Accuracy chemical sensors

The first group of sensor properties in Fig. 1.15 is concerned with the quality of results obtained in analytical processes involving a (bio)chemical sensor. All of them are obvious targets of analytical tasks [3]. As shown in the following section, the accuracy of the analytical results relies on a high reproducibility or repeatability, a steep slope of the calibration curve (or a low detection or quantification limit) and the absence of physical, chemical and physico-chemical interferences from the sample matrix. Sensors should ideally meet these essential requisites. Otherwise, they should be discarded for routine analytical use however great their academic interest may be. 

Chemical sensors, those that measure the presence or concentration of chemical species, are the subject of this book. Until recently, they received even less attention than other sensors in general, they are not as well developed. They have the same need to be small, inexpensive, and accurate as other sensors. However, accomplishing these requirements for chemical sensors is often more difficult than for other sensors because chemical sensors are noted for interferences. For example, a chloride sensor may be sensitive to other halides. One popular way to counter this limitation is to use an array of somewhat different sensors, each responsive to the same set of related compounds but with different sensitivity. The output of the sensor array can be processed by a computer to give greater accuracy than a single sensor for the concentration of one compound. Unfortunately, this approach tends to gain better accuracy at the expense of increased size and cost. 

Sensing chemical species is a much more difficult task than the measurement of mechanical variables such as pressure, temperature, and flow, because in addition to requirements of accuracy, stability, and sensitivity, there is the requirement of specificity. In the search for chemically-specific interactions that an serve as the basis for a chemical sensor, investigators should be aware of a variety of possible sensor structures and transduction principles. This paper adresses one such structure, the charge-flow transistor, and its associated transductive principle, measurement of electrical surface impedance. The basic device and measurement are explained, and are then illustrated with data from moisture sensors based on thin films of hydrated aluminum oxide. Application of the technique to other sensing problems is discussed. 

Abstract Development of new materials is needed for numerous applications in engineering, medical, and scientific arenas. In this chapter, we describe some of our research efforts that focus on developing strategies and tools for high throughput production and screening to create advanced biomaterials and chemical sensors. Using our developed tools, we are able to produce and screen a wide array of materials in a short period of time. In several current embodiments, the system can readily produce and fully screen 100-1,000 samples/day. Our developed automated systems can provide results with minimal user input, yet with better precision and accuracy in comparison to traditional manual methods. 

Sensor temperature coefficient and time response are also variables to be understood. Temperature may shift the pKa of the dye and change the cell thickness, and will certainly affect the actual value of the blood gas variables of the blood that is adjacent to the sensor. For these reasons, a complete blood gas sensor includes a local temperature sensor, particularly if the sensor is to be placed in a peripheral artery where local temperature may not be equal to central body temperature. The chemical sensor temperature coefficient must be well characterized so that it will accurately measure the local blood gas value. Bench analysers usually measure blood samples at 37 °C so the in vivo system must then adjust the measured value to that temperature. The temperature coefficient of the blood gas variables, in blood, may be several percent per degree, and, in the case of Foj, depend very strongly on the actual value. Thus, to make the in vivo sensor agree with bench analysers, local temperature sensing must have an accuracy of better than 1 °C. Size and accuracy requirements can be met by a miniature thermocouple. The system designer has to make sure that the temperature circuit can handle the microvolt signals with adequate accuracy and stability as well as meet patient electrical isolation requirements. 

Process Measurements. The most commonly measured process variables are pressures, flows, levels, and temperatures (see Flow LffiASURELffiNT Liquid-levell asurel nt PressureLffiASURELffiNT Temperaturel asurel nt). When appropriate, other physical properties, chemical properties, and chemical compositions are also measured. The selection of the proper instmmentation for a particular appHcation is dependent on factors such as the type and nature of the fluid or soHd involved relevant process conditions rangeabiHty, accuracy, and repeatabiHty requited response time installed cost and maintainabiHty and reHabiHty. Various handbooks are available that can assist in selecting sensors (qv) for particular appHcations (14—16). 

Typical analysis of blood and urine is based upon chemical reactions, one per chemical species (or analyte ) of interest. Low-accuracy, single-analyte tests such as for blood glucose can be performed using at-home kits comprehensive blood tests at hospitals are performed using dedicated analyzers that suck in a sample and pump portions of it to various sensor modules. The purpose of this chapter is to explore the possibility of using Raman spectroscopy to replace conventional present tests in certain circumstances. 

The ethanol concentration in the medium of a Saccharomyces cerevisiae cultivation can be monitored from its content in the gas phase by directly recording the current from a chemical MOS sensor [28]. The accuracy of such a measurement was significantly improved by using an electronic nose with five sensors in the array and recognizing the response pattern with ANN [29, 30]. The sensors were a combination of MOS and MOSFET sensors selected from a PCA loading plot. Data sets from three cultivations were used to train the ANN. When the trained net was applied on new cultivations the ethanol was predicted with a mean square error (RMSE) of 4.6% compared to the off-line determined ethanol (Fig. 6). With only one sensor the RMSE was 18%. 

---