Simple and sensitive determination of Cr (III), Cu (II) and Pb (II) in tea infusions using AgNPs-modified resin combined with laser-induced breakdown spectroscopy.

The detection of heavy metals in tea infusions is important because of the potential health risks associated with their consumption. Existing highly sensitive detection methods pose challenges because they are complicated and time-consuming. In this study, we developed an innovative and simple method using Ag nanoparticles-modified resin (AgNPs-MR) for pre-enrichment prior to laser-induced breakdown spectroscopy for the simultaneous analysis of Cr (III), Cu (II), and Pb (II) in tea infusions. Signal enhancement using AgNPs-MR resulted in amplification with limits of detection of 0.22 µg L-1 for Cr (III), 0.33 µg L-1 for Cu (II), and 1.25 µg L-1 for Pb (II). Quantitative analyses of these ions in infusions of black tea from various brands yielded recoveries ranging from 83.3% to 114.5%. This method is effective as a direct and highly sensitive technique for precisely quantifying trace concentrations of heavy metals in tea infusions.

Plasma parameters correction method based on plasma image-spectrum fusion for matrix effect elimination in LIBS.

Matrix effect is one of the obstacles that hinders the rapid development of laser-induced breakdown spectroscopy (LIBS), and it is currently a hot, challenging, and focal point in research. To eliminate the matrix effect, this study proposed a plasma parameters correction method based on plasma image-spectrum fusion (PPC-PISF). This method corrects the total number density, plasma temperature, and electron number density variations caused by matrix effect using effective features in plasma images and spectra. To verify the feasibility of this method, experiments were conducted on pressed and metal samples, and the results were compared with those corrected by image-assisted LIBS (IA-LIBS). For the pressed samples, after correction by PPC-PISF, the R2 of the calibration curves all improved to above 0.993, the average root-mean-square error (RMSE) decreased by 41.05%, and the average relative error (ARE) decreased by 59.35% evenly in comparison to IA-LIBS. For the metal samples, after correction by PPC-PISF, the R2 of the calibration curves all increased to above 0.997. Additionally, the RMSE decreased by 29.63% evenly, the average ARE decreased by 38.74% compared to IA-LIBS. The experimental results indicate that this method is an effective method for eliminating the matrix effect, promoting the further development of LIBS in industrial detection.

Sensitivity and stability improvement on slippery surface-aggregated substrate for trace heavy metals detection using NELIBS.

The sensitive and stable detection of trace heavy metals in liquid is crucial given its profound impact on various aspects of human life. Currently, nanoparticle-enhanced laser-induced breakdown spectroscopy (NELIBS) with dried droplet method (DDM) is widely applied for heavy metals detection. Nevertheless, the coffee ring effect (CRE) in DDM affects the stability, accuracy, and sensitivity of NELIBS. Here, we developed a slippery surface-aggregated substrate (SS substrate) to suppress the CRE and enrich analytes, and form a plasmonic platform for NELIBS detection. The SS substrate was prepared by infiltrating perfluorinated lubricant into the pores of PTFE membrane. The droplet, with targeted elements and gold nanoparticles, was dried on the SS substate to form the plasmonic platform for NELIBS analysis. Then, trace heavy metal elements copper (Cu) and manganese (Mn) were analyzed by NELIBS. The results of Cu (RSD = 5.60%, LoD = 3.72 µg/L) and Mn (RSD = 7.42%, LoD = 6.37 µg/L), illustrated the CRE suppression and analytes enrichment by the SS substrate. The results verified the realization of stable, accurate and sensitive NELIBS detection. And the LoDs succeeded to reach the standard limit of China (GB/T 14848-2017). Furthermore, the results for groundwater detection (relative error: 5.92% (Cu) and 4.74% (Mn)), comparing NELIBS and inductively coupled plasma mass spectrometry (ICP-MS), validated the feasibility of the SS substrate in practical applications. In summary, the SS substrate exhibits immense potential for practical application such as water quality detection and supervision.

All laser direct writing process for temperature sensor based on graphene and silver.

A highly sensitive temperature sensing array is prepared by all laser direct writing (LDW) method, using laser induced silver (LIS) as electrodes and laser induced graphene (LIG) as temperature sensing layer. A finite element analysis (FEA) photothermal model incorporating a phase transition mechanism is developed to investigate the relationship between laser parameters and LIG properties, providing guidance for laser processing parameters selection with laser power of 1-5 W and laser scanning speed (greater than 50 mm/s). The deviation of simulation and experimental data for widths and thickness of LIG are less than 5% and 9%, respectively. The electrical properties and temperature responsiveness of LIG are also studied. By changing the laser process parameters, the thickness of the LIG ablation grooves can be in the range of 30-120 µm and the resistivity of LIG can be regulated within the range of 0.031-67.2 Ω·m. The percentage temperature coefficient of resistance (TCR) is calculated as - 0.58%/°C. Furthermore, the FEA photothermal model is studied through experiments and simulations data regarding LIS, and the average deviation between experiment and simulation is less than 5%. The LIS sensing samples have a thickness of about 14 µm, an electrical resistivity of 0.0001-100 Ω·m is insensitive to temperature and pressure stimuli. Moreover, for a LIS-LIG based temperature sensing array, a correction factor is introduced to compensate for the LIG temperature sensing being disturbed by pressure stimuli, the temperature measurement difference is decreased from 11.2 to 2.6 °C, indicating good accuracy for temperature measurement.

Staging classification of omicron variant SARS-CoV-2 infection based on dual-spectrometer LIBS (DS-LIBS) combined with machine learning.

Effective differentiation of the infection stages of omicron can provide significant assistance in transmission control and treatment strategies. The combination of LIBS serum detection and machine learning methods, as a novel disease auxiliary diagnostic approach, has a high potential for rapid and accurate staging classification of Omicron infection. However, conventional single-spectrometer LIBS serum detection methods focus on detecting the spectra of major elements, while trace elements are more closely related to the progression of COVID-19. Here, we proposed a rapid analytical method with dual-spectrometer LIBS (DS-LIBS) assisted with machine learning to classify different infection stages of omicron. The DS-LIBS, including a broadband spectrometer and a narrowband spectrometer, enables synchronous collection of major and trace elemental spectra in serum, respectively. By employing the RF machine learning models, the classification accuracy using the spectra data collected from DS-LIBS can reach 0.92, compared to 0.84 and 0.73 when using spectra data collected from single-spectrometer LIBS. This significant improvement in classification accuracy highlights the efficacy of the DS-LIBS approach. Then, the performance of four different models, SVM, RF, IGBT, and ETree, is compared. ETree demonstrates the best, with cross-validation and test set accuracies of 0.94 and 0.93, respectively. Additionally, it achieves classification accuracies of 1.00, 0.92, 0.92, and 0.89 for the four stages B1-acute, B1-post, B2, and B3. Overall, the results demonstrate that DS-LIBS combined with the ETree machine learning model enables effective staging classification of omicron infection.

Visualization of laser-induced breakdown spectroscopy data of mouse organs based on the feature extraction method.

At present, there is no comprehensive and systematic research on laser-induced breakdown spectroscopy (LIBS) data visualization. In particular, the LIBS spectra of biological samples have large noise and weak signals, which seriously affect the feature visualization. Here, three commonly used sample visualization methods were compared, and a new method was applied for tissue sample visualization. We used the LIBS mapping technique to obtain LIBS spectra of different organ slice samples from mice. LIBS spectral distribution was visualized after extracting the region of interest. The three spectral visualization methods are compared, and the performance of visualization algorithms is quantitatively analyzed. The potential of heat-diffusion for the affinity-based transition embedding (PHATE) method highlights the details of the LIBS spectral distribution while maintaining the overall structure of the data. The correlation coefficient between dimensionality reduction data and raw data is 0.97, and the average distance between samples of different categories is 0.64, which are superior to those of traditional principal component analysis (PCA), multidimensional scaling (MDS), and t-distributed stochastic neighbor embedding (t-SNE). The results show that the PHATE method can serve as a very promising LIBS spectral visualization tool.

Line intensity calculation of laser-induced breakdown spectroscopy during plasma expansion in nonlocal thermodynamic equilibrium.

We present a new, to the best of our knowledge, simulation method for laser-induced breakdown spectroscopy during the plasma expansion phase in nonlocal thermodynamic equilibrium. Our method uses the particle-in-cell/Monte Carlo collision model to calculate dynamic processes and line intensity of nonequilibrium laser-induced plasma (LIP) in the afterglow phase. The effects of ambient gas pressure and type on LIP evolution are investigated. This simulation provides an added way to understand the nonequilibrium processes in more detail than the current fluid and collision radiation models. Our simulation results are compared with experimental and SimulatedLIBS package results and show good agreement.

Drying Acoustically Levitated Droplets as Signal-Amplifying Platforms for Ultrasensitive and Multimode Laser Sensing.

Ultrasensitive sensing to trace atomic and molecular analytes has gained interest for its intimate relation to industrial sectors and human lives. One of the keys to ultrasensitive sensing for many analytical techniques lies in enriching trace analytes onto well-designed substrates. However, the coffee ring effect, nonuniform distribution of analytes onto substrates, in the droplet drying process hinders the ultrasensitive and stable sensing onto the substrates. Here, we propose a substrate-free strategy to suppress the coffee ring effect, enrich analytes, and self-assemble a signal-amplifying (SA) platform for multimode laser sensing. The strategy involves acoustically levitating and drying a droplet, mixed with analytes and core-shell Au@SiO2 nanoparticles, to self-assemble an SA platform. The SA platform with a plasmonic nanostructure can dramatically enrich analytes, enabling enormous spectroscopic signal amplification. Specifically, the SA platform can promote atomic detection (cadmium and chromium) to the 10-3 mg/L level by nanoparticle-enhanced laser-induced breakdown spectroscopy and can promote molecule detection (rhodamine 6G) to the 10-11 mol/L level by surface-enhanced Raman scattering. All in all, the SA platform, self-assembled by acoustic levitation, can intrinsically suppress the coffee ring effect and enrich trace analytes, enabling ultrasensitive multimode laser sensing.

Visualization and accuracy improvement of soil classification using laser-induced breakdown spectroscopy with deep learning.

Deep learning method is applied to spectral detection due to the advantage of not needing feature engineering. In this work, the deep neural network (DNN) model is designed to perform data mining on the laser-induced breakdown spectroscopy (LIBS) spectra of the ore. The potential of heat diffusion for an affinity-based transition embedding model is first used to perform nonlinear mapping of fully connected layer data in the DNN model. Compared with traditional methods, the DNN model has the highest recognition accuracy rate (75.92%). A training set update method based on DNN output is proposed, and the final model has a recognition accuracy of 85.54%. The method of training set update proposed in this work can not only obtain the sample labels quickly but also improve the accuracy of deep learning models. The results demonstrate that LIBS combined with the DNN model is a valuable tool for ore classification at a high accuracy rate.

Individual Micron-Sized Aerosol Qualitative Analysis-Combined Raman Spectroscopy and Laser-Induced Breakdown Spectroscopy by Optical Trapping in Air.

The attribution of single particle sources of atmospheric aerosols is an essential problem in the study of air pollution. However, it is still difficult to qualitatively analyze the source of a single aerosol particle using noncontact in situ techniques. Hence, we proposed using optical trapping to combine gated Raman spectroscopy with laser-induced breakdown spectroscopy (LIBS) in a single levitated micron aerosol. The findings of the spectroscopic imaging indicated that the particle plasma formed by a single particle ablation with a pulsed laser within 7 ns deviates from the trapped particle location. The LIBS acquisition field of view was expanded using the 19-bundle fiber, which also reduces the fluctuation of a single particle signal. In addition, gated Raman was utilized to suppress the fluorescence and increase the Raman signal-to-noise ratio. Based on this, Raman can measure hard-to-ionize substances with LIBS, such as sulfates. The LIBS radical can overcome the restriction that Raman cannot detect ionic chemicals like fluoride and chloride in halogens. To test the capability of directly identifying distinctive feature compounds utilizing spectra, we detected anions using Raman spectroscopy and cations using LIBS. Four typical mineral aerosols are subjected to precise qualitative evaluations (marble, gypsum, baking soda, and activated carbon adsorbed potassium bicarbonate). To further validate the application potential for substances with indistinctive feature discrimination, we employed machine learning algorithms to conduct a qualitative analysis of the coal aerosol from ten different origin regions. Three data fusion methodologies (early fusion, intermediate fusion, and late fusion) for Raman and LIBS are implemented, respectively. The accuracy of the late fusion model prediction using StackingClassifier is higher than that of the LIBS data (66.7%) and Raman data (86.1%) models, with an average accuracy of 90.6%. This research has the potential to provide online single aerosol analysis as well as technical assistance for aerosol monitoring and early warning.

Self-absorption correction method for one-point calibration laser-induced breakdown spectroscopy.

As an important variant of calibration-free laser-induced breakdown spectroscopy (CF-LIBS), one-point calibration LIBS (OPC-LIBS) corrects the Boltzmann plot of the unknown sample by using one known sample and obtains higher quantitative accuracy than CF-LIBS. However, the self-absorption effect restricts its accuracy. In this work, a new self-absorption correction (SAC) method for OPC-LIBS is proposed to solve this problem. This method uses an algorithm to correct the self-absorption and does not require the calculation of the self-absorption coefficient. To verify the effectiveness of this SAC method, Ti, V, and Al elements in two titanium alloys were determined by classical OPC-LIBS and OPC-LIBS with SAC. The average relative errors (AREs) of all elements in the two samples were decreased from 8.78% and 9.28% to 8.07% and 7.56%, respectively. The results demonstrated the effectiveness of this SAC method for OPC-LIBS.

Time-Resolved Laser-Induced Breakdown Spectroscopy for Accurate Qualitative and Quantitative Analysis of Brown Rice Flour Adulteration.

To solve the adulteration problem of brown rice flour in the commodity market, a novel, accurate, and stable detection method based on time-resolved laser-induced breakdown spectroscopy (TR-LIBS) is proposed. Qualitative and quantitative analysis was used to detect five adulterants and seven different adulterant ratios in brown rice flour. Being able to excavate more information from plasma by obtaining time-resolved spectra, TR-LIBS has a stronger performance, which has been further verified by experiments. For the qualitative analysis of adulterants, the traditional machine learning models based on TR-LIBS, linear discriminant analysis (LDA), naïve Bayes (NB) and support vector machine (SVM) have significantly better classification accuracy than those based on traditional LIBS, increasing by 3-11%. The deep learning classification model based on TR-LIBS also achieved the same results, with an accuracy increase of more than 8%. For the quantitative analysis of the adulteration ratio, compared with traditional LIBS, the quantitative model based on TR-LIBS reduces the limit of detection (LOD) of five adulterants from about 8-51% to 4-19%, which effectively improves the quantitative detection performance. Moreover, t-SNE visualization proved that there were more obvious boundaries between different types of samples based on TR-LIBS. These results demonstrate the great prospect of TR-LIBS in the identification of brown rice flour adulteration.

A plasma image-spectrum fusion correction strategy for improving spectral stability based on radiation model in laser induced breakdown spectroscopy.

Spectral fluctuation is one of the main obstacles affecting the further development of LIBS, and it is also the current research hotspot and difficulty. To meet the requirements of industrial monitoring, a novel method named plasma image-spectrum fusion laser induced breakdown spectroscopy (PISF-LIBS) was proposed to correct the spectral fluctuation and improve the quantitative accuracy. In this method, by systematically analyzing the spectral radiation model, six main factors affecting the spectral stability were obtained. Further, the standard spectrum in the ideal plasma state which is not affected by these six factors was calculated, and the deviation from the actual spectrum was obtained. According to the above analysis, the calculated deviation was mainly affected by these six factors and can be estimated through them. Therefore, this study creatively proposed to use the effective information in the plasma images and spectra to indirectly characterize the deviation, so as to realize the correction of spectral fluctuation. To verify the wide applicability of PISF-LIBS in experimental conditions, the LIBS spectra of aluminum alloy obtained under four different experimental conditions were used. After PISF-LIBS correction, the R2 increased to more than 0.974, and the RMSE, MAPE and RSD of the prediction set decreased by 44.789%, 47.854% and 51.687% on average. To further verify the wide applicability of PISF-LIBS in experimental samples, alloy steel samples and pressed samples were also used. For alloy steel samples, after PISF-LIBS correction, the R2 increased to more than 0.996, and the RMSE, MAPE and RSD of the prediction set decreased by 48.337%, 52.856% and 25.819% evenly. For pressed samples, the R2 increased over 0.992, and the RMSE, MAPE and RSD of the prediction set decreased by 61.493%, 61.080% and 39.945% averagely. The experimental results prove the effectiveness and wide applicability of PISF-LIBS in spectral fluctuation correction.

Laser ultrasonic improvement and its application in defect detection based on the composite coating method.

Herein, we studied the increasing tendency of photoacoustic (PA) conversion efficiency of the Au/polydimethylsiloxane (PDMS) composite. The thickness of the Au layer was optimized by modeling the PA process based on the Drude-Lorentz model and finite element analysis method, and corresponding results were verified. The results showed that the optimal Au thickness of the Au/PDMS composite was 35 nm. Finally, the Au/PDMS composites were coated onto the surface of aluminum alloys, which improved the thermoelastic laser ultrasonic (LU) signals to near 100 times. Besides, the defect mapping was performed by thermoelastic LU signals with Au/PDMS coating and ablation LU signals without coating; the Pearson correlation coefficient was higher than 0.95. The application in the defect detection in metal could provide guides for nondestructive detection on metals by laser ultrasound.

Laser-Induced Fast Assembly of Wettability-Finely-Tunable Superhydrophobic Surfaces for Lossless Droplet Transfer.

Rose-petal-like superhydrophobic surfaces with strong water adhesion are promising for microdroplet manipulation and lossless droplet transfer. Assembly of self-grown micropillars on shape-memory polymer sheets with their surface adhesion finely tunable was enabled using a picosecond laser microprocessing system in a simple, fast, and large-scale manner. The processing speed of the wettability-finely-tunable superhydrophobic surfaces is up to 0.5 cm2/min, around 50-100 times faster than the conventional lithography methods. By adjusting the micropillar height, diameter, and bending angle, as well as superhydrophobic chemical treatment, the contact angle and adhesive force of water droplets on the micropillar-textured surfaces can be tuned from 117.1° up to 165° and 15.4 up to 200.6 µN, respectively. Theoretical analysis suggests a well-defined wetting-state transition with respect to the micropillar size and provides a clear guideline for microstructure design for achieving a stabilized superhydrophobic region. Droplet handling devices, including liquid handling tweezers and gloves, were fabricated from the micropillar-textured surfaces, and lossless liquid transfer of various liquids among various surfaces was demonstrated using these devices. The superhydrophobic surfaces serve as a microreactor platform to perform and reveal the chemical reaction process under a space-constrained condition. The superhydrophobic surfaces with self-assembled micropillars promise great potential in the fields of lossless droplet transfer, biomedical detection, chemical engineering, and microfluidics.

Accurate Identification and Quantification of Chinese Yam Powder Adulteration Using Laser-Induced Breakdown Spectroscopy.

As a popular food, Chinese yam (CY) powder is widely used for healthy and commercial purposes. Detecting adulteration of CY powder has become essential. In this work, chemometric methods combined with laser-induced breakdown spectroscopy (LIBS) were developed for identification and quantification of CY powder adulteration. Pure powders (CY, rhizome of winged yam (RY) and cassava (CS)) and adulterated powders (CY adulterated with CS) were pressed into pellets to obtain LIBS spectra for identification and quantification experiments, respectively. After variable number optimization by principal component analysis and random forest (RF), the best model random forest-support vector machine (RF-SVM) decreased 48.57% of the input variables and improved the accuracy to 100% in identification. Following the better feature extraction method RF, the Gaussian process regression (GPR) method performed the best in the prediction of the adulteration rate, with a correlation coefficient of prediction (Rp2) of 0.9570 and a root-mean-square error of prediction (RMSEP) of 7.6243%. Besides, the variable importance of metal elements analyzed by RF revealed that Na and K were significant due to the high metabolic activity and maximum metal content of CY powder, respectively. These results demonstrated that chemometric methods combined with LIBS can identify and quantify CY powder adulteration accurately.

Insertable, Scabbarded, and Nanoetched Silver Needle Sensor for Hazardous Element Depth Profiling by Laser-Induced Breakdown Spectroscopy.

Sensing of hazardous metals is urgent in many areas (e.g., water pollution and meat products) as heavy metals threaten people's health. Laser-induced breakdown spectroscopy (LIBS), as a rapid, in situ, and multielemental analytical technique, has been widely utilized in rapid hazardous heavy metal sensing. However, loose and water-containing samples (e.g., meat, plant, and soil) are hard to analyze by LIBS directly, and heavy metal depth profiling for bulk samples remains suspenseful. Here, inspired by the Needle, the sword of Arya Stark in Game of Thrones, we propose an insertable, scabbarded, and nanoetched silver (NE-Ag) needle sensor for rapid hazardous element sensing and depth profiling. The NE-Ag needle sensor features a micro-nanostructure surface for inserting into the bulk sample and absorbing hazardous analytes. For accurate elemental depth profiling, we design a stainless-steel scabbard to wrap and protect the NE-Ag needle from pollution (unexpected contaminant absorption) during the needle insertion and extraction process. The results for cadmium (Cd) show that the relative standard deviation equals to 6.7% and the limit of detection reaches 0.8 mg/L (ppm). Furthermore, the correlations (Pearson correlation coefficient) for Cd and chromium (Cr) depth profiling results are no less than 0.96. Furthermore, the total testing time could be less than 1 h. All in all, the insertable and scabbarded NE-Ag needle senor has high potential in rapid hazardous heavy metal depth profiling in different industries.

Time-resolved spectral-image laser-induced breakdown spectroscopy for precise qualitative and quantitative analysis of milk powder quality by fully excavating the matrix information.

A novel and effective method named time-resolved spectral-image laser-induced breakdown spectroscopy (TRSI-LIBS) was proposed to achieve precise qualitative and quantitative analysis of milk powder quality. To verify the feasibility of TRSI-LIBS, qualitative and quantitative analysis of milk powder quality was carried out. For qualitative analysis of foreign protein adulteration, the accuracy of models based on TRSI-LIBS was higher than those based on LIBS, with an accuracy improvement of about 5% to 10%. For the quantitative analysis of foreign protein adulteration and element content, the quantitative analysis models based on TSRI-LIBS also had better effect. For instance, limit of detection (LOD),determination coefficient of prediction (R2p), root-mean-square error of prediction (RMSEP) and average relative error of prediction (AREP) of quantitative model of calcium (Ca) content based on TRSI-LIBS improved from 1.47 mg/g, 0.95, 0.35 mg/g and 23.29% to 0.81 mg/g, 0.98, 0.20 mg/g and 12.60%.

Accuracy improvement of single-sample calibration laser-induced breakdown spectroscopy with self-absorption correction.

The single sample calibration laser-induced breakdown spectroscopy (SSC-LIBS) is quite suitable for the fields where the standard sample is hard to obtain, including space exploration, geology, archaeology, and jewelry identification. But in practice, the self-absorption effect of plasma destroys the linear relationship of spectral intensity and element concentration based on the Lomakin-Scherbe formula which is the guarantee of the high accuracy of the SSC-LIBS. Thus, the self-absorption effect limits the quantitative accuracy of SSC-LIBS greatly. In this work, an improved SSC-LIBS with self-absorption correction (SSC-LIBS with SAC) is proposed for the promotion of quantitative accuracy of SSC-LIBS. The SSC-LIBS with SAC can correct the intensity ratio of spectral lines in the calculation of SSC-LIBS through relative self-absorption coefficient K without complicated preparatory information. The alloy samples and pressed ore samples were used to verify the effect of the SSC-LIBS with SAC. Compared with SSC-LIBS, for alloy samples, the average RMSEP and average ARE of SSC-LIBS with SAC decreased from 0.83 wt.% and 13.75% to 0.40 wt.% and 4.06%, respectively. For the pressed ore samples, the average RMSEP and average ARE of SSC-LIBS with SAC decreased from 4.77 wt.% and 90.48% to 2.34 wt.% and 14.60%. The experimental result indicates that SSC-LIBS with SAC has a great improvement of quantitative accuracy and better universality compared with traditional SSC-LIBS, which is a mighty promotion of the wide application of SSC-LIBS.

Classification of steel using laser-induced breakdown spectroscopy combined with deep belief network.

The identification of steels is a crucial step in the process of recycling and reusing steel waste. Laser-induced breakdown spectroscopy (LIBS) coupled with machine learning is a convenient method to classify the types of materials. LIBS can generate characteristic spectra of various samples as input variable for steel classification in real time. However, the performance of classification model is limited to the complex input due to similar chemical composition in samples and nonlinearity problems between spectral intensities and elemental concentrations. In this study, we developed a method of LIBS coupled with deep belief network (DBN), which is suitable to deal with a nonlinear problem, to classify 13 brands of special steels. The performance of the training and validation sets were used as the standard to optimize the structure of DBN. For different input, such as the intensities of full-spectra signals and characteristic spectra lines, the accuracies of the optimized DBN model in the training, validation, and test set are all over 98%. Moreover, compared with the self-organizing maps, linear discriminant analysis (LDA), k-nearest neighbor (KNN) and back-propagation artificial neural networks (BPANN), the result of the test set showed that the optimized DBN model performed second best (98.46%) in all methods using characteristic spectra lines as input. The test accuracy of the DBN model could reach 100% and the maximum accuracy of other methods ranged from 62.31% to 96.16% using full-spectra signals as input. This study demonstrates that DBN can extract representative feature information from high-dimensional input, and that LIBS coupled with DBN has great potential for steel classification.