Portable chemical detection platform for on-site monitoring of odorant levels in natural gas.

The adequate odorization of natural gas is critical to identify gas leaks and to reduce accidents. To ensure odorization, natural gas utility companies collect samples to be processed at core facilities or a trained human technician smells a diluted natural gas sample. In this work, we report a detection platform that addresses the lack of mobile solutions capable of providing quantitative analysis of mercaptans, a class of compounds used to odorize natural gas. Detailed description of the platform hardware and software components is provided. Designed to be portable, the platform hardware facilitates extraction of mercaptans from natural gas, separation of individual mercaptan species, and quantification of odorant concentration, with results reported at point-of-sampling. The software was developed to accommodate skilled users as well as minimally trained operators. Detection and quantification of six commonly used mercaptan compounds (ethyl mercaptan, dimethyl sulfide, n-propylmercaptan, isopropyl mercaptan, tert­butyl mercaptan, and tetrahydrothiophene) at typical odorizing concentrations of 0.1-5 ppm was performed using the device. We demonstrate the potential of this technology to ensure natural gas odorizing concentrations throughout distribution systems.

Non-destructive method to classify walnut kernel freshness from volatile organic compound (VOC) emissions using gas chromatography-differential mobility spectrometry (GC-DMS) and machine learning analysis.

Analysis of volatile organic compounds (VOCs) can be an effective strategy to inspect the quality of horticultural commodities and following their degradation. In this work, we report that VOCs emitted by walnuts can be studied using gas chromatography-differential mobility spectrometry (GC-DMS), and those GC-DMS data can be analyzed to predict the rancidity of walnuts, i.e., classify walnuts into grades of freshness. Walnut kernels were assigned a class n depending on their level of freshness as determined by a peroxide assay. VOC samples were analyzed using GC-DMS. From these VOC data, a partial least square regression (PLSR) model provided a freshness prediction value m, which corresponded to the rancid class n when m=n±0.5. The PLSR model had an accuracy of 80% to predict walnut grade and demonstrated a minimal root mean squared error of 0.42 for the m response variables (representative of walnut grade) with the GC-DMS data. We also conducted gas chromatography-mass spectrometry (GC-MS) experiments to identify volatiles that emerged or were enhanced with more rancid walnuts. The findings of the GC-MS study of walnut VOCs align excellently with the GC-DMS study. Based on our results, we conclude that a GC-DMS device deployed with a pre-trained machine learning model can be a very effective device for classifying walnut grades in the industry.

Machine learning and signal processing assisted differential mobility spectrometry (DMS) data analysis for chemical identification.

Differential mobility spectrometry (DMS)-based detectors are being widely studied to detect chemical warfare agents, explosives, chemicals, drugs and analyze volatile organic compounds (VOCs). The dispersion plots from DMS devices are complex to effectively analyze through visual inspection. In the current work, we adopted machine learning to differentiate pure chemicals and identify chemicals in a mixture. In particular, we observed the convolutional neural network algorithm exhibits excellent accuracy in differentiating chemicals in their pure forms while also identifying chemicals in a mixture. In addition, we propose and validate the magnitude-squared coherence (msc) between the DMS data of known chemical composition and that of an unknown sample can be sufficient to inspect the chemical composition of the unknown sample. We have shown that the msc-based chemical identification requires the least amount of experimental data as opposed to the machine learning approach.

Complex temperature dependence of coherent and incoherent lattice thermal transport in superlattices.

Currently, it is still unclear how and to what extent a change in temperature impacts the relative contributions of coherent and incoherent phonons to thermal transport in superlattices. Some seemingly conflicting computational and experimental observations of the temperature dependence of lattice thermal conductivity make the coherent-incoherent thermal transport behaviors in superlattices even more elusive. In this work, we demonstrate that incoherent phonon contribution to thermal transport in superlattices increases as the temperature increases due to elevated inelastic interfacial transmission. On the other hand, the coherent phonon contribution decreases at higher temperatures due to elevated anharmonic scattering. The competition between these two conflicting mechanisms can lead to different trends of lattice thermal conductivity as temperature increases, i.e. increasing, decreasing, or non-monotonic. Finally, we demonstrate that the neural network-based machine learning model can well capture the coherent-incoherent transition of lattice thermal transport in the superlattice, which can greatly aid the understanding and optimization of thermal transport properties of superlattices.

Quenching Thermal Transport in Aperiodic Superlattices: A Molecular Dynamics and Machine Learning Study.

Random multilayer (RML) structures, or aperiodic superlattices, can localize coherent phonons and therefore exhibit drastically reduced lattice thermal conductivity compared to their superlattice counterparts. The optimization of RML structures is essential for obtaining ultralow thermal conductivity, which is critical for various applications such as thermoelectrics and thermal barrier coatings. A higher degree of disorder in RMLs will lead to stronger phonon localization and, correspondingly, a lower lattice thermal conductivity. In this work, we identified several essential parameters for quantifying the disorder in layer thicknesses of RMLs. We were able to correlate these disorder parameters with thermal conductivity, as confirmed by classical molecular dynamics simulations of conceptual Lennard-Jones RMLs. Moreover, we have shown that these parameters are effective as features for physics-based machine learning models to predict the lattice thermal conductivity of RMLs with improved accuracy and efficiency.

Ultralow Lattice Thermal Conductivity of the Random Multilayer Structure with Lattice Imperfections.

Randomizing the layer thickness of superlattices (SL) can lead to localization of coherent phonons and thereby reduces the lattice thermal conductivity κ l . In this work, we propose strategies that can suppress incoherent phonon transport in the above random multilayer (RML) structure to further reduce κ l . Molecular dynamics simulations are conducted to investigate phonon heat conduction in SLs and RMLs with lattice imperfections. We found that interfacial species mixing enhances thermal transport across single interfaces and few-period SLs through the phonon "bridge" mechanism, while it substantially reduces the κ l of many-period SLs by breaking the phonon coherence. This is a clear manifestation of the transition from incoherent-phonon-dominated to coherent-phonon-dominated heat conduction in SLs when the number of interface increases. In contrast, interfacial species mixing always increases the κ l of RMLs owing to the dominance of incoherent phonons. Moreover, we found that doping a binary RML with impurities can reduce κ l significantly, especially when the impurity atom has an atomic mass lower or higher than both of the two base elements. This work reveals the critical effect of lattice imperfections on thermal transport in SLs and RMLs, and provides a unique strategy to hierachically suppress coherent and incoherent phonon transport concurrently.