Detection of pesticide residues on fruit surfaces using laser induced breakdown spectroscopy

Abstract

The detection of pesticide residues on fruit surfaces is highly relevant to people’s lives. Based on our previous research, we further explored the detection of chlorpyrifos residues on apple surfaces by laser induced breakdown spectroscopy (LIBS) in this paper. We observed the characteristic peaks of P at 213.62 nm, 214.91 nm, 253.56 nm, and 255.33 nm and the characteristic peak of Cl at 837.59 nm. We studied the influence of pesticide concentration and argon on the intensity of the LIBS signals. The intensity of the LIBS signal showed a linear relationship with the pesticide concentration, and argon could enhance the intensity of the LIBS signal. In the case of purging with argon, the characteristic peak of P was observed in the LIBS spectrum of a chlorpyrifos solution of 1 : 1000 dilution. Then we studied the differences in the LIBS signals of pesticide residues among different matrixes and pesticides. Finally, we performed the quantitative detection of the pesticide residues with LIBS, which will provide a reference for quantitative detection. Our work gives a new method for the fast detection of pesticide residues on fruit.

---

Introduction

With the developments in modern agriculture, more and more chemicals and fertilizers are widely used in agricultural production and the phenomenon of pesticide residues on the surfaces of agricultural products is serious. Pesticide residues have threatened the health of human beings. Therefore, the detection of pesticide residues on the surfaces of agricultural products is becoming a research hot spot.

Conventional detection methods for pesticide residues on the surfaces of agricultural products including gas chromatography, high-performance liquid chromatography, gas chromatography coupled with mass spectrometry, and high-performance liquid chromatography coupled with mass spectrometry^1–3 are time-consuming and laborious and can hardly achieve real-time detection.^2,3 Moreover, some of the organic solvents used in these methods are toxic and unfriendly to the environment.¹ Considering the above shortcomings of the traditional methods, some scholars have studied optical methods to analyze agricultural products, such as surface-enhanced Raman spectroscopy.^4–6 Shende et al. studied the detection of a chlorpyrifos-methyl solution in orange juice using surface-enhanced Raman spectroscopy coupled with solid-phase extraction and realized the detection of chlorpyrifos-methyl at 50 ppb.⁴ Shende et al. studied the detection of organophosphorus pesticides on fruit surfaces by surface-enhanced Raman spectroscopy and successfully detected 0.1% fonofos on apple surfaces within approximately 5 min, with a sensitivity of 10 ppm.⁵ Liu et al. studied the detection of pesticide residues on fruit surfaces by surface-enhanced Raman spectroscopy coupled with gold nanostructures and achieved limits of detection (LODs) for carbaryl, phosmet, and azinphos-methyl of 4.51 ppm, 6.51 ppm, and 6.66 ppm on apple surfaces and 5.35 ppm, 2.91 ppm, and 2.94 ppm on tomato surfaces, respectively.⁶ Although the limit of detection (LOD) of surface-enhanced Raman spectroscopy is low, it cannot realize quick online detection.

Since the method using ruby maser-induced plasma was proposed by Brech in 1962,^7,8 LIBS has been applied in solid, liquid, and gas fields.^9,10 Moreover, LIBS can realize the fast real-time detection of multiple components without a sample preparation procedure and LIBS is developing rapidly.^11–13 Kim et al. studied the detection method for nutrient elements and agricultural products contaminated by pesticides with LIBS and quantitatively analyzed the nutrient elements in rice and spinach.¹² The LODs of Mg, Ca, Na, and K were 29.63 mg kg⁻¹, 102.65 mg kg⁻¹, 36.36 mg kg⁻¹, and 44.46 mg kg⁻¹ in spinach, respectively. The LODs of Mg, Ca, Na, and K were 7.54 mg kg⁻¹, 1.76 mg kg⁻¹, 4.19 mg kg⁻¹, and 6.70 mg kg⁻¹ in unpolished rice, respectively. They also distinguished the samples contaminated by pesticides from uncontaminated samples and the mistaken determination percentage was less than 2%. Multari et al. studied the method of LIBS to distinguish samples contaminated by different pesticides in complex matrixes (tissue fats and fatty oils).¹⁴ The detection concentration of pesticides ranged from 0.005 to 0.100 μg g⁻¹. The detection method distinguished not only the samples of different pesticides, but also the samples of different concentrations. Our previous study described the detection of chlorpyrifos residues on apple surfaces by LIBS.¹⁵ We distinguished the high concentration samples (1 : 1 and 1 : 20) accurately. However, it was difficult to distinguish the samples with lower concentrations (1 : 100 and 1 : 1000).

In this paper, based on the previous study, we further studied the precise detection method of pesticide residues on fruit surfaces with LIBS in three aspects. Firstly, we explored the influence of the concentration of pesticides and purging with argon on the intensity of the LIBS signal. Secondly, we studied the differences of the LIBS signal among different matrixes and pesticides. Thirdly, we quantitatively analyzed the pesticide residues on apple surfaces. This paper provides the technical support for the application and development of LIBS in the detection of pesticide residues on agricultural products.

Materials and methods

Preparation of samples

Preparation of pesticide samples. The chemical name of chlorpyrifos is O,O-diethyl-O-(3,5,6-trichloro-2-pyridyl)phosphorothioate (C₉H₁₁C_l3NO₃PS). Chlorpyrifos was from Dow AgroScience Company and the content of the active ingredients was 480 g L⁻¹. We dissolved chlorpyrifos in deionized water according to 9 dilution ratios (1 : 50, 1 : 80, 1 : 90, 1 : 100, 1 : 110, 1 : 120, 1 : 130, 1 : 500, and 1 : 1000).

The chemical name of omethoate is O,O-dimethyl-S-methylcarbamoylmethyl phosphorothioate (C₅H₁₂NO₄PS). Omethoate was produced by Zhengzhou Labor Agrochemicals Co., Ltd and the content of the active ingredients was 40%. We dissolved omethoate in deionized water according to a dilution ratio of 1 : 100.

Preparation of fruit samples. Red Fuji apples and Hosui pears were bought from GuoXiangSiYi Supermarket. We first cleaned the fruit with water and then dried them in the natural environment. Then we cut the fruit into small pieces with a weight of each piece of about 100 g and an area of about 4 cm². We smeared 100 μL of pesticide solution of each concentration on the surface of the fruit pieces as homogeneously as possible and dried them in the natural environment for about 30 min. Therefore, the samples can be considered fresh but the pesticide solution had been dried. As the laser spot size was only 100 μm, the apple surface can be considered flat. So, if the pesticide was diluted 1000 times, its average concentration in a 100 g sample is about 1 mg kg⁻¹.

Experimental devices

Fig. 1 shows a schematic diagram of the LIBS system. A Nd:YAG laser induces a laser beam, which is reflected by the mirror. Then the laser is collimated by the lens and focused on the sample surface. The materials on the sample surface are ionized and vaporized by the high-power laser. Then a high-temperature plasma is formed on the sample surface. The vaporized materials will be broken down into atoms and ions. At the end of the laser pulse, the plasma will be cooled and spread to the surrounding environment. The atoms and ions in the excited state will relax from a high energy level to a low energy level and optical radiation of a specific wavelength will be released.^16,17 A quartz fiber is used to collect the spectrum signal and transmit it to the spectrometer.

Fig. 1 A schematic diagram of the LIBS system.

A delay generator (Fig. 1) is integrated into the equipment to produce the delay time from laser emission to spectrometer collection. A motorized rotation stage is used to adjust the position of the samples. Through the adjustment, the focus of the lens just falls on the sample surfaces. The laser generator produced by Beamtech Optronics Co., Ltd can deliver monochromatic light at 1064 nm. The maximum energy of the laser beam is 200 mJ and the frequency is 20 Hz. The laser energy was set as 160 mJ in the experiment. The spectrometer (HR2000+) was produced by Ocean Optics Company. The spectral range of the spectrometer is from 200 nm to 1000 nm. The resolution is 0.2 nm and the signal-to-noise ratio is 250 : 1.

The delay time was set as 2 μs, which was the optimal value obtained in the experiment. Compared with our previous studies, this study optimized the focusing lens group used to collect the plasma signal. We also optimized the focusing system of the laser and enhanced the laser energy. All these optimizations have improved the detection performance of the system that will be presented in the Results and analysis section.

Experimental design

We used the LIBS system to collect the signals of the samples, including the apple samples contaminated with chlorpyrifos solutions of 9 dilution ratios (1 : 50, 1 : 80, 1 : 90, 1 : 100, 1 : 110, 1 : 120, 1 : 130, 1 : 500, and 1 : 1000), the clean apple sample, the apple sample contaminated with the omethoate solution of a dilution ratio of 1 : 100, the pear sample contaminated with the chlorpyrifos solution of a dilution ratio of 1 : 100, and the apple sample contaminated with the chlorpyrifos solution of a dilution ratio of 1 : 1000 and purged with argon. We collected 12 data values for each sample and averaged the 12 data values of each sample to reduce the errors. The influence of the heterogeneity of the pesticide on the samples can also be reduced.

Then, we analyzed the pesticide LIBS signal from four aspects. Table 1 shows the experimental design.

Table 1 The design of the experiments

Experiments		Matrix	Pesticide	Concentration
Characteristics of the pesticide LIBS signal		Apple	Chlorpyrifos	1 : 100, clean apple
Influencing factors of the pesticide LIBS signal	Pesticide concentration	Apple	Chlorpyrifos	1 : 50, 1 : 100, 1 : 500, 1 : 1000
	Purging with argon	Apple	Chlorpyrifos	1 : 1000
Different matrixes		Apple	Chlorpyrifos	1 : 100
		Pear	Chlorpyrifos	1 : 100
Different pesticides		Apple	Chlorpyrifos	1 : 100
		Apple	Omethoate	1 : 100
Quantitative analysis of the pesticide LIBS signal		Apple	Chlorpyrifos	1 : 80, 1 : 90, 1 : 100, 1 : 110, 1 : 120, 1 : 130

---

Results and analysis

LIBS signal analysis of chlorpyrifos

From the molecular formula of chlorpyrifos, we know that chlorpyrifos consists of 7 different elements: C, H, O, N, P, S, and Cl. Because the elements C, H, O and N are in air, which will interfere with the LIBS signal analysis of chlorpyrifos, we chose the characteristic peaks of the LIBS signals of P, S, and Cl to analyze the chlorpyrifos residues on the apple surfaces.

Fig. 2 shows the LIBS spectra of P and Cl. The red lines represents the spectra of the apple sample contaminated with chlorpyrifos solution (1 : 100). The black lines represents the spectra of the clean apple. We observed the characteristic peaks of P at 213.62 nm and 214.91 nm (Fig. 2a), which were the same as our previous results. In addition, we found the other characteristic peaks of P at 253.56 nm and 255.33 nm (Fig. 2b), which were not found in our previous study. As Fig. 2c shows, our previous study showed that peak A was located at 393.33 nm and peak B was located at 396.89 nm, which were regarded as the characteristic peaks of S. After careful analysis, we found that peaks A and B were not the characteristic peaks of S. We also found the characteristic peak of the Cl element at 837.59 nm (Fig. 2d), which was also observed in our previous study.

Fig. 2 The LIBS spectra of apple samples contaminated with the chlorpyrifos solution (1 : 100) and clean apple: (a) the characteristic peaks of P at 213.62 nm and 214.91 nm; (b) the characteristic peaks of P at 253.56 nm and 255.33 nm; (c) the characteristic peaks, A and B; (d) the characteristic peak of Cl at 837.59 nm.

Influencing factors of the LIBS signal of chlorpyrifos

Different solution concentrations. Fig. 3 shows the Cl and P peaks in the LIBS signal spectra of apple samples contaminated with chlorpyrifos solution at different concentrations (1 : 50, 1 : 100, 1 : 500, and 1 : 1000). As shown in Fig. 3a, we can see that the intensities of the LIBS characteristic peaks show a stepped change. In addition, with the increase of the solution concentration, the intensity of the LIBS signal increased as well. Fig. 3b and c show the same tendency. Therefore, we concluded that the intensity of the LIBS signal of chlorpyrifos for the apple sample showed a linear relationship with the concentration of the chlorpyrifos solution.

Fig. 3 The LIBS spectra of apple samples contaminated with chlorpyrifos solution of different dilution ratios (1 : 50, 1 : 100, 1 : 500, 1 : 1000) and clean apple: (a) the characteristic peaks of P at 213.62 nm and 214.91 nm; (b) the characteristic peaks of P at 253.56 nm and 255.33 nm; (c) the characteristic peak of Cl at 837.59 nm.

In addition, we found the LIBS characteristic peaks of P in the apple sample which was contaminated with the chlorpyrifos solution of 1 : 500 (Fig. 3a and b). However, in the previous study, the LIBS characteristic peaks of P were found in the apple samples which were contaminated with chlorpyrifos solution which had a concentration higher than 1 : 100. We also found the LIBS characteristic peak of Cl in the apple sample which was contaminated with the chlorpyrifos solution of 1 : 100 (Fig. 3c), while the characteristic peak of Cl was only found in the apple sample which was contaminated with the chlorpyrifos solution of 1 : 1 in the previous study. We attributed this change to the optimization of the laser energy and the focusing lens group.

Purging with argon. Fig. 4 shows the spectra of the apple sample contaminated with the chlorpyrifos solution of 1 : 1000 and the chlorpyrifos solution of 1 : 1000 purged with argon. As shown in Fig. 4, we could not find the characteristic peaks of P and Cl in the apple sample which was contaminated by the chlorpyrifos solution of 1 : 1000. After purging with argon, the characteristic peaks of P (213.62 nm, 214.91 nm, 253.56 nm, and 255.33 nm) were observed clearly. However, the intensity of the characteristic peak of Cl showed a slight change, because it was difficult to excite the peak of Cl at 837.59 nm. Therefore, we concluded that the argon could enhance the LIBS signal. This conclusion was also mentioned in previous studies by some scholars.^18,19 Ref. 18 reports that argon gas could provide the best environment to enhance the intensity of the LIBS signal and diminish the interface. In our study, the characteristic peaks of P were found in the apple sample contaminated with chlorpyrifos solution of 1 : 1000 purged with argon, which was not mentioned in our previous study.

Fig. 4 The LIBS spectra of the apple sample contaminated with chlorpyrifos solution of 1 : 1000 after purging with argon and without purging. (a) The characteristic peaks of P at 213.62 nm and 214.91 nm; (b) the characteristic peaks of P at 253.56 nm and 255.33 nm; (c) the characteristic peak of Cl at 837.59 nm.

Differences in the LIBS spectra among different matrixes and pesticides

In order to study the method of differentiating pesticides by LIBS, we analyzed the differences between the LIBS spectra among different pesticides and matrixes.

Different matrixes. Fig. 5 shows the LIBS spectra of apple and pear surfaces contaminated with chlorpyrifos solution (1 : 100).

Fig. 5 The LIBS spectra of samples contaminated with chlorpyrifos solution in different matrixes (apple and pear): (a) the characteristic peaks of P at 213.62 nm and 214.91 nm; (b) the characteristic peaks of P at 253.56 nm and 255.33 nm; (c) the characteristic peak of Cl at 837.59 nm.

The characteristic peaks of P in the apple matrix contaminated with chlorpyrifos solution are stronger than those in the pear matrix contaminated with chlorpyrifos solution. We attributed this difference to the difference in physical properties between an apple and a pear. The peel of the Red Fuji apple was very smooth and the flesh was firm. Therefore, the chlorpyrifos solution smeared on the surface of the apple was not easily absorbed. On the contrary, the peel of the Hosui pear was coarse and the flesh was sparse. Therefore, the chlorpyrifos solution smeared on the surface of the pear was easily absorbed. The difference in physical properties led to the difference in intensity of the LIBS signal of the pesticide residues on the surfaces of different fruits. From the analysis, we concluded that the intensity of the pesticide LIBS signal varied with the matrix.

Different pesticides. Fig. 6 shows the LIBS spectra of apple samples contaminated with solutions of chlorpyrifos and omethoate in the dilution ratio of 1 : 100. Both chlorpyrifos and omethoate contain the elements P and S, but chlorpyrifos also contains Cl. As shown in Fig. 6a and b, the characteristic peaks of P of chlorpyrifos are obviously higher than those of omethoate. We attributed the distinction to the physical properties of the pesticides, such as viscosity and volatility. The viscosity and volatility of different pesticides made the residual quantity different. So the LIBS spectrum presented differences in intensity. As shown in Fig. 6c, the characteristic peak of Cl at 837.59 nm is observed in chlorpyrifos, whereas it cannot be observed in omethoate because omethoate does not contain Cl. The characteristic peaks of the LIBS spectra show significant differences among different pesticides, as well as the intensities of the characteristic peaks.

Fig. 6 The LIBS spectra of different pesticides (chlorpyrifos and omethoate): (a) the characteristic peaks of P at 213.62 nm and 214.91 nm; (b) the characteristic peaks of P at 253.56 nm and 255.33 nm; (c) the characteristic peak of Cl at 837.59 nm.

According to the above analysis, we can conclude that the LIBS signal is closely related to the matrixes and pesticides used.

Quantitative analysis of pesticide residues

We quantitatively analyzed the relationship between the pesticide concentration and the intensity of the characteristic peaks (213.62 nm, 214.91 nm, 253.56 nm, 255.33 nm, and 837.59 nm), and we fitted the calibration curves using the apple samples contaminated by chlorpyrifos solution, where the concentrations used were 1 : 80, 1 : 90, 1 : 100, 1 : 110, 1 : 120 and 1 : 130. Fig. 7 shows the calibration curves of the LIBS characteristic peaks of chlorpyrifos in apple fitted by the method of single variable linear regression.

Fig. 7 The quantitative analysis of the LIBS characteristic signals of chlorpyrifos in apple samples: (a) the calibration curve of P at 213.62 nm; (b) the calibration curve of P at 214.91 nm; (c) the calibration curve of P at 253.56 nm; (d) the calibration curve of P at 255.33 nm; (e) the calibration curve of Cl at 837.59 nm.

All the correlation coefficients (R²) are larger than 0.88, which indicates that the fitted curves are reliable. From the curves we know that the intensity of the LIBS spectra has a linear relationship with the solution concentration. Therefore, we can use the calibration curves to detect chlorpyrifos quantitatively. It should be noted that the slopes of the calibration curves at different wavelengths are different. We think this may be due to the following reasons: (1) in terms of the spectral characteristics, at each wavelength there is a different detection limit and signal-to-noise ratio, and if the concentrations in the calibration curve are near the detection limit, this may lead to different slopes in the calibration curves. (2) The pre-processing method (e.g. background subtraction etc.) may have caused the slopes of the calibration curves at different wavelengths to vary.

In order to explore the detection ability of LIBS for pesticide residues, we calculated the LODs of the characteristic peaks. First, we subtracted the intensity of the background from the original LIBS signal to acquire the actual intensity of the characteristic peak. Then we calculated the LODs by the equation which is defined as 2σ/k, where σ is the standard deviation of the background and k is the slope of the calibration curve.^12,20–22 Table 2 shows the LODs of the characteristic peaks. We can see that the LODs of P at 213.62 nm, 214.91 nm, 253.56 nm and 255.33 nm are 4.3 mg kg⁻¹, 2.1 mg kg⁻¹, 1.5 mg kg⁻¹, and 6.9 mg kg⁻¹, respectively, and the LOD of Cl at 837.59 nm is 3.0 mg kg⁻¹. The Chinese standard for chlorpyrifos residue on apple is 1 mg kg⁻¹.²³ Although the LODs we calculated are a little bigger than the Chinese standard, we can observe the characteristics of P in the apple sample contaminated with chlorpyrifos of 1 : 1000 purged with argon, which is equivalent to 1 mg kg⁻¹. So the method of LIBS is reliable for the detection of pesticide residues.

Table 2 The LODs of P and Cl of chlorpyrifos

Elements	Wavelength (nm)	LOD
P I	213.62 nm	4.3 mg kg^−1
P I	214.91 nm	2.1 mg kg^−1
P I	253.56 nm	1.5 mg kg^−1
P I	255.33 nm	6.9 mg kg^−1
Cl I	837.59 nm	3.0 mg kg^−1

---

If a wide range of concentrations of pesticide solution is used, the calibration curve and the LOD will be more accurate. This section is only an exploration of quantitative detection using LIBS.

The quantitative analysis of pesticide residues and the calculation of the LOD provide a reference for the quantitative LIBS detection of pesticide residues on the surfaces of agricultural products.

Conclusions

Based on our previous study, we analyzed the characteristics of the chlorpyrifos LIBS signal in this paper. We observed the characteristic peaks of P at 253.56 nm and 255.33 nm, which were not observed in our previous study. We also discussed the influence of pesticide concentration and argon on the LIBS spectra. The intensity of the LIBS signal had a linear relationship with the concentration of pesticide. Argon could enhance the intensity of the LIBS signal of the pesticides. We also analyzed the differences in intensity of the LIBS signal among different matrixes and pesticides. Finally, we discussed the quantitative detection of pesticide residues. However, the detection limits of the method are higher than the Chinese standard if no argon is used. We believe some methods could lower the LOD: (1) a more sensitive calibration model may be established using some chemometric methods; (2) the LIBS spectral characteristics in the ultraviolet region of P and S are much more obvious than in the visible band and thus can achieve a much lower LOD. We will further study the above methods and present the results in our future publications.

Overall, this paper provides a reference for the detection of pesticide residues on the surfaces of agricultural products as well as a reference for the application of LIBS.

Acknowledgements

This work was supported by ​National Key Technologies R&D Program of China (2013BAD19B02) and National Natural Science Foundation of China (No. 31271614 &​ 41201299).

References

1. P. X. Zhang, X. Zhou, A. Y. S. Cheng and Y. Fang, J. Phys.: Conf. Ser., 2006, 28, 7.
2. J. Li, L. Xue, M. Liu, X. Wang and C. Luo, Chin. Opt. Lett., 2010, 8, 1050.
3. X. Jiang, D. Li, X. Xu, Y. Ying, Y. Li, Z. Ye and J. Wang, Biosens. Bioelectron., 2008, 23, 1577.
4. C. Shende, F. Inscore, A. Sengupta, J. Stuart and S. Farquharson, Sens. Instrum. Food Qual. Saf., 2010, 4, 101.
5. C. Shende, A. Gift, F. Inscore, P. Maksymiuk and S. Farquharson, Monitoring, Food Safety, Agriculture, and Plant Health, 2004, p. 28.
6. B. Liu, P. Zhou, X. Liu, X. Sun, H. Li and M. Lin, Food Bioprocess Technol., 2013, 6, 710.
7. F. Brech and L. Cross, Appl. Spectrosc., 1962, 2, 59.
8. D. A. Rusak, B. C. Castle, B. W. Smith and J. D. Winefordner, TrAC, Trends Anal. Chem., 1998, 8, 453.
9. W. B. Lee, J. Wu, Y. Lee and J. Senddon, Appl. Spectrosc. Rev., 2004, 1, 27.
10. Y. Lee and J. Sneddon, ISIJ Int., 2002, 42, S129.
11. V. S. Burakov, N. V. Tarasenko, M. I. Nedelko, V. A. Kononov, N. N. Vasilev and S. N. Isakov, Spectrochim. Acta, Part B, 2009, 64, 141.
12. G. Kim, J. Kwak, J. Choi and K. Park, J. Agric. Food Chem., 2012, 60, 718.
13. J. Kwak, C. Lenth, C. Salb, E. Ko, K. Kim and K. Park, Spectrochim. Acta, Part B, 2009, 64, 1105.
14. R. A. Multari, D. A. Cremers, T. Scott and P. Kendrick, J. Agric. Food Chem., 2013, 61, 2348.
15. F. Ma and D. Dong, Food Anal. Method., 2014, 7, 1858.
16. D. Yuan, M. Zhou, C. Liu, F. Yan, J. Dai and N. Ren, Spectrosc. Spectral Anal., 2008, 28, 2019.
17. C. Pasquini, J. Cortez, L. M. C. Silva and F. B. Gonzaga, J. Braz. Chem. Soc., 2007, 18, 463.
18. V. I. Babushok, F. C. DeLucia, P. J. Dagdigian and A. W. Miziolek, Spectrochim. Acta, Part B, 2005, 60, 926.
19. A. K. Pathak, V. K. Singh, N. K. Rai, A. K. Rai, P. K. Rai, P. K. Rai, S. Rai and G. D. Baruah, Laser Med. Sci., 2011, 26, 531.
20. H. Li, Chin. J. Spectrosc. Lab., 2010, 27, 2465.
21. J. Huang, H. Liu and K. Lin, Anal. Chim. Acta, 2007, 581, 303.
22. A. W. Miziolek, V. Palleschi and I. Schechter, Laser-induced breakdown spectroscopy (LIBS), Cambridge University Press, Cambridge, 2006.
23. J. T. Wang, 2010 Gap International Forum, Beijing, 2010.

---