Highly sensitive and selective trimethylamine sensor using one-dimensional ZnO–Cr₂O₃ hetero-nanostructures

The ZnO NWs were grown directly on an alumina substrate (area, 1.5 × 1.5 mm²; thickness, 0.25 mm) with two gold electrodes (electrode length, 1 mm; separation, 0.2 mm) by thermal evaporation using mixed powders of ZnO (99.9%, Aldrich), graphite ( < 20 μm, Aldrich) and Sn (99.8%, Acros). The source (ZnO:graphite:Sn  =  1:1:0.01 wt%) was loaded in an Al₂O₃ boat which was located in the centre of a quartz tube (diameter: 2.5 cm) and the alumina substrates were placed 5 cm downstream from the source. After evacuating the quartz tube to ∼9 × 10⁻² Torr using a rotary pump, the furnace temperature was increased to 900 °C. The ZnO NWs were grown for 20 min by a reaction between the source and an Ar–O₂ mixture gas (Ar, 100 sccm; O₂, 1 sccm). The gold electrodes acted as catalysts for the growth of ZnO NWs by the vapour–liquid–solid mechanism, which facilitated networking between NWs with highly porous structures and increased the connectivity between the electrodes and NWs.

The semielliptical Cr₂O₃ nanoparticles (lateral width, 3–8 nm; height, 1–5 nm) were deposited on the ZnO NW networks by the following procedures. The as-grown ZnO NWs on the alumina substrate and the CrCl₂ (99.99%, Aldrich) were placed in the right and left parts of the Al₂O₃ boat (length, 4 cm), respectively. After evacuating the quartz tube to ∼9 × 10⁻² Torr using a rotary pump, the furnace temperature was increased to 630 or 670 °C and then was kept at the reaction temperature for 20 min. Ar gas (100 sccm) was flowed during the reaction. The discrete Cr₂O₃ nanoparticles were coated on ZnO NWs at 630 °C, while ZnO–Cr₂O₃ core–shell NCs were prepared at 670 °C.

The phase and crystallinity of the pristine ZnO NWs, Cr₂O₃-decorated ZnO NWs, and ZnO–Cr₂O₃ core–shell NCs were analysed by x-ray diffraction (XRD, Rigaku D/MAX-2500V/PC), the morphologies of the NWs by field-emission scanning electron microscopy (FE-SEM, S-4800, Hitachi, Japan) and transmission electron microscopy (TEM, FEI Titan), and the chemical state of the NWs by x-ray photoelectron spectroscopy (XPS, Thermo, MultiLab 2000).

Sensor elements with NW networks were heat treated at 550 °C for 2 h prior to the measurement of their gas sensing characteristics. The sensor was placed in a quartz tube and the temperature of the furnace was stabilized at 400 °C. A flow through technique with a constant flow rate of 500 cm³ min⁻¹ was used. The gas concentration was controlled by changing the mixing ratio of the parent gases (5 ppm TMA, 100 ppm C₂H₅OH, 5 ppm p-xylene, 10 ppm NH₃, 5 ppm benzene, 100 ppm C₃H₈, 5 ppm toluene, 100 ppm CO, and 100 ppm H₂, all in air balance) and dry synthetic air. The dc-probe resistance of the sensor was measured using an electrometer interfaced with a computer.

Single crystalline ZnO NWs were directly grown on an alumina substrate with two Au electrodes by thermal evaporation (figures 1(a) and (b)). The diameters of the NWs ranged from 30 to 80 nm and the NWs were several tens of micrometres long (figures 1(c) and (d)). The selected area electron diffraction (SAED) pattern confirmed that the single crystalline ZnO NWs were grown in the [$0 1\bar {1}0$] direction (figure 1(e)). The lattice-resolved image of a single NW showed highly crystalline ($0 1\bar {1}0$) fringes separated by 2.81 Å (figure (f)).

Zoom In Zoom Out Reset image size

The catalytic Cr₂O₃ layers were deposited on the ZnO NW networks by the thermal evaporation of CrCl₂ powders at 630 °C. The overall network configuration of ZnO NWs in the sensor remained similar (figure 2(a)). The surface of ZnO NWs was made rougher by the coating of Cr₂O₃ via thermal evaporation of CrCl₂ (figure 2(b) and inset in figure 2(b)). The small semielliptical particles on the surface of ZnO NWs (figure 2(c)) were identified as Cr₂O₃ in the high magnification TEM image (figure 2(d)). The highly crystalline Cr₂O₃ nanoparticles with semielliptical shapes were uniformly coated on the ZnO NWs, which can be confirmed by the (110) fringes of Cr₂O₃ separated by 2.48 Å. The lateral sizes of semielliptical Cr₂O₃ nanoparticles ranged from 3 to 8 nm and their heights ranged from 1 to 5 nm. The high-angle annular dark field (HAADF) scanning TEM (STEM) image and the elemental line scanning of the Cr₂O₃-decorated ZnO NWs by energy dispersive x-ray spectroscopy (EDS) indicated again that the semielliptical nanoparticles on the surfaces of the ZnO NWs were indeed Cr₂O₃ (figure 2(e)).

Zoom In Zoom Out Reset image size

As the thermal evaporation temperature was increased to 670 °C, the ZnO NWs were completely coated with a continuous Cr₂O₃ shell layer, which leads to the formation of ZnO–Cr₂O₃ core–shell NCs (figure 3(a)). The lattice fringes of the ZnO core are hardly visible in the high magnification TEM image due to the presence of the Cr₂O₃ shell layer. The formation of a crystalline Cr₂O₃ shell layer was confirmed by the presence of (110) fringes of Cr₂O₃ separated by 2.48 Å and the fast Fourier transform pattern (figure 3(b)). The shell layer was 30–40 nm thick. The formation of ZnO–Cr₂O₃ core–shell NCs was verified again by the high-angle annular dark field (HAADF) scanning TEM (STEM) image and the elemental line scanning profile (figure 3(c)). The presence of Cr in the central region can be attributed to the Cr₂O₃ shell layer located at the front and back of the ZnO core NWs. However, the significant amount of Zn in the shell region indicates that the Cr₂O₃ in the shell region contains a small amount of ZnO. The ionic radius of Cr³⁺ with the coordination number of 6 in Cr₂O₃ is 0.755 Å, while that of Zn²⁺ with the coordination number of 6 is 0.88 Å. Thus, the substitution of Zn²⁺ in the sites of Cr³⁺ can be given as a plausible reason for the presence of a Zn component in the Cr₂O₃ shell layer.

Zoom In Zoom Out Reset image size

The pristine ZnO NWs were identified as wurtzite phase (JCPDS 36-1451), characterized by primary (101), (100), and (002) peaks (figure 4(a)). Cr₂O₃ with a corundum structure (JCPDS 38-1479) and wurtzite ZnO phases were found to coexist in Cr₂O₃-decorated ZnO NWs (figure 4(b)) and ZnO–Cr₂O₃ core–shell NCs (figure 4(c)). Compared to Cr₂O₃-decorated ZnO NWs, the XRD pattern of ZnO–Cr₂O₃ core–shell NWs showed a slightly stronger intensity of Cr₂O₃ peaks (figures 4(b) and (c)). Further characterization in order to confirm the oxidation states of the Cr in both Cr₂O₃-decorated ZnO NWs and ZnO–Cr₂O₃ core–shell NCs has been carried out using XPS (figure 5). The Cr (2p_2/3) and Cr (2p_1/2) peaks were centred near 577 and 586 eV, respectively (figure 5), indicating that the bonding states of chromium in both Cr₂O₃-decorated ZnO NWs and ZnO–Cr₂O₃ core–shell NCs are in the form of Cr₂O₃. The Cr concentrations of Cr₂O₃-decorated ZnO NWs and ZnO–Cr₂O₃ core–shell NCs were determined by XPS to be 11.8 and 18.8 at.%.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The temperature dependences of the TMA responses in ZnO NWs, Cr₂O₃-decorated ZnO NWs, and ZnO–Cr₂O₃ core–shell NCs were measured to investigate the morphological and catalytic effects of Cr₂O₃ on the surfaces of ZnO NWs (figure 6(a)) The pristine ZnO and Cr₂O₃-decorated ZnO NWs showed n-type gas sensing behaviours, that is, a resistance decrease in response to reducing gases (figure 6(b)). This indicates that the conduction in the Cr₂O₃-decorated ZnO NWs is dominated by the n-type ZnO NW core rather than the p-type Cr₂O₃ surface modification. The R_a/R_g values (S = R_a/R_g: R_a, resistance in air; R_g, resistance in gas) were used as the gas responses. In contrast, the ZnO–Cr₂O₃ NCs showed p-type gas sensing characteristics (figure 6(b)); that is, the resistance is increased by reducing gases. This indicates that the conduction and chemoresistive variation in ZnO–Cr₂O₃ core–shell NCs are dominated by the p-type Cr₂O₃ shell layers with thicknesses of 30–40 nm on the surfaces of n-type ZnO NWs. Thus, in ZnO–Cr₂O₃ core–shell NCs, the R_g/R_a values were used as the gas responses.

Zoom In Zoom Out Reset image size

The response of pure ZnO NWs to 5 ppm TMA was increased from 2.25 to 6.88 as the sensor temperature increased from 300 to 450 °C. The TMA responses in Cr₂O₃-decorated ZnO NWs showed a maximum value (17.79) at 400 °C. The responses to TMA were very small in ZnO–Cr₂O₃ core–shell NCs and tended to decrease with increasing sensor temperature.

The gas responses to TMA, C₂H₅OH, p-xylene, NH₃, benzene, C₃H₈, toluene, CO, and H₂ of three different NWs were measured (figure 7) at 400 °C. In pristine ZnO NWs, although the response to 5 ppm TMA (R_a/R_g = 5.41) was the highest among the responses to all the gases (figure 7(a)), it was comparable to those to C₃H₈ (R_a/R_g = 3.80) and C₂H₅OH (R_a/R_g = 3.52). The decoration of Cr₂O₃ nanoparticles on ZnO NWs significantly enhanced the response to 5 ppm TMA up to 17.79 (figure 7(b)). This value is 2.4 times higher than the response to C₂H₅OH (R_a/R_g = 7.35) and significantly higher than the cross-references to other gases such as p-xylene, NH₃, benzene, C₃H₈, toluene, CO, and H₂. ZnO–Cr₂O₃ NCs showed similar responses to all the gases (R_g/R_a = 1.06–1.68) (figure 7(c)). In order to quantify the selective detection of TMA, the ratios of the gas responses to 5 ppm TMA and 5 ppm of another gas (S_TMA/S_gas: S_TMA, response to TMA; S_gas, response to other gas) were calculated (figures 7(d)–(f)). In pure ZnO NWs, the S_TMA/S_gas values ranged from 1.54 to 2.81 (figure 5(d)). In Cr₂O₃-decorated ZnO NWs, most of the S_TMA/S_gas values, with the exception of S_TMA/S_ethanol, were in the range from 4.24 to 8.36 (figure 7(e)). Note that the S_TMA/S_ethanol value was also enhanced significantly from 1.54 to 2.42 by the decoration of Cr₂O₃ on the ZnO NWs. Finally, the S_TMA/S_gas values of ZnO–Cr₂O₃ NCs ranged from 1.05 to 1.58 (figure 7(f)). These results clearly show that the coating of discrete Cr₂O₃ nanoparticles on ZnO NWs is a very effective way to achieve selective and sensitive detection of TMA.

Zoom In Zoom Out Reset image size

The sensing transients of Cr₂O₃-decorated ZnO NWs to 0.05–5 ppm TMA at 400 °C exhibited stable and reproducible sensing behaviours (figure 8(a)). The 90% response times, defined as the time to reach a 90% variation of resistance upon exposure to gas, ranged from ∼1 to 10 s, while the 90% recovery times were longer (358–398 s). The TMA responses were plotted as a function of concentration (figure 8(b)). To date, the responses to high concentrations of TMA have been investigated [20, 30, 32, 38–41]. Most of these works have measured the responses to TMA gas for concentrations from 50 ppm up to as high as 1000 ppm, which is significantly beyond the maximum TMA level for fish freshness. Zhu et al [35] have shown responses of 2.0, 3.3, and 49.6 to 0.5, 1.0 and 10 ppm TMA, respectively, using Cu-doped WO₃. By contrast, the present Cr₂O₃-decorated ZnO NWs showed a remarkable response (R_a/R_g = 5.14) to 0.05 ppm TMA. From the fitting of the gas response as a function of TMA concentration, R_a/R_g values as high as 2 could be expected upon exposure to 1 ppb. Although the emittance of [TMA] > 10 ppm is often used as the criterion to evaluate the fish decay [23], the more precise evaluation of fish freshness requires the detection of [TMA] < 10 ppm. Accordingly, the present TMA sensor with ppb-level-precision can be used to evaluate extremely fresh fishes.

Zoom In Zoom Out Reset image size

The resistance in air (R_a) of the pure ZnO NW sensor was gradually increased from 0.10 to 0.18 MΩ as the sensor temperature increased (figure 8(c)). It should be noted that the R_a values of Cr₂O₃-decorated ZnO NW and ZnO–Cr₂O₃ core–shell NC sensors are 60–300 times higher than that of the pure ZnO NW sensor. The variation of the Schottky junction between the Au contact and ZnO NWs is less probable as the reason for the change of R_a because both ZnO NWs and Cr₂O₃-decorated ZnO NWs make similar contacts between ZnO and Au electrodes. The R_a values of several sensors prepared under identical growth conditions were similar. Precise estimation of p–n heterojunction structures between n-type ZnO and p-type Cr₂O₃ is difficult because many different band gap energy (E_g) values of ZnO [42, 43] Cr₂O₃ [44, 45] have been reported and, in particular, little is known about the electron affinity and work function of Cr₂O₃. However, the electron depletion layer will be formed near the junction. This shows that not only the decoration of Cr₂O₃ on ZnO NWs but also the formation of ZnO–Cr₂O₃ core–shell NCs increases the R_a value due to the extension of the depletion layer by the formation of p–n junctions.

In an n-type gas sensor such as the Cr₂O₃-decorated ZnO NW sensor, the increase of R_a will lead to enhanced responses to reducing gases because the gas response is defined as R_a/R_g. However, in ZnO–Cr₂O₃ core–shell NCs, the gas sensing reaction is not simple and depends on the contact configuration between NCs. When p-type Cr₂O₃ layers completely cover the ZnO NWs, the chemoresistive change will occur along the p-type shell layers. Most p-type oxides are known to form a charge accumulation layer near the surface by the adsorption of negatively charged surface oxygen [46]. The oxidation of reducing gases by negatively charged surface oxygen releases electrons, which thins the near-surface hole accumulation layer and increases the gas resistance. This can explain the p-type gas sensing behaviours observed in the present study. However, the formation of longitudinal p–n junctions between the end parts of the ZnO core and the Cr₂O₃ shell cannot be excluded completely. In this case, conduction through the p-type Cr₂O₃ shell layer seems to dominate over conduction via the n-type ZnO core. The incorporation of the ZnO component into the Cr₂O₃ shell layer, as shown in figure 3c, can be also taken into account. In order to investigate the effect of ZnO incorporation on the gas sensing characteristics of crystalline Cr₂O₃ NWs, single crystalline or pseudo-single-crystalline Cr₂O₃ NWs are needed as counterparts for comparison. However, no report on the preparation of single crystalline NWs via vapour phase synthesis is available in the literature, probably due to the high melting temperature (2435 °C) and low vapour pressure of the Cr source. This should be studied further after the preparation of crystalline Cr₂O₃ NWs.

The enhancement of TMA selectivity by decorating Cr₂O₃ onto the ZnO NW sensor should be understood in relation to the catalytic effect of Cr₂O₃ and the extension of the p–n junction between Cr₂O₃ and the ZnO NW. The former is related to the chemical affinity between Cr₂O₃ and amine-containing gases. TMA has a very similar molecular structure to methylamine (MA), the simplest primary amine. TMA has two additional methyl groups replacing the hydrogens in MA. The nitrogen in both MA and TMA has a lone pair of electrons that can be donated and bonded. The adsorption and dissociation of MA on the surface of several transition metals including Cr has been experimentally investigated [47]. Borck et al [35] investigated the adsorption and dissociation of methylamine on the surface of Cr₂O₃ using density functional theory. It has been confirmed that MA adsorption on Cr₂O₃ involves the bonding of the lone-pair electrons on the nitrogen atom to an exposed cation (Cr in this case) atom on the oxide surface. In detail, the bonding at the surface of Cr₂O₃ results from the donation of the methylamine lone-pair electron density to a partially empty cation orbital. This explains the promotion of TMA adsorption by the presence of the Cr₂O₃ layer.

The p–n junction is another important factor in the determination of the gas response [48, 49]. In order to explain this, the schematic gas sensing mechanisms of ZnO NW sensors in relation to the presence and/or configuration of p-type Cr₂O₃ overlayers are illustrated in figure 9. In pristine ZnO NWs, the adsorption of negatively charged oxygen establishes a carrier (electron) depletion layer (CDL(n)) near the surface (figure 9(a), left). The oxidation of reducing gases by the reaction with negatively charged oxygen (O⁻) injects electrons into the semiconducting core (n-type semiconducting layer, SL(n)), which decreases the sensor resistance in proportion to the concentration of reducing gas (figure 9(a), right). When a discrete configuration of Cr₂O₃ nanoparticles is uniformly coated on the surface of the ZnO NWs, the CDL(n) within the ZnO NWs will extend along the radial direction from beneath the semielliptical Cr₂O₃ nanoparticles (figure 9(b), left), which significantly narrows the conducting core part. The electron injection into ZnO NWs upon exposure to reducing gases would enlarge the conduction path to a great extent by the shrinkage of the electron depletion layer (figure 9(b), right). Negatively charged oxygen is also known to be formed at the surface of p-type Cr₂O₃, which induces the formation of a carrier (hole) accumulation layer (CAL(p) in figure 9(b), left) near the surface of the p-type Cr₂O₃. The CAL(p) can be thinned by the reaction between the reducing gas and the negatively charged surface oxygen (figure 9(b), right), which can provide an additional reason for chemoresistive variation. However, considering the discrete configuration of p-type Cr₂O₃ nanoparticles, conduction along the continuous n-ZnO NWs is more plausible than that along the discrete Cr₂O₃ nanoparticles, which is supported and verified by the n-type gas sensing characteristics shown in figures 7(a) and (b). This explains the overall enhancement of gas response by the decoration of Cr₂O₃ nanoparticles. The highly sensitive and selective detection of TMA in Cr₂O₃-decorated ZnO NW sensors, accordingly, can be understood as the combined effect of the catalytic function of Cr₂O₃ and the radial extension of the electron depletion layer in the ZnO NWs.

Zoom In Zoom Out Reset image size

The ZnO–Cr₂O₃ core–shell NCs do not show high responses to TMA or other gases. Although further investigation is necessary to clarify the reason for this, the following explanation can be considered plausible if one assumes a complete covering of the Cr₂O₃ layer on the ZnO NWs. In ZnO–Cr₂O₃ core–shell NCs, the adsorption of negatively charged oxygen establishes a charge (hole) accumulation layer (CAL(p) in figure 9(c), left) near the surface of the p-type Cr₂O₃. In addition, carrier depletion layers will be formed near the interfaces between the p-type Cr₂O₃ shell layer and the n-type ZnO NWs (CDL(p) and CDL(n), figure 9(c), left). The p-type gas sensing characteristics of ZnO–Cr₂O₃ core–shell NCs in figure 6(b) strongly indicate that conduction occurs not along the SL(n) channel in the ZnO NWs but along the thin CAL(p) channel in the Cr₂O₃ shell layers. The high R_a values of ZnO–Cr₂O₃ core–shell NCs (figure 8(c)) can be attributed to the narrow cross-sectional area for conduction along the very thin CAL(p) channel and the formation of a carrier depletion layer near the p–n junction. The high R_a value leads to low gas response (R_g/R_a) characteristics. In this configuration, the chemoresistive change, which originates from further thinning of the accumulation layer by the reducing gas (figure 9(c)), will be relatively small compared to that caused by the significant shrinkage of the electron depletion layer in the Cr₂O₃-decorated ZnO NW sensor (figure 9(b)).