Efficient sensing approaches for high-density memristor sensor array

Abstract

Recent research shows ever growing interest in the potential applications of memristive devices. Among the many proposed fields, sensing is one of the most interesting as it could lead to unprecedented sensor density and ubiquity in electronic systems. In this paper, a framework for efficient gas detection using mem-ristor crossbar array is proposed and analysed. A novel Verilog-A based memristor model that emulates the gas sensing behaviour of doped metal oxides is developed for simulation and integration with design automation tools. Using this model, we propose and analyse three different gas detection structures based on array of memristor-based sensors. Gas presence together with some of its properties can be detected using resistance changes and spatial information from one or group of memristive sensors. Our simulation results show that depending on the organisation of the memristive elements and the sensing method, the response of the sensor varies providing a broader design space for future designers. For instance, with a 8 × 8 memristor sensor array, there is a ten times improvement in the accuracy of the sensor's response when compared with a single memristor sensor but at the expense of extra area overhead.

Sensing with Memristive Complementary Resistive Switch: Modelling and Simulations

“…One of the advantages of the VTEAM approach is based on voltage thresholds, unlike the previous characterizations reported in [20], [21], [22], and apart from this, the greater fidelity [19] to the real device behavior allows us to accurately design and simulate logic based circuits with memristors [23]. Our VTEAM model also has advantages as we have shorter computational times, generic behaviour, sufficient accuracy for the practical memristive device and less complexity [22] as compared to the previously reported sensing model presented in the [13], [17], [18]. All the simulation has been performed by using LTSpice simulator as it has an intuitive interface and able to see netlist by circuit designer, easy to use, and its free accessibility.…”

“…Array of sensors also open up the possibility to detect, quantify and differentiate multiple gases in real time [13]. In traditional approaches, bidirectional flow of current in oxidebased memristors causes sneak path leading to errorneous sensing in sensor crossbar architecture.…”

“…This is mostly due to the formation of a parallel resistance path between resistance of unselected cells and the desired cell(s) [14]. Several (b) Cross-section view of a memristor device for sensing applications [13].…”

“…approaches were explored in [13], [15] to minimize the sneak path current but had several disadvantages. For example in [15], if one of the faulty sensors in the crossbar exhibits a very low resistance, the output voltage is altered drastically and it become very difficult to identify the faulty sensor in a larger array.…”

“…For example in [15], if one of the faulty sensors in the crossbar exhibits a very low resistance, the output voltage is altered drastically and it become very difficult to identify the faulty sensor in a larger array. In [13], 1T1M architecture interposes a MOSFET transistor between the sensing memristor and the row, enabling and disabling the connection of the sensor to the circuit. However, MOSFET transistors show short channel effects at lower technology nodes [16].…”