In addition to organic contaminants, the inorganic toxic ions impose another threat to the drinking water quality as well. Generally, toxic ions includes heavy metal ions such as Pb²⁺, Cd²⁺, Cu²⁺, Ni²⁺, Hg²⁺, and As³⁺. These toxic ions can be removed by some well-developed techniques including chemical precipitation, reverse osmosis, membrane filtration and adsorption technology and so on (Fu and Wang, 2011). Adsorption technology has various merits including cost-effectiveness, relatively high efficiency, and convenient utilization, which make it one of the most promising approaches for ions removal from contaminated water. The efficient adsorption materials are usually materials of porous structure associated with a large surface area. For example, charcoal has been used in the water filtration for hundreds of years. In modern days, activated carbons with a broad range of pore sizes resulting from intrinsic cracks are most commonly utilized. Many literatures regarding the traditional toxic ion absorption by carbon-based materials have been published (Kadirvelu et al., 2001; Meena et al., 2005). We will therefore, exclude carbon-based materials in the following discussion but only concentrate on other inorganic hierarchical nanostructures mainly metal oxide designed for better toxic ions removal.

The nanostructured materials designed for metal ions trapping and absorbing has received great research interest and many typical nanostructure systems have been demonstrated since the end of last century. Basically, the advantages “nano” could provide include high surface to bulk ratio associated with high porosity, large metal ion absorbing capacity, and good regeneration ability. In this section, we will review the latest progress in nanomaterials development for ion trapping and adsorption and elaborate the rational hierarchy design.

The selection and design for absorbent materials are required to meet several criteria as follows. First, the absorbent itself should be non-toxic in nature. Second, the absorbent should have selectivity and sensitivity as high as possible toward specific type of ion in addition to the large adsorption capability. Third, the adsorbed ions must be easy to collect off the surface of absorbent materials so that the absorbents could be used for repeated times.

Many metal oxides are low cost and non-toxic materials and the hierarchical nature makes them promising for ion removal in the waste water. Typical examples include magnesium oxide (MgO) (Schiller and Khalafalla, 1984), aluminum oxide (Al₂O₃) (Zhang et al., 2012b), iron oxide (Fe₂O₃) (Gadde and Laitinen, 1974; Phuengprasop et al., 2011), titanium oxide (TiO₂) (Chen et al., 2012), and so on. Much research progress has been made in the synthesis of these metal oxides with controlled nanoscale morphology (Hua et al., 2012). Yu et al. reported the synthesis of hierarchical micro/nanostructured MgO which exhibits excellent As(III) and As(V) removal capability (Yu et al., 2011). Different amount of potassium carbonate was found to tune the structure from flower-like shape to nest-like geometry. It is generally alleged that the removal of As(V), which is less toxic compared with As(III), is easy to achieve. However, the hierarchical MgO microstructure demonstrated large As(III) trapping and adsorption. By detailed XPS analysis, the whole adsorption process involved formation of Mg(OH)₂ by partial reaction of MgO with the aqueous environment, which provided excessive adsorption sites. The adsorption capacities of MgO with both shapes are proved to be higher than similar hierarchical structures of other materials.

Hierarchical iron oxide nanostructures have been extensively studied among all the metal oxide nanostructures for the heavy metal ion removal (Zhong et al., 2006; Li et al., 2011b; Cao et al., 2012). In addition to the large adsorption capacity, high surface area and low cost, an important advantage over other metal oxides is the magnetic property which makes it easily recoverable from water after toxic ion removal. Magnetite (Fe₃O₄) and maghemite (γ-Fe₂O₃), which are ferromagnetic and ferrimagnetic in nature, have been proved to act as efficient ion absorbents. The relatively strong magnetism makes them promising as magnetic absorbents which can be recycled by small magnetic gradient. Various hierarchical structures including hollow spheres and nanoflowers have been achieved by facile wet chemical approach. The removal efficiency was found to be influenced by particle size. For arsenic ion removal, for example, the efficiency increased by orders of magnitude with the reduced particle diameter. However, the nanoparticles are going to aggregate in water when the size is small and surface modification is thus, important. The humic acid coated Fe₃O₄ nanoparticles reported by Liu et al. were found to be stable in tap water, natural waters, and acidic/basic solutions ranging from 0.1 M HCl to 2 M NaOH with low leaching of Fe (<3.7%) and HA (<5.3%) (Liu et al., 2008b). Zhong et al. synthesized self-assembled flower-like iron oxide nanostructures (Zhong et al., 2006). Different calcination conditions were applied to realize the phase transformation that leads to the formation of α-Fe₂O₃, γ-Fe₂O₃, and Fe₃O₄, all of which exhibit good adsorption capability in Ar(V) and Cr(VI) removal. Specifically, the as-prepared iron oxides have larger surface area compared with their bulk counterparts. The removal capacity for Ar(V) (4.65–5.3 mg/g) is slightly better than that of commercial TiO₂ powders (4.11 mg/g). In the Cr(VI) adsorption, however, the removal capacity of three types of iron oxides is nearly doubled compared with TiO₂.

In addition to heavy metal ions, the majority of water and soil contamination is of organic compounds, which impose severe threat to environmental and human health. It is therefore, of high necessity to develop environmental sensors of high sensitivity and selectivity that detect trace organic compound(s) rapidly and accurately (Rogers, 2006; Badihi-Mossberg et al., 2007; Zhang and Fang, 2010). Traditional analytical chemistry strategy is usually time consuming and less cost-effective in spite of good analytical accuracy. Biosensing technique enabled by nanomaterials represents a low cost, simple, and fast analytical method that complements the current traditional analytical instrumentations for trace organic compound detection and monitoring (Wang, 2005; Zhang and Fang, 2010; Aragay et al., 2012).

The biosensors can be categorized into several types based on the working mechanisms and the transducing components. Typical biosensors include enzyme and antibody based biosensors. Enzymatic biosensors have several advantages such as stable source of materials, tunable catalytic property, and surface selectivity enabled by gene engineering and modulation of enzyme activity (Rogers, 2006). On the other hand, antibody-based biosensors are of higher precision and selectivity toward certain pollutants since antibodies are able to specifically bind to individual chemical compound or group. Besteman et al. demonstrated the enzyme coated carbon nanotube could act as a single molecule biosensor (Besteman et al., 2003). The attachment of the redox enzyme glucose oxidase (GOx) to the nanotube sidewall was found to induce a clear change of the conductance. With increased complexity in geometry, hierarchical nanostructures have been proved lately to work as potential candidates for biosensor devices. For example, Wei et al. presented an enzymatic glucose biosensor using ZnO nanorod arrays (Wei et al., 2006). The negatively charged GOx was successfully immobilized on positively charged ZnO nanorods through electrostatic interaction. The response time for such prototype biosensors is less than 5 s and the glucose concentration detection limit reaches as low as 0.1 mM. In addition to semiconductor nanowires, hierarchical metallic nanowire array has also been demonstrated to work for glucose detection. Delvaux et al. prepared gold nanotube arrays and GOx was immobilized to the modified gold surface by cross-linking or covalent attachment (Delvaux and Demoustier-Champagne, 2003). Recently, much research effort has been focused on one-pot self-assembly of bionanocomposite which can be directly applied to biosensing application. Zeng et al. reported the self assembled graphene–enzyme hierarchical nanostructures in aqueous solution for electrochemical biosensing (Zeng et al., 2010). The bionanocomposite consists of sodium dodecyl benzene sulphonate (SDBS) functionalized graphene sheets and horseradish peroxidase, which exhibits high electrocatalytic activity toward H₂O₂ detection with high sensitivity, low detection limit and fast response. Shao et al. demonstrated single wall carbon nanotube functionalized with IGF1R-specific and Her2-specific antibodies for breast cancer cell detection (Shao et al., 2008).

Different nanostructured biosensors have been proposed based on different working mechanism and physical principles. For example, surface plasmon resonance, an important phenomenon taking place on metal nanoparticle surface as described in previous section of plasmonic photocatlaysis, has also found great potential in biosensor application for trace amount of molecules detection (Shankaran et al., 2007). Other physical principles for biosensors include chemically sensitive field-effect transistors (Janata, 1994) and microcantilevers (Hansen and Thundat, 2005). An apt example is the silicon nanowire developed by Lieber group (Patolsky et al., 2006). The characteristic nanoscale feature of Si nanowire enables higher sensitivity than traditional planar FET sensors because the nanoscale dimension is comparable to the size of biomolecules. The intrinsic oxide coating surface facilitates the linking of receptors on the surface. When such nanowire based sensors are put into specific solutions with certain macromolecules, the binding will lead to changes of surface charges on the nanowire, therefore, the conductance variation easily detected through direct electrical measurement. Avarez et al. reported a nanomechanical biosensor to detect pesticide DDT by measuring the tiny bending of a microcantilever produced by differential surface stress (Alvarez et al., 2003).

The conventional gas sensor fabrication has witnessed a revolutionary transition from powder based thick film preparation to later thin film technology. Compared with powder thick film technology, thin films prepared by physical or chemical vapor deposition enable more controls over the materials microstructures such as grain size and boundary. Generally, gas sensors are composed of pristine or doped metal oxide with semiconducting, photoconducting, piezoelectric, pyroelectric, or luminescent properties, which can realize transduction upon external stimuli in different ways (Azad et al., 1992). In this review we will only discuss conductometric semiconducting gas sensors since they are cost-effective, flexible in manufacturing and most widely used with a large number of detectable gases. The performance of thin film sensors, however, is reversible and reproducible only when relatively high temperature is maintained. This is because the variation in environmental temperature usually leads to grain coalescence and grain boundary modification which largely affect the electrical property of the semiconducting thin films. The development in nanofabrication has allowed large scale integration of nanoscale building blocks in a controlled fashion that offers possibility to manipulate the microscopic geometry and physical properties. Different semiconducting nanowire arrays of single crystalline nature have been prepared by either chemical vapor deposition or low temperature solution approach. The single crystallinity of semiconductor nanowires is able to improve the thermal stability without grain coalescence and grain boundary alteration. Moreover, the adjustable spacing between each individual nanowire allows for porosity manipulation in a controlled way, which is challenging in thin film technology, to achieve better gas sensing performance.

Other advantageous features of hierarchical nanostructures include the high surface-to-volume ratio, which indicates a large fraction of atoms on the surface making surface reaction more kinetically favorable (Sun et al., 2012). Secondly, the nanoscale dimensions of the building blocks are very close or at least comparable to the Debye length, which measures the field penetration depth in the materials. For example, in most semiconducting oxides the Debye length is usually around several nanometers. When the nanoparticle size or the nanowire diameter approaches this dimension, space charge region will constitute majority of the particle interior or nanowire cross section (Wang et al., 2010). With most of the electrons trapped in the surface states, the electronic property of both nanoparticles and nanowires is thus, going to be greatly influenced by the surface process. Such phenomenon is critical to achieve higher sensitivity in the gas sensor design. In the conductometric semiconducting gas sensors, the sensitivity is defined as:

R_g is the resistance of gas sensing materials at the target gaseous atmosphere, which is either reducing or oxidative, while R_a stands for the resistance of gas sensing materials in the ambient atmosphere, which is air condition in most situations.

The gas sensing mechanism in conductometric semiconducting metal oxide sensors is illustrated by using n-type semiconductor as an example. Generally the trapping of electrons in the surface state upon gas molecules adsorption and the consequent electronic band changes contribute to the conductivity variation. When the n-type semiconductor is exposed to the air, certain oxygen gas molecules are adsorbed on the surface with various extent of coverage depending on the surface condition. These adsorbed oxygen molecules are going to extract electrons from the metal oxide surface and lead to a band structure bending upward at the surface. In semiconductor physics, this created electron-depleted area from band bending is called the space charge layer. The conductivity of the n-type semiconductor with surface trapped electrons is thus, decreased compared with the semiconductor of a clean surface. However, when a reducing type of gas or another oxidative gas is introduced to the surface, the surface reaction with the reducing gas or the competitive adsorption from the other oxidative gas molecules is able to change the original oxygen molecules coverage, which results in the band structure completely restored or less bended. Therefore, the conductivity of the gas sensing semiconductor is recovered to different degree and triggers the resistance change.

As discussed above, much research effort has been devoted to the metal oxide based semiconductors for the gas sensing applications. Wang et al. categorize these metal oxides for gas sensing application into two groups which are transition metal oxide and non-transition metal oxides (Wang et al., 2010). The first type of non-transition metal oxides is pre-transition metal oxides such as MgO and Al₂O₃ which demonstrates little electronic conductivity with a large band gap and these materials are therefore, rarely used in gas sensing. The other type of non-transition metal oxides is post-transition metal oxides in periodic table. Typical examples involve Ga₂O₃ and SnO₂. The transition metal oxides are more sensitive to the environment because of their multiple valence states. With the d orbitals empty or fully occupied, metal oxides such as TiO₂, V₂O₅, ZnO, WO₃, and SnO₂ represent widely used gas sensing materials for reliable performance (Wan et al., 2004; Ponzoni et al., 2006; Zuruzi et al., 2006; Young-Jin et al., 2008). Based on the gas sensing mechanism discussed previously, one of the key parameters for designing high performance gas sensing materials is large surface area and porosity. Hierarchical nanostructures such as hollow architectures and flower shaped structures will satisfy the requirements (Lee, 2009). Typical synthetic strategies including both templated growth and self-assembly to prepare architectures with hollow and flower-like geometry have been introduced in the photocatalytic water treatment section. We will therefore, review the gas sensing performance to illustrate the improvement enabled by these typical geometries. In addition, gas sensors targeted at different toxic gas detections will be covered.

For organic gas sensing, Huang et al. developed hierarchical SnO₂ nanostructure with porous flower-like shape by a facile hydrothermal method followed with thermal annealing (Huang et al., 2010). The hierarchical SnO₂ nanoflowers exhibit good response and reversibility toward several organic gases including ethanol, n-butanol, methanol, 2-propanol and acetone at different working temperatures. The sensitivity and response time is better compared with solid thin film, porous thin film and the nanoparticles. Zhang et al. reported the one-pot synthesis to prepare the Cu₂O hollow spheres with multiple shells (Zhang et al., 2007). The as-prepared hierarchical Cu₂O had the sensitivity of 1.9 toward 10 ppm ethanol with no significant decay detected after more than 100 cycling tests. The high sensitivity can be ascribed to the improved surface accessibility enabled by multiple layers. These works well illustrates the advantages hierarchical nanostructures have brought in high performance gas sensor design.

Gas sensor development for inorganic toxic gas detection is of crucial importance to public health and safety. Toxic gaseous chemicals usually include hydrogen sulfide, ammonia, carbon monoxide, nitric oxide and flammable hydrogen etc. The detection of these toxic gases should have sufficiently high sensitivity because trace amount of some gases will impose severe health threat or even cause death. Ponzoni et al. fabricated the WO₃ nanowire networks by thermal evaporation of tungsten powders at 1400–1450°C in the presence of oxygen (Ponzoni et al., 2006). The three-dimensional hierarchical WO₃ nanowire networks display ultrahigh sensitivity toward NO₂ detection with concentration as low as 50 ppb. The decent selectivity at different working temperatures is observed for various gases including ammonia, carbon monoxide and hydrogen sulfide. Chen et al. prepared porous ZnO nanoflakes by a microwave hydrothermal method and demonstrate the hierarchical structure exhibits better NO₂ sensing performance compared with ZnO nanoparticles (Chen et al., 2011). The nanoflakes show high selectivity toward NO₂ detection with concentration of 0.5 ppm in the atmosphere accompanied by 85 ppm other gases including ethanol, carbon monoxide, propylene, methane, diethyl ether, formaldehyde, acetone, cyclohexane, and dimethylbenzene, some of which are from automobile emission as air pollutants. For trace detection of toxic H₂S, Kaur et al. used a physical evaporation method to prepare the hierarchical In₂O₃ nanostructures with whisker and bipyramidal shapes (Kaur et al., 2008). The single crystal whiskers were found to detect very low concentrations (200 ppb) of hydrogen sulfide. Larger response of single crystal whiskers may result from the presence of defects formed during their growth.

A common strategy to enhance the sensitivity of the gas sensors is to incorporate noble metal nanoparticles such as Pt and Au to the hierarchical nanostructures. In carbon monoxide sensing, for example, Au nanoparticles are deposited onto the ZnO nanowire surface and the sensitivity with 50 ppm inlet CO gas can be enhanced from 4.2 to 46.5% at 350°C (Shoou-Jinn et al., 2008). The role of Au nanoparticles played during the gas sensing process is to improve the interaction of adsorbed oxygen with the reducing CO gas. Basically, the noble metal enables preferred adsorption and activation sites for the target gas; the activated atoms can thus, be spilled over onto the semiconductor and react with the adsorbed oxygen. More details on the function of noble metal in gas sensors can be found in a nice review article by Kohl (1990). However, there are other methods to improve the sensitivity without catalyst decoration on the surface. For instance, oxygen plasma treatment can be applied to lower the carrier concentration in the ZnO nanowires and result in a larger population of the O²⁻ ions on the nanowire surface (Law and Thong, 2008). With the reduced conductance, the oxygen plasma treated ZnO nanowires demonstrate four times higher sensitivity toward ammonia detection compared with the ZnO nanowires before treatment. The increased availability of surface O²⁻ ions will contribute to more effective NH₃ to O²⁻ ion interaction and enhance the sensitivity during gas sensing. Hierarchical composite nanostructure with multiple metal oxides configuration provides another possibility to improve sensitivity. Kim et al. prepared the CuO/ZnO by deposition of CuO nanostructures on the hydrothermally grown ZnO nanorods using a photochemical method (Kim et al., 2012). Compared with bare ZnO nanorods, the composite nanostructure display much improved gas sensitivity. The H₂S gas response of CuO/ZnO nanorod sensors are found to increase exponentially with temperature. The chemical conversion of CuO to Cu₂S during H₂S sensing is responsible for the enhanced gas sensing. Recently we prepared large scale Ag₂O/Zn₂SnO₄ hybrid periodic nanowires at 650°C by using a unique one-step silver oxide catalyzed vapor-solid-solid (VSS) growth process and the ethanol sensing property was investigated (Cai et al., 2010). Figure 7 presents the sensor configuration and ethanol sensing performance of the hybrid periodic nanowires. These Ag₂O/Zn₂SnO₄ hybrid nanowires were found to respond to ethanol at 150°C upon 150 ppm ethanol pulses by one order of magnitude reduced conductivity. This abnormal behavior is ascribed to ethanol pulses associated with Ag₂O nanoparticles on nanowire surfaces that leads to a catalytic activation of ambient oxygen detection. The 150 ppm ethanol pulse frees a large number of surface donors due to the bounded ambient oxygen molecules at 150°C, which therefore, activated the metal oxide surface sensing activity to the ambient oxygen. With the removal of Ag₂O nanoparticles by argon plasma treatment, the detection of ethanol molecules instead of ambient oxygen is enabled. This reversible catalytic ambient ethanol/oxygen detection mechanism can be potentially used for gas sensor design targeted at multiple transient gas detection.

Hierarchical composite nanostructures can also be utilized in humidity monitoring and detection. Similar to the ZnO/LSCO photocatalyst preparation, we have grown ZnO/LSCO core-shell nanorod arrays using facile hydrothermal growth followed by the sol-gel deposition (Gao et al., 2010). The uniform coating of perovskite is achieved by carefully controlled solution pH. A good rectifying characteristic have been observed in the ZnO/LSCO heterojunction arrays, which is similar to the characteristic of p-n junctions in semiconductors. As demonstrated in Figure 8, negative photoconductivity response is detected upon UV illumination on the diode arrays resulting from desorption of surface absorbed water moisture. Besides, the forward current of the LSCO/ZnO nanofilm–nanorod diode increases significantly with increasing relative humidity. With increasing humidity, more and more water molecules will adsorb onto the oxide surface through hydrogen bonding. The conductance thus, can be increased according to Grotthuss's chain reaction (Wenmin and Jörg-Uwe, 1997).

In this section, we reviewed the gas sensor development based on hierarchical nanostructures. The gas sensing mechanisms and relevant issues associated with high sensitivity and selectivity have been discussed. Either semiconducting metal oxide with various geometry or hierarchical composite structures have been introduced for various toxic gas detection. The gas sensitivity and selectivity can be improved by noble metal decoration as well as multiple components integration. However, there are still challenges in gas sensor development especially for gas sensing and monitoring in harsh environment. These conditions include high temperature sensing in power plant and combustion engine, gas regulation in automobile emissions as well as gas monitoring in radioactive environment such as nuclear reaction. Recently, we fabricated high temperature O₂ and CO sensors operating at 800 and 1000°C using CeO₂ nanofibers (Liu et al., 2012a). The real time in situ gas detection is realized with sensitive, reversible, and reproducible response. Similar high temperature oxygen sensor was also prepared using Sr-doped lanthanum manganese oxide (LSMO) nanofibers by electrospinning technique (Liu et al., 2012b).

The air pollution has become a worldwide crisis resulting from burning of fossil fuels. A major portion of the air pollutants comes from the automobile exhaust and power plant. The waste gas usually constitutes nitrogen oxide and sulfur dioxide, which either triggers photochemical pollution (Atkinson et al., 1982) or contributes to acidic rain (Mohnen, 1988), as well as hydrocarbon and carbon dioxide, which lead to green-house effect (Rodhe, 1990). Specific catalyst is needed to deal with each pollution gas component. In this section we will discuss the catalysts which deal with the major components contributing to air pollution such as carbon monoxide, hydrocarbons, and nitrogen oxide. The catalytic mechanisms, catalytic techniques as well as the latest progress in preparation of nanocatalyst toward different gas treatment will be reviewed in the first part. Within all the discussion we will introduce the materials selections and discuss how nanoscale geometry matters. Later we will concentrate on catalytic reactor design which rationally integrates the free standing nanocatalysts ranging from two-dimensional planar arrays to a much complex three-dimensional hierarchical fashion.

The best way to treat carbon monoxide and hydrocarbons is the catalytic oxidation. Catalytic CO oxidation has been extensively studied either from catalyst design or reaction mechanism interpretation (Haruta et al., 1993; Lopez and NøRskov, 2002; Kung et al., 2003; Lopez et al., 2004; Herzing et al., 2008; Xie et al., 2009). As discussed in the gas sensing section, the surface area and the porosity greatly influence the gas-solid interactions. Therefore, similar principles should be obeyed for the nanocatalysts design. The catalyst should have large surface to volume ratio so as to expose majority of active sites and facilitate the gas adsorption at the same time. Several nanocatalysts in the powder form have been discovered to have high efficiency toward CO oxidation. For example, Co₃O₄ nanocrystals have been demonstrated to achieve 100% catalytic CO conversion to CO₂ at temperature as low as −77°C and the activity can be maintained in moisture stream of feed gas (Xie et al., 2009). Before Co₃O₄ nanopowders are demonstrated, the catalytic reaction of CO oxidation usually involves gold nanoparticles as the catalyst. Haruta et al. systematically studied gold nanoparticles supported on various metal oxides and their catalytic CO oxidation performance (Haruta et al., 1993). The supported gold nanoparticles are found to become the sites for the reversible adsorption of CO. At the same time, they are able to increase the amount of oxygen adsorbed on the support oxides. A generally accepted model for catalytic CO oxidation mechanism is that the adsorbed CO molecules on Au nanoparticles are going to travel toward the support oxides and react with adsorbed oxygen to form carbonate species. The decomposition of the carbonate intermediate gives rise to carbon dioxide as the final product. With the development of solution chemistry, many Co₃O₄ nanostructures with various geometries have been prepared (Hu et al., 2008b, 2010; Teng et al., 2011) and different geometric anisotropy is believed to influence the catalytic activity by different crystal facets exposed. Other materials systems such as CeO₂(Ho et al., 2005; Pan et al., 2008; Wu et al., 2012), MnO₂ (Liang et al., 2008; Wang et al., 2009; Ching et al., 2011), CuO (Jernigan and Somorjai, 1994; Feng and Zheng, 2010) have also been investigated. A nice review article by Royer et al. can be referred to for more details on transition metal oxide used for catalytic CO oxidation (Royer and Duprez, 2011).

As the simplest hydrocarbon, methane is widely used in power generation but it is also a greenhouse gas 20 times effective than carbon dioxide (Farrauto, 2012). The low temperature oxidation of methane is thus, critical for emission regulation and control (Choudhary et al., 1999; Eguchi and Arai, 2001). Cargnello et al. reported a hierarchical design of Pd@CeO₂ core-shell catalyst which is deposited onto a modified hydrophobic Al₂O₃ (Cargnello et al., 2012). Metal-support interaction was discovered to cause complete conversion of methane below 400°C (Cargnello et al., 2012) and the Pd@CeO₂ core-shell structure has excellent thermal stability (Cargnello et al., 2009; Adijanto et al., 2013). For nitrogen oxide removal, selective catalytic reduction (SCR) and NOx storage and reduction (NSR) are widely used technologies. In SCR, reducing gases such as ammonia (Wu et al., 2008; Metkar et al., 2011) or hydrocarbon (Amiridis et al., 1996; Shibata et al., 2003; Brosius and Martens, 2004; Liu et al., 2012a,b,c) are utilized to react with nitrogen oxide to produce nitrogen and water. However, NSR technique is slightly more complex (Takahashi et al., 1996; Liu and Gao, 2011). The NSR catalyst usually consists of storage materials such as barium oxide or carbonate, noble metal (Pt, Pd) and porous support (Al₂O₃). The nitric oxide (NO) will be first converted into NO₂ through catalytic oxidation by Pt during lean condition where oxygen is rich. The barium component absorbs the nitrogen oxide and transforms into barium nitrate. When switched to rich cycle where fuel is sufficient, the decomposition of barium nitrate and reduction take place simultaneously to convert nitrogen oxide into nitrogen and water.

In practical application for catalytic gas treatment, the nanocatalysts must be deposited onto substrate to make a catalytic reactor or converter. Compared with powder form catalyst, catalytic reactors are generally more efficient and cost-effective due to several advantages enabled by the monolithic configuration such as low pressure drop, high surface area, efficient mass transfer and relatively low catalytic materials usage (Williams, 2001). However, the traditional procedures for constructing a catalytic reactor usually involve the wash coating of nanocatalysts (Irandoust and Andersson, 1988; Choi et al., 1996). The wash-coated nanopowders lack well-defined structural features and geometrical size, which leads to compromised materials usage and lower catalytic activity. Another issue associated with the catalytic reactor is the inevitable use of noble metals. Although some other nanomaterials have been discovered to exhibit excellent catalytic performance, their thermal and mechanical stability under practical working condition are usually far from satisfactory, which prevents them from large scale application in industry. A new nanomaterial manufacturing process is therefore, necessary to realize rational configuration of nanostructure without compromised catalytic activity.

Although many types of nanostructures have been proved to have great catalytic activity in their powder-form, devices based on hierarchical nanowire arrays have rarely been reported in the catalytic gas treatment. This is because of the lack of synthetic strategy to rationally integrate nanocatalysts on a large scale. We recently demonstrated the possibility to prepare a catalytic converter by hierarchical assembly of one-dimensional nanowires as building blocks. Two-dimensional planar assembly was first investigated and ordered TiO₂/(La,Sr)MnO₃ composite nanorod array catalysts have been successfully fabricated as shown in Figure 9 (Guo et al., 2012). Instead of using template, the TiO₂ nanorod arrays on silicon or glass substrate were prepared by a one-pot facile hydrothermal synthesis. The radio-frequency (RF) magnetron sputtering method was applied to deposit a uniform coating of (La,Sr)MnO₃ on the TiO₂ nanorod surface. TiO₂ is a traditional oxide support for immobilizing noble metal catalyst, while it can be served as a catalyst itself in photo-exited reactions due to the semiconducting nature. In our experiments the hydrothermally prepared TiO₂ nanorod arrays were found to be rutile TiO₂ in crystal phase. But the as-prepared TiO₂/LSMO composite nanorod array does not exhibit perovskite LSMO peak due to the poorly crystallized as-deposited LSMO shell. The thermal annealing at 800°C improved the crystallinity and X-ray diffraction results revealed that {110} planes are the dominant crystal facet in the LSMO shell. The ratio between La and Sr in LSMO is close to 1:1. An important rationale behind such materials combination is that perovskite have the oxidation activity comparable to that of the platinum group metal catalysts. It has been demonstrated to have an improved performance for automobile exhaust emission control such as lean NO_X trapping (Kim et al., 2010) CO oxidation was used as a probe reaction to characterize the catalytic performance of the hierarchical TiO₂/LSMO composite nanorod arrays. The maximum CO conversion efficiency of pure rutile TiO₂ nanorod array was very low and was only around 20%, while the TiO₂/LSMO composite nanorod array achieved 100% CO conversion efficiency at 400°C. The catalytic performance was also better than that of perovskite LSMO thin film catalyst from literature (Karita et al., 2007).

With successful demonstration of two-dimensional assembly, we further explored the possibility of three-dimensional integration of nanowires onto the ceramic cordierite honeycomb substrate to develop a prototype catalytic converter (Guo et al., 2013). Figure 10 illustrates the materials synthesis process and the electron microscopy characterization of the nanostructures. The three-dimensional ordered nanoarchitectures provide well-defined geometry with controlled porosity and surface area which allows for efficient gas diffusion and facilitates the gas-solid interaction. Different metal oxide nanowire arrays with various morphologies including ZnO nanowires, TiO₂ nanorods, Co₃O₄ porous nanowires and CeO₂ nanotubes have been in situ grown on the honeycomb substrate by a facile hydrothermal synthesis. The nanowires array based catalytic honeycombs were discovered to have good thermal stability under 800°C for 100 h and decent mechanical robustness exposed to constant air flow with rate of 50 L/min for 10 days. Pt nanoparticles were then wash coated onto the nanostructures with controlled weight ratio (Pt/nanowires = 1 wt %). The catalytic performance of these nanowire arrays based honeycombs are studied by using CO oxidation as the probe reaction. Comparable and even better catalytic performances were achieved compared to traditional powder catalyst prepared by washcoating. However, the materials usages of both support metal oxide and the noble metal were reduced by 10–40 folds with sacrificing the catalytic performance. It is therefore, plausible to conclude that the ordered nanoscale geometry with well-defined shape and arrangement will reduce the manufacturing cost of catalyst preparation by improving the materials utilization efficiency. The low materials consumption as well as the maintained high catalytic activity makes the hierarchical nanostructure based honeycombs a potentially promising design for better catalytic converter. Another advantageous feature of the nanostructured catalytic honeycombs is that it allows for more control over the catalytic performance by tailoring nanoscale structures. The well-defined geometry and special arrangement help elucidate the structure-property relation, which in this case is the correlation between nanostructure and the catalytic activity, and thus, provide rationales for catalytic materials design. For instance, in our nanostructured honeycombs, we are able to tune the catalytic CO oxidation performance by adjusting different shape and structure of the ZnO support. Specifically, longer ZnO nanowires allow for more uniform dispersion of Pt nanoparticles that favors the reaction kinetics. Besides, when the aspect ratio of single crystalline ZnO nanowire decreases and the shape transforms from one-dimensional nanowires to two-dimensional nanoplates, the catalytic activity toward CO oxidation is much improved. This can be ascribed to the different interaction of Pt nanoparticles with specific crystal facets of ZnO nanostructures. The interaction of Pt with ZnO (0001) is found to favor the catalytic CO oxidation (Roberts and Gorte, 1990; Petrie and Vohs, 1994). When the ZnO (0001) planes are highly involved in the nanostructure, the catalytic performance will be improved.

In this section, we start with typical catalyst materials toward different polluting gas and introduced the progress in nanocatalysts preparation in different gas treatments. Later we summarized the recent research effort on integration of nanoscale building blocks to achieve hierarchical nanostructures for catalytic device fabrication and illustrate the advantages of ordered nanostructure arrangement over the traditional wash-coated powder catalyst. For the future development of hierarchical nanostructured catalytic converter, however, the design of hierarchical nanostructures with rational combination of multiple components targeted at complex chemical reactions is still challenging. Much research effort should be devoted regarding better materials selection, further reduction of noble metal usage and industrial scale fabrication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.