Rapid Analysis of Logging Wellhead Gases Based on Fourier Transform Infrared Spectroscopy

Gas logging provides essential compositional data on the gas within the mud (wellhead gas) during drilling operations, enabling the discovery and evaluation of hydrocarbon resources. However, the long analysis period for wellhead gas leads to lagging results and tends to miss thin oil and gas layers.Therefore, Fourier transform infrared (FT-IR) spectroscopy was used in this paper to rapidly analyze seven light alkanes (methane, ethane, propane, n-butane, i-butane, n-pentane, and i-pentane) in wellhead gases.Spectral analysis involves denoising, baseline correction, feature selection, and partial least squares (PLS) modeling to obtain the concentrations of the wellhead gas components. Finally, the efficiency of this method was validated on actual offshore rigs. It is shown that FT-IR has an analysis period of 15 s, is fast enough to recognize thin oil layers, can work at the wellhead, and is easy to operate and equip. This work improves the real-time performance of well logging and provides an application of FT-IR spectroscopy in actual oil and gas exploration sites.

Well logging analyzes the solids, liquids, and gases returned from the wellbore during drilling. It is the key to oil and gas exploration and is known as “the eyes that discover hydrocarbon resources” (1). Gas logging quantitatively analyzes the gases degassed from the drilling fluid at the wellhead (wellhead gases) to evaluate the quality of the reservoir (2). The wellhead gas analyzed mainly consists of seven light alkanes, including CH₄, C₂H₆, C₃H₈, i-C₄H₁₀, n-C₄H₁₀, i-C₅H₁₂ and n-C₅H₁₂. With the development of new types of resources, such as shale gas and shale oil, the increasing speed of oil and gas exploration, rapid gas logging has become more demanding (3–5). If the analysis time lags, thin oil and gas layers may be missed, and the coincidence rate of well logging is reduced. Therefore, rapid analysis of wellhead gases from logging is crucial.

Gas chromatography (GC) (6–8) and mass spectrometry (MS) (9) are mainly used to analyze wellhead gases in oil and gas fields at present. Although they can accurately analyze alkanes in wellhead gases, the analysis speed is slow, with the analysis period normally longer than 120 s. In addition, complex equipment and the risk of gas explosions require analyzers to be located remotely. This standoff distance inherently creates a time lag, making it challenging to obtain the component concentrations of the wellhead gases corresponding to the drilled layer in time. GC is also expensive to operate and difficult to maintain. Raman spectroscopy (10–12) has also been used to analyze gas components in logging muds. However, it is difficult to apply widely due to the weak signal, limited sensitivity, and susceptibility to fluorescence interference.

The target components for gas logging are light alkanes, which are polar molecules, and Fourier transform infrared (FT-IR) spectroscopy responds to them all. In addition, FT-IR has the advantages of being fast, sensitive, and non-destructive. It has been widely used for gas analysis (13–16). Therefore, many infrared (IR) spectral analysis methods have been proposed. Zhang and associates (17) proposed an adaptive iteratively reweighted penalized least squares (airPLS) automatic baseline correction method, which is widely used in spectral preprocessing. On this basis, asymmetric reweighted penalised least squares (arPLS) proposed by Baek and associates (18) and adaptive extended Gaussian peak derivative reweighted penalised least squares (agdPLS) proposed by Li and colleagues (19) can improve the accuracy of baseline estimation at low signal-to-noise (S/N) ratios. Yun and co-authors (20) proposed a method for selecting the feature variables based on the successive contraction of the variable space, which can reduce the data dimensionality and improve the accuracy of the analysis. Tang and associates (21) proposed Tikhonov regularized feature selection to eliminate cross-interference for IR spectral analysis of multi-component gases. In spectral analysis modeling, Partial least squares (PLS) regression is currently the most widely used model for quantitative analysis of spectra (22,23). In recent years, non-linear methods such as artificial neural networks have been used for spectral modeling (24). However, absorption spectra of alkane gases overlap with each other because they have similar molecular structures. It is also challenging to distinguish their IR spectral absorption peaks. Meanwhile, FT-IR spectroscopy tends to be used in the laboratory and less applied to the actual engineering sites.

In this paper, FT-IR spectroscopy is used to analyze seven light alkanes in wellhead gases rapidly, with the analysis period less than 15 s. In addition, the method was applied on rigs in the Bohai Sea, China. Compared with GC and MS, the proposed method has three advantages. Firstly, the real-time analysis is greatly improved. Secondly, it can be placed at the wellhead for online analysis. Finally, it does not require carrier gas, features simpler operation, and facilitates easier maintenance. This work provides an application of FT-IR spectroscopy in oil and gas exploration.

Materials and Methods

FT-IR Wellhead Gas Rapid Analysis System

The diagram of the FT-IR wellhead gas rapid analysis system is shown in Figure 1. Wellhead gases from the drilling fluid entered a sample chamber with an optical path length of 80 cm. An FT-IR spectrometer was placed at the wellhead to scan the IR spectra of the wellhead gases and transmit them to a wireless data transmission system. The sensors measured the temperature and pressure of the environment. The controller is used to set the scanning parameters of the spectrometer and control the scanning background or sample. The wavenumbers range from 4000 cm^-1 to 500 cm^-1 with a resolution of 4 cm^-1. The wireless data transmission system transmits the wellhead gas spectra, temperature, and pressure data to the monitoring room 50 m away from the wellhead. The upper computer in the monitoring room is used to analyze the IR spectra and obtain the concentration of each component of the wellhead gases. The wellhead gas spectra, temperature and pressure data are saved in storage. The power supply provides power to the units.

Analytical Method for Wellhead Gas IR Spectra

The absorption peaks of the IR spectra contain the concentration information of target gas components, while the position of the absorption peaks is related to the gas component and its intensity is related to the concentration. The basic principle for quantitative analysis of wellhead gas IR spectroscopy is Lambert-Beer law (25,26), as in equation 1:

where T(v) and K(v) denote the transmittance and absorption coefficient at the wavenumber v, respectively. L denotes the optical path in the sample chamber. c denotes the molar concentration of the gas. For a specific gas, K is fixed. The same sample cell has a fixed length L. Therefore, T and c maintain a definite relationship. c can be obtained from T at the absorption.

The IR absorption band positions of gases are related to their chemical bonding patterns, which results in the proximity of absorption peaks for gases with similar molecular structures. IR spectra usually have broad absorption peaks. Alkanes have similar molecular structures, so their IR absorption spectra overlap, leading to severe cross-interference. In addition, the IR spectra obtained by the spectrometer inevitably contain noise and baseline drift, which may reduce the accuracy of analytical results. This noise and baseline drift may be severe when the FT-IR instrument operates in situ and for a long time. Although they can be eliminated partly by subtracting the background, this will introduce additional analysis time. Therefore, it is essential to process wellhead gas IR spectra data, the flow of which is shown in Figure 2.

Raw spectra are first pre-processed, including denoising and baseline correction to remove noise and baseline drift. Multiple scans of the spectra and averaging can reduce the noise, but this adds additional time consumption. Therefore, the Savitzky-Golay filter (27) was used to eliminate the noise in this paper. Baseline drift will cause the absorption peaks to be elevated or pulled down, causing additional measurement errors. The baseline is corrected using polynomial fitting to improve the analysis efficiency (28). The polynomial fitting equation is established by equation 2:

where y denotes the spectral vector, X denotes the wavenumber vector, and a denotes the fitting coefficient vector.

Then, the baseline is estimated by equation 3:

where b^{^} denotes the estimated baseline vector.

Tikhonov regularization (TR) (29) was used to select feature variables in the pre-processed spectra to eliminate cross-interference and improve accuracy and efficiency of regression modeling. The criterion for variable selection was to have higher sensitivity for the measured component and as low cross-sensitivity as possible for the other components. The model established using the TR is shown in equation 4:

where m denotes the gas concentration vector, Y denotes the spectral matrix, e denotes the error vector, and h denotes the coefficient vector based on the principle of equation 5:

where a and b are set to 1, ||×|| denotes the norm calculation, m denotes the regularization weight parameter with a value of 0.5, and L denotes the regularization operator.

Then, the selected feature variables are shown in equation 6:

where f_i denotes the feature variable of the ith component, h_ij denotes the jth coefficient of the ith component, and T_x denotes the xth spectral value.

PLS regression obtains the best model by simultaneously considering the relationship between the spectral and concentration matrices in the decomposition. Therefore, quantitative analysis was modeled by PLS regression (23) using the selected feature variables as inputs.

Finally, the results were calibrated with temperature and pressure to eliminate the effects of environmental variations. Combining the data measured by the temperature and pressure sensors, the measurements were calibrated by equation 7:

where C_c(n) and C₀(n) denote the concentration of the nth component after and before calibration, t denotes temperature in K, and p denotes pressure in atm.

Results

Calibration Spectra

The calibrated IR spectra obtained by standard concentration gases (background was nitrogen with 99.999% vol concentration) are shown in Figure 3. The concentration of each component was 0.05% vol. The absorption peaks of the target component in the wellhead gases were located near 3000 cm^-1 and 1250 cm^-1. The absorption peaks of alkanes overlap with each other at these locations. It is worth mentioning that absorption peaks opposite to the absorption direction occur near 3980 cm^-1 to 3340 cm^-1, 2040 cm^-1 to 1280 cm^-1, and 2360 cm^-1. Of these, the absorption peak near 2360 cm^-1 was from CO₂, and the other two ranges were water vapor. An open space was observed between the sample chamber and the FT-IR spectrometer. Water vapor and CO₂ variations in this space resulted in the absorption peaks when scanning the background and the sample. These absorption peaks are in the opposite direction from the alkanes, and the percentage transmittance is greater than 100%T since the background contains more water vapor and CO₂ than the sample. However, this has little effect on alkane analysis because the interfering absorption peaks overlap less with the alkane peaks.

The absorption peaks of the calibration spectra ranging from 3200 cm^-1 to 2700 cm^-1 are shown in Figure 4. The absorption of C₂H₆ is relatively weak in this range. In addition, the alkane absorption peaks overlap severely. However, the shapes and positions of these absorption peaks are still significantly different. For example, in the range of 3140 cm^-1 to 3084 cm^-1, a strong absorption of CH₄ is observed, while other alkane gases are almost absent.

IR Spectra of Actual Wellhead Gases

The FT-IR wellhead gas rapid analysis system was fixed at the wellhead of a rig in the Bohai Sea region of China to analyze the wellhead gas component concentrations. The wellhead gas IR spectra obtained at five different moments (with an interval of 1 hr) are shown in Figure 5. The IR absorption peaks of the wellhead gas show characteristics consistent with alkanes, indicating that alkanes dominate the wellhead gas components.

The absorption peaks around 3000 cm^-1 in Figure 5 are like those of methane, but the shapes are different because of the higher methane and superimposed other alkane peaks such as ethane. In addition, five spectra from the two wells each show synchronous variations, indicating that the alkanes in the analyzed formations are synchronous. The absorption peaks in the opposite direction in Figure 5a appear in the ranges of 3980 cm^-1 to 3340 cm^-1 and 2040 cm^-1 to 1280 cm^-1, which are related to the water vapor concentrations in the background and samples. The water vapor absorption peak points upward when the background contains more water vapor than the sample and faces downward otherwise (30). The intensities of the absorption peaks are related to the wellhead gas component concentrations. The alkane absorption peaks in Figure 5b are stronger than those in Figure 5a, indicating that Well #2 contains more alkanes than Well #1 at these moments. The strongest alkane absorption peaks at 4:00:00 in Well #1 suggest the formation has a high alkane content. The weakest alkane absorption peaks at 2:00:00 in Well #1 correspond to the formation with fewer alkanes. The alkane absorption peaks in Figure 5b are the strongest at 19:00:00 and close to saturation, indicating that more alkanes are in the formation at this time in Well #2.

Analytical Results for Actual Wellhead Gases

The continuous analysis results of the wellhead gas component concentrations for the two wells measured are shown in Figure 6. The analysis lasted 357.75 min for Well #1 and 103 min for Well #2, with 15 s intervals. The overall concentration of CH₄ is the highest, with the highest concentration exceeding 3.5% vol in Well #1 and 10% vol in Well #2. This indicates that alkanes are abundant in both wells. As the molecular weight increases, the overall concentration decreases.

In Figure 6a, the n-C₄H₁₀ concentrations are higher than those of i-C₄H₁₀, and the n-C₅H₁₂ concentrations are higher than i-C₅H₁₂. Overall, the C₂H₆ is less than the C₃H₈, and show synchronous changes among the components. Meanwhile, the alkane concentrations show box-like changes, indicating uniformity of hydrocarbons. In Figure 6b, the n-C₅H₁₂ concentrations are too low and rapidly varying to show synchronization with other alkanes. The alkane concentrations show mountainous variations, indicating that the hydrocarbons are high but not homogeneous.

Conclusion

This work demonstrated the rapid quantification of seven alkane components in wellhead gas using FT-IR spectroscopy during well logging. The novel approach provides a significant advancement for real-time gas analysis at the wellsite, effectively enhancing the timeliness of gas logging. It is relevant to applying IR spectroscopy in oil and gas exploration. The following conclusions can be drawn:

First, a single FT-IR analysis takes less than 15 s, which significantly improves the real-time performance of gas logging. Compared with GC, FT-IR analysis can be carried out at the wellhead, avoiding the data time lag caused by the sample gas transmission in the pipeline.

Second, the seven alkane concentrations were obtained from the in situ spectra of the wellhead gases by spectral analysis. Noise interference was eliminated by SG filtering. The PLS model was established after baseline correction and feature selection, which was able to analyze the wellhead gas component concentrations quantitatively.

Finally, FT-IR was applied to quantitatively analyze actual wellhead gases at offshore oil field sites. Seven component gas concentrations in wellhead gas at the rig site were analyzed. This provides a field application of FT-IR in oil and gas exploration.

Based on the results of this work, future research will focus on adding components for wellhead gas analysis, optimizing background scanning, and conducting application tests at more drilling sites.

Acknowledgements

This work was supported by the Key Research and Development Projects of Shaanxi Province (Grant number: 2021GXLH-Z-043) and the National Key Research and Development Program of China (Grant number: 2016YFF0102805).

Declaration of Interest Statement

The authors have no relevant financial or non-financial interests to disclose.

References

1. Seren, H. R.; Buzi, E.; Al-Maghrabi, L.; Ham, G.; Bernero, G.; Deffenbaugh, M., An Untethered Sensor Platform for Logging Vertical Wells. IEEE Trans. Instrum. Meas. 2018, 67 (4), 798–803.DOI: 10.1109/TIM.2017.2774183
2. Li, X. S.; Tang, X. J.; Huang, Z. J.; Guo, M. Y.; Chen, H. Q.; Zhang, Z. Y.; Wu, T., Methanol Infrared Spectral Characteristics in Offshore Gas Logging. Fuel 2024, 361, 130716. DOI: 10.1016/j.fuel.2023.130716
3. Lai, J.; Wang, G. W.; Fan, Q. X.; Pang, X. J.; Li, H. B.; Zhao, F.; Li, Y. H.; Zhao, X.; Zhao, Y. D.; Huang, Y. Y.; Bao, M.; Qin, Z. Q.; Wang, Q. Q., Geophysical Well-Log Evaluation in the Era of Unconventional Hydrocarbon Resources: A Review on Current Status and Prospects. Surv. Geophys. 2022, 43 (3), 913–957. DOI:10.1007/s10712-022-09705-4
4. Meng, X. G.; Wang, W. J.; Shen, Z. Z.; Xiong, J. Y.; Zhang, H., Production Logging via Coiled Tubing Fiber Optic Infrastructures (FSI) and its Application in Shale Gas Wells. Arabian J. Geosci. 2019, 12 (24), 782. DOI: 10.1007/s12517-019-4885-z
5. Torghabeh, A. K.; Rezaee, R.; Moussavi-Harami, R.; Pradha, B.; Kamali, M. R.; Kadkhodaie-Ilkhchi, A., Electrofacies in Gas Shale from Well Log Data via Cluster Analysis: A Case Study of the Perth Basin, Western Australia. Central Eur. J. Geosci. 2014, 6 (3), 393–402. DOI: 10.2478/s13533-012-0177-9
6. Weed, B. M.; Brambila, G. D.; Doezema, L. A., Natural Seepage of Methane and Light Alkanes at Three Locations in Southern California. Atmosphere 2020, 11 (9), 979. DOI: 10.3390/atmos11090979
7. Antoniv, M.; Zhu, S. S.; Chang, S. O. O.; Poitzsch, M. E.; Jabri, N. M.; Marsala, A.; O'Brien, J., Method for Detecting Nanoparticles on Cuttings Recovered from a Gas Reservoir. Energy & Fuels 2021, 35 (9), 7708–7716. DOI: 10.1021/acs.energyfuels.0c04304
8. Pollo, B. J.; Alexandrino, G. L.; Augusto, F.; Hantao, L. W., The Impact of Comprehensive Two-Dimensional Gas Chromatography on Oil & Gas Analysis: Recent Advances and Applications in Petroleum Industry. Trac-Trends Anal. Chem. 2018, 105, 202–217. DOI: 10.1016/j.trac.2018.05.007
9. Alimzhanova, M.; Abdykarimov, B., Analysis of Kazakhstan Crude Oil Biomarkers by Gas Chromatography in Combination with Mass Spectrometry. Separations 2023, 10 (11), 561. DOI: 10.3390/separations10110561
10. Han, X.; Huang, Z. X.; Chen, X. D.; Li, Q. F.; Xu, K. X.; Chen, D., On-Line Multi-Component Analysis of Gases for Mud Logging Industry Using Data Driven Raman Spectroscopy. Fuel 2017, 207, 146–153. DOI:10.1016/j.fuel.2017.06.045
11. Cai, Y. Y.; Xu, G. R.; Yang, D. W.; Tian, H. Y.; Zhou, F. J.; Guo, J. J., On-Line Multi-Gas Component Measurement in the Mud Logging Process Based on Raman Spectroscopy Combined with a CNN-LSTM-AM Hybrid Model. Anal. Chim. Acta 2023, 1259, 341200. DOI: 10.1016/j.aca.2023.341200
12. Petrov, D. V.; Tanichev, A. S., 13CH4/12CH4 Sensing Using Raman Spectroscopy. Spectrochim. Acta A Mol. Biomol. Spectrosc. 2024, 315, 124253. DOI:10.1016/j.saa.2024.124253
13. Tang, X. J.; Liang, Y. T.; Dong, H. Z.; Sun, Y.; Luo, H. Z., Analysis of Index Gases of Coal Spontaneous Combustion Using Fourier Transform Infrared Spectrometer. J. Spectro. 2014, 2014, 414391. DOI:10.1155/2014/414391
14. Tang, X. J.; Wang, W. J.; Zhang, X. L.; Wang, E. Z.; Li, X. J. N., On-Line Analysis of Oil-Dissolved Gas in Power Transformers Using Fourier Transform Infrared Spectrometry. Energies 2018, 11 (11), 3192. DOI:10.3390/en11113192
15. Mayorova, V.; Morozov, A.; Golyak, I.; Golyak, I.; Lazarev, N.; Melnikova, V.; Rachkin, D.; Svirin, V.; Tenenbaum, S.; Vintaykin, I.; Anfimov, D.; Fufurin, I., Determination of Greenhouse Gas Concentrations from the 16U CubeSat Spacecraft Using Fourier Transform Infrared Spectroscopy. Sensors 2023, 23 (15), 6794. DOI:10.3390/s23156794
16. Weise, D. R.; Johnson, T. J.; Myers, T. L.; Hao, W. M.; Baker, S.; Palarea-Albaladejo, J.; Scharko, N. K.; Bradley, A. M.; Banach, C. A.; Tonkyn, R. G., Comparing Two Methods to Measure Oxidative Pyrolysis Gases in a Wind Tunnel and in Prescribed Burns. Int. J. Wildland Fire 2023, 32 (1), 56–77. DOI: 10.1071/WF22079
17. Zhang, Z. M.; Chen, S.; Liang, Y. Z., Baseline Correction Using Adaptive Iteratively Reweighted Penalized Least Squares. Analyst 2010, 135 (5), 1138–1146.DOI:10.1039/b922045c
18. Baek, S. J.; Park, A.; Ahn, Y. J.; Choo, J., Baseline Correction Using Asymmetrically Reweighted Penalized Least Squares smoothing. Analyst 2015, 140 (1), 250–257. DOI:10.1039/c4an01061b
19. Li, X. S.; Tang, X. J.; Wang, B.; Lu, Y. S.; Chen, H. Q., An Adaptive Extended Gaussian Peak Derivative Reweighted Penalised Least Squares Method for Baseline Correction. Anal. Methods 2023, 15 (44), 6048–6060. DOI:10.1039/d3ay01389h
20. Yun, Y. H.; Bin, J.; Liu, D. L.; Xu, L.; Yan, T. L.; Cao, D. S.; Xu, Q. S., A Hybrid Variable Selection Strategy Based on Continuous Shrinkage of Variable Space in Multivariate Calibration. Anal. Chim. Acta 2019, 1058, 58–69. DOI: 10.1016/j.aca.2019.01.022
21. Tang, X. J.; Li, Y. J.; Zhu, L. J.; Zhao, A. X.; Liu, J. H., On-Line Multi-Component Alkane Mixture Quantitative Analysis Using Fourier Transform Infrared Spectrometer. Chemometrics Intell. Lab. Syst. 2015, 146, 371–377. DOI: 10.1016/j.chemolab.2015.06.010
22. Li, M. P.; Liu, Y. D.; Hu, J.; Su, C. T.; Xu, Z.; Cui, H. Z., Detection of the Early Fungal Infection of Citrus by Fourier Transform Near-Infrared Spectra. Spectroscopy 2023, 38 (s8), 12–22, 28. DOI: 10.56530/spectroscopy.dx3267q6
23. Lee, L. C.; Liong, C. Y.; Jemain, A. A., Partial Least Squares-Discriminant Analysis (PLS-DA) for Classification of High-Dimensional (HD) Data: A Review of Contemporary Practice Strategies and Knowledge Gaps. Analyst 2018, 143 (15), 3526–3539. DOI: 10.1039/c8an00599k
24. Petrini, J.; Salgado, R. M.; Rodriguez, M. A. P.; Machado, P. F.; Mourao, G. B., Machine Learning Methods for the Prediction of Milk Fatty Acid Content. Int. J. Dairy Technol. 2022, 75 (3), 553–562. DOI: 10.1111/1471-0307.12877
25. Mayerhöfer, T. G.; Pahlow, S.; Popp, J., The Bouguer-Beer-Lambert Law: Shining Light on the Obscure. Chem. Phys. Chem 2020, 21 (18), 2029–2046. DOI: 10.1002/cphc.202000464
26. Mayerhöfer, T. G.; Pahlow, S.; Popp, J., Understanding the Limits of the Bouguer-Beer-Lambert Law. Spectroscopy 2023, 38 (s8), 29–30, 58. DOI: 10.56530/spectroscopy.iq2368r9
27. Sadeghi, M.; Behnia, F.; Amiri, R., Window Selection of the Savitzky-Golay Filters for Signal Recovery From Noisy Measurements. IEEE Trans. Instrum. Meas. 2020, 69 (8), 5418–5427. DOI: 10.1109/TIM.2020.2966310
28. Gan, F.; Ruan, G. H.; Mo, J. Y., Baseline Correction by Improved Iterative Polynomial Fitting with Automatic Threshold. Chemometrics Intell. Lab. Syst. 2006, 82 (1-2), 59–65.
29. Randolph, T.; Ding, J. M.; Kundu, M. G.; Harezlak, J., Adaptive Penalties for Generalized Tikhonov Regularization in Statistical Regression Models with Application to Spectroscopy Data. J. Chemom. 2017, 31 (4), e2850. DOI: 10.1002/cem.2850
30. Li, X. S.; Tang, X. J.; Huang, Z. J.; Wu, L. W.; Guo, M. Y.; Lu, Y. S.; An, T. R.; Zhang, Z. Y.; Liu, C. Z.; Chen, H. Q.; Wu, T., Water vapor absorption peaks in infrared spectroscopic gas logging and correlation with water layers. Gas Sci. Eng. 2025, 138, 205608. DOI: 10.1016/j.jgsce.2025.205608.

Xiaoshan Li, Xiaojun Tang, Houqing Chen, Tong Wu, Zeyu Zhang, Chongzhi Liu, and Youshui Lu are with the State Key Laboratory of Electrical Insulation and Power Equipment of the School of Electrical Engineering at Xi’an Jiaotong University, in Shaanxi, China. Tang and Lu are also with the State Key Laboratory for Manufacturing Systems Engineering of the School of Instrument Science and Technology at Xi’an Jiaotong University, in Shaanxi, China. Zijian Huang is with Panjin Zhonglu Oil and Gas Technology Service Co., Ltd., in Liaoning, China. Liwei Wu and Mingyu Guo are with the Tianjin Branch of the China National Offshore Oil Co., Ltd., in Tianjin, China. Tongrui Anis with King’s College London, in London, United Kingdom. Direct correspondence to Xiaojun Tang at xiaojun_tang@xjtu.edu.cn