Nanoparticles as the Active Element of High-Temperature Metal±Insulator±Silicon Carbide Gas Sensors** By Oliver Böhme,* Anita Lloyd Spetz, Ingemar Lundström, and Dieter Schmeiûer The sensor performance of metal±insulator±silicon carbide (MISiC) diode devices depends on temperature pretreatment and can be divided into three different types. As-fabricated structures (Pt±TaSi x ±SiC) have very low sensitivity. An activa- tion process at 600 C results in fast-responding sensors with extraordinarily high signals of several volts. The devices fail after prolonged operation at elevated temperatures (above 700 C). The electrical consequences of the three different states of MISiC diode sensors can be related to a model that involves the size-dependent interface dipole moment of oxide-covered metallic nanoparticles. X-ray photoemission spectroscopy (XPS) data of MISiC diode sensors based on nanoparticles as the driving force were compared to MISiC sensors in capacitor configuration with no nanoparticles pres- ent. The physical effects observed were applied to describe the loss of rectifying properties of many metal±semiconductor interfaces reported in the literature. Nanoparticles have evoked great interest because of their unique properties, which differ from the corresponding bulk materials' properties. [1±6] A variety of systems have been reported in which the nanoparticles' properties are superior to those of their bulk counterparts. [7±13] The potential use of nanoparticles in many technologically important applications has been demonstrated. [14±17] Recently, we proposed a model that involves the size-dependent interface dipole moment of oxide-covered nanoscale metal particles. [18] We gave spectro- scopic evidence of such nanoparticles at the interfaces of sev- eral semiconducting oxide materials. In the present study, we focus on the key role of nanoscale particles in high-tempera- ture MISiC gas sensor applications. Such fast-responding devices are of considerable interest for cylinder-specific moni- toring and control of exhaust gases of combustion engines. [19] MISiC sensors in capacitor configuration [20] will be compared to those in Schottky diode configuration. [21] The latter devices show extraordinarily high response signals of several volts and are preferable because of their simpler fabrication and less expensive electronic instrumentation. However, they break more easily and cannot be operated above 700C. We will explain the observed differences by the presence of nanopar- ticles, which form the active element of MISiC devices in Schottky diode configuration. The MISiC capacitor sensors consist of n-type SiC wafers with a 100 nm SiO 2 surface layer, a 10 nm TaSi x buffer layer, and a 100 nm Pt gate electrode (see inset to Fig. 1). In Schott- ky diode configuration, the MISiC sensors consist of n-type SiC wafers (without a SiO 2 layer), a 10 nm TaSi x buffer layer, and a 100 nm Pt gate electrode (see inset to Fig. 2). The gas sensitivity of the MISiC structures is based on the dissociation of hydrogen and hydrogen-containing gases on the catalytic metal surface. Hydrogen atoms rapidly diffuse through the metal and form a dipole layer at the active inner interface. [20] The dipole layer causes a voltage shift of the capacitance± voltage curve of the capacitor sensors and the current±voltage curve of the sensors in Schottky diode configuration. Details of the gas sensing principle of the sensors in diode configura- tion will be given in the present paper. In particular, we dem- onstrate that the active inner interface is not the Schottky contact but the nanoparticle interface. The MISiC capacitors were investigated earlier. [20] Here, we only review the main findings. The TaSi x of the as-received (fresh) samples is converted by the activation process to a mixed phase, predominantly containing Ta 2 O 5 . Due to the for- mation of Ta 2 O 5 , the activation of MISiC sensors results in non-drifting faster signals. It forms, together with the TaSi x / SiO 2 interface, the active part of the capacitor sensors. In the present study, we used the capacitors as reference samples. The X-ray photoemission spectrum of an activated capacitor, with about 60 % of the covering Pt gate removed, is shown in Figure 1. The spectrum of Figure 1 is composed of two main components, with Ta 4f 7/2 positions at 23.3 eV and 26.9 eV. They are assigned to TaSi x [22] and Ta 2 O 5 , [22,23] respectively. According to earlier findings, [20] the Ta 2 O 5 component domi- nates the spectrum of the activated sample. However, there is also a small contribution of lower oxides between the Ta 2 O 5 and the TaSi x components. For reasons of clarity, the lower Adv. Mater. 2001, 13, No. 8, April 18 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim,2001 0935-9648/01/0804-0597 $ 17.50+.50/0 597 COMMUNICATIONS ± [*] Dr. O. Böhme, Prof. D. Schmeiûer Applied Physics IIÐSensor Technology Technical University of Cottbus D-03044 Cottbus (Germany) E-mail: boehme@icmm.csic.es Dr. A. Lloyd Spetz, Prof. I. Lundström Swedish Sensor Centre, S-SENCE and Applied Physics Linköping University S-58183 Linköping (Sweden) [**] We acknowledge the skillful experimental assistance of G. Beukert. Fig. 1. XPS (Al Ka 1,2 ) Ta 4f spectrum of an activated capacitor MISiC sensor after removal of about 60 % of the Pt gate upon depth profiling. The inset shows a schematic cross section of the MISiC sensors in capacitor configuration.