Chemometrics-assisted calibration transfer strategy for determination of three agrochemicals in environmental samples: Solving signal variation and maintaining second-order advantage

doi:10.1016/j.chemolab.2019.103869

Chemometrics and Intelligent Laboratory Systems

Volume 194, 15 November 2019, 103869

https://doi.org/10.1016/j.chemolab.2019.103869 Get rights and content

Highlights

•
PDS/SWANRF is firstly proposed for transferring EEM signals measured on two instruments.
•
SWANRF method is employed to resolve the highly overlapping fluorescence signals and achieve second-order advantage.
•
PDS is carried out to compensate the signal instability and variation recorded on different instruments.
•
Results from the PDS/SWANRF strategy are well consistent with those from the complete recalibration.

Abstract

This manuscript proposes a chemometrics-assisted calibration transfer strategy to determine three agrochemicals (indole-3-acetic acid, 1-naphthylacetic acid and thiabendazole) in environmental samples by excitation-emission matrix (EEM) fluorescence detection. The fluorescence landscapes of target compounds are heavily overlapped with each other and also with the extracts of real samples, making it impractical to be quantified by conventional spectroscopy methods. The second-order calibration methods based on “mathematical separation” are carried out to handle these headaches as its potential capacity to get successful resolution and quantification even in the presence of overlapped peaks and unknown interferences, known as the famous “second-order advantage”. Piecewise direct standardization (PDS) method, one calibration transfer strategy, is applied for updating chemometric multivariate calibration model to compensate for the signal instability and variation in responses recorded on different instruments, which avoids overhead of a full recalibration usually involves considerable effort, cost, and time. Both root-mean-square error of prediction (RMSEP) and average recoveries calculated from simulated and experimental data are analyzed to validate the feasibility and applicability of this strategy. Result reveals that the number of calibration standard used was reduced from 10 to 3 for later determinations while producing comparable analytical results in terms of RMSEPs and recovery values to those obtained by a full recalibration strategy. The developed strategy takes advantage of less experimental effort and cost, which can be an alternative one for fast screening of agrochemicals and long-term process analysis in environmental samples or related food matrices.

Introduction

Agrochemicals are profusely used in agriculture to increase production and improve the quality of plants and crops. It is necessary to monitor and quantify the levels of their residues in environmental samples as they can accumulate in the environment and show potentially detrimental effects on mammals and plants [1]. Thiabendazole (TBZ), a systemic benzimidazole fungicide, which is commonly used as a fungicide to prevent fruits such as pears, citrus and bananas from mould, rot, blight and stain, thus keeping their freshness before waxing step during storage [2]. Indole-3-acetic acid (IAA) and 1-Naphthylacetic acid (NAA) are two of the representatives of auxins widely used as plant growth regulators in agriculture, primarily in many kinds of fruit and vegetables [3,4]. Commonly, two or even more types of agrochemicals are used simultaneously in combination to obtain synergistic effects, requiring effective and sensitive methods for determining their residue levels in fruits and environmental samples including soils and sewage waters.

Methods developed to determine these agrochemical residues are mainly based on chromatographic and luminescent techniques [[5], [6], [7], [8], [9]]. However, chromatographic techniques usually involve large amounts of organic solvents consumption and long analysis time, making it uneconomic and less competitive to some extent. By contrast, luminescent techniques, such as fluorescence spectroscopy, are generally regarded as a green strategy for its simple, cheap, non-destructive and sensitive. Nevertheless, one of the main difficulties in the use of luminescent techniques in quantitative analysis is the resolution of overlapping signals, especially for multi-component mixtures, which require prior separation processes to reach the goal. Meanwhile, some efforts have been made to increase selectivity and achieve successful resolution by using synchronous fluorescence spectrometry (SFS) for qualitative and quantitative analysis [10,11].

Luckily, the combination of suitable chemometric tools along with excitation-emission matrix fluorescence may handle these troubles with good resolution and high selectivity. So far, chemometric multivariate calibration coupled with high order analytical instruments have been successfully applied in many fields, either for quantitative or qualitative purposes [[12], [13], [14], [15], [16], [17], [18]]. In multivariate calibration, the potential interferences in the complex samples can be modeled by chemometric algorithms and pure signals of target analytes can be extracted out accurately. Therefore, quantification can be achieved successfully even in the presence of unknown components, known as the famous “second-order advantage”.

An ideal state in chemometric analysis is that the multivariate models should be made to last a long time and “second-order advantage” can be maintained because much time and effort is necessary to develop them, e.g. a sufficient number of samples should be prepared and analyzed by reference methods [19]. It is particularly useful especially for long-term process analysis and quality control in environment and food analysis. Nevertheless, the signals of instrument may be changed when original instrument is replaced by a new one, or the analytical signals may be changed by using the same instrument but changing over time due to variations in spectrographic components (e.g., aging of xenon lamp) or in experimental conditions (e.g., temperature, humidity, operator) among others. The most direct way is to recalibrate for the new measurement conditions or to expand the original model for the new situation, which are rather costly and in conflict with principle of the green analytical chemistry.

Calibration transfer strategy may provide good compromise in the costly and analytical accuracy. It applies chemometric tools to find a mathematical transformation that makes the response of samples measured on the secondary conditions corrected to that obtained on the primary conditions so that the response signals collected from two different conditions/instruments resemble each other. One of the most widely used calibration transfer strategies is piecewise direct standardization (PDS) [20], which requires the selection of a subset of calibration samples from the primary situation and measure them on the secondary situation. So far, this strategy has been widely used in combination with NIR [21,22], electrochemical [23,24], Raman [25,26], UV–Vis [27,28] and fluorescence data [[29], [30], [31]]. At the same time, there are also several recent works dealing with the joint use of chemometric multivariate calibration and calibration transfer strategy. Gu et al. [32] firstly applied PDS strategy with ATLD algorithm to compensate for the signal instability of LC-MS data acquired from one instrument with different days. Recently, Montemurro et al. [31] presented a similar strategy that combined PDS with second-order PARAFAC algorithm to overcome the matrix effect and reduce analysis time for the determination of pirimiphos-methyl in maize grains.

Hence, the aim of this work was to develop a chemometric second-order calibration strategy named self-weighted alternating normalized residue fitting (SWANRF) for simultaneous determination of IAA, NAA and TBZ in complex environmental samples by using excitation-emission matrix fluorescence. Moreover, piecewise direct standardization (PDS) method was applied to compensate for the variations in the instrument responses on two process EEM fluorescence spectrophotometers of the same manufacturer. They are produced to be alike, but due to differences in the optical fibers used and inherent, small differences in the filters e.g. due to aging/bleaching, there is a clear distinction between the spectra the two spectrometers record [29]. The applicability of the proposed strategy was first verified using one set of simulated EEM fluorescence data that contains three components of interest and one uncalibrated interference, and one set of real EEM fluorescence data produced by three agrochemicals in soil and sewage samples. Both analytical figures of merit and statistical parameters including average recoveries and root-mean-square error of prediction were calculated to investigate the quantitative results obtained from the proposed strategy and that from the complete recalibration. To the best of knowledge, it is the first time that PDS and SWANRF are combined together for transferring and resolving EEM fluorescence signals measured on two instruments. A schematization of the global chemometric analysis flow is shown in Fig. 1.

Section snippets

Trilinear component model

For the type of EEM fluorescence data, a two-way matrix with the size of I excitation wavelength points and J emission wavelength points can be acquired for each analytical sample. Given K samples containing calibration and prediction samples, a three-way data array X (I × J × K) could be obtained by stacking them along the sample dimension. According to the theory of trilinear component model, the three-way array X can be expressed as: $x_{i j k} = \sum_{n = 1}^{N} a_{i n} b_{j n} c_{k n} + e_{i j k}; f o r i = 1, \dots, I; j = 1, \dots, J; k = 1, \dots, K$ here x_ijk

Simulated EEM fluorescence data

A set of EEM fluorescence data containing 23 samples with four components was simulated to evaluate the performance of the developed strategy. Both excitation and emission profiles were generated by the Gaussian equation. The first 15 samples were used as a calibration set containing only three components of interest. The remaining 8 samples, served as a prediction set, contained three components of interest and an uncalibrated interference component. The simulated fluorescence peaks of four

Simulated EEM fluorescence data

Herein, the calibration transfer has been carried out from situation A to B, which simulates a calibration transfer of signal responses from different instruments. The objective of the standardization strategy is to get satisfactory analytical results with the significant reduction of experimental efforts, cost, and time for a full recalibration. To investigate the feasibility of the proposed strategy, two kinds of simulated EEM fluorescence data sets were created and divided into four possible

Conclusion

This work presented a chemometrics-assisted calibration transfer strategy to determine three agrochemicals in environmental samples by excitation-emission matrix fluorescence detection. With the aid of prominent “second-order advantage” provided by the SWANRF method, successful resolution together with satisfactory quantification results can be achieved, regardless of highly overlapped fluorescence signals of analytes and unknown interferences. Moreover, a standardization approach named PDS was

Declaration of competing interest

The authors declare that they have no conflict of interest.

Acknowledgments

The authors gratefully acknowledge the National Natural Science Foundation of China (Grant no. 21575039 and no. 21775039) and the Foundation for Innovative Research Groups of NSFC (Grant no. 21521063) for financial supports.

References (37)

F.K. Alsammarraie et al.
Rapid determination of thiabendazole in juice by SERS coupled with novel gold nanosubstrates
Food Chem.
(2018)
M. Fernandez et al.
Liquid chromatographic–mass spectrometric determination of post-harvest fungicides in citrus fruits
J. Chromatogr., A
(2001)
G. De Armas et al.
Sequential injection analysis for the determination of fuberidazole and thiabendazole by variable-angle scanning fluorescence spectrometry
Anal. Chim. Acta
(2001)
E.J. Llorent-Martínez et al.
Fluorimetric determination of thiabendazole residues in mushrooms using sequential injection analysis
Talanta
(2012)
X.-D. Sun et al.
Target-based metabolomics for fast and sensitive quantification of eight small molecules in human urine using HPLC-DAD and chemometrics tools resolving of highly overlapping peaks
Talanta
(2019)
L. Rubio et al.
Determination of cochineal and erythrosine in cherries in syrup in the presence of quenching effect by means of excitation-emission fluorescence data and three-way PARAFAC decomposition
Talanta
(2019)
R. Ríos-Reina et al.
Data fusion approaches in spectroscopic characterization and classification of PDO wine vinegars
Talanta
(2019)
M. Vosough et al.
Development of a fast HPLC-DAD method for simultaneous quantitation of three immunosuppressant drugs in whole blood samples using intelligent chemometrics resolving of coeluting peaks in the presence of blood interferences
J. Chromatogr. B
(2018)
N. Shekari et al.
Chromatographic fingerprinting through chemometric techniques for herbal slimming pills: a way of adulterant identification
Forensic Sci. Int.
(2018)
Y. Hu et al.
Interference-free spectrofluorometric quantification of aristolochic acid I and aristololactam I in five Chinese herbal medicines using chemical derivatization enhancement and second-order calibration methods
Spectrochim. Acta Part A Mol. Biomol. Spectrosc.
(2017)

F. Sales et al.

Standardization of a multivariate calibration model applied to the determination of chromium in tanning sewage

Talanta

(2000)

M.F. Abdelkader et al.

Calibration transfer of partial least squares jet fuel property models using a segmented virtual standards slope-bias correction method

Chemometr. Intell. Lab. Syst.

(2012)

A. Herrero et al.

Modelling the background current with partial least squares regression and transference of the calibration models in the simultaneous determination of Tl and Pb by stripping voltammetry

Talanta

(1998)

M. Khaydukova et al.

Multivariate calibration transfer between two different types of multisensor systems

Sens. Actuators B Chem.

(2017)

D. Brouckaert et al.

Calibration transfer of a Raman spectroscopic quantification method from at-line to in-line assessment of liquid detergent compositions

Anal. Chim. Acta

(2017)

F. Sales et al.

Multivariate standardization techniques using UV-Vis data

Chemometr. Intell. Lab. Syst.

(1997)

J. Thygesen et al.

Calibration transfer for excitation–emission fluorescence measurements

Anal. Chim. Acta

(2011)

R.S. Valverde et al.

Determination of tetracyclines in surface water by partial least squares using multivariate calibration transfer to correct the effect of solid phase preconcentration in photochemically induced fluorescence signals

Anal. Chim. Acta

(2006)

Cited by (13)

A calibration model transfer method based on quadrilinear model for excitation-emission matrix fluorescence measurements on different fluorescence spectrophotometers
2024, Microchemical Journal
In this work, a calibration model transfer method based on quadrilinear model for excitation-emission matrix fluorescence (EEM) data measurements on different fluorescence spectrophotometers was proposed. The proposed method can realize the calibration model transfer of EEM data on different fluorescence spectrophotometers through only one modeling, provided that the shape of the fluorescence spectrum of each target analyte in these data does not change significantly. The calibration sample data acquired on the 2^nd or 3^rd instrument was transposed to that obtained from the 1^st instrument using the calibration model transfer. Subsequently, the transferred data was used to build a calibration model and it was employed to predict the results of spiked prediction samples obtained from the 1^st instrument. The reliability of the method was demonstrated through the simulated EEM data and the real EEM data that came from three different fluorescence spectrophotometers (HITACHI F-7000, Edinburgh FS5 and HITACHI F-4600). In addition, when unknown interference was accidentally introduced during the measurement process of calibration samples, this method support the elimination of interference before implementing AQLD and subsequent calibration model transfer was not affected. The obtained EEM data after the calibration model transfer could be combined with the second-order calibration algorithm for the quantitative analysis of target analytes in complex systems.
Data standardization strategy to correct the effect of seasonality in the authentication of virgin olive oil
2023, Microchemical Journal
This study proposes a standardization strategy for dealing with seasonal variability in the authentication of extra virgin oils from the PDOs Les Garrigues and Siurana. Samples from four harvests were measured by fluorescence spectroscopy. A PLS-DA two-class model was developed and validated from samples from one of the harvests. When samples from three other harvests were predicted with the model developed, it was observed that the sensitivity and specificity were lower than when the model was validated. For the standardization process, we adapted the PDS technique to obtain the transfer function. The results obtained from the transformed spectra show that standardization is a good strategy for extending the usefulness of the models if the samples to be predicted are subject to seasonal variability.
Piecewise direct standardization assisted with second-order calibration methods to solve signal instability in high-performance liquid chromatography-diode array detection systems
2022, Journal of Chromatography A
Citation Excerpt :
In PDS, transfer parameters are calculated piecewise in a multivariate way by regressing the signal at each variable in the primary condition with signals of neighboring variables in a moving window on the other conditions, which can solve the problem of time shifts and signal intensity variances by using the moving window and the mathematical transfer formula [15]. According to relevant literatures, multivariate calibration transfer methods are often combined with different instruments such as near-infrared spectroscopy (NIR), electrochemical, UV-visible spectrophotometry, Raman spectroscopy, fluorescence spectroscopy and liquid chromatography-mass spectrometry (LC-MS) [17,22–30]. Zhang et al. proposed a parameter-free framework for calibration enhancement of NIR based on correlation constraint method, which solve the problem of NIR spectroscopy inconsistency in different environments and improve the predictive ability of the calibration model [22].
In this paper, in order to solve signal instability of high-performance liquid chromatography-diode array detection (HPLC-DAD) data and maintain the second-order advantage, this work proposed piecewise direct standardization (PDS) assisted with second-order calibration methods to analyze two different complex HPLC-DAD data with signal instability, including simulated HPLC-DAD data and the data of pesticide residues in saffron. Accurate quantitative results of target analytes can be obtained by PDS combined with alternating trilinear decomposition algorithm (ATLD) and alternating trilinear decomposition assisted multivariate curve resolution algorithm (ATLD-MCR) in the presence of signal instability with time shifts and changes of peak shape. Quantitative results of the model after calibration transfer are better than those of the model established by calibration sets and prediction sets with signal instability using ATLD algorithm and ATLD-MCR algorithm under different situations. Meanwhile, t-test was used to judge whether there are significant differences between these quantitative results of models. The performances of MCR-ALS algorithm were compared with that of PDS-ATLD method and PDS-ATLD-MCR method and the proposed methods have greater potential in dealing with the case of signal instability with time shifts and changes of peak shape. In a word, this methodology can reduce the number of calibration samples for recalibration and modeling, improve the efficiency of experiment, conform to the principle of green chemistry, and obtain satisfactory quantitative results in the presence of signal instability with time shifts and changes of peak shape.
Quantitative analysis of carbaryl and thiabendazole in complex matrices using excitation-emission fluorescence matrices with second-order calibration methods
2022, Spectrochimica Acta - Part A: Molecular and Biomolecular Spectroscopy
Citation Excerpt :
Meanwhile, other methods generally require complex pretreatments of the actual sample, which may result in the quality loss of target substances. By reading the relevant literatures, it is really that the excitation-emission fluorescence matrices combined with the second-order calibration method has not yet been used to simultaneously determine the concentrations of CAR and TBZ in fruit and environment samples [14,20–23]. This method can cleverly take advantage of the “second-order advantage” of the second-order calibration method to solve the aforementioned problems [24–26].
In this paper, a fast and efficient analytical strategy was proposed that chemometrics assisted with excitation-emission fluorescence matrices was used to quantify carbaryl (CAR) and thiabendazole (TBZ) in peach, soil and sewage. Even if there are serious overlapped peaks and unknown interferences in fluorescence analysis, the second-order calibration method based on alternating trilinear decomposition (ATLD) algorithm can be used to analyze CAR and TBZ in peach, soil and sewage. The recoveries of CAR and TBZ in peach are 110.4% and 99.7% and their standard deviations are lower than 2.1% and 0.3%, respectively. In addition, the accuracy of the method was assessed with figures of merit as well as intra-day and inter-day precision. The limit of detection, the limit of quantitation of CAR and TBZ in peach are 1.2 ng mL⁻¹ and 0.3 ng mL⁻¹, 3.5 ng mL⁻¹ and 0.8 ng mL⁻¹, respectively. And their root-mean-square error of prediction are 17.0 ng mL⁻¹ and 5.0 ng mL⁻¹ and there are high sensitivity and selectivity in this method. Meanwhile, the results obtained by ATLD algorithm were compared with those obtained by the self-weighted alternate trilinear decomposition algorithm (SWATLD) and the parallel factor analysis (PARAFAC) algorithm, and statistical methods such as the t-test, F-test and the elliptic joint confidence region were used to evaluate for analysis. There were no significant differences among these methods. At last, high performance liquid chromatography-fluorescence detector (HPLC-FLD) was used to evaluate the accuracy and reliability of the proposed method. These results are satisfactory and indicate that the proposed method can be used for accurate and rapid determination of pesticides in complex systems.
Deltamethrin determination in natural water samples via photochemically-induced fluorescence coupled to third-order multivariate calibration
2020, Microchemical Journal
Citation Excerpt :
For this reason, the coupling of this spectroscopic technique with modern chemometric tools, such as multivariate calibration, offers a powerful alternative for the determination of various chemicals. This approach has been used to satisfactorily determine pesticides [16,17], agrochemicals [18], and pharmaceuticals [19] in different biological and environmental matrices. In accordance with the modelled data type, different calibration strategies can be proposed, where the second-order/three-way calibration allows the analyte information (e.g., analyte concentration) to be obtained from an unselective instrumental signal (e.g., analyte signal overlapped with the responses of unexpected interferences), a property known as a second-order advantage [20].
In this work, an analytical method for deltamethrin determination in natural waters based on photochemically-induced fluorescence coupled to third-order/four-way calibration was evaluated and compared with second-order/three-way calibration. The four-way data were obtained during the photodegradation of deltamethrin in the form of excitation-emission fluorescence matrices and modelled by unfolded partial least squares coupled to residual trilinearization (U-PLS-RTL). According to the results, the third-order model resulted in a satisfactory fit and better figures of merit, even in the presence of unexpected interferences due to the additional dimension. In this way, the method presents a limit of detection of 2.9 µg L⁻¹ and a relative error of prediction of 15.8%. The optimization of a dispersive liquid-liquid microextraction (DLLME) procedure reached an enrichment factor (EF) of 5, improving the detection and quantification limits. Finally, the analytical method based on third-order multivariate calibration was applied to quantify this analyte in spiked natural water samples, both directly and after preconcentration. In all cases, the third-order property allowed us to satisfactorily model the data and quantify this compound in these complex matrices, demonstrating the superior analytical performance of the high-order data evaluated.
Recent applications of multiway calibration methods in environmental analytical chemistry: A review
2020, Microchemical Journal
Citation Excerpt :
Lozano et al. carried out the simultaneous determination of isoproturon, linuron, monuron and rimsulfuron in underground, tap water, and soil samples by using EEMs combined with U-PLS/RBL [135]. Sun et al. developed a second-order calibration method assisted EEMs strategy for simultaneous quantification of thiabendazole, indole-3-acetic acid and 1-naphthylacetic acid in complex soil and sewage samples [136]. In this work, the second-order calibration method effectively addressed the headaches of overlapping peaks and unknown interference and achieved successful resolution and acceptable quantitative results.
There has been an increasing concern on environmental issues in the worldwide over the last decade. Environmental pollution has diverse and substantial damages to public health. Multiway calibration has become a powerful tool in complex environmental matrices due to the distinctive “second-order advantage”. This review presents the recent progress of multiway calibration methods in environmental analytical chemistry. Various data types, available calibration models and corresponding algorithms, some fundamental issues and the recent application of multiway calibration methods in quantitative analysis of environmental samples were introduced in detail.

View all citing articles on Scopus

View full text

Chemometrics-assisted calibration transfer strategy for determination of three agrochemicals in environmental samples: Solving signal variation and maintaining second-order advantage

Highlights

Abstract

Introduction

Section snippets

Trilinear component model

Simulated EEM fluorescence data

Simulated EEM fluorescence data

Conclusion

Declaration of competing interest

Acknowledgments

Food Chem.

J. Chromatogr., A

Anal. Chim. Acta

Talanta

Talanta

Talanta

Talanta

J. Chromatogr. B

Forensic Sci. Int.

Spectrochim. Acta Part A Mol. Biomol. Spectrosc.

Talanta

Chemometr. Intell. Lab. Syst.

Talanta

Sens. Actuators B Chem.

Anal. Chim. Acta

Chemometr. Intell. Lab. Syst.

Anal. Chim. Acta

Anal. Chim. Acta