Introduction Nowadays, environment, health and safety problems caused by NO2 are attracting more and more attention. Extensive efforts have been made to investigate and develop various novel detectors including resistance-type sensors, mixed potential sensors and optical sensors to satisfy the increasing demand of NO2 detection. Among these different kinds of detectors, resistance-type gas sensors, primarily based on metal oxide semiconductors, have received wide attention due to their high sensitivity, fast response, and low cost. As one of typical metal oxide semiconductors, ZnO has been widely investigated for detecting NO2 gas [1]. However, NO2 detection at the ppb level is rarely achieved for ZnO-based sensors. Recent studies have shown that gas-sensing properties of ZnO are highly related with the microstructure and can be further enhanced via selective doping [2]. Among different doping elements, indium (In) has attracted great attention. The advantages of In-doping are the followings: In2O3 is an important material for NO2sensing; In3+ (80 pm) can partly substitute Zn2+ (74 pm) due to their similar ionic radii, generating more free electrons. In this work, pure and In-doped ZnO porous cages were synthesized by a ZIF-8-based template method and their gas-sensing performance was systematically studied. Method In-doped ZnO porous cages were prepared via a one-pot synthesis route. Briefly, 4.05 g of Zinc nitrate hexahydrate was dissolved into 200 mL methanol. Appropriate amounts of Indium(III) acetylacetonate and 200 mL of a methanol solution containing 3.34 g 2-methylimidazole were added under stirring. The resulting solutions were stirred for 20 min and aged for 18 h at room temperature. After precipitation, obtained ZIF-8 derivativeswere washed with methanol via centrifugation. The precipitates were further dried at 60 °C for 24 h and calcined at 400 °C for 3 h to obtain pristine or In-doped ZnO powders. The obtained samples are hereafter denoted as In/ZnO-X, where X represents the nominal In/Zn atomic ratio (multiplied by 100) used during the synthesis, e.g., In/ZnO-0 for un-doped ZnO, In/ZnO-2 for In/Zn of 2.0 at.%, In/ZnO-10 for In/Zn of 10.0 at.%. Results and Conclusions Typical SEM images of the samples are presented in Fig. 1. For pristine ZnO, well-separated nanocages were observed (Fig. 1a). As can be seen from Fig1b-c, the size of the nanocages appeared larger with the increase of In doping content. Fig. 2a shows the electrical resistance for the fabricated gas sensors measured in air. Typical semiconductor behavior was observed, i.e., resistance decreased as temperature rose. Furthermore, the resistance remarkably decreased with increase of Indium content. Fig. 2b presents the temperature dependence of the gas response to 10 ppm NO2 for all the obtained sensors. For the un-doped sample (In/Zn-0), the response first increased with temperature, reached a maximum at 300 °C, and then decreased. The increasing addition content of In to ZnO resulted in significant increase of response to NO2. For In/ZnO-10, the response was 339 at 300 °C, 22 times higher than that of In/ZnO-0. Fig. 2c depicts sensor response as a function of NO2 concentration for In-doped ZnO sensors at 300 °C. It can be clearly seen that the gas response increased with increase of NO2 concentration for all the sensors, and at each concentration the response for In/ZnO-10 was much higher than those of the other sensors. Furthermore, the gas response varied almost linearly with the NO2 concentration until the measured concentration exceeded 1 ppm. This linear dependence is favorable for calibration and determination of the gas concentration in practical applications. Fig. 2d depicts the dynamic response and recovery curves for the In/ZnO-10 sensor in the concentration range of 10-100 ppb, which were measured by alternately exposing the sensor to ambient air and air/NO2 mixture at the optimal operating temperature of 300 °C. The response to 10 ppb NO2 was 3.7, which identified the excellent performance of In/ZnO for ppb-level NO2 detection. Unlike the greatly increased NO2response, the response to the main interfering gas, H2S, only slightly increased with In doping, indicating marked increase of the NO2 selectivity.