Terahertz Time-domain Spectra Research Articles

Wood-species identification based on terahertz spectral data augmentation and pseudo-label guided deep clustering

ABSTRACT In order to address the problem that existing methods combining spectral data and machine learning for wood species identification rely on labeled samples, this study introduces unsupervised learning into the field of wood identification. It proposes a novel wood-identification model called DCVAE (deep conditional variational autoencoder)-PLCAE (pseudo-label convolutional autoencoders). Terahertz time-domain spectra of wood samples at breast height of five broadleaf and five coniferous species were obtained (40 samples of each species of wood, 400 in total). A conditional variational autoencoder was applied to augment the terahertz spectroscopy dataset. Subsequently, a pseudo-label-guided deep clustering model was developed to extract more discriminative deep features. The model was compared with three traditional clustering algorithms and four deep clustering methods. Clustering experiments and visualization results show that the comprehensive clustering performance of DCVAE-PLCAE is better than the other comparative algorithms and that the extracted low-dimensional features has a more straightforward structure. The algorithm in this study can solve the problems of fewer labeled samples and the difficulty of extracting discriminative features by traditional clustering algorithms.

Infrared and Terahertz Spectra of Pu’er White Tea with Different Degrees of Oxidation

In this work, the Fourier transform infrared spectroscopy (FTIR) and Terahertz time domain spectra (THz-TDS) were employed to study Pu’er white tea with different oxidation levels by changing the withering time. The strongest absorption peak of 3359 cm-1 in FTIR spectra comes from the proteins and polysaccharides. Several intensity ratios of the characteristic absorption peaks with different oxidation levels were analyzed. The white tea with different oxidation levels can be distinguished from the absorbance of THz-TDS in the range of 0.5–1.5 THz. The refractive index calculated from THz-TDS was between 1.42 and 1.44. Principal component analysis (PCA) using FTIR absorbance data showed that the scattered points of PC1 and PC2 can be well distributed in different districts. Hierarchical cluster analysis (HCA) helped to carry out the clustering results of different white tea samples. The sample with overoxidation level can be distinguished by PCA and HCA using THz-TDS absorbance data. Therefore, FTIR might be a convenient, fast, and nondestructive strategy for identifying white tea with different oxidation levels. THz-TDS combined with some software algorithms can be applied to distinguish the different oxidation-processed tea.

Non-destructive detection of adulteration of weight-loss drugs in the field of spectral feature fusion

To crack down on criminals using the delivery channel to transport weight-loss drugs doped with toxic and harmful nonfood raw materials, a pattern recognition method of weight-loss drugs based on terahertz time-domain spectroscopy was proposed. Compared with traditional methods, terahertz spectrum had high signal-to-noise ratio in time-domain spectrum, which was fast, time-saving and lossless. In this study, seven kinds of weight-loss drugs were selected as experimental samples. The terahertz time-domain spectra of the samples were collected. Three characteristic frequency intervals of 0–0.19, 1.75–2.14 and 2.23–2.5 (THz) were found by automatic peak finder. The characteristic frequency intervals were processed by Hilbert transform, Butterworth low-pass filter, fast Fourier transform low-pass filter and the first-order derivatives after standard normal transform, the feature data was fused with the original spectra, and the original data and the data fused by the four methods were classified and recognized by particle swarm optimization least squares support vector machine and extreme learning machine model optimized by Cuckoo algorithm. The experimental results showed that the particle swarm optimization least squares support vector machine model had the best recognition effect on the spectral feature fusion data after Hilbert transform, and the accuracy can reach 100 %. It had a certain reference significance for the identification of weight-loss drugs in forensic science.

Study on the laser-induced damage of thin films by terahertz time-domain spectroscopy

The undamaged and damaged areas of the ITO thin film deposited on a silicon substrate induced by laser with different energies were tested with the transmission terahertz time-domain spectroscopy system. Their time-domain spectra and frequency-domain spectra of 0.4–1.0 THz were obtained, and the differences were analyzed between the undamaged and damaged areas in these two spectra. The results show that the peak-to-peak value in the time domain and the amplitude in the frequency domain of the damaged area evidently increased compared with those of the undamaged area, and with the increase in laser-induced energy, resulting in the larger damaged area and heavier surface roughness, these two parameters gradually decrease. For the damaged area of the ITO thin film with the undamaged substrate, the refractive index and absorption coefficient are lower than those of the undamaged area. Therefore, the variation in the terahertz time-domain and frequency-domain spectra could be utilized to distinguish the damage of the optical thin film irradiated by laser, which provides a new approach for the laser-induced damage identification and the technical support in effect of the laser-induced damage on the properties of the optical thin film element in the terahertz band.

Evaluating quality consistency of cigarette by 3 kinds of quantum fingerprints

Cigarette has a history of more than 40 years and once sold well all over the country. The purpose of this paper is to evaluate the quality consistency of 30 batches of cigarettes. Differential scanning calorimetry (DSC) curves, Fourier Transform Infrared (FT-IR) spectra and terahertz time-domain spectra of 30 batches of cigarettes were collected. In this paper, a new quality evaluation method was explored, the obtained fingerprints were quantized, the t-test was used to analyze the correlation between the original fingerprint and the quantum fingerprint. The quantized fingerprints were evaluated by the systematically quantitative fingerprint method (SQFM), the macro qualitative similarity and the macro quantitative similarity were used to reflect the internal differences of the samples. Finally, the antioxidant activities of 30 batches of samples were investigated, and the correlation between quantum fingerprints and IC50 values were analyzed by Pearson correlation coefficient (r).The result shows that there was no significant difference between the quantum fingerprint and the original fingerprint; SQFM can effectively calculate the similarity of sample curves and classify samples into different grades; the FT-IR fingerprints had a good correlation with the IC50 value. 30 samples performed a good quality consistency within an acceptable grades range. In addition, quantum fingerprints have broad application prospects in spectral analysis.

Lattice dynamics and terahertz response of microwave dielectrics: A case study of Al-doped Ca0.6Sm0.27TiO3 ceramics

Low-loss microwave dielectric ceramics are of great significance for fifth-generation telecommunication (5G) devices used for higher frequencies, while reducing the dielectric loss in the microwave band is still a technical challenge. Lattice dynamics and terahertz response are closely related to the polarization of dielectrics, and can be used as effective methods to explore the loss mechanism of microwave dielectrics. Here, Al3+ was doped into the conventional high-permittivity (εr) Ca1-xSm2x/3TiO3 (x = 0.4) ceramic system, enhancing the Q×f values significantly. Raman spectra, thermally stimulated depolarization currents (TSDC) and terahertz time-domain spectra (THz-TDS) together showed that the stiffness of cation vibration became stronger when z value was larger, indicating a smaller damping of the lattice vibration, and leading to lower extrinsic losses and higher Q×f values. The relationship between lattice dynamics, defect behavior and microwave dielectric loss will provide a significant reference for the development of low-loss microwave dielectric ceramics.

Cryogenic blackbody calibration source for superconducting terahertz detectors

Blackbody radiation source has been widely used as a calibration source for terahertz (THz) radiometers in recent decades with the applications of THz detection technology in the fields of aerospace, astronomy and remote sensing. We develop a THz blackbody calibration source capable of working in the cryogenic environment and having adjustable radiation power for the calibration of THz superconducting detectors. The ideal blackbody source has an emissivity and absorptivity of 1 and the reflectance coefficient is used to indirectly characterise the performance of the developed blackbody source. In this work, we use a mixture of epoxy, catalyst, carbon black and glass beads as blackbody absorbing material. The real part and imaginary part of the complex dielectric constant of Berkeley blackbody material are extracted from the THz time-domain spectra, and its reflection coefficient is measured. We use this material to design a conical blackbody radiation source , and simulate it as well. The simulation result show that it has low reflectivity below –35 dB in a frequency range of 0.2–0.5 THz. We fabricate a conical blackbody radiation source that is mounted in a dilution refrigerator, and use filters and light-guiding systems to make the detector for measuring the radiation by the THz light of a specific wavelength. The radiation power can be tuned by changing its temperature. The relationship between radiation power and temperature shows a power tuning range of 10<sup>–12</sup>–10<sup>–9</sup> W in the frequency range of 0.2–0.5 THz with a minimum power value of 2.13 × 10<sup>–12</sup> W. The designed blackbody radiation source can meet the calibration requirements of THz superconducting detectors, and will contribute to the development and application of highly sensitive THz radiometers.

Thermally-stimulated defect relaxations and microwave/terahertz dielectric response of La,Al co-doped (Ba,Sr)La4Ti4O15 ceramics

The thermally-stimulated relaxation of defects in La,Al co-substituted (Ba,Sr)La4Ti4O15 dielectric ceramics with hexagonal perovskite structure were investigated systematically with the thermally stimulated depolarization current (TSDC) measurements. The ion substitution decreased the lattice parameters and the bond lengths of the cations and O2−. TSDC spectra indicated the main extrinsic defects in the ceramics were oxygen vacancies and the concentration of oxygen vacancies decreased with the increasing ion substitutions, which was derived from the decrease in the lattice parameters. The terahertz time-domain spectra indicated the damping factor of lattice vibration was reduced with the decline of the concentration of oxygen vacancies, which decreased the microwave/terahertz dielectric loss. High-performance (Ba0.2Sr0.8)0.75La4.25Ti3.75Al0.25O15 (εr = 43.9, Q × f = 84300 GHz, τf = -18.0 ppm/°C) ceramics were obtained. The correlations between crystal structures, defect behaviors and dielectric properties discussed in this work could provide guidance for the modification of microwave dielectric ceramics.

Crystal Structure-Free Method for Dielectric and Polarizability Characterization of Crystalline Materials at Terahertz Frequencies.

Terahertz (THz) time-domain spectroscopy provides a direct and nondestructive method for measuring the dielectric properties of materials directly from the phase delay of coherent electromagnetic radiation propagating through the sample. In cases when crystals are embedded within an inert polymeric pellet, the Landau, Lifshitz, and Looyenga (LLL) effective medium model can be used to extract the intrinsic dielectric constant of the crystalline sample. Subsequently, polarizability can be obtained from the Clausius-Mossotti (CM) relationship. Knowledge of the crystal structure density is required for an analytical solution to the LLL and CM relationships. A novel crystal structure-free graphical method is presented as a way to estimate both dielectric constants and polarizability values for the situation when the crystal structure density is unknown, and the crystals are embedded within a pellet composed of a non-porous polymer. The utility of this crystal structure-free method is demonstrated by analyzing THz time-domain spectra collected for a set of amino acids (L-alanine, L-threonine, and L-glutamine) embedded within pellets composed of polytetrafluoroethylene. Crystal structures are known for each amino acid, thereby enabling a direct comparison of results using the analytical solution and the proposed crystal structure-free graphical method. For each amino acid, the intrinsic dielectric constant is extracted through the LLL effective medium model without using information of their crystal structure densities. THz polarizabilities are then calculated with the CM relationship by using the determined intrinsic dielectric constant for each amino acid coupled with its crystal density as determined graphically. Comparison between the analytical and graphical solutions reveal relative differences between dielectric constants of 3.7, 5.1, and 13.6% for threonine, alanine, and glutamine, respectively, and relative differences between polarizability of 0.6, 0.9, and 5.4%, respectively. These values were determined over the 10-20 cm-1 THz frequency range. The proposed method requires no prior knowledge of crystal structure information.

MXene-based ultra-thin film for terahertz radiation shielding

We have successfully fabricated Ti-based MXenes flakes, Ti3C2Tx, by chemical etching, then prepared it as an organic dispersion and finally spin-coated it on polyimide plastic substrate for terahertz wave shielding. The shielding effectivity of the 12 μm ultra-thin film can reach up to 17 dB measured by the terahertz time-domain spectra. We can attribute the excellent phenomenon to the intrinsic absorption of triple-layered Ti3C2, due to the similar double-peak type refraction curves, which have been respectively observed from the experimental samples and the simulation ones. High conductivity and strong THz absorption indicate the Ti3C2Tx MXene is the absorptive electromagnetic shielding material. Comparing with other kinds of THz shielding materials, the Ti-based MXenes might be a potential candidate for the next generation of ultra-thin and lightweight THz shielding.

Study on terahertz spectrum analysis and recognition modeling of common agricultural diseases.

Diseases are critical factors that affect the yield and quality of crops. Therefore, it is of great research value to develop rapid and quantitative methods for identification of common agricultural diseases. This exploratory study involved data analysis of common fungal pathogens using identification modeling based on terahertz spectrum technology. The selected pathogens were Physalospora piricola, Erysiphe cichoracearum, and Botrytis cinerea, which are common fungal pathogens that cause apple ring rot, cucumber powdery mildew, and grape gray mold blight, respectively. Taking polyethylene as the control, the terahertz time-domain spectra, and frequency-domain spectra of samples of the three pathogens were both measured. The absorption and refraction characteristics of these samples in the range of 0.1-2.0 THz were calculated and analyzed, and samples were then divided using the KS algorithm. Terahertz spectrum-image data blocks of the pathogen samples were preprocessed, and the dimensions of data were reduced using non-local mean filtering and the SPA algorithm, respectively. K-nearest neighbors (KNN), support vector machine (SVM), and BP neural network (BPNN), and other algorithms were used for analysis of terahertz images at characteristic frequencies, and for investigating the identification model. The model was quantitatively evaluated, and its imaging visualization was studied. The results suggest that there are significant differences among P. piricola, E. cichoracearum, and B. cinerea in absorption and refraction in the terahertz band. SVM modeling identification results of the three pathogens at the frequency of 1.376 THz were satisfactory, with an Rp of 0.9649, RMSEP of 0.0273, and a high (93.8212%) comprehensive evaluation index F1-score, and a clearly identifiable visualization effect. This study demonstrated the potential of terahertz spectroscopy to be used for identification of common crop pathogens and has provided technical references for the rapid diagnosis and early warning of agricultural diseases.

Application of terahertz spectrum and interval partial least squares method in the identification of genetically modified soybeans.

Genetically modified soybeans are the world's most important genetically modified agricultural product. At present, the traditional methods for identifying genetically modified and non-transgenic soybeans are time-consuming, costly, and complicated to operate, which cannot meet the needs of practical applications. Therefore, it is necessary to discover a fast and accurate method for identifying transgenic soybeans. Terahertz (THz) time domain spectra were collected in sequence from 225 transgenic and non-transgenic soybean samples. Fourier transform was used to convert the terahertz time domain spectrum into a THz frequency domain spectrum with a frequency range of 0.1-2.5THz. Firstly, the interval partial least squares (iPLS) method was used to remove interference spectral bands and select appropriate spectral intervals. Secondly, 168 samples were selected as the calibration set. Discriminant partial least squares (DPLS), Grid Search support vector machine (Grid Search-SVM) and principal component analysis back propagation neural network (PCA-BPNN) were used to establish a qualitative identification model. Afterwards, 57 test set samples were predicted. By comparing the experimental results, it was found that iPLS could effectively screen and remove the interference THz band, which was more helpful to improve the efficiency and accuracy of the identification model. After the iPLS and mean center pre-treatment technology, the Grid Search-SVM identification model had the best identification effect, with a total accuracy rate of 98.25% (transgenic identification rate was 96.15%, non-transgenic identification rate was 100%). This study shows that after selecting spectra from iPLS, THz spectroscopy combined with chemometrics can more accurately, quickly, and efficiently identify transgenic and non-transgenic soybeans.

Optical characteristics of the undamaged and laser damaged K9 glass in terahertz band

K9 glass is an important optical element of the high energy laser system. In an attempt to investigate the variation of optical properties of K9 glass before and after laser induced damage, based on the transmission terahertz time-domain spectral system, the undamaged and the damaged K9 glasses induced by laser were tested to obtain the terahertz time-domain and frequency-domain spectra and the refractive index and absorption coefficient of the samples were calculated in terahertz band for further analysis. Results show that with the increase of laser energy, from the undamaged to the damaged K9 glasses to different extent, the peak to peak value in time domain and the amplitude value in frequency domain present a decreasing trend in varying degrees, and in the 0.4 ∼ 0.5 THz band, the absorption coefficient increases continuously with a slight change of refractive index, which indicates that K9 glass not only appears the melting and fracture in the macroscopic morphology after laser-induced damage, but also forms non-bridged oxygen atoms in its microstructure with more ion bonds and free electrons and the change of ion polarization. In this regard, this paper provides a technical basis for exploring the micro characteristics of K9 glass after laser-induced damage, and lays a good foundation for the application of terahertz time-domain spectroscopy to laser-induced damage analysis.

Investigations of dielectric properties of wolframite A0.5Zr0.5NbO4 ceramics by bond theory and far-infrared spectroscopy

In this study, intrinsic dielectric properties of monoclinic A0.5Zr0.5NbO4 ceramics (A = Mn, Zn, Mg, Co, Ni) with wolframite structure were investigated by chemical bond theory, infrared and terahertz spectroscopy. According to bond theory, Zr–O bonds and A-O bonds with larger bond susceptibility values provide majority contributions to relative permittivity; while Nb–O bonds dominate the structural stability and oxygen polyhedron distortion. Intrinsic dielectric properties calculated via far-infrared reflectivity spectroscopy are close to experimental properties, which indicates that most of dielectric contributions come from absorptions of structural phonon oscillation and infrared modes of dielectric response in A0.5Zr0.5NbO4 ceramics vary with A-site ions. Furthermore, intrinsic properties estimated by terahertz time domain spectra also testify the effectiveness of theoretical dielectric properties.

Investigation of aging characteristics in explosive using terahertz time-domain spectroscopy

The terahertz absorption spectrum of the five aging explosive samples (PETN, RDX, HMX, LLM-105 and TATB) was measured and calculated by Terahertz time-domain spectroscopy system (THz-TDS) and air-biased coherent detection system (ZAP-ABCD), respectively. In this paper, compared with the unaging explosive, each aging explosive sample’s terahertz time-domain spectra were obtained and the terahertz absorption spectra were calculated by using Fourier transform and Lambert’s law. The results show that there are several terahertz absorption peaks which were called “fingerprint spectra” for different aging explosive samples in the range of 0.3–6.0 THz spectrum. Meanwhile, the results also show that the locations of the characteristic absorption peaks are not the same. Moreover, the unaging and aging explosive samples have obviously different terahertz absorption spectra.

Quantitative analysis of alum based on Terahertz time-domain spectroscopy technology and Support vector machine

In this paper, the Terahertz time-domain spectroscopy (THz-TDS) technique was used to quantitatively analyze the alum proportion in sweet potato starch. Terahertz time-domain spectra and frequency-domain spectra of pure sweet potato starch and pure alum and their mixtures were obtained. The absorption coefficient spectrum and refractive index spectrum of the samples were calculated. Results show that alum exhibit unique absorption peaks in the terahertz range, and can thus be identified with its absorption fingerprints in the detection from Sweet potato starch. Differences in the refractive index can also be observed between the alum and the sweet potato starch. The Partial Least Squares (PLS) and the Support vector machine (SVM) algorithm were used to establish the mathematical model between the absorption coefficient and the proportion. Results show that the prediction accuracy can reach 99.8% and 99.9%. Comparing the two predictive models, SVM has a better prediction effect. In all, the THz-TDS system plus SVM method is very promising for the further quantitative analysis of alum in sweet potato starch, with the characteristics of being nondestructive and laborsaving compared to other analytical tools.

Heteroleptic Cu(I) complexes with aromatic diimines and phosphines: Synthesis, structure, photophysical properties and THz time domain spectroscopy

Nine novel copper(I) complexes with diphosphine and diimine ligands, namely [Cu(dpq)(xantphos)]BF4 (1), [Cu(dpq)(xantphos)]I (2), [Cu(dpq)(dppp)]BF4 (3), [Cu(dppz)(dppp)]BF4 (4), [Cu(dppz)(dppp)]I (5), [Cu(dppz)(pop)]I (6), [Cu(dpq)(pop)]I (7), [Cu(dpq)(pop)]Br (8), [Cu(dpq)(pop)]SCN (9) (dpq = pyrazino[2,3-f][1,10]phenanthroline, dppz = dipyrido[3,2-a:2′,3′-c]phenazine, xantphos = 9,9-dimethyl-4,5-bis(diphenylphosphanyl)xanthene, dppp = 1,3-bis(diphenylphosphino)propane, pop = 1,1′-[(Oxydi-2,1-phenylene)]bis[1,1-diphenylphosphine]), were characterized by single crystal X-ray diffraction, IR, elemental analysis, 1H NMR, 31P NMR, fluorescence spectra and terahertz time domain spectroscopy (THz-TDS). These nine complexes were synthesized by the reactions of copper salts, diimine ligands and various of P-donor ligands through one-pot method. Single crystal X-ray diffraction reveals that complex 9 is of a simple mono-nuclear structure while complexes 6 and 7 are of dimer structures. For complex 8, hydrogen bonds and CH⋯π interactions lead to the formation of a 1D infinite chain structure. Interestingly, complexes 1–5 show novel 2D or 3D network structures through CH⋯π interactions. In addition, complexes 1–3 and 6–9 exhibit interesting fluorescence in the solid state at room temperature. Among the nine complexes, complex 1 shows the highest quantum yield up to 37% and the lifetime of 1 is 6.0 μs. The terahertz (THz) time-domain spectra of these complexes were also studied.

Cancellation between auto- and mutual correlation contributions of protein/water dynamics in terahertz time-domain spectra.

Terahertz time-domain spectra (THz-TDS) were investigated using the results of molecular dynamics (MD) simulations of Staphylococcal nuclease at two hydration states in the temperature range between 100 and 300 K. The temperature dependence of THz-TDS was found to differ significantly from that of the incoherent neutron scattering spectra (INSS) calculated from the same MD simulation results. We further examined contributions of the mutual and auto-correlations of the atomic fluctuations to THz-TDS and found that the negative value of the former contribution nearly canceled out the positive value of the latter, resulting in a monotonic increase of the reduced absorption cross section. Because of this cancellation, no distinct broad peak was observed in the absorption lineshape function of THz-TDS, whereas the protein boson peak was observed in INSS. The contribution of water molecules to THz-TDS was extremely large for the hydrated protein at temperatures above 200 K, in which large-amplitude motions of water were excited. The combination of THz-TDS, INSS and MD simulations has the potential to extract function-relevant protein dynamics occurring on the picosecond to nanosecond timescale.

New discovery in crystallography: correlation of terahertz time-domain spectra with crystal structures and photoluminescence properties of mononuclear/binuclear diimine–Cu(i)-phosphine complexes

A method was provided for qualitatively determining the photoluminescence quantum yield of diimine–Cu(i) complex by the waveform of terahertz time-domain spectroscopy.

Wood species identification using terahertz time-domain spectroscopy

A new method for wood identification using terahertz (THz) time-domain spectroscopy (TDS) is presented. THz-TDS technology is an emerging method of spectral analysis method that shows great advantages in wood non-destructive testing. Wood is opaque in the optical and near infrared spectrum, and relatively transparent in the THz band. In this experiment, four wood species (Pinus densiflora, Pinus sylvestris, Pseudotsuga menziesii, and Cunninghamia lanceolata) were selected as samples. Optical parameters such as absorption coefficient and refractive index of the measured samples can be extracted from THz time-domain spectra. Principal component analysis was introduced to reduce the dimensionality of the THz spectral data of the wood. The principal component score matrix of the wood was obtained and used as a characteristic matrix. A wood classification model was established using a support vector machine. The experimental results showed that THz-TDS had excellent performance in wood recognition. This research lays a theoretical and application foundation for the application of THz technology in the field of wood detection and recognition. It also provides a new direction and technical scheme in the wood identification field.