-volume ratio of nanoparticles makes them inherently more reactive. In other words sensor sensivity of nanomaterials is higher than conventional sensors. Consequently many investigations have been carried out in order to improve the performance of sensors by using particles with reduced size, therefore sensors based on nanomaterials developed in the recent years [22].

Response of sensor is the primary property that comes to mind when discussing about sensors. Response of sensor is proportional to the change in the resistance of the material on exposure to gas. The response (S) in the presence of gases is usually defined in several different forms including S = R_a/R_g, S = R_g/R_a, S = ΔR/R_g, and S = ΔR/R_a, where R_a is the sensor resistance in ambient air, R_g is the sensor resistance in the target gas, and ΔR = |R_a − R_g|. Another key sensor property is selectivity, which reflects the often enormous challenge of selectively detecting small numbers of a specified molecule suspended in a sea of other chemical species, e.g. the surrounding atmosphere. Response time is other important sensor parameter. An automotive exhaust gas sensor must respond within the order of 10 ms to a change in gas composition in order to enable feedback control of the air-to-fuel ratio needed for proper operation of the catalytic convertor. Response time is defined as the time required for a sensor to reach 90 % of the total response upon exposure to the target gas. Recovery time which is defined as the time required for a sensor to return to 90 % of the original baseline signal upon removal of the target gas is not as important as response time, but it is obvious that shorter recovery times are more desired. Stability is another key property, without a stable and reliable sensor readings become impossible. The stability is becoming more challenging to achieve, as some sensors require to operate under harsh temperature and environmental conditions [23]. Working temperature of sensor is the temperature at which the sensor material must be heated to obtain the most optimal response. The sensor selects a particular gas at a particular temperature. Thus by setting the temperature, one canuse the sensor for particular gas detection. A high-performance sensor also should be able to detect a tiny amount of certain gas, and the lowest amount of the gas which the sensor could have a response to it, is called the detection limit [24]. Overall, a good gas sensor is characterized by high sensitivity, high selectivity, short response/recovery time, and low working temperature [25].

2 Alpha iron oxide (α-Fe₂O₃)

Metal oxide nanomaterials, which are considered to be some of the most fascinating functional materials, have been widely exploited in various important applications. The electrostatic interactions between the positive metallic and negative oxygen ions result in firm and solid ionic bonds. The s-shells of metal oxides are completely filled, so that most of the metal oxides have good thermal and chemical stability. However, their d-shells may not be completely filled, giving them a variety of unique properties that make them potentially of great use in electronic devices. These unique properties include wide band gaps, high dielectric constants, reactive electronic transitions, optical, and electrochromic characteristics, as well as gas sensing properties [26].

Among different metal oxides, iron (III) oxide has been one of the extensively investigated transition metal oxides. Besides to amorphous phase [27] Fe₂O₃ has four polymorphs: (1) α-Fe₂O₃; (2) β-Fe₂O₃; (3) γ-Fe₂O₃; and (4) ε-Fe₂O₃ [28]. While the highly crystalline α-Fe₂O₃ and γ-Fe₂O₃ occur in nature, β-Fe₂O₃ and ε-Fe₂O₃ are generally synthesized in the laboratory [29]. Recently Sakurai et al. [30], reported that the threshold sizes of the phase transformations for Fe₂O₃ phases had the following values: γ → ε at 7 nm, ε → β at 35 nm, and β → α at 80 nm. However it seems more studies in this aspect are needed.

Among four polymorphs, α-Fe₂O₃ (mineralogically known as hematite) is the oldest known of the iron oxides and is widespread in rocks and soils. It is also known as ferric oxide, iron sesquioxide, red ochre, specularite, specular iron ore, kidney ore, and martite. It is blood-red in color if finely divided, and black or grey if coarsely crystalline (Fig. 3a, b) [31]. It has a plenty of some fascinating features such as low cost, abundance, environmental friendly nature [26] easy of production, easy of storage [32] high corrosion resistance [33] and excellent substrate adherence [34]. These features make it very interesting for researchers. Additionally, α-Fe₂O₃ nanomaterials with low toxicity, chemical inertness and biocompatibility show a tremendous potential in biotechnology [35].

Fig. 3

a α-Fe₂O₃ (particle size 400 nm), b α-Fe₂O₃ (particle size 100 nm), c Fe₃O₄, d γ-Fe₂O₃ [36]

α-Fe₂O₃ is a transition metal oxide with a moderate band gap of 2.1 eV [37] allowing the capture of ∼40 % of sunlight [38], which has received considerable attention due to its potential application in many technological areas such as magnetic materials [39] lithium ion rechargeable batteries [40], photocatalytic degradation [41] and electrode materials [42], pigments [43], field effect transistor [44], heavy metals removal from water/wastewater [45] and water splitting [46]. It has been used as humidity sensor since the 60’s of the last century [47]. Additionally, it is considered as a good candidate in the gas sensing field and there are many reports about gas sensing properties of α-Fe₂O₃-based nanomaterials (Tables 1, 2, 3, 4, 5).

Table 1 Gas sensing properties of pristine α-Fe₂O₃ gas sensors with different morphologies

Table 2 Gas sensing properties of metal (as dopant or surface catalyst)/α-Fe₂O₃ gas sensors

Table 3 Gas sensing properties of metal oxide/α-Fe₂O₃ nanocomposite

Table 4 Gas sensing properties of polymer/α-Fe₂O₃ nanocomposite gas sensors

Table 5 Gas sensing properties of graphene/α-Fe₂O₃ nanocomposite gas sensors

Generally α-Fe₂O₃ shows n-type behavior because it has tendency to become oxygen-deficient with oxygen vacancies [29]. The defect formation can be described as:

$$Fe_{2} O_{3} \leftrightarrow 2Fe_{\text{int}}^{N + } + 2Ne + \frac{3}{2}O_{2} \quad N = 2 {\text{ or }}3$$

(1)

and

$$Fe_{2} O_{3} \leftrightarrow 2Fe_{Fe}^{ \times } + 3V_{O}^{ \cdot \cdot } + 6e + \frac{3}{2}O_{2}$$

(2)

It means that hematite has a complex defect structure in which three types defects species (oxygen vacancies, Fe³⁺ interstitials, Fe²⁺ interstitials) are present. The presence of the defects gives rise to semiconducting properties. Loss of oxygen leaves behind extra electrons and produces an n-type semiconductor; extra oxygen (entering the lattice as O²⁻) creates a deficit of electrons (i.e. introduces electronic holes), which produces p-type behavior. The electrical properties of α-Fe₂O₃ depend on the nature of the precursors and thermal conditions that influence the defect chemistry of the hematite. α-Fe₂O₃ annealed in air was found to be a n-type semiconductor, and annealed in oxygen a p-type one. The p-type conductivity is mainly due to the presence of cationic impurities, e.g. presence of Mg²⁺ or Ni²⁺ (0.019–0.17 %) [48]. A good example showing role of thermal treatment on the n or p type behavior of was investigated by Fan et al. [49]. They fabricated Zn-doped hematite nanobelts by annealing in a quartz tube inside a furnace using zinc powder as the dopant source. Interestingly, the electronic properties of the hematite belts could be tuned by varying the doping conditions. To obtain p-type character, hematite nanobelts were annealed at 700 °C for 5 min. On the other hand, the enhanced n-type nanobelts obtained by annealing at the substantially lower temperature of 350 °C for 1 h. Owning to many application of α-Fe₂O₃, there are great scientific interests in the synthesis of α-Fe₂O₃ particles and it is known that depending on the addition of various organic or inorganic additives during synthesis procedure, the formation processes of hematite particles are usually thought to contain a phase transfer process either from akaganeite (β-FeOOH) to α-Fe₂O₃ by using FeCl₃, or from goethite (α-FeOOH) to α-Fe₂O₃ by using Fe (NO₃)₃. However, detailed structure and morphology evolution of α-Fe₂O₃ nanostructures is still not fully understood [50] Also there are many papers about the modifications of size, morphology and porosity of α-Fe₂O₃ particles and over the past decade, various techniques such as the sol–gel method [51, 52], hydrothermal method [53], microemulsion [54], laser ablation [55] atomic layer deposition (ALD) [56] etc., have been employed to prepare α-Fe₂O₃ particles and various α-Fe₂O₃ structures such as nanoparticles, nanorods [57], nanowires [58], nanotubes [59] nanobelts [49], nanoflakes [60], nanodendrites, and hollow nanoparticles [61] have been successfully prepared.

α-Fe₂O₃ has hexagonal crystal structure consisting of iron atoms surrounded by six oxygen atoms where the iron and oxygen atoms are arranged in a corundum structure (Fig. 4), which is in fact a trigonal-hexagonal scalenohedral with space group $R\overline{3} c$, lattice parameters a = 5.0356 nm, c = 13.7489 nm, and there are six formula units per its unit cell [62, 63]. In α-Fe₂O₃, the oxide ions (O²⁻) are arranged along the (0 0 1) plane of a hexagonal closed-packed lattice, whereas two-thirds of the octahedral interstices are occupied by the cations (Fe³⁺) in the (0 0 1) basal planes and the tetrahedral sites are unoccupied. This cationic arrangement generates pairs of FeO₆ octahedrons, in which the edges are shared by three neighboring octahedrons in the same plane and one face with an octahedron in an adjacent plane in the (0 0 1) direction. The coordinances of iron and oxygen are six and four respectively (Fe/O = 6 and O/Fe = 4 [62]).

Fig. 4

α-Fe₂O₃ crystal structure [29]

3 Pristine α-Fe₂O₃ gas sensors

Although there are some reports about gas sensing properties of another well-known polymorph of Fe₂O₃ namely γ-Fe₂O₃ in pure [64] or doped form [65], but the application of γ-Fe₂O₃ as gas sensor is limited. This is because of its low phase transition temperature from γ-Fe₂O₃ to α-Fe₂O₃ (between 523 and 873 K as a function of its previous history [66]), which causes the resistance change of the sensor that eventually affects the stability of the sensor [67]. Furthermore the phase transformation which occurs during heating γ-Fe₂O₃ can results in considerable aggregation and grain growth of transformed α-Fe₂O₃ [68]. Also there are few reports about gas sensing properties of Fe₃O₄ (magnetite) in pure [69, 70] or doped form [71], but the low oxidation stability of Fe₃O₄ restrict its application to oxygen and humidity free atmospheres [69]. Furthermore, recently Peeters et al. [72] reported a NO₂ gas sensor based on Au/ε-Fe₂O₃ nanocomposites for the first time. In spite of above mentioned iron oxide based gas sensors, In fact most studied phase of iron (iii) oxide for gas sensing applications is α-phase and in this review focus will be on the gas sensing properties of α-Fe₂O₃ based nanomaterials. As it can be seen from Fig. 5 among a lot of metal oxides, Fe₂O₃ is the sixth most studied metal oxide for gas sensing applications.

Fig. 5

Metal oxides studied for gas sensors [73]

α-Fe₂O₃ formations occur naturally as the mineral hematite where the crystals do not cleave. Their predominant growth faces are (0001) and $\left( {101\overline{1} } \right)$. Nearly stoichiometric α-Fe₂O₃ (0001) surfaces are quite inert for surface adsorption of gaseous species, probably due to the stability of the half-filled d-band configuration. Only weak interactions are possible with pure material. Defective surfaces or defect creation by incorporating other impurity species is the best way to use this material for gas-sensing applications [74].

For gas sensing applications, α-Fe₂O₃ sensors are generally more sensitive to alcohols and NO₂, Their operation temperature is between 250 and 450 °C, their stability is medium, compatibility of them with standard IC fabrication is medium and synthesis and deposition of them on the surfaces of substrate is easy [75].

Gas sensors should work in atmospheres which containing water vapors. It was established that adsorption of water is a dominant factor in the surface characteristics forming, both with respect to adsorption of other species and to surface catalysis. A hydroxylated surface is formed at an oxide by the chemisorption of a monolayer of water. Water may also catalyze reactions, taking place at the surface of a gas sensor. The adsorption of water also has an effect on the electronic properties of α-Fe₂O₃, usually acting as a donor. It is intermediate between hydration of the surface and physical adsorption of water. Long exposure to humidity can lead to hydration of the surface layer, and correspondingly to drift of chemical sensors characteristics. Therefore low tendency to hydration is an important factor for good gas sensors. For α-Fe₂O₃ based gas sensors, as they work at relatively high temperature (250–450 °C) water adsorption is minimal. Also one effective solution for reducing effect of water vapor is use of filters. In this regards, various coatings have been used to either consume water vapor that one does not wish to pass to the gas sensor or to permit the passage of selected gases to the sensor. For example it is found that that Teflon was helpful in stopping H₂O reaching the sensor [76].

A recent study about temperature program reduction (TPR) of iron oxide sensors showed that transformation of hematite to magnetite occurred at 455 °C, so for α-Fe₂O₃ sensors (Working temperature 250–450 °C) after exposure to H₂ gas probability of reduction of α-Fe₂O₃ to Fe₃O₄ is very unlikely [77].

General gas sensing mechanism of α-Fe₂O₃ can be explained as follow: When the α-Fe₂O₃ is exposed in air, oxygen will be adsorbed on surface vacancies and form anionic oxygen (Fig. 6a) which will cause the formation of depletion region and band bending on the surface. The band bending on α-Fe₂O₃ surface will bring a barrier for carrier transport between the particles, and then cause the decrease of α-Fe₂O₃ conductivity. When reducing gases such as H₂, CO, NH₃, H₂S, and VOCs are adsorbed on the surface anionic oxygen, electrons will be injected into α-Fe₂O₃ surface, which will reduce width of the depletion region (Fig. 6b) and release the band bending, hence, the α-Fe₂O₃ conductivity will be improved. On the contrary, when oxidative gases like NO₂ or O₃ are adsorbed on α-Fe₂O₃ surface, they will gain electrons from anionic adsorbed oxygen, which will increase width of the depletion region (Fig. 5c), leading to the decrease of conductivity. The conductivity change can be easily transferred into resistance signal, which is the best-known sensor output signal. In some cases, the measurement is operated at constant temperature by direct current measurement [25].

Fig. 6

a Metal oxide surface in presence of air, b metal oxide surface in presence of reducing gas, c metal oxide surface in presence of oxidizing gas [25]

It is well known that the gas sensing properties of α-Fe₂O₃ dependent strongly on its morphologies, such as shape, size, orientation and crystal density [78]. Thus, over the past decades, considerable efforts have been made to improve the sensitivity, selectivity, stability and reproducibility of α-Fe₂O₃ sensors by controllable synthesis of various α-Fe₂O₃ nanostructures, doping by noble metals, UV light activation and composite with other metal oxides. Table 1 shows gas sensing properties of pristine α-Fe₂O₃ gas sensors with various morphologies. Although there are some researches about humidity sensing of iron oxide gas sensors in the form of pure [79] composite (NiFe₂O₄–Fe₂O₃) [80], or doped [81], however in this review we will focus on gas sensing properties of α-Fe₂O₃ based gas sensors. As it can be seen a large variety of gases and VOCs such as CO, H₂S, LPG, ethanol, acetone, formaldehyde and n-butanol can be detected by pristine α-Fe₂O₃ gas sensors at different temperature ranging from Room temperature up to 350 °C.

In following we briefly review some papers relating to pristine α-Fe₂O₃ gas sensors. Liang et al. [91], reported a facile hydrothermal method to prepare the α-Fe₂O₃ nanoparticles in the presence of graphene sheets, afterwards a heat treatment process was performed to produce well-dispersed α-Fe₂O₃ nanoparticles along with the elimination of graphene sheets [91]. Figure 6 (left) shows dynamic response of pure α-Fe₂O₃ (synthesized without graphene) and α-Fe₂O₃–G (0.2) (synthesized with graphene) powders at 280 °C. They attributed the improvement of sensing performance of α-Fe₂O₃–G (0.2) nanoparticles to the well dispersion of α-Fe₂O₃ nanoparticles with the help of graphene, which facilitated the diffusion of ethanol gas and improved the reaction of ethanol gas with surface adsorbed oxygen. They pointed out that, the Fe₂O₃–G (0.2) nanoparticles (Fig. 7 (right-b) possessed a porous and loose structure in comparison to the compact pure Fe₂O₃ nanoparticles (Fig. 7 (right-a). Therefore loose structures were of great benefit to numerous oxygen molecules adsorbed onto the Fe₂O₃–G (0.2) surface and consequently higher response to ethanol vapor.

Fig. 7

Left response of Fe₂O₃–G (0.2) and Fe₂O₃ sensors to ethanol with different concentrations. Right schematic illustrations of two different sensor materials: a pure Fe₂O₃ and b Fe₂O₃–G (0.2)

Sun et al. [88], synthesized 3D urchin-like α-Fe₂O₃ nanostructure (Fig. 8a) for ethanol vapor sensing with high sensivity and short response and recovery times (Fig. 8b). The high response and fast response-recovery of the α-Fe₂O₃ nanorods gas sensor were attributed to large specific surface area and unique architectures of 3D urchin-like nanostructures. There was a network of interconnected pores in sensing layer of the α-Fe₂O₃ sensor, where the diffusion of target gas toward all of the surfaces of α-Fe₂O₃ nanorods was effective so the large specific area provided more active sites for gas sensing reactions. Moreover, the as-prepared urchin-like α-Fe₂O₃ nanostructures were composed of many well-aligned nanorods, and many nanorods/nanorods grain boundaries formed in it, so the sensor was expected to have short response-recovery time and high sensivity, because grain boundaries or grain junctions are considered as the active sites and they acted positively on the performance of sensor.

Fig. 8

a Field emission scanning electron microscopy (FESEM) images of urchin-like α-Fe₂O₃ nanostructures after calcining the precursors at 600 °C for 2 h, b response of sensor upon exposure to different concentrations of ethanol at 260 °C

In another report, three-dimensional nanostructures of α-Fe₂O₃ materials, including solid nanospheres, hollow nanosphese and yolk-shell (core–shell) nanospheres (Fig. 9), were synthesized via an environmentally friendly hydrothermal approach by Wang et al. [94].

Fig. 9

FESEM and TEM images of each step α-Fe₂O₃ products: a, b irregular nanoparticles, c, d solid nanospheres, e, f hollow nanospheres, and g, h yolk-shell nanospheres [94]

Ethanol sensing of different synthesized powders were investigated, where yolk-shell nanoparticles showed the best ethanol sensing performance among all other nanostructures (Fig. 10). They pointed out that the improved ethanol sensing properties of the sensor based on the yolk-shell α-Fe₂O₃ nanospheres were primarily due to the perfect architecture, i.e., hollow structure with the porous double shell. Yolk shell nanospheres had a protective hollow shell, in which a small hollow core was encapsulated. The outer shell layer not only could separate hollow core nanospheres but it can prevent them from aggregation. Also the outer shell layer possessed a porous structure which offered a sufficiently active surface and good permeability to gas adsorption and gas diffusion. On the other side, the inner shell possessed a hollow structure with larger specific surface area. Therefore, the high response might be due to the larger specific surface area of the yolk-shell and hollow nanospheres (98.3 m² g⁻¹), which allowed the outer and inner surface to absorb more target gases. Furthermore, the shell layer possessed a porous structure and gap developed to accelerate the in-diffusion of the target gases toward the materials surfaces, then enhancing the response rate. It hence revealed a large amount of enhanced performances compared to conventional solid and hollow structured nanomaterials.

Fig. 10

a–d Response transients (240 °C) of four sensors to 100 ppm ethanol [94]

Wu et al. [96], synthesized α-Fe₂O₃ hollow spheres (Fig. 11a), with uniform distribution of mesoporous on the shell by a smart complex precursor method.

Fig. 11

a SEM image of the hollow spheres with higher magnification, b room temperature sensitivity of the sensors made of as prepared α-Fe₂O₃ samples to ethanol [96]

The as-prepared α-Fe₂O₃ mesoporous hollow spheres possessed two kinds of porous architecture: one within the spheres and the other on the shell of the hollow spheres. The interconnected honeycomb-like pores formed a network that provided more sufficient space and active sites enabling both the target species and the background gas to access all the surfaces of α-Fe₂O₃ mesoporous hollow spheres contained in the sensing unit. Therefore they reported that the high sensitivity and reversibility under ambient conditions (Fig. 11b) was due to the intrinsic mesoporosity and high surface-to-volume ratio associated with the mesoporous hollow spheres [96].

Hao et al. [105], reported anomalous conductivity-type transition sensing behaviors of n-type porous α-Fe₂O₃ nanostructures toward H₂S. Interestingly, they observed abnormal n–p transition sensing behavior induced by the variation of working temperature and p–n transition of sensing layer was related to the increase of H₂S concentration (Fig. 12).

Fig. 12

a The response of the sensors against 10 ppm H₂S at various temperatures, b typical dynamics response to various concentrations of H₂S at 400 °C [105]

Generally, the oxygen deficiencies inside α-Fe₂O₃ result in an n-type conductivity nature, and the corresponding band diagram is shown in Fig. 13a, where the Fermi level (E_F) is above the intrinsic Fermi level (E_i). It is assumed that a certain amount of oxygen is adsorbed on the surface of the α-Fe₂O₃ at certain temperature. The adsorbed oxygen molecules capture free electrons from α-Fe₂O₃ and become ionized molecular (O₂ ⁻) or atomic (O⁻ or O²⁻) oxygen species. The thermally stimulated processes of oxygen adsorption, dissociation, and charge transfer involve only the electrons in conduction band, and the reaction kinematics can be explained by the following reactions:

$$O_{2} \left( {\text{g}} \right) \leftrightarrow O_{2} \left( {ads} \right)$$

(3)

$$O_{2} \left( {\text{ads}} \right) + {\text{e}}^{ - } \leftrightarrow {\text{O}}_{2}^{ - } \left( {ads} \right)$$

(4)

$${\text{O}}_{2}^{ - } \left( {ads} \right) + {\text{e}}^{ - } \leftrightarrow 2{\text{O}}^{ - } \left( {ads} \right)$$

(5)

$${\text{O}}^{ - } \left( {ads} \right) + {\text{e}}^{ - } \leftrightarrow {\text{O}}^{2 - } \left( {ads} \right)$$

(6)

In these processes, there may exist a transition temperature (∼300–350 °C), below which oxygen adsorbed on the surface are mainly in the form of O₂ ⁻, whereas above which chemisorbed oxygen dominate in the form of O⁻ or O²⁻. At low working temperatures (150–300  °C), the absorbed oxygen captures electrons from the conduction band and converts to ionosorbed oxygen that is mainly dominated by O₂ ⁻, which leads to the upward of the intrinsic level and the formation of a depletion layer near the surface of α-Fe₂O₃, as shown in Fig. 13b. Reducing gas like H₂S react with O²⁻ at the surface and release electrons to conduction band as following:

$$H_{2} S \left( {ads} \right) + 3O^{ - } \left( {ads} \right) \to H_{2} O + SO_{2} + 3e^{ - }$$

(7)

$$H_{2} S \left( {ads} \right) + 3O^{2 - } \left( {ads} \right) \to H_{2} O + SO_{2} + 6e^{ - }$$

(8)

The electron concentration increases and the position of intrinsic level are brought downward (Fig. 12c). During these processes, α-Fe₂O₃ nanostructure based sensor exhibits typically n-type sensing behavior. So the resistance decreases upon exposure to H₂S, and the decreased magnitude depends strongly on the concentration of H₂S. By contrast, at high temperatures (350 °C or even higher), O⁻ or O²⁻ are dominated forms instead of O₂ ⁻, which capture more electrons from the conduction band. Such behavior is more pronounced due to the increased unstable surface states of the nanostructures. Because α-Fe₂O₃ has a narrow band gap of 2.1 eV, the strong absorption of atomic oxygen species results in the formation of an inversion layer near the surface, as shown in Fig. 13d, where Fermi level is below the intrinsic level. Holes become the majority carriers, which result in a surface p-type conductivity. This accounts for the n-type to p-type transition, as shown in Fig. 12a. During the reductive process at low concentrations, H₂S is oxidized to be H₂O and SO₂ by the chemisorbed O⁻ or O²⁻ and some electrons transfer back to the semiconductor, resulting in a reduced inversion layer near the surface and an increase in resistance, but Fermi level is still below the intrinsic level (Fig. 13e) and the majority carriers are still holes (p-type). With the introduction of high concentration H₂S, H₂S further react with the chemisorbed O⁻ or O²⁻, and more electrons transfer back to the semiconductor. The change is large enough to invert electron to be the majority carriers (n-type), which cause a depletion layer instead of the inversion layer, as shown in Fig. 13f. Then p-type to n-type transition behavior is observed, and the resistance decreases upon exposure to H₂S.

Fig. 13

Schematic energy-level diagrams of: a an n-type semiconductor caused by oxygen vacancies; b an n-type semiconductor with depletion layer caused by mild surface absorption; c an n-type semiconductor with reduced depletion layer caused by surface reaction with reducing gas; d a p-type semiconductor through formation of an inversion layer caused by strong surface absorption due to the large density of surface defect states; e a p-type semiconductor with reduced inversion layer caused by surface reaction with low concentrations of reducing gas; f an n-type semiconductor with depletion layer caused by surface reaction with high concentrations of reducing gas [105]

In another study, Long et al. [106], reported a p-type α-Fe₂O₃ a gas sensor. The polyhedral α-Fe₂O₃ particles were synthesized via a modified polyol method with the addition of an extra amount of NaBH₄ in ethylene glycol, and fabricated to a thick gas sensing film after calcination at 500 and 900 °C (Fig. 14). It can be seen than sensors showed p-type behavior upon exposure to H₂ gas. They attributed p-type nature of iron oxide to low concentration of Na ions from NaBH₄ which affected the major carrier of synthesized α-Fe₂O₃ nanoparticles. It was reported that when Na ions from NaBH₄ for the synthesis of samples were doped into α-Fe₂O₃ bulk, Na ions in the α-Fe₂O₃ lattice generated holes as follows:

$$Na_{2} O \to 2Na_{Fe}^{{\prime \prime }} + O_{O} + 2V_{O} + 4h^{.}$$

(9)

The holes generated by the above equation acted as a source of p-type nature of obtained α-Fe₂O₃ nanoparticles.

Fig. 14

The response transients to H₂ at 300 °C

Dai et al. [107], reported pristine iron oxide with abnormal p–n transition. Figure 15a shows that upon exposure to 0.25–1 ppm NO₂ sensor shows p-type behavior, whereas after exposure to higher concentrations shows n-type behavior. It was interesting that this abnormal behavior was not due to presence of impurities in the sample (Fig. 15b).

Fig. 15

a Dynamic response of α-Fe₂O₃ sensor to 0.25–5 ppm NO₂ gas (inset shows dependence of p and n behavior to NO₂ concentration), b XRD pattern of pure α-Fe₂O₃ [107]

Abnormal behavior explained as follows. When the sensor was placed in air, oxygen molecules adsorbed on the surface to form O⁻ (ads.) ions by capturing electrons from the conductance band, leading to the formation of an electron depletion layer and thus to an increase in the resistance of the sensing body. NO₂ gas directly adsorbed on the surface leading to formation of $NO_{2}^{ - } \left( {ads} \right)$. For low NO₂ concentrations, the adsorbed O⁻ and NO₂ ⁻ ions were normally too far apart to interact. Because α-Fe₂O₃ porous film was superrich in O (ads.)⁻, with [O_ads.]/[O_lattice] > 1 (Fig. 15c), so there were no surface active sites available for NO₂ gas adsorption. Therefore, the formation of NO₂ (ads.)⁻ mainly occurs by substitution of O (ads.)⁻ and, the balance between reaction (10) and (11) governs the final sensing behavior.

$$O_{2} \left( g \right) + 2e^{ - } \to 2O_{{\left( {ads} \right)}}^{ - }$$

(10)

$$NO_{2} \left( g \right) + e^{ - } \to NO_{{2\left( {ads} \right)}}^{ - }$$

(11)

The general process is described in the following reaction whereby electrons were simultaneously fed back into the sensing body, leading to increased conductance and thereby to abnormal p-type NO₂ sensing behavior.

$$NO_{2} \left( g \right) + 2O_{{\left( {ads} \right)}}^{ - } \to NO_{{2\left( {ads} \right)}}^{ - } + O_{2} \left( g \right) + {\text{e}}^{ - }$$

(12)

When the sensor was exposed to high concentrations of NO₂, the NO₂(ads.)⁻ not only replaced the O(ads.)⁻, but also led to surface adsorbed active site redistribution. The distance between the O(ads.)⁻ and NO₂(ads.)⁻ ions were shorter, such that they interact as described in reaction (13). The lower partial pressure of oxygen makes reaction (13) more favorable, whereby electrons were captured from the sensor, lowering its conductance as in a normal n-type NO₂ sensor.

$$NO_{{2\left( {ads} \right)}}^{ - } + 2O_{{\left( {ads} \right)}}^{ - } \to NO\left( {gas} \right) + \frac{1}{2}O_{2} \left( g \right) + 2O_{{\left( {ads} \right)}}^{ - 2}$$

(13)

The competition between reactions (12) and (13) governs the transitions between p- and n-type sensing. At high NO₂ concentrations, reaction (13) dominates so that n-type sensing is observed (Fig. 16).

Fig. 16

A schematic illustration of the mechanism governing the transitions from p- to n-type NO₂ sensing in the porous Fe₂O₃ thin film [107]

According to above mentioned papers, n- or p-type nature of Fe₂O₃ sensors was attributed to the different types of impurities or vacancies present in the films, target gas concentration as well as thermal treatment conditions [82].

4 Metal/α-Fe₂O₃ gas sensors

Despite good results have been obtained for pristine α-Fe₂O₃ sensors, the development of more highly sensitive and markedly selective gas sensors based on α-Fe₂O₃ nanostructures remains a challenge. Nowadays many efforts such as metal doping, heterostructure constructing and adding catalyst have been taken in order to meet the increasing demands for making sensors with higher performance [108]. Among different strategies for improving the sensor characteristics, one of the most simple, feasible and exciting possibility is metal doping (on oxide surface or in its volume). The metal nanoparticles play both passive and active roles in the sensing process, but the most studied metals used in gas sensing applications are active metals such as Au [109, 110], Pt [111, 112], Pd [113, 114] and Ag [115].

Metal–metal oxide nanocomposite films combine the catalytic property of metal and gas reactivity of semiconducting metal oxide, thereby possess unique gas sensing properties unavailable from either the metal or metal oxide alone. Metallic nanoparticles activate or dissociate the detected gas on their surface. These activated products are easier to react with the adsorbed oxygen species on the metal oxide surface, resulting in a change of resistance. In addition, direct exchange of electrons between the semiconductor metal oxide and metallic nanoparticles causes a change in the width of the depletion layer of the semiconductor oxide, leading to a change in sensing properties. The metal nanoparticles can reduce the sensing temperatures, improve the selectivity, and increase the surface area of the metal oxide.

In most cases, the metal nanoparticles are anchored on the surfaces of the semiconductors as isolated islands to produce heterointerfaces due to the fact that catalytic activity dependent on the size of the metal nanoparticles (Fig. 17).

Fig. 17

TEM micrograph of the 3 % Pt-In₂O₃ nanopowders. Small platinum particles, some of them are indicated by arrows, are clearly visible as island on the surface of cube-like grains of In₂O₃ [116]

Although doping has been used for a long time in preparation of commercial gas sensors, the working principle of additive-modified metal-oxide materials is still not completely understood, but enhanced gas sensing performances using noble metal is generally explained by electronic mechanism and chemical mechanism (Fig. 18).

Fig. 18

Mechanism of sensitization by metal or metal oxide additives a electronic sensitization, where the additive is an acceptor for electrons and the redox state/chemical potential is changed by reaction with the analyte; b chemical sensitization by activation of the analyte (H₂) followed by spill over and change of the surface oxygen concentration [117]

In the electronic mechanism the noble metal acts as an electron acceptor on semiconductor oxide surfaces, which contributes to the increase of the depletion layer. Therefore, as compared with the pristine oxide, the change in resistance is larger, leading to the increase in response.

In the chemical mechanism the noble metal catalytically activate the dissociation of molecular oxygen, so the additive role is to increase the reaction rate of the gas molecules, which are firstly adsorbed on the metallic cluster and, later, moved to the oxide surface in a so-called, spill-over process [118, 119]. Catalyst particles should be finely dispersed on the metal oxide matrix so that they are available near all the intergranular contacts. In an open atmosphere the oxygen molecules are first adsorbed on the catalyst and then spillover to the metal oxide matrix. At appropriate temperatures, the reducing gases are first adsorbed on to the surface of additive particles and then migrate to the oxide surface to react with surface oxygen species thereby increasing the surface conductivity. It was established that for attaining the optimal effect, surface cluster size should not exceed 1–5 nm [120].

In fact, as gases reach the surface of the material, they absorb energy and become activated. Then, they react with the particles absorbed on the material surface, following the reaction rate described by the Arrhenius equation

$$r = {\text{A}}\exp \left( { - {\raise0.7ex\hbox{${E_{\text{act}} }$} \!\mathord{\left/ {\vphantom {{E_{\text{act}} } {\text{kT}}}}\right.\kern-0pt} \!\lower0.7ex\hbox{${\text{kT}}$}}} \right)$$

(14)

Where r is the reaction rate constant, A is the pre-exponential factor, E_act is the activation energy, k is the Boltzmann constant, and T is the absolute temperature. The reaction rate constant (r) increases with a decreasing activation energy. In other words, the rate of response becomes higher by lowering the activation energy at the same operating temperature. The experiment showed that platinum treatment condition had a notable effect on the activation energy. Thus, noble metal helps in the improvement of sensing performance [121]. As metallic surface additive, Pd, like Ag, is typically considered to be related with an electronic mechanism, whereas Pt is supposed to lead to the chemical one [119]. The interfacial region between metal nanoparticle and metal oxide also has very different electron band structure than inside the bulk semiconducting metal oxide, which also contribute to the unique gas sensing properties of this type of nanocomposite.

Operating temperature is one of the most important factors in gas sensing. Increase of operating temperature of metal oxide sensor could accelerate the diffusivity of the target gas atoms into the surface of metal oxide surfaces and thus lead to a higher sensitivity. However too high of a temperature is not acceptable in view of energy cost or environment safety. Therefore, trying to decrease the working temperature for sensing is another important focus. Noble metals (e.g., Pt, Pd, Au, or Ag) are highly valued because they could effectively improve the interaction between metal oxide surface and gas molecules, thereby reducing the operation temperature. An enhanced performance is achieved by a catalytic boost via the so-called spill-over effect as noble metal particles favor gas molecule adsorption and desorption, achieving higher concentration of adsorbed ionic oxygen at lower temperature [25].

Table 2 shows gas sensing properties of metal/α-Fe₂O₃ gas sensors, where different metals as La, Ag, Au, Pt and Pd have been doped with α-Fe₂O₃ to detect different gases and VOCs mostly ethanol and acetone.

Neri and co-workers investigated the influence of water vapor (0 and 50 % relative humidity) on the CO response of Au/Fe₂O₃ nanopowders. They reported that Au/Fe₂O₃ had higher sensivity to CO in humid air at 200 °C (Fig. 19).

Fig. 19

Log-log plot of sensitivity versus CO at two different humidity levels at T = 200  °C

As shown in Fig. 20, they proposed a mechanism for such observation. In dry air, the surface of the semiconductor oxide was covered by oxygen species and by reaction of adsorbed oxygen with CO adsorbed on the noble metal electrons were donor in the conduction band of the iron oxide. Due to the promoting effect of Au, the maximum of sensitivity to CO occurs at relatively low temperature, around 200 °C. In wet air the surface almost entirely covered with surface hydroxyl, so CO reacted with OH groups to form CO₂, H₂O and oxygen vacancies. Moreover, the minor concentration of oxygen species on the surface minimized the probability that the CO directly react with adsorbed oxygen on the surfaces. The variation of the electrical conductivity in the latter case was related to the oxygen vacancies, a temperature as high as 300 °C is required for CO-sensing. Such high temperature causes the ionization of oxygen vacancies providing electrons in the bulk [127].

Fig. 20

Schematization of the surface reactions occurring under sensor working in a dry and b wet air [127]

Mirzaei et al. [129], synthesized Ag@α-Fe₂O₃ nanocomposites as ethanol sensor. They showed that core–shell sensor had better ethanol sensing than pristine α-Fe₂O₃ (Fig. 21a). They reported that Ag core had three crucial rules (1) an electron donates to gaseous oxygen, and consequently accelerate these reactions on the surface; (2) an electron reservoir releases electrons from surface, and (3) decrease of ethanol adsorption and desorption energy on the iron oxide shell, therefore, in the case of Ag@α-Fe₂O₃, sensor response was higher than pristine α-Fe₂O₃ sensor. A possible sensing mechanism for Ag@α-Fe₂O₃ sensor is presented in Fig. 21b, where silver donate electron for adsorption of oxygen.

Fig. 21

a Sensing response of Ag@α-Fe₂O₃ and α-Fe₂O₃ towards different concentrations of ethanol vapor at 250 °C, b possible ethanol sensing mechanism by Ag@α-Fe₂O₃ core–shell nanocomposites [129]

5 Metal oxide/α-Fe₂O₃ nanocomposite gas sensors

Transition metal oxides have promising gas sensing performance due to their catalytic properties. Some transition metal oxides are stable, have low electric resistance and good gas sensing properties at low operating temperature, while others have high electric resistivity and require high operating temperatures.

Owning to peculiar properties originated from their individual phases, composite materials are of special interest. Nowadays, considerable attention has been paid to the synthesis of semiconductor composite nanostructures and their great potential in sensing applications. Some of the metal oxides are more sensitive to oxidizing gases while others are more to reducing gases. It is therefore a natural approach to combine metal oxides with different properties with an appropriate proportion so that gas sensor performance can be modified as desired. Table 3 summarizes gas sensing properties of metal oxide/α-Fe₂O₃ nanocomposite. The most studied system is SnO₂/α-Fe₂O₃ nanocomposite and ethanol and acetone are most studied gases by metal oxide/α-Fe₂O₃ nanocomposite sensors. Working temperature of these sensors is from RT to 400 °C.

Zhao et al. [143], synthesized SnO₂/α-Fe₂O₃ (Fig. 22a) nanotubes by a facile single-capillary electrospinning route followed by heat treatment at 500 °C for 2 h for the improvement sensing properties. The binary SnO₂/α-Fe₂O₃ composite material contains two kinds of n-type oxides, which showed a typical n-type semiconductor gas-sensing characteristic towards ethanol and acetone vapors (Fig. 22b). They reported that compared with the gas-sensing properties of pristine α-Fe₂O₃ nanotubes in their previous work, obviously SnO₂/α-Fe₂O₃ nanotube based sensors showed greatly enhanced gas-sensing performance. The enhancement of gas-sensing was attributed to three factors: (1) grain refinement of α-Fe₂O₃ caused by SnO₂ additives because the addition of SnO₂ increased the crystal defects, and restrained the growth of α-Fe₂O₃ nanograins. (2) Decreasing the sensor resistances in air (R_a) by introducing proper amount of tin components. (3) Formation of heterostructures constructed by α-Fe₂O₃ and SnO₂ nanograins.

Fig. 22

a Typical TEM and HRTEM images, and an EDX spectrum of SnO₂/α-Fe₂O₃ nanotubes calcined at 500 °C, b typical dynamic response curves of SnO₂/α-Fe₂O₃ nanotubes toward ethanol and acetone with increasing gas concentrations at 200 °C, the inset shows the corresponding response transient at 100 ppm [143]

Kaneti et al. [165], reported a series α-Fe₂O₃ based nanocomposites as an n-butanol sensor (Fig. 23). The ternary α-Fe₂O₃–ZnO–Au nanocomposites were found to show higher sensitivity/responses of 113 toward 100 ppm n-butanol compared to single α-Fe₂O₃ (S = 11.7) and binary α-Fe₂O₃–ZnO (S = 54.4) sensing materials, and lower optimal operating temperature, i.e., 225 °C. They postulated that in the case of the ternary nanocomposites, the electron transfer was likely to occur in the order of Au → ZnO → α-Fe₂O₃. This was because the work function of ZnO (5.2 eV) is slightly higher than that of Au (5.1 eV) and therefore its Fermi energy level is lower than that of Au, so electrons flowed from Au to ZnO, to ensure similar Fermi energy levels between the two materials. This subsequently created an electron depletion layer at the Au/ZnO heterojunction interface. Next, as the Fermi level of ZnO ((E_f(ZnO) = 5.2 eV > E_f (α-Fe₂O₃) = 5.0 eV)) was not yet equalize to that of the base α-Fe₂O₃ nanorods, the transfer of electrons occurred from ZnO to α-Fe₂O₃, and hence, an additional electron depletion layer was formed at the α-Fe₂O₃/ZnO heterojunction interface. The presence of these multiple depletion layers greatly depleted the number of electrons which subsequently enhances the resistivity of the α-Fe₂O₃–ZnO–Au sensor when exposed to the gas, ultimately producing a significantly higher response. Also decorated Au nanoparticles on the surfaces of the α-Fe₂O₃–ZnO nanocomposites significantly increased the quantity of active oxygen ions via the catalytic dissociation of molecular oxygen. As a consequence, a high and fast degree of electron depletion occurred in the α-Fe₂O₃–ZnO–Au sensor, which significantly enhanced its response.

Fig. 23

a HRTEM image of α-Fe₂O₃-ZnO–Au nanocomposites, b response of different iron-based sensors to n-butanol at 225 °C [139]

Neri et al. [166], reported CeO₂–Fe₂O₃ thin films prepared by a liquid-phase method (LPD) and tested as methanol gas sensor. They proposed that methanol can adsorb on the sensor both molecularly (1) and dissociatively, forming a methoxy species at room temperature (2). Also the adsorbed methoxy group can lose another hydrogen(s) to a neighboring oxygen site creating formaldehyde type adsorbed specie (3) and/or more oxidized species, with the abstraction of H in α-carbon being the rate determining step (Fig. 24).

Fig. 24

Scheme of methanol adsorption mechanism on the surface of metal oxides

CeO₂–Fe₂O₃ sensor showed higher response to methanol compared to pristine iron oxide (Fig. 25a, b). They attributed role of Ce for higher response to methanol as following: (a) form a solid solution and reduce the grain size; (b) enhance the number of basic sites and methanol adsorption and (c) promote the methanol oxidation activity of the sensing layer.

Fig. 25

a Dynamic responses of CeO₂–Fe₂O₃ films to 100 ppm of methanol at 300 °C, b methanol calibration curves on CeO₂–Fe₂O₃ films at 400 °C [166]

Lou et al. [167], synthesized a novel hierarchical heterostructure of α-Fe₂O₃ nanorods/TiO₂ nanofibers with branch-like nanostructures (Fig. 25) using a simple two-step process called the electrospinning technique and hydrothermal process. α-Fe₂O₃/TiO₂, the sensor exhibited better sensing performance than that of the pristine ones (Fig. 26a). They reported two possible mechanisms that can cause such enhancement, first, α-Fe₂O₃/TiO₂ heterostructure generated an enhanced charge separation at the interface between the α-Fe₂O₃ nanorods and TiO₂ nanofibers, resulting in the enhanced conductance modulation. In the pristine TiO₂ nanofibers network, the electron depletion layer was formed due to the adsorption of oxygen ions (O⁻, O²⁻, O₂ ⁻), which led to a decrease in the sensor conductivity (Fig. 26a).

Fig. 26

FESEM images of the α-Fe₂O₃/TiO₂ heterostructures: a, b panoramic and c, d magnified

The electron transfer from the adsorbed TMA molecule to the TiO₂ was energetically feasible, as illustrated in Fig. 27a-left, resulting in a decrease of electron depletion layer and an increase of the charge-carrier density and thus an increase of the conductivity. After the TiO₂ nanofibers surface was decorated by the α-Fe₂O₃ nanorods, the sensor response to the TMA gas was markedly enhanced. The synergistic effect of TiO₂ nanofibers and α-Fe₂O₃ nanorods was a key factor for improving sensing performances. The Schottky barrier and an additional depletion layer between α-Fe₂O₃ nanorods and TiO₂ nanofibers was formed since the work function of α-Fe₂O₃ (5.88 eV) is greater than that of TiO₂ (4.2 eV), making it easy for the electrons in TiO₂ nanofibers to transfer to α-Fe₂O₃ nanorods (Fig. 27a right). Moreover, the surface area of the α-Fe₂O₃/TiO₂ heterostructure (37 m² g⁻¹) was larger than that of the bare TiO₂ nanofibers (21 m² g⁻¹) and α-Fe₂O₃ nanorods (7 m² g⁻¹), which allows them to absorb more gas molecules. Thus, it can make the conductivity have greater changes and improve the gas sensing performance. Also, the potential barriers formed at junctions between nanofibers, making it modulating for the electrons to travel between adjacent electrodes. Compared with the pristine one, the TMA response enhancement of α-Fe₂O₃/TiO₂ hierarchical heterostructure was due to the presence of α-Fe₂O₃/TiO₂ heterojunctions and α-Fe₂O₃/α-Fe₂O₃ homojunction (Fig. 27b right). These junctions acted as additional active sites, leading to improvement of sensing performances.

Fig. 27

a Response of the α-Fe₂O₃ nanorods, TiO₂ nanofibers, and α-Fe₂O₃/TiO₂ heterostructures to 10–2000 ppm of TMA (b), a, b Schematic illustration of sensing mechanism of the bare TiO₂ nanofibers and the α-Fe₂O₃/TiO₂ hierarchical heterostructure

Sun et al. [152], prepared a hierarchical nano-heterostructure consisted of inner SnO₂ hollow spheres surrounded by an outer α-Fe₂O₃ nanosheets. Deposition of the α-Fe₂O₃ on the SnO₂ hollow spheres outer surface was achieved by a facile microwave hydrothermal reaction to generate a double-shell SnO₂@ α-Fe₂O₃ nanostructure. They investigated ethanol sensing properties of SnO₂@ α-Fe₂O₃ nanostructures at 225 °C and compared the results with pristine SnO₂ nanoparticles (Fig. 28).

Fig. 28

a Responses of two sensors versus C₂H₅OH concentrations at 225 °C, b, c response transients of hollow SnO₂ nanospheres and hierarchical α-Fe₂O₃/SnO₂ heterostructures to diverse concentrations of ethanol at 225 °C [152]

The enhancement of performance was owning to novel heterostructure of the as-synthesized composite rather than simply a result of the addition of the other sensing component. They reported that the surface reactions between tested gases and adsorbed oxygen species depend on the acid–base property of oxide semiconductors. In terms of ethanol gas, the base property will be beneficial for the reaction with adsorbed oxygen. The addition of α-Fe₂O₃ enhanced the basicity of the composite, because α-Fe₂O₃ and SnO₂ are basic oxides. Therefore, the heterostructure of composites exhibited higher response to ethanol than that of pristine SnO₂. On the other hand, the band configuration at the interface of the α-Fe₂O₃/SnO₂ composites in different atmospheres was proposed. For composite, the electron transport strongly modulated by the heterojunction barrier, which has been generally researched for many heterostructure of devices. When the composites were exposed to air, the electrons in conduction bands of α-Fe₂O₃ and SnO₂ were trapped by oxygen to form adsorbed oxygen species, which results in the increase of barrier height, thereby decreasing the conductivity owing to the lower free electron concentration. Upon exposure to C₂H₅OH, which reacted with oxygen species, the electrons trapped in the adsorbed oxygen species were released back into the conduction bands of α-Fe₂O₃ and SnO₂. As such, the barrier height decreased owing to the higher concentration of electrons. Therefore, the resistance of the heterostructures was greatly decreased. Thus, the synergetic effect and the change of barrier height at the diverse gas atmosphere were origin of the improvement in gas sensing performances.

Wang et al. [153] synthesized hierarchical α-Fe₂O₃ (n-type)/NiO (p-type) composites with a hollow nanostructure (Fig. 29a–f) by a facile hydrothermal method and investigated toluene sensing of them (Fig. 29g). The enhanced gas sensing properties of α-Fe₂O₃/NiO composites were result of the following factors. First, the structural characteristics of hierarchical α-Fe₂O₃/NiO composites provide a high surface area and good permeability. The surface area of hierarchical α-Fe₂O₃/NiO composites (129.97 m² g⁻¹) was larger than that of pure flowerlike NiO (109.64 m² g⁻¹). This means that more oxygen absorbed and ionized on surfaces of this sensor. Meanwhile, the effective and rapid gas diffusion toward both the inner and surface regions of the hollow spheres easily accomplished because of the hollow structure, porous nanosheets, and large pores constituted by the space between nanosheets (Fig. 30a). Therefore, a strong response and a short response time were obtained. Second, the p–n junction between NiO and α-Fe₂O₃ was the principal factor in the enhanced response of composites. Because the Fermi level of α-Fe₂O₃ is higher than NiO (Fig. 30b, c), electrons flowed from α-Fe₂O₃ to NiO while holes will flow along the contrary direction of electrons, which will result in the formation of the hole depletion layer and the increase in the amount of the adsorbed oxygen species In this case, more electrons released back to the composites and the enhanced response obtained eventually. In addition, it is well-known that the lattice mismatch is unavoidable in the heterojunction. For this reason, a number of dangling bonds were produced in the material with a smaller lattice constant near the interface and form a large amount of interface states. For a p-type semiconductor, the dangling bonds performed an electron donor function. NiO possessed a lattice constant smaller than that of α-Fe₂O₃, therefore, the number of holes in NiO decreased near the interface, which was beneficial to further increase the amount of adsorbed oxygen species. Therefore, the enhanced gas response mainly attributed to the formation of the heterojunction.

Fig. 29

a–b TEM images of α-Fe₂O₃/NiO, c HRTEM image of α-Fe₂O₃/NiO, d–f scanning TEM (STEM) image and corresponding elemental mapping images, g calibration curve for α-Fe₂O₃/NiO and NiO [153]

Fig. 30

Schematic diagrams a illustrating the plausible reason for fast response and recovery and the energy band structure of b pure NiO and c α-Fe₂O₃/NiO heterostructures in air [153]

Vallejos et al. [156], reported α-Fe₂O₃@WO_3−x nanoneedles (Fig. 31) as toluene sensors. They reported that these nanocomposites had better toluene sensing than WO_3−x. In fact owning to the differences in the energy bands of WO_3−x and α-Fe₂O₃, heterojunctions emerged at the interface of these materials, inducing electron migration from α-Fe₂O₃ to WO_3−x and the formation of a larger electron density (accumulation layer) in WO_3−x, which facilitated oxygen adsorption at the functionalized WO_3−x nanoneedles surface (Fig. 32a) as supported by the larger resistance changes registered when exposing the Fe₂O₃@WO_3−x films to N₂ and air.

Fig. 31

a STEM and b HRTEM of the WO_3−x nanoneedles functionalized with α-Fe₂O₃. Insets in a display the localized EDX spectra and the size distribution of α-Fe₂O₃

Fig. 32

Possible interface-dependent mechanism involved in the gas sensors based on Fe₂O₃@WO_3−x. The left side shows the conduction channel mechanism in a nanoneedle cross-section (a) and the interfacial mechanism in the film comprising a network of nanoneedles (c) when exposed to air. The right side shows the same mechanisms when the films are exposed to a reducing gas such as toluene (b, d). D_cond is the diameter of the nondepleted region available for charge conduction through the core, E_CB represents the conduction band minimum, E_F is the Fermi level, E_VB is the valence band maximum, and LD is the Debye length or depth of the depletion region from the surface [156]

By subsequent oxygen adsorption during exposition of the films to air, the accumulation layer at WO_3−x depleted, decreasing the conduction channel and as a consequence reducing the conductivity along the nanoneedles. Alternatively, when the functionalized nanoneedles were exposed to toluene, it reacted with the pre-adsorbed oxygen and released electrons into the conduction band of the structure, which shrank the depletion layer, increasing the conduction channel and the conductivity along the nanoneedles, as illustrated in Fig. 32b. Hence, the nanoscale heterojunctions formed at the nanoneedles modulate the conduction channel mechanism of each nanoneedles, potentially enhancing the resistance changes during the gas–solid interactions. Also, conductance of the entire film extended across the electrodes of the device depends on a second interfacial mechanism modulated by the potential barriers originated during oxygen adsorption/desorption at the interface of different nanoneedles (Fig. 32c, d). This mechanism is present in nonfunctionalized and functionalized films, although the height of the potential barriers formed at the interface of the nanoneedles is also expected to be influenced by the incorporation of second-phase NPs at the surface of the nanoneedles. The changes in the activation energy of conduction noticed for Fe₂O₃@WO_3−x films, in relation to WO_3−x films, may be linked to these interface-dependent mechanisms. The functionalization of the WO_3−x with Fe₂O₃ NPs also indicated the presence of a surface-dependent mechanism, which coexisted with the interfacial mechanism enhancing the effects at the heterojunction of these functionalized material.

The term ferrite is commonly used to describe a class of magnetic oxide compounds that contain iron oxide as a principal component. The general formula is ${\text{M}}_{1}^{ + 2} {\text{M}}_{2}^{ + 3} {\text{O}}_{4}$, where M₁ generally is Fe²⁺, Mg²⁺, Ni²⁺, Co²⁺, Zn²⁺, Cu²⁺, and M₂ generally is Fe⁺³, Cr⁺³ and Mn³⁺. When M³⁺ is Fe⁺³, i.e. MFe₂O₄ (M = Mg, Zn, Ni, Co, Cd, Mn, etc.), the resulting spinel ferrites, are widely used as sensing materials. A great advantage of ferrites is their porosity, which is necessary for a humidity sensor [168]. The gas sensing properties of the ferrites are dependent on their chemical composition and nanostructural characteristics, which can be controlled in the synthesis and fabrication processes. For the gas sensing studies, there is a great need for developing synthesis and fabrication techniques that are relatively simple and yield controlled particle sizes [169].

Singh et al. [157], synthesized a series of nanostructured cobalt ferrite systems (CoFe₂O₄) in different compositions via chemical co-precipitation method and test their LPG sensing properties at room temperature. LPG sensing at room temperature is interesting for industrial applications as most of the currently available LPG sensors are operated above room temperature. They reported that LPG sensing investigations of the fabricated pellets illustrate that the cobalt ferrite synthesized in 1:1 M ratio possesses an improved response in comparison to other compositions. The maximum sensitivity of cobalt ferrite film sensor was 2.0 MΩ/s. The response and recovery times were 30 and 60 s, respectively. The sensor was 95 % reproducible after 3 months of fabrication of the film, showing the stability of the fabricated sensor.

In another attempt, Singh et al. [161], reported nanorods and mixed shaped (nanospheres/nanocubes) copper ferrite (CuFe₂O₄) for liquefied petroleum gas (LPG) sensing at room temperature. Their results show that the mixed shaped CuFe₂O₄ had an improved sensing performance (R_a/R_g = 60 for 5 % vol LPG) over that of the CuFe₂O₄ nanorods (R_a/R_g = 35 for 5 % vol LPG). They attributed enhanced LPG sensing to higher surface area of mixed shaped copper ferrites.

6 Polymer/α-Fe₂O₃ nanocomposite gas sensors

Conducting polymers can easily synthesized through chemical or electrochemical processes and their molecular chain structure can be modified conveniently by copolymerization or structural derivations. In addition, conducting polymers have good mechanical properties, which allow a facile fabrication of sensors.

Several conducting polymers (organic semiconductors), such as polyaniline (PANI) [170], polypyrrole, polythiophene and polyacetylene have gas sensitivity at normal temperature [171], but they have long response time due to the highly ordered structure. Owning to the high affinity of conducting polymers toward volatile organic compounds (VOCs) and moisture present in the environment, they are sometimes unstable and exhibit poor sensitivity so their commercialization as gas sensors has limited [172, 173].

Issues related to low conductivity and poor stability of organic materials, and the high-temperature operation and complicated processability of inorganic materials, hinder their use in gas sensor fabrication. Conducting polymer matrices embedded with nano-scale metal or metal oxide particles forms a new class of nanocomposite materials that has shown better gas sensing features. The reasons for such an improvement can be attributed either to catalytic behaviors of metal particles or to the formation of metal oxide-polymer junction [174].

The use of hybrid nanocomposites of these two classes of materials may result in gas sensors with improved and efficient gas sensing characteristics. In fact use of polymers and metal oxide counterparts in hybrid nanocomposite forms may help to eliminate their particular drawbacks due to synergetic or complementary effects, leading to development of improved gas sensing devices. These nanocomposite materials have aroused extensive interest in gas sensing applications and recently to complement the characteristics of pure inorganic and organic materials, organic–inorganic sensing hybrids have been developed [173, 175]. The sensitivity enhancement of such type sensors is due to a synergic effect of organic and inorganic materials [176]. Furthermore These composites minimize the need to operate at high temperature of typical of bulk metal oxides and sensing materials with low operating temperatures can inhibit structural changes, reduces the power consumption and enables safer detection of combustible gases [177]. Mechanism of gas sensing behavior of organic–inorganic hybrid nanocomposite sensors is illustrated in Fig. 33. Exposure of reducing/oxidizing gas molecules onto nanocomposite film results in a change of resistance. Simultaneously, gas flow is cut off and fresh air is introduced into a test chamber, thus leading to restoration of the original resistance value. The observed change in the resistance is due to physical adsorption of the gaseous molecules onto the surface of the nanocomposite film. The interaction of gaseous molecules with a π electron network of a polymer that has embedded metal/metal oxide nanoparticles results in capture/donation of electrons, depending upon the nature of gaseous molecules, leading to the decrease/increase of the resistance. The n-type metal oxide nanoparticles form a barrier layer with a polymer matrix leading to the formation of the depletion region. The interaction of the nanocomposite film surface with target gas molecules causes a change in the width of the depletion region (W₀) and therefore modulates the conductivity of the sensing element. The space charge region (W_gas) varies with the decrease or increase of electrons (by adsorption of gas molecules) in the polymers, which in turn results in conductivity change. The modulation of the space charge region at the interface of the metal oxide nanoparticles with the polymer matrix results in enhanced sensitivity for the desired gas [173].

Fig. 33

Energy band gap diagram of organic–inorganic hybrid nanocomposites on gas exposure [173]

Table 4 shows gas sensing properties of polymer/α-Fe₂O₃ nanocomposite gas sensors which are mainly composites of polypyrrole with iron oxide. Polypyrrole (PPy) is the most studied conducting polymers for gas sensing applications, but the self-limitations reduce the application largely. This is mainly due to sensitivity of PPy to electrochemical and chemical degradation. Furthermore the conductivity exhibits a long-term irreversible decay due to the irreversible attack of oxygen present in the ambient, and the long response time due to the highly ordered structure [178].

Suri et al. [172], reported polypyrrole/α-Fe₂O₃ nanocomposites with different amounts of PPY (1, 5, 10, 15 %) for gas sensing application. The response of the sensors to different gases, namely CO₂, N₂ and CH₄ at various pressures was investigated. Figure 34a, b shows response of the sensors with varying polypyrrole concentration to various CO₂ and CH₄ gases respectively.

Fig. 34

a Variation of sensivity with pressure of CO₂ gas and with polypyrrole concentration at three different CO₂ gas pressure, b variation of sensivity with pressure of CH₄ gas and with polypyrrole concentration at three different CH₄ gas pressure [172]

It is seen that sensor with 15 % polypyrrole shows maximum response for all gases. This composite showed the presence of fibrillar morphology with microscopic voids. As the permeability depends on the kinetic diameter of the gas molecule and also on the size of pores in the sensor. It was expected that this structure to be highly permeable to all the gases. The response of all sensors was maximum to CO₂ gas as compared to other gases which could be due the fact that larger the molecule, lesser is permeability of the gas. As the kinetic diameter of CO₂ molecule (3.3 Å) was lesser than of N₂ (3.64 Å) and CH₄ (3.8 Å), therefore, permeability of CH₄ was minimum and CO₂ was maximum, resulting in highest response and sensitivity of all sensors to CO₂ and least to CH₄ gas.

Navale et al. [179], reported PPy/α-Fe₂O₃ nanocomposite as NO₂ sensor. They found that PPy/α-Fe₂O₃ hybrid nanocomposites can complement the drawbacks of pure PPy and α-Fe₂O₃ to some extent. As shown in Fig. 35a, b, they observed that PPy/α-Fe₂O₃ (50 %) hybrid sensor operating at room temperature could detect NO₂ at low concentration (10 ppm) with very high selectivity (18 % compared to C₂H₅OH) and high sensitivity (56 %), with better stability (85 %).

Fig. 35

a Response of PPy/α-Fe₂O₃ (10–50 %) nanocomposite film for100 ppm of NO₂, b selectivity of PPy, α-Fe₂O₃ and PPy/α-Fe₂O₃ (50 %) sensor films [179]

Among many organic sensitive materials, polyaniline (PANI) has cheap raw material, simple synthesis procedure, excellent environmental stability, high conductivity and potential solutions melt processing possibilities. In addition, polyaniline can easily build soft and strong films with good electricity denaturation [170]. Bandgar et al. [175], demonstrated that α-Fe₂O₃ nanoparticles (50 %): PANI hybrid nanocomposites films were highly selective to NH₃ with high sensitivity (50 at 100 ppm), fast response time (29 s) and highly reproducible response curves at 160 °C (Fig. 36a).

Fig. 36

a Response of PANI/α-Fe₂O₃ (50 %) nanocomposite film for various concentrations of NH₃, and b proposed energy band diagram for PANI/α-Fe₂O₃ hybrid nanocomposite film with the interaction of NH₃ gas [175]

Figure 36b shows proposed energy band diagram to investigate the NH₃ gas sensing mechanism of PANI/α-Fe₂O₃ hybrid nano-composite thin films, where HOMO presents the highest occupied molecular orbital level, and LUMO is the lowest unoccupied molecular orbital level. Charge separation can be enhanced due to well energy band gap matching between the conduction band of α-Fe₂O₃ and the LUMO level of PANI for charge transfer. Therefore such enhancement promotes NH₃ gas-sensing ability of the PANI/α-Fe₂O₃ hybrid nanocomposite.

7 Graphene/α-Fe₂O₃ nanocomposite gas sensors

Currently graphene is becoming a “rising star” material after its successful production by a simple scotch tape approach using readily available graphite in 2004 by Andre Geim and his coworkers [186]. One of the most fascinating properties of graphene is that it has only one atom thick and thus instead of being a 3D material, this material is a 2D material. As a new carbon material, graphene possesses remarkable electrical and thermal conductivity, high surface area (theoretical value of 2630 m²/g), high chemical stability and excellent adsorptivity which make it a good material for gas sensing [187].

The gas sensing characteristics of graphene were first investigated by Schedin et al. [188] and since many researchers have reported graphene functionalized metal oxide NPs to enhance gas sensing performance, i.e., sensitivity, selectivity, or response/recovery times. Although metal oxide NPs have good electron transfer properties and large surface area, they suffer from degradation of sensitivity and long response/recovery times due to aggregation between particles, thereby reducing the effective surface area for the reaction with gas molecules. Therefore, combining NPs and graphene/reduced graphene oxide (rGO) sheets to further increase the gas sensing performance by inhibiting aggregation of NPs and graphene sheets is a powerful strategy. For example ZnO/graphene composites [189] have been reported as gas sensor with improved sensivity. In general the synergistic effect between graphene and metal oxides can be summarized as follows: (1) graphene can control the size and the morphology of metal oxides during synthesis; (2) graphene increases the conductivity of metal oxides, which may rapidly transfer the electrons acquired from the surface reaction of gas molecules and metal oxides to electrodes (3) there are p–n junctions between p-type graphene and n-type metal oxides, which may modulate the space-charged layers at the interfaces between graphene and metal oxides; and (4) metal-oxide nanostructures can prevent aggregation of graphene [190].

Table 5 shows nanocomposites of graphene based/α-Fe₂O₃ nanocomposites. As it can be seen there are few reports about gas sensing properties of graphene base/α-Fe₂O₃ nanocomposites. Therefore it seems there are a lot of rooms for research in this kind of composites as gas sensor materials.

Liang et al. [187], reported α-Fe₂O₃ nanoparticles anchored on graphene as gas sensor material (Fig. 37a). The α-Fe₂O₃@graphene nanocomposite exhibited very impressive sensitivity toward ethanol at high concentration (1000 ppm), which was about three times higher than that of its counterpart of Fe₂O₃ nanoparticles (Fig. 37b).

Fig. 37

a The transmission electron microscopy micrographs of Fe₂O₃–G, b response of Fe₂O₃-G (0.02) and pure Fe₂O₃ sensors to ethanol with different concentration [187]

They reported the improvement of sensing performance of α-Fe₂O₃@graphene nanocomposite to the following three reasons. Firstly, the high surface area distributed in the two-dimensional space facilitated the diffusion of ethanol vapor and improved the reaction of the ethanol gas with surface adsorbed oxygen. The α-Fe₂O₃@graphene nanostructures (Fig. 38b) composed of multiple nanosheets possessed a porous and loose structure in comparison to the compact pure α-Fe₂O₃ nanoparticles (Fig. 38a). These kinds of loose structures provided a large specific surface area, which had great benefit to numerous oxygen molecules adsorbed onto the α-Fe₂O₃@graphene surface. Secondly, graphene exhibit outstanding electrical conductivity, which improved conductivity of composites and result in electrons quickly spreaded to surface of the semiconductor, leading to quick response and recovery time. Finally, graphene created a Schottky contact at the interface with α-Fe₂O₃. The α-Fe₂O₃@graphene interface was a forward-biased Schottky barrier, thus resulting in the easy capture and migration of electrons from α-Fe₂O₃ nanocomposite to graphene. As a result, the gas sensing performance had been greatly improved in the presence of graphene.

Fig. 38

Schematic illustrations of three different sensor materials: a pure Fe₂O₃, b Fe₂O₃–G (0.02), c Fe₂O₃–G (0.06)

Also they reported that with increase of graphene content, response value of sensor decreased. In fact, as a kind of excellent electric conductor, the graphene sheet were connected with each other with the increase of graphene content, and the micro electric bridges formed both on the surface and in the body of the α-Fe₂O₃@graphene nanocomposite coating when the content of graphene was too high (Fig. 38c). This made α-Fe₂O₃@graphene nanocomposite coating radically reduced the semiconductor’s resistance and making it into a full-blown conductor. Consequently, it was hard to feel changes of the component’s resistance for the computer-controlled gas-sensing measurement system. As a result, the response value to ethanol decreased with the increase of graphene contents.

Dong et al. [192], reported nanosphere-like α-Fe₂O₃ modified reduced graphene oxide (rGO) nanosheets prepared by a simple hydrothermal method without any surfactant or template. They showed good NO₂ response of sensor in room temperature (Fig. 39a). They proposed a mechanism for gas sensing as follows: when α-Fe₂O₃/rGO nanocomposites were exposed to NO₂ (Fig. 39b), an electron of rGO was captured by NO₂, which led to the decrease of resistance. At the same time, NO₂ reacted with O²⁻ (ads) on the surface of α-Fe₂O₃ of the nanocomposites, formed an intermediate complex NO₃:

$$2{\text{NO}}_{2} ({\text{gas}}) + {\text{O}}_{2}^{ - } ({\text{ads}}) \to 2{\text{NO}}_{3}^{ - } ({\text{ads}})$$

(15)

The reaction of NO₂ and O²⁻ (ads) led to unbalance of charge on the surface of α-Fe₂O₃. rGO provided more electrons to α-Fe₂O₃ to form O²⁻ (<100 °C) on the surface of α-Fe₂O₃, consequently more holes produced in rGO resulting in the decrease of the nanocomposites resistance. When the nanocomposites were exposed to air again, NO₂ (ads) species desorbed with leaving the electrons to the nanocomposites. Electrons combined with holes again, which caused the resistance of the nanocomposites increased to the starting value. Also uniformly distributed α-Fe₂O₃ nanospheres separated by rGO layers perfectly so the specific surface area of the composites increased greatly compared with that of rGO, which was benefit for more NO₂ molecules to adsorb and reacted on the surface of the nanocomposites. As a consequence, the nanocomposites exhibit high response to NO₂ even in room temperature.

Fig. 39

a Dynamic responses of α-Fe₂O₃/rGO nanocomposites to different concentrations of NO₂, b proposed sensing mechanism of α-Fe₂O₃/rGO nanocomposites to NO₂ [192]

8 Stability

An important issue of metal-oxide gas sensor materials is their low stability and long-range signal drift. This problem leads to in uncertain results, false alarms and the need to frequently recalibrate sensors. Little attention is paid in the literature to the problems of stability and only a few papers, report the stability of sensor response in a period of several days. There are two types of stability. One is connected with reproducibility of sensor characteristics during a certain period of time at working conditions, which may include high temperature and the presence of a known analyte. Such stability may be referred to as active stability. Another type of sensor stability, which can be called conservative stability, is connected with retaining the sensitivity and selectivity during a period of time at normal storage conditions, such as room temperature and ambient humidity [194]. Gas sensors should provide long-term performance, even at high operation temperature and corrosive media. In general, any gas sensing device should exhibit a stable and reproducible signal for the period of at least 2–3 years. This means that stability of gas sensor is one of most important factors determining the practical uses of gas sensors [4].

Generally, nanostructured oxides with small grains as well as nanotubes, nanorods etc. are subject to degradation because of their high reactivity. There is no unified approach to increasing the stability of metal-oxide gas sensors. To some extent, stability can be increased by calcination and annealing as the post-processing treatment and by reducing the working temperature of the sensor element. Here we present some α-Fe₂O₃ based sensors with good stability.

Figure 40 shows the initial stabilization curve of the α-Fe₂O₃ film at 200 °C. The decrease in resistance observed at the beginning of the stabilization period can be attributed to the generation of electrons due to thermal excitation. The resistance of the film decreased gradually in the first 3 min, for next 6.5 min, the resistance decreased slowly, and then it remained constant. After 30 min the α-Fe₂O₃ thin film attained a stable constant resistance 200 °C. Then sensor was at a stable baseline for long time at 400 °C, showing high stability of iron oxide sensor.

Fig. 40

Resistance stabilization curve with time for α-Fe₂O₃ film sensor [195]

Figure 41a shows that PPy/α-Fe₂O₃ nanocomposites have good stability compared with pure PPy and α-Fe₂O₃ sensors. Therefore PPy/α-Fe₂O₃ sensors can detect target gases over a long term period. Figure 41b shows excellent stability of α-Fe₂O₃/ZnO nanocomposites toward 100 ppm ethanol at 200 °C during 120 days, where no obvious attenuation was found. Thus the α-Fe₂O₃/ZnO sensors had good reproducibility and long-time stability. This indicated that the α-Fe₂O₃/ZnO sensor can apply to detect various kinds of gases and vapors safely and stably.

Fig. 41

Stability of a PPy/α-Fe₂O₃ sensor film to 100 ppm of NO₂ [179], b the long-term response values of α-Fe₂O₃/ZnO to 100 ppm ethanol at 200  °C [133]

Figure 42a shows stability of α-Fe₂O₃–ZnO–Au sensor toward 100 ppm of n-butanol over a period of 30 days. As it can be seen, repeated experiments were carried out every day, and the sensors showed little variations in resistance that demonstrated good response performance even after 30 days. Figure 42b presents sensivity variations of 0.75 % Pd α-Fe₂O₃ sensor where little variations are observed and despite this certainly sensor can work over many days.

Fig. 42

a Stability of α-Fe₂O₃–ZnO–Au sensor toward 100 ppm of n-butanol over a period of 30 days [139], b long-time stability of sensor 0.75 % Pd α–Fe₂O₃ to 8071 ppm of LPG [128]

9 Conclusions

α-Fe₂O₃ nanomaterials possess large specific surface area which is essential for gas sensing applications. Moreover, they are nontoxic and have environmentally friendly nature, excellent biocompatibility and high stability which allow α-Fe₂O₃ nanomaterials to be adopted in a wide range of applications especially in the field of gas sensing applications. As is known, α-Fe₂O₃ based gas sensors are sensitive to many types of gases based on the resistance change, ensuring a wide application in gas monitoring areas. In this review paper, α-Fe₂O₃ based nanomaterials as gas sensors were discussed. The sensitivity of a gas sensor is the major concern in most research in this field. The sensitivity as well as the dynamic response of α-Fe₂O₃ based gas sensors can be improved significantly when nanostructured α-Fe₂O₃ is employed. Therefore, the newly developed nanostructured α-Fe₂O₃ could be the potential candidate for highly sensitive gas sensor. In other ways, modification with noble metals (e.g., Pt, Pd, Au, or Ag) could effectively improve the interaction between α-Fe₂O₃ surface and gas molecules, thereby reducing the operation temperature. Also introducing other materials (such as metal oxides, polymers, graphene) could enhance gas adsorption or electron transfer and consequently improve the sensitivity of sensor.

In terms of gas sensor, α-Fe₂O₃-based sensors is widely applicable for the detection of various gases such as H₂, O₂, CO, H₂O, VOCs, etc. However, several obstacles have to be overcome for its future application. For example, the working temperatures are still high (∼200 °C), and the recovery time is too long. These shortcomings can be partly avoided or improved by depositing noble metals, by composing with other semiconductors, and, most importantly, by introducing newly developed nanostructured α-Fe₂O₃ to the sensors. Furthermore, the current research focuses merely on the target analyte in a simple matrix instead of complex matrixes and real samples.

Pure α-Fe₂O₃ gas sensors are very cheap and are show higher sensivity to ethanol and acetone, however they suffer from lack of selectivity and sometime low sensivity. Metal/α-Fe₂O₃ sensors have higher sensivity than pure ones and can be used to detect of wider range of gases. Metal oxide/α-Fe₂O₃ nanocomposite gas sensors are principally compounds of α-Fe₂O₃ with other n-types metal oxides such as ZnO, SnO₂ and In₂O₃ that show high sensivity to different gases, high selectivity and high stability, so they are promising materials for industrial gas sensors. Reports about polymer/α-Fe₂O₃ gas sensors and graphene/α-Fe₂O₃ gas sensors are still rare, but these kinds of α-Fe₂O₃ based gas sensors are promising materials for fabrication of sensors which works at room temperatures.

Regarding stability of α-Fe₂O₃ based gas sensors, there are some reports that show they have high stability in pure, doped or composite form, but it seems that stability should be study more in future researches as it is one of the main parameters of gas sensors. Research on α-Fe₂O₃ based gas sensors gas sensors relates to many fields, such as physics, chemistry, electronics and materials science and addressing their problems is a great challenge, for which it is important to enhance interdisciplinary collaboration.