Multi-analyte quantisation via time-frequency analysis of surface acoustic wave device response

<p>Perturbation of acoustic wave velocity and attenuation within piezoelectric media has long been understood, enabling the development of a range of acoustic wave devices for sensing applications. Such devices have been widely employed to address a range of practical problems, though they are particularly well suited to accurate quantification of the concentration of targeted analytes in gases. The high sensitivity achievable with Surface Acoustic Wave (SAW) devices in particular make them suitable for detecting even trace concentrations of such compounds, often down to several parts-per-billion with a high degree of accuracy.</p> <p>When employing these devices instrumentation is required to produce measurable output in response to acousto-electric perturbation via physical or chemical interaction. Interface electronics allow for the minute changes in a SAW devices electrical characteristics induced via this interaction to be monitored and correlated to observed operating conditions enabling characterisation, calibration, and integration into complex systems. However, such sensing platforms have thus far found limited adoption outside esoteric sensing applications. While SAW devices exhibit high sensitivity to targeted analytes such sensitivity often extends to other physical parameters such as device temperature, operating pressure, or toward other contaminant compounds limiting use in uncontrolled environments.</p> <p>Such limitations stem from the interface electronics and analysis methodology employed to produce measurable output from SAW device response. Conventional analysis typically condenses complex device response in the time and frequency domains down to a single measurand. For example, the feedback oscillator, arguably the most common low-cost interface, outputs a single measurement correlated to the response of a single acoustic mode in the frequency-domain. This discards potentially useful information from both the time-domain and any additional acoustic modes. In applications where multiple response mechanisms may drive device response it becomes impossible to distinguish between them when only a single measurand is available.</p> <p>While other approaches may overcome some limitations facing use of SAW devices in real-world sensing applications this typically requires additional, expensive instrumentation. Therefore, an alternative approach based upon modern Digital Signal Processing (DSP) methods is necessary to provide greater insight into the device response in both the time and frequency-domains for sensing applications. This work details the development of a novel device response analysis technique utilising time-frequency analysis. This method enables study of signals properties both the time and frequency-domains simultaneously, making use of information that is typically discarded in conventional analysis.</p> <p>To establish the developed time-frequency method as a viable alternative to conventional analysis it was employed to study temperature sensitivity of a SAW device between 30 °C and 150 °C. The same device was also examined using feedback oscillator and network analysis topologies. Frequency-domain measurements extracted using the time-frequency method were shown to deviate by less than 3% from the more established methods, while demonstrating measurement stability on-par with the feedback oscillator. Furthermore, measurement noise is demonstrated to increase only 8% above that measured when using the feedback oscillator and a consistent stimulus signal is used, thus a similar limit of detection may be achieved. The device tested additionally supports a secondary Surface Skimming Bulk Wave (SSBW) mode which cannot be reliably analysed using the feedback oscillator, however the time-frequency method is demonstrated as capable of extracting useful information from this previously inaccessible mode.</p> <p>Refinement of the method enables extraction and analysis of additional features generated by all supported acoustic modes. The time-frequency method was applied to a layered InOx/ZnO/XZ-LiNbO3 SAW structure supporting both Rayleigh and Shear Horizontal SAW (SH-SAW) acoustic modes. This layered device was exposed to varying concentrations hydrogen and humidity; exciting conductometric and mass-loading response mechanisms respectively. The behaviour of both supported acoustic modes in the time-frequency domain is observed, and several key measurands derived from frequency and time-domain information, as well as acoustic peak energy, are extracted. These measurands provide a means of quantifying each analyte, and provide insight into the response mechanisms at work in this sensing application. These measurements are extracted from a single set of experiments, demonstrating the ability to simultaneously monitor multiple acoustic modes using the novel time-frequency method.</p> <p>Through extraction of several key measurands from the layered SAW device - peak location in the time and frequency axes, and peak energy - for both the supported modes, a multi-dimensional characterisation curve was generated for concentrations of hydrogen up to 0.8%, and concentrations of humidity up to 40% relative humidity. This curve is then tested against several unknown combined concentrations of both analytes, and quantisation of the concentration of each analyte from a single experiment utilising a single SAW device is achieved.</p>