Quantitative analysis of bisphenol analogue mixtures by terahertz spectroscopy using machine learning method

Green construction of magnetic azo porous organic polymer for highly efficient enrichment and detection of phenolic endocrine disruptors

Porous organic polymers (POPs) are prominent sorbents for effective extraction of endocrine disrupting chemicals (EDCs). However, green and sustainable construction of functional POPs is still challenging. Herein, we developed a magnetic azo POP (Mazo-POP) for the first time using hydroxy-rich natural kaempferol and low-toxic basic fuchsin as monomers through a diazo coupling reaction. The Mazo-POP exhibited excellent extraction capabilities for EDCs with a phenolic structure. Consequently, it was used as a magnetic sorbent for extracting phenolic EDCs from water and fish samples, followed by ultrahigh-performance liquid chromatography-tandem mass spectrometric detection. The Mazo-POP based analytical method afforded a good linearity of 0.06-100 ng mL-1 and 0.3-500 ng g-1 for water and fish samples respectively, with detection limits (S/N = 3) of 0.02-0.5 ng mL-1 and 0.1-1.5 ng g-1, respectively. The method recovery was from 85.2% to 109% and relative standard deviation was less 5.3%. Moreover, the effective adsorption was mainly contributed by hydrogen bond, π-π interaction, pore filling and hydrophobic interaction. This work not only provides an efficient method for sensitive determination of phenolic EDCs, but also highlights the significance of green preparation of environmentally friendly sorbents for enriching/adsorbing pollutants.

Detection of biomarkers using terahertz metasurface sensors and machine learning

To achieve classification and concentration detection of cancer biomarkers, we propose a method that combines terahertz (THz) spectroscopy, metasurface sensors, and machine learning. A metasurface sensor suitable for biomarker detection was designed and fabricated with five resonance frequencies in the range of 0.3-0.9 THz. We collected biomarkers of five types and nine concentrations at 100 sets of time-domain spectra per concentration. The spectrum is processed by noise reduction and fast Fourier transform to obtain the frequency-domain spectrum. Five machine learning algorithms are used to analyze time- and frequency-domain spectra and ascertain which algorithm is more suitable for the classification of the biomarker THz spectrum. Experimental results show that random forest can better distinguish five biomarkers with an accuracy of 0.984 for the time-domain spectrum. For the frequency-domain spectrum, the support vector machine performs better, with an accuracy of 0.989. For biomarkers at different concentrations, we used linear regression to fit the relationship between biomarker concentration and frequency shift. Experimental results show that machine learning can distinguish different biomarker species and their concentrations by the THz spectrum. This work provides an idea and data processing method for the application of THz technology in biomedical detection.

Chemometric Analysis of a Ternary Mixture of Caffeine, Quinic Acid, and Nicotinic Acid by Terahertz Spectroscopy

Caffeine, quinic acid, and nicotinic acid are among the significant chemical determinants of coffee quality. This study develops a chemometric model to quantify these compounds in ternary mixtures analyzed by terahertz time-domain spectroscopy (THz-TDS). A data set of 480 THz spectra was obtained from 80 samples. Combinations of data preprocessing methods, including normalization (Z-score, min-max scaling, Mie baseline removal) and dimensionality reduction (principal component analysis (PCA), factor analysis (FA), independent component analysis (ICA), locally linear embedding (LLE), non-negative matrix factorization (NMF), isomap), and prediction models (partial least-squares regression (PLSR), support vector regression (SVR), multilayer perceptron (MLP), convolutional neural network (CNN), gradient boosting) were analyzed for their prediction performance (totaling to 4,711,685 combinations). Results show that the highest quantification performance was achieved at a root-mean-square error of prediction (RMSEP) of 0.0254 (dimensionless mass ratio), using min-max scaling and factor analysis for data preprocessing and multilayer perceptron for prediction. Effects of preprocessing, comparison of prediction models, and linearity of data are discussed.

Exploiting total internal reflection geometry for deep broadband terahertz modulation using a GaAs Schottky diode with integrated subwavelength metal microslits

We developed a GaAs Schottky diode with integrated periodic subwavelength metal microslits with total internal reflection (TIR) geometry to achieve deep broadband THz modulation at high frequency with low insertion loss. The non-resonant electric field enhancement effect in the subwavelength microslits intensifies the evanescent wave in TIR, which increases broadband absorbance of THz light signals by free carriers in the GaAs Schottky diode. Devices with various microslit spatial periods and gap widths were fabricated and measured. Among the devices, that with a microslit period of 10 µm and gap width of 2 µm produced ∼70% modulation depth at frequencies of 0.2 to 1.2 THz, while in the range of 0.25 to 0.4 THz, ∼90% modulation depth was achieved. By encapsulating the device in high refractive index material, ∼100% modulation depth was achieved in the range of 0.4 to 0.6 THz, the 3 dB bandwidth operational frequency was ∼160 kHz, and the insertion loss introduced by the device was less than 8 dB, which is much lower than existing metasurface-based THz modulators. In general, our first-generation device has improved modulation depth, operational bandwidth, insertion loss, and operational frequency. Optimization of the metal microslits, TIR geometry, and doped layer could further improve the performance of our design.

Terahertz spectroscopy of temperature-induced transformation between glutamic acid, pyroglutamic acid and racemic pyroglutamic acid

Under heating conditions, L-Glutamic acid (L-Glu) can be dehydrated to form L-pyroglutamic acid (L-PGA), and L-PGA can racemize to form DL-PGA. Here, we characterized this transformation at different temperatures and times by terahertz time domain spectroscopy (THz-TDS). By Powder X-ray diffraction (PXRD), the validity of THz spectroscopy is verified. The results prove that the reaction rate of dehydration and racemization is significantly affected by temperature. The THz spectra divided the reactions into three stages. At 150-155 °C, the reaction changes drastically. Furthermore, we found that the absorption intensity at 0.97 and 1.55 THz has a good dependence on the reaction temperature and time, showing a non-linear relationship (R² > 0.98). Our findings suggest that the chemical transformation and reaction rate can be sensitively probed by terahertz spectroscopy, which provides a potential method for the quantitative analysis of reaction products.

Ultrahigh sensitivity nitrogen-doping carbon nanotubes-based metamaterial-free flexible terahertz sensing platform for insecticides detection

With the rapid advances in terahertz (THz) spectroscopy, metamaterial-free THz sensors have been of importance due to efficient cost, high sensitivity and overcoming the limited tunability of the optical constants of metals. Here, a metamaterial-free and flexible THz sensor based on nitrogen-doping carbon nanotubes (N-CNTs) coupled with signal-enhancing Au NPs was proposed for detecting nereistoxin-related insecticides (NRIs). Sensitivity and selectivity for NRIs detection have been realized over the range of 3.3-100 μg/L with good linear fitting (R² ≥ 0.9003) and LOD was 1.33 μg/L. Accuracy was validated by the recovery rates of 105.87-109.75% of NRI in spiked food-matrix sample. These results indicated the developed signal-enhancing THz method, validated by LC-MS/MS, exhibited high sensitivity and simplicity detection, which has noteworthy potential for applications in food safety and environment monitoring.

Quantitative determination of acacia honey adulteration by terahertz-frequency dielectric properties as an alternative technique

The dielectric characteristics in the terahertz region contribute to a revealing insight into the material components and provide intermolecular information. The dielectric properties of adulterated honey, described as the real and imaginary parts of the complex dielectric constant (Re[ε] and Im[ε]), were obtained from 0.3 to 1.5 THz. The relationship between invert syrup proportions and complex dielectric constants at different frequencies implied the possibility of using the dielectric property as an indicator of honey authenticity. The selected effective dielectric variables of Re[ε] and Im[ε] and their combination were chosen by stability competitive adaptive reweighted sampling (SCARS) algorithm and then used to establish PLS models. The accuracy and uncertainty result revealed SCARS-PLS model based on the combination of Re[ε] and Im[ε] is the best model relatively. These findings indicated the potential utility of this rapid, non-destructive, and on-site method for authenticity verification.

Simultaneous quantitative determination of low-concentration ternary pesticide mixtures in wheat flour based on terahertz spectroscopy and BPNN

Terahertz spectroscopy has been widely applied in the quantitative analysis of pesticides, however, it still encounters challenge pursuing high prediction accuracy in multi-component mixtures analysis with ultra-low concentration. Here, back propagation neural network (BPNN) was applied on the determination of ternary pesticide mixtures in wheat flour. By spectral pre-processing and model parameter optimization, high-quality spectra and complete network frame was achieved. On this basis, a novel wavelength selection method was presented and the most efficient peak width was given. Our method here achieved the optimal results, the correlation coefficient of prediction sets (R_P) were 0.9913, 0.9948, 0.9923, and corresponding root mean square error (RMSE) were 0.0211%, 0.0176%, 0.0191%. More importantly, the concentration of pesticides in this study was extremely low compared with similar quantitative analysis based on terahertz spectroscopy, which can promote the application of this technology into actual production.

Enhanced sensitivity of dilute aqueous adrenaline solution with an asymmetric hexagonal ring structure in the terahertz frequencies

Quantitative detection of neurotransmitters in aqueous environment is crucial for the early diagnosis of many neurological disorders. Terahertz waves, as a non-contact and non-labeling tool, have demonstrated large potentials in quantitative biosensing. Although the detection of trace-amount analyte has been achieved with terahertz metamaterials in the recent decades, most studies have been focused on dried samples. Here, a hexagonal asymmetric metamaterial sensor was designed and fabricated for aqueous solution sensing with terahertz waves in the reflection geometry. An absorption enhancement of 43 was determined from the simulation. Dilute adrenaline solutions ranging from 30 µM to 0.6 mM were measured on our sensor using a commercial terahertz time-domain spectroscopy system, and the effective absorption was found to be linearly correlated with the concentration (R²= 0.81). Furthermore, we found that as the concentration becomes higher (>0.6 mM), a non-linear relationship starts to take place, which confirmed the previous theory on the extended solvation shell that can be probed on the picosecond scale. Our sensor, without the need of high-power and stable terahertz sources, has enabled the detection of subtle absorption changes induced by the solvation dynamics.

Molecularly imprinted polymer-based electrochemical sensor for the determination of endocrine disruptor bisphenol-A in bovine milk

Bisphenol A (BPA) contamination from food packaging material has been a major concern in recent years, due to its potential endocrine-disrupting effects on humans, especially infants and children. This paper reports the detection of BPA using an electrochemical sensor based on molecularly imprinted polymer (MIP). Electrochemically reduced graphene oxide coated glassy carbon electrode used for this study. Density functional theory (DFT) at B3LYP/6-31 + G (d,p) level was used to calculate the molecular-level interaction between BPA and MIP. The pyrrole electrochemically polymerized in the presence of template molecule BPA on the electrode surface. BPA imprinted cavities were formed by removing entrapped BPA molecules from the polypyrrole film. MIP electrode was used for the determination of BPA in standard and real samples by differential pulse voltammetry. The peak current shows the linear relationship to the logarithmic concentration of BPA between 750 and 0.5 nmolL^-1 with a correlation coefficient, R² = 0.992. The limit of detection was found to be 0.2 nmolL^-1 (S/N = 3). The reproducibility and repeatability of the sensor were also studied.