Fingerprint terahertz spectroscopy combined with machine learning for multicomponent mixture analysis

Summary form only given. The terahertz (THz) band of the electromagnetic spectrum has long held interest in the science and engineering community for its unique applications in imaging and spectroscopy. However, due to the difficulty in generating and receiving THz radiation, only very recently practical and efficient THz systems become available. Due to recent advances in the development and manufacturing of commercially available THz spectroscopy and imaging system many of these applications are coming to fruition. As such, the practical implementation of THz in applications needs to be investigated in order to determine the advantages of THz technology. In this work, THz spectroscopy of photovoltaic semiconductor devices will be performed throughout various critical stages in the fabrication process. Due to the few millimeter penetrating and none-destructive properties of THz radiation it will be possible to monitor semiconductor wafers using THz spectroscopy. This technology has a potential to provide insight into the development of wafer defects earlier in the fabrication process. Using a pulsed Terahertz imaging and spectroscopy system at the University of Arkansas, our group will investigate the electrical properties of semiconductor photovoltaics and will conduct THz imaging of defected ones. Testing the wafers throughout various key steps in fabrication is a process known to produce a functional device. This will then be repeated for the same process, only now with a critical fatal defect such as a buried crack, surface crack or a defect in the doping level. These defects will be intentionally inserted at an early stage to determine if THz spectroscopy can be used to detect them early on. THz characterization of the semiconductors for the impurity doping levels and crystal orientation to be studied will be performed to determine the electrical properties in this frequency range. This will allow for accurate numerical simulation to compare with the experimental results.

Terahertz (THz) spectroscopy has high sensitivity for the interaction between molecules and vibration of large molecules. Besides, it is well known that infrared (IR) spectroscopy can be used to detect the vibration of functional groups. Hence, the comparison of the conventional diagnosis method using IR spectroscopy and THz spectroscopy is expected for more potential diagnosis. In this study, the comparison between THz spectroscopy and IR spectroscopy for deteriorated hydrocarbon liquids has been conducted for the degraded insulating oils. The surveyed samples are pure dodecane and mineral oil. The degraded samples are made by heating in the oxygen atmosphere. In addition, we also measured the breakdown voltage and the volume resistivity to know the states of degraded insulating oil. The experimental results show that oxidation causes spectral changes in the both of THz and IR regions. The spectral change in the IR region is due to carbonyl group and hydroxyl group vibrations. Whereas, in the THz region, the spectral change originates from the modification of intermolecular interaction. The degradation phenomena of the insulating oil contribute to higher carbonyl/hydroxyl content and associated by stronger intermolecular interaction that leads to its lower breakdown voltage.

In this paper, we present a novel muscle synergy extraction method based on multivariate curve resolution–alternating least squares (MCR-ALS) to overcome the limitation of the nonnegative matrix factorization (NMF) method for extracting non-sparse muscle synergy, and we study its potential application for evaluating motor function of stroke survivors. Nonnegative matrix factorization (NMF) is the most widely used method for muscle synergy extraction. However, NMF is susceptible to components’ sparseness and usually provides inferior reliability, which significantly limits the promotion of muscle synergy. In this study, MCR-ALS was employed to extract muscle synergy from electromyography (EMG) data. Its performance was compared with two other matrix factorization algorithms, NMF and self-modeling mixture analysis (SMMA). Simulated data sets were utilized to explore the influences of the sparseness and noise on the extracted synergies. As a result, the synergies estimated by MCR-ALS were the most similar to true synergies as compared with SMMA and NMF. MCR-ALS was used to analyze the muscle synergy characteristics of upper limb movements performed by healthy (n = 11) and stroke (n = 5) subjects. The repeatability and intra-subject consistency were used to evaluate the performance of MCR-ALS. As a result, MCR-ALS provided much higher repeatability and intra-subject consistency as compared with NMF, which were important for the reliability of the motor function evaluation. The stroke subjects had lower intra-subject consistency and seemingly had more synergies as compared with the healthy subjects. Thus, MCR-ALS is a promising muscle synergy analysis method for motor function evaluation of stroke patients.

Metal-organic frameworks (MOFs) are porous crystalline materials. Their large pores make them particularly interesting for membranes, gas separation, and gas storage. Furthermore, MOFs are ultralight, making them suitable for a large realm of exciting applications ranging from wearable devices to space technology. Optimizing MOFs for these applications demands a detailed understanding of their low energy dynamics and photophysics, which can be provided by terahertz (THz) spectroscopy. MOFs exhibit structural modes, or phonons, with energies in the meV range, which corresponds to the THz spectral range (0.1–10 THz, 0.4–40 meV). Understanding these modes is crucial in determining how a MOF interacts with guest molecules in the process of gas capture and storage. In this perspective, we discuss how gas-MOF interactions alter the MOFs’ spectral fingerprints. We demonstrate that THz spectroscopy can be used for gas adsorption monitoring and explain how density functional theory, together with THz spectra, can illuminate the dynamic structure of MOFs, providing unique insight into their functionality. THz is also a contact free probe for conductivity and allows us to measure short range conductivity within an individual MOF crystal. We will discuss the advantages of THz as a conductivity probe for MOFs as compared to more established direct current techniques. We will then expand our view to incorporate ultrafast photoconductivity in MOFs measured via optical pump-THz probe spectroscopy, in comparison to more established ultrafast spectroscopic tools such as optical transient absorption and photoluminescence. We will supplement this section with a discussion of THz studies on perovskites, which unveiled electron–phonon interactions not yet explored in MOFs.

Terahertz (THz) wave science and technology have been found countless applications in biomedical imaging, security screening, and non-destructive testing as they approach maturity. However, due to the challenge of high ambient moisture absorption, the development of remote open-air broadband THz spectroscopy technology is lagging behind the compelling need that exists in homeland security, astronomy and environmental monitoring. Furthermore, the underlying physical mechanisms behind the interaction between the THz wave and laserinduced plasma which responds strongly to electromagnetic waves have not been fully understood. This review aims to explain the light-plasma interaction at THz frequencies within a semiclassical framework along with experimental study of the femtosecond-laserinduced nitrogen plasma fluorescence under the illumination of single-cycle THz pulses. The results indicate that THz-radiation-enhanced-emission-of-fluorescence (THz-REEF) is dominated by electron kinetics in the THz field and the electron-impact excitation of gas molecules/ions. The information of the time-dependent THz field can be recovered from the measured time-resolved THz-REEF from single-color laser induced plasma with the help of the bias as local oscillator. The calculations and experimental verification lead to complete understanding of the science behind these effects and push forward to extend their capabilities in related applications such as remote THz sensing, plasma diagnostics and ultrafast photoluminescence modulation. Systematic studies in selected gases including neon, argon, krypton, xenon, methane (CH4), ethane (C2H6), propane (C3H8), and n-butane (C4H10) gases were performed to obtain an improved understanding of the THz-REEF. The dependences of the enhanced fluorescence on the THz field, laser excitation intensity, gas pressure, and intrinsic atomic properties were experimentally characterized. Both narrow line emission and broad continuum emission of the gas plasma were enhanced by the THz field. Their fluorescence enhancement ratios and time-resolved enhanced fluorescence were largely dependent on the scattering cross section and ionization potential of atoms. For the first time, we demonstrated a novel ‘all-optical’ technique of broadband THz wave remote sensing by coherently manipulating the fluorescence emission from asymmetrically ionized gas plasma that interacted with THz waves. By studying the ultrafast electron dynamics under the single cycle THz radiation, we found that the fluorescence emission from laser-induced air plasma was highly dependent on the THz electric field and the symmetry of the electron drift velocity distribution created by two-color laser fields. The time-resolved THz-REEF can be tailored by switching the relative two-color phase and laser polarizations. Owing to the high atmospheric transparency and omni-directional emission pattern of fluorescence, this technique can be used to measure THz pulses at standoff distances with minimal water vapor absorption and unlimited directionality for optical signal collection. The coherent THz wave detection at a distance of 10 m had been demonstrated. The combination of this method and previously demonstrated remote THz generation would eventually make remote THz spectroscopy available. We also introduced a unique plasma diagnostic method utilizing the THz-wave-enhanced fluorescence emission from the excited atoms or molecules. The electron relaxation time and plasma density were deduced through applying the electron impact excitation/ionization and electron-ion recombination processes to the measured time-delay-dependent enhanced fluorescence. The electron collision dynamics of nitrogen plasma excited at different gas pressures and laser pulse energies were systematically investigated. This plasma diagnostic method offers picosecond temporal resolution and is capable of omnidirectional optical signal collection. The ultrafast quenching dynamics of laser-pulse-induced photoluminescence in semiconductors under the radiation of single-cycle THz pulses was studied. It was found that the quenching in both cadmium telluride (CdTe) and gallium arsenide (GaAs) was linearly proportional to the intensity of incident THz waves and reaches up to 17% and 4% respectively at the peak intensity of 13 MW/cm2. The THz-wave-induced heating of the carriers and lattice and the subsequent decreased efficiency of photocarrier generation and recombination were most likely to be responsible for the quenching. This is potentially useful for the applications of a non-invasive ultrafast light modulator for photoluminescence devices with picoseconds switching time in the fields of the light-emitting devices and optical communication.

Effects of Hydration Levels on the Bandwidth of Microwave Resonant Absorption Induced by Confined Acoustic Vibrations

The timely detection of information on crop nutrition is of great significance for improving the production efficiency of facility crops. In this study, the terahertz (THz) spectral information of tomato plant leaves with different nitrogen levels was obtained. The noise reduction of the THz spectral data was then carried out by using the Savitzky-Golay (S-G) smoothing algorithm. The sample sets were then analyzed by using Kennard-Stone (KS) and random sampling (RS) methods, respectively. The KS algorithm was optimized to divide the sample sets. The stability competitive adaptive reweighted sampling (SCARS), uninformative variable elimination (UVE), and interval partial least-squares (iPLS) algorithms were then used to screen the pre-processed THz spectral data. Based on the selected characteristic frequency bands, a model for the detection of the nitrogen content of tomato based on the THz spectrum was established by the radial basis function neural network (RBFNN) and backpropagation neural network (BPNN) algorithms, respectively. The results show that the root-mean-square error of correction (RMSEC) and root-mean-square error of prediction (RMSEP) of the BPNN model were respectively 0.1722% and 0.1843%, and the determination coefficients of the correction set (Rc 2) and prediction set (Rp 2) were respectively 0.8447 and 0.8375. The RMSEC and RMSEP values of the RBFNN model were respectively 0.1322% and 0.1855%, and the Rc 2 and Rp 2 values were respectively 0.8714 and 0.8463. Thus, the accuracy of the model established by the RBFNN algorithm was slightly higher. Therefore, the nitrogen content of tomato leaves can be detected by THz spectroscopy. The results of this study can provide a theoretical basis for the research and development of equipment for the detection of the nitrogen content of tomato leaves.

Terahertz (THz) spectroscopy and imaging were used for probing defects in potential imaging phantoms. Imaging phantoms are materials that can be used to mimic biological systems or general industrial materials. Polyvinyl acetate, silicone and plaster of Paris were used to simulate thin, gel-like, and solid systems, respectively. Defects and impurities were introduced for each material. The THz properties and images of the samples were determined using THz time-domain spectroscopy (THz-TDS). The THz absorption spectra and images were able to reveal variations between the control samples and samples with defects. This clearly shows that these variations are correlated with the defects introduced into the material. These results demonstrate the effectiveness of THz-TDS and imaging in detecting defects even across a wide variety of materials. Furthermore, using principal component analysis (PCA) of the THz images, it was observed that the THz properties varied depending on the type of defect introduced into the samples. PCA results showed distinct clusters for different defects and impurities in the samples. These results highlight a promising non-destructive method for identifying material defects, which can complement other non-destructive evaluation methods, ensuring the quality and safety of materials and products across various industries.

Soil shear strength is an important indicator of soil erosion sensitivity and the tillage performance of the cultivated layer. Measuring soil shear strength at a field scale is difficult, time-consuming, and costly. This study proposes a new method to predict soil shear strength parameters (cohesion and internal friction angle) by combining cone penetration test (CPT) data and soil properties. A portable CPT measuring device with two pressure sensors was designed to collect two CPT data in farmland, namely cone tip resistance, and cone side pressure. Direct shear tests were performed in the laboratory to determine the soil shear strength parameters for 83 CPT data collection points. Two easily available soil properties (water content and bulk density) were determined via the oven-drying method. Using the two CPT data and the two soil properties as predictors, three machine learning (ML) models were built for predicting soil cohesion and the internal friction angle, including backpropagation neural network (BPNN), partial least squares regression (PLSR), and support vector regression (SVR). The prediction performance of each model was evaluated using the coefficient of determination (R2), the root-mean-square error (RMSE), and the relative error (RE). The results suggested that among all the evaluated models, the BPNN model was the most suitable prediction model for soil cohesion, and the SVR model performed best in predicting soil internal friction angle. Thus, our findings provide a foundation for the convenient and low-cost measurement of soil shear strength parameters.

The brown rice rate (BRR), milled rice rate (MRR), and head rice rate (HRR) are important indicators of rice milling quality. The simultaneous detection of these three metrics holds significant economic value for rice milling quality assessments. In this study, hyperspectral imaging was employed to estimate the rice milling quality attributes of two rice varieties (Xiushui121 and Zhehujing26). Partial Least Squares Regression (PLSR), Support Vector Regression (SVR), Convolutional Neural Networks (CNNs), and Backpropagation Neural Networks (BPNNs) were used to establish both single-task and multi-task models for the prediction of milling quality attributes. Most multi-task models demonstrated a higher prediction accuracy compared with their corresponding single-task models. Among single-task models, BPNNs outperformed the others in predicting BRR and HRR, with correlation coefficients (r) up to 0.9. SVR excelled in forecasting the MRR. In multi-task learning, BPNNs exhibited relatively better performance, with r values exceeding 0.81 for all three indicators. SHapley Additive exPlanations (SHAP) analysis was used to explore the relationship between wavelength and rice milling quality attributes. This study confirmed that this nondestructive detection method for rice milling quality using hyperspectral imaging combined with machine learning and deep learning algorithms could effectively assess rice milling quality, thus contributing to breeding and growth management in the industry.

The aim of this work was to develop an accurate model to swiftly forecast the ultimate conditions and the stress–strain curves of ultra-high-performance concrete (UHPC) confined by fibre-reinforced polymer (FRP). Four neural network predictive models based on machine learning (back-propagation neural network (BPNN), support vector regression (SVR), extreme learning machine (ELM) and generalised regression neural network (GRNN)) were constructed to forecast the ultimate conditions. A BPNN-based model to predict stress–strain curves was developed. A database of 193 compression test results of FRP-confined UHPC members was collected from the open literature and a thorough evaluation of data quality was conducted. The results showed that the SVR, BPNN, GRNN and ELM prediction models exhibited excellent accuracy when compared with a design-oriented model. The GRNN exhibited the highest predictive precision, followed by SVR. The developed prediction models establish a foundation for the design and quick prediction of FRP-confined UHPC.

Investigations were initiated for developing a rapid and nondestructive detection method to measure the illegal additive of melamine into milk powder by using terahertz (THz) spectroscopy and the correlation analysis algorithm. The absorption coefficients exhibited a maximum absorption peak at 2.04 THz, which would normally increase along with the concentration of melamine additive. In the current study, correlation analysis was carried out to select a pair-variable at 2.04 and 2.34 THz for improving the predictive ability of the multiple linear regression (MLR) model. Compared with the partial least square (PLS) model in full spectrum, the MLR model for powder samples could be considered successful in terms of quality control of milk powder with correlation coefficient R2 of 0.97 and root mean square error of prediction (RMSEP) of 1.38%. At the same time, the MLR model was simple and easier to interpret than the PLS one. The results of the research suggested that THz spectroscopy in combination with the correlation analysis algorithm has a significant potential in the quantitative analysis of the illegal additive of melamine in milk powder.

Characterization of Pesticide Residue, cis-Permethrin by Terahertz Spectroscopy

In this work, the terahertz (THz) spectroscopy system was used for the detecting and evaluation of hidden damages in a glass fiber reinforced plastic (GFRP). The interaction between THz and the GFRP was analyzed including the effects of reflecting, scattering and absorption of THz radiations with respect to the type of hidden damage. Both the transmission and reflective configurations were used to investigate the hidden damages including the delamination, fiber fracture and moisture absorption. Finally, the hidden damages inside of the composite laminates were successfully imaged simultaneously based on the time-domain spectroscopy of THz radiation. Additionally, the moisture absorption damage in the GFRP could be detected by analyzing of the frequency domain spectrum. It is expected that the developed THz nondestructive evaluation (NDE) technique can be widely used to evaluate the health of the composite structures.

The role of water in biomolecule dynamics has attracted much interest over the past decade, due in part to new probes of biomolecule–water interactions and developments in molecular simulations. Terahertz (THz) spectroscopy, among the most recent experimental methods brought to bear on this problem, is able to detect even small solute‐induced changes of the collective water network dynamics at the biomolecule–water interface. When studying the properties of water in aqueous solutions via THz absorption, proteins show a long‐ranged influence on water network dynamics, and even small saccharides influence the dynamics of several hydration layers. The THz spectrum of water solvating a protein is sensitive to mutations and depends on the surface charge and flexibility of the protein. Influence on the solvation shell appears most pronounced for native wild‐type proteins and decreases upon partial unfolding or mutation. THz spectra of solvated saccharides reveal that the number of water molecules coupled dynamically to a saccharide forming a dynamic hydration shell around it is related to the number of exposed oxygen atoms on the solute surface. The thickness of this layer appears correlated with the bioprotection efficiency of the saccharide. All findings support the thesis of a long‐range dynamic coupling between biomolecules and solvent. Kinetic terahertz absorption (KITA) studies of protein folding recently revealed that solvent dynamics are coupled to secondary structure formation of the protein. The solvent water network is dynamically rearranged in milliseconds before the protein folds to its native state in the following seconds. THz spectroscopy gives experimental evidence that collective long‐range dynamics are a key factor of biomolecular hydration.