Errors of Integrated Hydrogen Sensors Based on FETs with Structure Pd (Ag)–Ta2O5 –SiO2 –Si

Abstract

Introduction There are a lot of hydrogen sensors commercially available or in development, which can be used in the environmental and fire-explosion safety monitoring devices [1]. Among solid-state gas sensitive elements, the capacitor and transistor elements based on metal-insulator-semiconductor (MIS) structures possess the best compatibility with the integrated circuits’ elements. Therefore, such sensitive elements seem promising to develop the integrated hydrogen sensors and gas analysis microsystems-on-chip. The sensors based on MIS-devices (MIS-capacitors and field-effect transistors called as FETs) have been studied by many investigators. A great contribution to the developments of gas-sensitive MIS-devices has been made by the researchers at Linköping University [2] since their pioneering work [3]. The researchers at NRNU MEPhI developed the MIS sensors based on structures Pd (or Pt)-SiO2-Si, Pd/Ti-SiO2-Si, Pd (or Pt)-Ta2O5-SiO2-Si. The best characteristics belong to the integrated cells containing on silicon chip two gas-sensing elements (Pd-resistor and FET with structure Pd(Ag)-Ta2O5-SiO2-Si), temperature sensor and heater. Effects of electrical modes, chip temperature and radiation on the performance of these sensors were studied in previous works (e.g., [4] – [6]). However, sensors’ errors of were not investigated. The motivation of this work is to estimate errors and its components of FET-based hydrogen sensors, using the experimental data and the models proposed for calculation of sensors’ errors. Experimental technique and results The layout of sensor chip, simplified wiring diagrams of elements and parameters of FET are presented in Figure 1 and in Table 1. The n-channel FET element based on Pd(Ag)-Ta2O5-SiO2-Si (1) structure was fabricated on single chip together heater-resistor (2) and with temperature sensor (3) by means of conventional n-MOS-technology using laser evaporation Pd(Ag)-films. The Pd(Ag)-resistor was not investigated because it was previously shown that it has worse characteristics than FET. The measuring circuit provides the constant drain current ID (0.1 mA) and source-drain voltage VD (0.2 V), and the output voltage V is the gate voltage VG . The chip temperature T being equal to (130 ± 2 °С) was supported by the temperature-stabilization circuitry using on-chip thermo-sensor and heater. Five sensors with similar characteristics were selected (dispersion of initial output voltage V 0 is not more than 10%). There were repeatedly measured hydrogen responses of sensors for different concentrations Ci (e.g., ordinary j-cycle in Figure 2a). The model’s parameters ΔVCM and k of the function ΔVC (C) based on the approximation of the averaged values of the experimental points ΔVCi were determined as in [7] (Figure 2b). Averaged values of FET’s parameters in Table 1, values of Vai (Ci ), values of ΔVCi (Ci ) and variation indices θ Vi (Ci ), presented in Table 2, were calculated using Equations (1). Modeling of errors and discussion To estimate errors of FET-based sensors there were proposed general and numerical models of V(D,C), hydrogen sensitivities Si and Sd , absolute ΔC and relative errors δC, the maximum absolute error Δ(V) of the output voltage V, which are given by Equations (2) – (7). According to (3), the component V 0 of model V (D,C) depend on primary voltage V 00, the initial time drift ΔV 0t (t) and the drift ΔV 0(D) associated with the total hydrogen dose D = ∫C(t)dt, and as errors of “zero”, can be compensated by calibrating the sensor measuring circuit. Value V 00 depends on VT 0, ID and b (3). According to models the measurement errors depend on the instrumental errors of etalon concentration ∆C 0 and of voltage measurement ΔVV , the random errors ΔVai associated with the voltage V variations of various sensors, fluctuations of chip temperature ΔVT [4], of electrical circuits’ parameters ΔVID [5] and additional errors ΔVZ due to the influence of other gases and radiation [6]. Besides, errors depend on hydrogen concentration, sensitivity Sd and total hydrogen dose D. Approximated dependences of relative errors on the concentration for different conditions (e.g., Figure 3) can be used to estimate operating concentration ranges for given δC, and hydrogen sensitivity threshold CT (e.g., as in Table 3). Conclusions The proposed methodology of estimation errors of sensors based on FET with structure Pd(Ag)-Ta2O5-SiO2-Si, using numerical models with parameters determined from experimental hydrogen responses, can be applied to estimate errors of integrated sensors with other types of FETs and measuring circuits.