Sensor Array Complexity and Analytical Capability

Abstract

Introduction Chemical sensors are generally associated with tasks involving detection of chemical stimuli while conventional analytical laboratory instruments are associated with analysis of chemical mixtures. While sensing tasks may not necessarily be perceived as involving chemical mixture analysis and evaluated as such, there is always an underlying analytical context in any sensing problem. This context becomes progressively more important as the complexity of chemical environment in which the sensing task is to be conducted increases, through both the number of potential intereferents as well as uncertainty regarding the composition of the chemical background.Design of individual sensors can be reasonably predicated on maximizing the sensitivity and selectivity of a narrowly-scoped chemical stimulus against expected backgrounds. However, as one may expect, design considerations can become significantly more complicated for systems based on sensor arrays. Arrays consisting of sensors that exhibit diverse, partially-selective response to a wide range of chemical stimuli allow, in theory, selective detection of a range of chemical stimuli against more complex backgrounds. [1] This is generally supported by analogy to biological olfaction as well as experimental work over the past few decades. Intuitively, a clear consequence of moving towards applications involving more complex and uncertain sensing tasks is a requirement that the system become less “sensor-like” and more “instrument-like.” A significant challenge in sensor array design has been that these concepts are not well-understood in a quantitative and device-agnostic fashion, making it difficult appropriately match sensor array and sensing task complexity in a predictable and systematic fashion. This presentation will discuss how analytical capabilities can be characterized on a continuum between simple sensor arrays and classical analytical instruments and the implications for sensor array design and selection. Approach Sensor array capability for determining composition of fixed, finite domains of independent chemical stimuli can be expressed and predicted using information-theoretic measures and knowledge of individual chemical response functions for each chemical stimulus/sensor pair. [2-4] Moving towards expression of more general-purpose analytical capability (i.e. dynamic, high-dimensional domains of independent chemical stimuli) requires a shift from consideration of the span of specific chemical stimuli to one of the span of molecular parameters probed by the sensor technology. Sensor array capability for addressing generalized chemical analysis problems can then be expressed in terms of the adjacency in molecular parameter space induced by the underlying response mechanism of a sensor technology and the manner in which specific sensor array configurations induce particular array sensitivities across this same domain of molecular parameters. Hypothetical sensor systems based on optical detection and chemisorption are used to illustrate these concepts. In particular, this framework allows consideration of the manner in which sensor arrays and laboratory instruments occupy a continuum of complexity with which they interact with the chemical environment. Conclusions This work demonstrates the value in developing improved first-principles models of sensor response grounded in molecular parameters rather than specific chemical stimuli. Such an understanding allows expression and prediction of the extent to which a sensor array offers “instrument like” general purpose capability as opposed to narrowly scoped sensing tasks. The complexity of a given sensor technology’s interaction with the chemical environment ultimately provides a limit to the complexity of chemical sensing tasks that are addressable by sensor arrays based on this technology. Considering these aspects of chemical sensor array design allows more reliable selection and design of sensor arrays for complex, real-world applications.