Refine
Document Type
- conference proceeding (article) (7)
- Article (6)
- Preprint (3)
Is part of the Bibliography
- no (16)
Keywords
- Photoacoustic spectroscopy (3)
- 2D materials (1)
- Acetone (1)
- Acoustic resonance monitoring (1)
- Atemluft (1)
- Breath analysis (1)
- Calibration-free method (1)
- CoNRad (1)
- Deep learning (1)
- Digital Twin (1)
- Micro-opto-electro-mechanical Systems (1)
- Miniaturization (1)
- Photoakustische Spektrometrie (1)
- Pressure broadening (1)
- Pseudo-line lists (1)
- Quantenkaskadenlaser (1)
- Quantum cascade laser (1)
- SPR-imaging (1)
- Sensor (1)
- Simulation (1)
- Smart sensor (1)
- Spectral simulation (1)
- Spectroscopy (1)
- Surface Plasmon Resonance Spectroscopy (1)
- Transfer learning (1)
- acetone (1)
- breath analysis (1)
- functional surface (1)
- graphene oxide (1)
- pattern recognition (1)
- quantum cascade laser (1)
- reduced graphene oxide (1)
- small-molecule sensing (1)
- spectral simulation (1)
- surface plasmon resonance imaging (1)
Institute
Begutachtungsstatus
- peer-reviewed (7)
In process analytics or environmental monitoring, the real-time recording of the composition of complex samples over a long period of time presents a great challenge. Promising solutions are label-free techniques such as surface plasmon resonance (SPR) spectroscopy. They are, however, often limited due to poor reversibility of analyte binding. In this work, we introduce how SPR imaging in combination with a semi-selective functional surface and smart data analysis can identify small and chemically similar molecules. Our sensor uses individual functional spots made from different ratios of graphene oxide and reduced graphene oxide, which generate a unique signal pattern depending on the analyte due to different binding affinities. These patterns allow four purine bases to be distinguished after classification using a convolutional neural network (CNN) at concentrations as low as 50 μM. The validation and test set classification accuracies were constant across multiple measurements on multiple sensors using a standard CNN, which promises to serve as a future method for developing online sensors in complex mixtures.
With an atmospheric concentration of approximately 2000 parts per billion (ppbV, 10−9) methane (CH4) is the second most abundant greenhouse gas (GHG) in the atmosphere after carbon dioxide (CO2). The task of long-term and spatially resolved GHG monitoring to verify whether climate policy actions are effective, is becoming more crucial as climate change progresses. In this paper we report the CH4 concentration readings of our photoacoustic (PA) sensor over a five day period at Hohenpeißenberg, Germany. As a reference device a calibrated cavity ringdown spectrometer Picarro G2301 from the meteorological observatory was employed. Trace gas measurements with photoacoustic instruments promise to provide low detection limits at comparably low costs. However, PA devices are often susceptible to cross-sensitivities related to environmental influences. The obtained results show that relaxation effects due to fluctuating environmental conditions, e.g. ambient humidity, are a non-negligible factor in PA sensor systems. Applying algorithm compensation techniques, which are capable of calculating the influence of relaxational effects on the photoacoustic signal, increase the accuracy of the photoacoustic sensor significantly. With an average relative deviation of 1.11 % from the G2301, the photoacoustic sensor shows good agreement with the reference instrument.
Successful transfer of photoacoustic gas sensors from laboratory to real-life applications requires knowledge about potential cross-sensitivities towards environmental and gas matrix changes. Multi-dimensional calibration in case of cross-sensitivities can become very complex or even unfeasible. To address this challenge, we present a novel algorithm to compute the collision based non-radiative efficiency and phase lag of energy relaxation on a molecular level (CoNRad) for photoacoustic signal calculation. This algorithmic approach allows to calculate the entire elaxation cascade of arbitrarily complex systems, yielding a theoretical photoacoustic signal. In this work the influence of varying bulk compositions, i.e. nitrogen (N2), oxygen (O2) and water (H2O) on the photoacoustic signal during methane (CH4) detection is demonstrated. The applicability of the algorithm to other photoacoustic setups is shown exemplary by applying it to the relaxational system investigated in [1]. Hayden et al. examined the effect of water on photoacoustic carbon monoxide (CO) detection.
We describe how to build a homogeneous light source for Surface Plasmon Resonance Imaging (SPRi) which mainly finds its applications in pharmaceutical screening and biotechnology so far. SPR spectroscopy is a label-free, non-destructive and highly sensitive measurement principle for detecting changes in the refractive index in close vicinity of a gold surface. A transfer of this technology to a miniaturized sensor will broaden the range of possible applications. Commercial SPR assays are mainly working with a small number of sensing spots. In contrast, the SPR imaging system shown here will allow the use of an array of many sensing spots. In combination with chemical receptors designed as an artificial nose or an electronic tongue, the simultaneous detection of many analytes is envisioned. So far, lasers or other inhomogeneous light sources were used to illuminate the sensing surface, which is decreasing the systems sensitivity. We show a compact ( 300mm2) homogeneous top hat profile. The combination of a high bit-resolution camera with our new light source enables a reflectivity based surface plasmon resonance imaging system with a high refractive index unit (RIU) resolution.
Confidence in results of an Artificial Neural Networks (ANNs) is increased by preferring to reject data, that is not trustful, instead of risking a misclassification. For this purpose a model is proposed that is able to recognize data, which differs significantly from the training data, during inference. The proposed model observes all activations of the hidden layers, as well as input and output layers of an ANN in a grey-box view. To make ANNs more robust in safety critical applications, this model can be used to reject flawed data, that is suspected to decrease the accuracy of the model. If this information is logged during inference, it can be used to improve the model, by training it specifically with the missing information. An experiment on the MNIST dataset is conducted and its results are discussed.
Ultrasound measurement technology has advancedin previous decades due to positive developments in computing power and lower cost for the necessary hardware. Correlation-based processing can be advantageous for high-accuracy measurements, especially in noisy environments. Higher computational load makes correlation less attractive for industrialapplications, though, favouring traditional processing techniques. A platform for comparing various processing techniques hasbeen developed. It was found that both zero-crossing detectionas well as correlation can achieve very high accuracy over a hightemperature range but that correlation-based processing can lead to a significant precision bias depending on system bandwidth.
A Deep Learning System to Transform Cross-Section Spectra to Varying Environmental Conditions
(2022)
Absorption cross-sections provide a basis for many gas sensing applications. Therefore, any error in molecular cross-sections caused by varying environmental conditions propagates to spectroscopic applications. Original molecular cross-sections in varying environmental conditions can only be simulated for some molecules, whereas for most multi-atom molecules, one must rely on high-precision measurements at certain environmental configurations. In this study, a deep learning system trained with simulated absorption cross-sections for predicting cross-sections at a different pressure configuration is presented. The system’s capability to transfer to measured, multi-atom cross-sections is demonstrated. Thus, it provides an alternative to (pseudo-) line lists whenever the required information for simulation is unavailable. The predictive performance of the system was evaluated on validation data via simulation, and its transfer learning capabilities were demonstrated on actual measurement chlorine nitrate data. From the comparison between the system and line lists, the system shows slightly worse performance than pseudo-line lists but its predictive quality is still deemed acceptable with less than 5% relative integral change with a highly localized error around the peak center. This opens a promising way for further research to use deep learning to simulate the effect of varying environmental conditions on absorption cross-sections.
Digital Twin of a Photoacoustic Trace Gas Sensor for Monitoring Methane in Complex Gas Compositions
(2022)
The digitalization of industrial processes requires smart sensor systems. Photoacoustic spectroscopy is well suited for this purpose as it allows for small-sized and low-cost trace gas analysis. However, the method is susceptible to changes in measurement conditions and standard calibration routines often fail to correct for all changes. We therefore created a Digital Twin (DT) of a photoacoustic trace gas sensor for methane and evaluated it regarding variations in gas composition (CH4, N2, O2, CO2, H2O), temperature and pressure. With a mean absolute percentage error of 0.8 % the accuracy of the sensor after DT compensation significantly exceeds the 24 % achieved based on standard calibration in nitrogen. For the first time, we can fully analytically compute the photoacoustic signal under moderate ambient conditions with an error in the ppbV range by taking a holistic approach. Assuming knowledge of the underlying energy transfer processes, the model of this Digital Twin can be adapted to any microphone based photoacoustic sensor for monitoring any analyte species.
Creating a Digital Twin of a Photoacoustic Gas Sensor for Methane Detection in Complex Gas Matrices
(2022)
We present the calculation results from a digital twin (DT) of our photoacoustic (PA) sensor for methane detection, regarding gas composition, temperature and pressure variations.
Stickstoffdioxid (NO2) wird hauptsächlich durch anthropogene Verbrennungsprozesse in Verkehr und Industrie in die Umgebungsluft eingebracht [1]. Die stark reaktive, oxidative Wirkung von NO2 kann bei Exposition mit höheren Konzentrationen Reizungen und allergische Reaktionen in den Atemwegen hervorrufen. Weiter führt NO2 zur Bildung von saurem Regen, Smog und Ozon (O3) die sich ebenso negativ auf die menschliche Gesundheit und Umwelt auswirken [2,3]. Aufgrund dieser Effekte wird die Spurengaskonzentration von NO2 in der Luft in Deutschland an ausgewählten, verkehrsreichen Orten gemessen.
Trace gas analysis in breath is challenging due to the vast number of different components. We present a highly sensitive quantum cascade laser based photoacoustic setup for breath analysis. Scanning the range between 8260 and 8270 nm with a spectral resolution of 48 pm, we are able to quantify acetone and ethanol within a typical breath matrix containing water and CO2. We photoacoustically acquired spectra within this region of mid-infra-red light and prove that those spectra do not suffer from non-spectral interferences. The purely additive behavior of a breath sample spectrum was verified by comparing it with the independently acquired single component spectra using Pearson and Spearman correlation coefficients. A previously presented simulation approach is improved and an error attribution study is presented. With a 3σ detection limit of 6.5 ppbV in terms of ethanol and 250 pptV regarding acetone, our system is among the best performing presented so far.
Trace gas analysis in breath is challenging due to the vast number of different components. We present a highly sensitive quantum cascade laser based photoacoustic setup for breath analysis. Scanning the range between 8263 and 8270 nm with a spectral resolution of 48 pm, we are able to quantify acetone and ethanol within a typical breath matrix containing water and CO2. We photoacoustically acquired spectra within this region of mid-infra-red light and prove that those spectra do not suffer from non-spectral interferences. The purely additive behavior of a breath sample spectrum was verified by comparing it with the independently acquired single component spectra using Pearson and Spearman correlation coefficients. A previously presented simulation approach is improved and an error attribution study is presented. With a 3σ detection limit of 6.5 ppbv in terms of ethanol and 250 pptv regarding acetone, our system is among the best performing presented so far.
Digital Twin of a Photoacoustic Trace Gas Sensor for Monitoring Methane in Complex Gas Compositions
(2023)
The digitalization of industrial processes requires smart sensor systems. Photoacoustic spectroscopy is well suited for this purpose as it allows for small-sized and low-cost trace gas analysis. However, the method is susceptible to changes in measurement conditions and standard calibration routines often fail to correct for all changes. We therefore created a Digital Twin (DT) of a photoacoustic trace gas sensor for methane and evaluated it regarding variations in gas composition (CH4, N2, O2, CO2, H2O), temperature and pressure. With a mean absolute percentage error of 0.8 % the accuracy of the sensor after DT compensation significantly exceeds the 24 % achieved based on standard calibration in nitrogen. For the first time, we can fully analytically compute the photoacoustic signal under moderate ambient conditions with an error in the ppbV range by taking a holistic approach. Assuming knowledge of the underlying energy transfer processes, the model of this Digital Twin can be adapted to any microphone based photoacoustic sensor for monitoring any analyte species.
With an atmospheric concentration of approximately 2000 parts per billion (ppbV, 10−9), methane (CH4) is the second most abundant greenhouse gas (GHG) in the atmosphere after carbon dioxide (CO2). The task of long-term and spatially resolved GHG monitoring to verify whether climate policy actions are effective is becoming more crucial as climate change progresses. In this paper we report the CH4 concentration readings of our photoacoustic (PA) sensor over a 5 d period at Hohenpeißenberg, Germany. As a reference device, a calibrated cavity ring-down spectrometer, Picarro G2301, from the meteorological observatory of the German Weather Service (DWD) was employed. Trace gas measurements with photoacoustic instruments promise to provide low detection limits at comparably low costs. However, PA devices are often susceptible to cross-sensitivities related to fluctuating environmental conditions, e.g. ambient humidity. The obtained results show that for PA sensor systems non-radiative relaxation effects induced by varying humidity are a non-negligible factor. Applying algorithm compensation techniques, which are capable of calculating the influence of non-radiative relaxation effects on the photoacoustic signal, increase the accuracy of the photoacoustic sensor significantly. With an average relative deviation of 1.11 % from the G2301, the photoacoustic sensor shows good agreement with the reference instrument.