High-order calibration for the spectrofluorimetric determination of pesticides based on photochemical derivatization. A solution of the problems of inner-filter effects and matrix interferences in complex environmental water
Introduction
The increasing worldwide contamination of surface and ground waters with pesticides, as a result of the widespread use, becomes one of the severe environmental problems facing humanity [1], [2], [3], [4]. Although most of pesticides are present at low concentrations, they are toxic by design and the potential hazard for aquatic life and human health should not be neglected, such as chronic health effects including carcinogenesis, neurotoxicity as well as reproductive and developmental effects caused by long-term exposure to pesticides [5], [6], [7]. Therefore, it is of very great significance to explore highly sensitive and accurate quantification methods for analytically monitoring the transport, fate and effect of trace-level pesticides in aquatic environment.
To date, there are a growing number of analytical methods for the determination of pesticide residues in various environmental samples including water, sediments, air, and soil [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19]. Among them, liquid chromatography coupled to mass spectrometry (LC–MS) as a high-performance technique has gained wide acceptance for trace analysis [9], [10], [11], [12], [13], [14]. However, it is well-known that the LC–MS method requires skilled technicians and the use of expensive instrumentation both of which are not in the access of many laboratories in developing countries. The spectrofluorimetry methods as an excellent alternative have proved capable of providing high sensitivity and clean spectrum acquisition, and its cost is so much cheaper than MS detection that it can be easily available in the general case. In addition, the spectrofluorimetric equipment is relatively easier to run and especially appropriate for inaccessible regions in the area of that the trained technical personnel are not enough to be distributed to the analytical systems. The fluorescence method opens another opportunity for trace analysis of pesticides. Although the number of pesticides presenting an intrinsic fluorescence is limited, there are various kinds of derivatization techniques such as photochemical, chemical and electrochemical methods, through which the analyte with no or weak native fluorescence can converse into strongly fluorescent species and then be measured via the fluorescence detection (FD) [4], [19], [20], [21], [22]. Compared with chemical/electrochemical derivatization, the photochemical derivatization is a better technique to yield high the fluorescent photoproduct due to these two advantages: (1) UV radiation for relevant analysis is sufficiently simple and it can make the analyte(s) exhibit strong fluorescence with the use of the direct irradiation of the liquid solution containing the analyte(s), and (2) their chemical structure(s) of the fluorescent product(s) formed after UV irradiation is/are not necessary to be identified, provided that photo-chemically induced fluorescence signals are reproducibly yielded [23]. However, if without thorough previous separation procedures, when using the spectrofluorimetric method to directly analyze the pesticides in complex environmental samples, it will face the challenges such as serious interferences or the matrix effect (fluorescence quenching/enhancement which probably exists in systems).
To address the above-mentioned challenges, it could be a feasible and valid strategy by introducing more extra factors (i.e., analytical time, pH, sample volume, etc.) as new analytical dimensions to construct three-way or higher-way data which exploits more information inherent in sample in conjunction with chemometrics tools [16], [17], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48]. In the context of chemometrics, second-order calibration as a famous method, which allows the accurate quantification of analyte(s) in samples with unexpected components that are potential interferences, has been obtained wide application [24], [25], [26]. However, since third- or higher-order data introduce more information, third- or higher calibration methods can allow for better performance than second-order calibration, such as a theoretical increase in the predictive ability of the models, a high tolerance for heavily collinear data, the growing sensitivity and selectivity, and so on. In recent years, a series of new third-order calibration methods have been developed to resolve third-order data of complex samples [16], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48]. Some of them have also been extended for five-way data arrays in particular cases [17], [18]. Generally, they can be classified into two types. One is built on the direct least-squares procedure, for example, trilinear least-squares algorithm coupled with residual trilinearization (TLLS/RTL) [19], unfolded and N-way partial least-squares (N-PLS) and unfolded principal component analysis (U-PCA) combined with RTL (U-PLS/RTL [46], N-PLS/RTL [47] and U-PCA/RTL [48], respectively). The other is based on the iterative least squares, including 4-way parallel factor analysis (FPARAFAC) [5], four-way self-weighted alternating normalized residue fitting (4-way SWANRF) [14], alternating penalty quadrilinear decomposition (APQLD) [15], alternating weighted residue constraint quadrilinear decomposition (AWRCQLD) [17], alternating quadrilinear decomposition (AQLD) [16] as well as regularized self-weighted alternating quadrilinear decomposition (RSWAQLD) [8]. These algorithms have been used to successfully extract the qualitative and quantitative information of single component or multi-components in complex practical systems. However, the algorithms belonging to the second type whose objection functions are constructed according to the fully stretched matrices along single mode present relatively slow convergence rate, such as FPARAFAC and 4-way SWANRF. In addition, they are generally sensitive to the over-estimated number of components used in calculation. But this problem can be addressed to some extent when the exact number of components is provided by some kinds of diagnostics, such as core consistency [49], split half analysis [50], the ADD-ONE-UP truncating and fitting method (ADD-ONE-UP) [51] and so on. By contrast, the approaches, involving APQLD, AWRCQLD, AQLD and RSWAQLD, which are based on the pseudo-fully unfolded matrices along single mode have been considered as great alternatives to overcome the above-mentioned drawbacks. However, the penalty factors of chosen in both of APQLD and AWRCQLD need to be very large (more than 106 as the authors suggested), sometimes the convergence of them may be slow and meaningless, especially when the collinearity is extremely serious. And the application of AQLD is restricted when the noise level is too high. It follows that no best alternative with all the advantages can successfully resolve different third-order data problems in any cases, up to now. In other words, the further development of third-order calibration algorithm is urgently needed and full of significance.
In this paper, we propose a new third-order calibration method, namely alternating fitting weighted residue quadrilinear decomposition (AFWRQLD), whose four objective functions are built according to the pseudo-fully unfolded matrices along two different directions. To the best of our knowledge, this is the first attempt to develop a third-order algorithm based on the incompletely unfolded matrix forms of the quinquelinear model which are stretched along different modes. A detailed study comparing AFWRQLD with FPARAFAC and AWRCQLD by using a set of simulated data and an already published experimental data array was carried out to reveal the merits of the proposed algorithm. In addition, combined with photochemical derivatization, the newly developed method has been employed to explore the spectrofluorimetric analysis of quantifying the level of a model pesticide, imidacloprid, with no native fluorescence in environmental water. Since the matrix effect from environmental water leads to the partial fluorescence quenching of the fluorescent photochemical derivative of imidacloprid, AFWRQLD has been further extended for the resolution of the emission–excitation–UV irradiation time–volume of environmental water fourth-order data array. The satisfactory results demonstrate that the combination of the new algorithm and photochemical derivatization can achieve highly sensitive and accurate quantitative detection of the pesticide with no native fluorescence in complex environmental water.
Section snippets
Different forms of the quadrilinear component model
An emission–excitation fluorescence matrix (EEFM) with size of I × J for the lth sample at the kth UV irradiation time point (UVITP) can be experimentally obtained with I emission wavelengths and J excitation wavelengths (Fig. 1A, left). If the EEFMs for the samples with the same concentration level at K different UVITPs are collected along the K mode, a three-way data array (I × J × K) can be produced (Fig. 1A, middle). If L samples, consisting of L0 calibration samples and LP prediction samples,
Simulated data
A simulated four-way data array produced by a two-dimensional liquid chromatography system with diode array detection on ten samples with four species which has been given in Ref. 42 was employed to evaluate the performance of the newly proposed algorithm. The first chromatographic profiles a1–a4 were produced by:where i = 1, 2,. .., 30; gs (x, a, b, i) refers to the value at x of a Gaussian function with standard deviation a and centre b, i.e.gs
Choice of constraint coefficients of pa, pb, pc and pd
The constraint coefficients of pa, pb, pc and pd were firstly determined according to the results yielded by using the AFWRQLD algorithm to decompose the simulated four-way data arrays when running 100 times. The results indicate that the algorithm is insensitive to the excess factors used for calculations and exhibits fast convergence rate no matter how the parameters vary. However, when the parameters pa = pb = pc = pd < 104, the AFWRQLD algorithm sometimes suffered from “swamps” in the convergence
Conclusion
In summary, a quantitative spectrofluorimetric method by photochemical derivatization and high-order calibration has been developed for the determination of imidacloprid as a model pesticide in river water. A new third-order calibration algorithm AFWRQLD with excellent performance has been firstly proposed by constructing four objective functions according to the pseudo-fully unfolded matrices of quadrilinear model along two different directions. In addition, the fourth-order AFWRQQLD algorithm
Acknowledgments
This work was supported by the National Natural Science Foundation of China (Nos. 21565012, 21205021 and 21365009), the Guangxi Natural Science Foundation (No. 2014GXNSFAA118033), the Scientific Research Fund of Guangxi Education Department (No. 201203YB090), the Open Foundation of State Key Laboratory of Chemo/biosensing and Chemometrics in Hunan University (No. 2014005) as well as the Project of High Level Innovation Team/Outstanding Scholar and Key Laboratory of Food Safety and Detection in
References (52)
- et al.
Ionic liquid functionalized gold nanoparticles: synthesis, rapid colorimetric detection of imidacloprid
Sensors Actuators B
(2014) - et al.
Development of a novel alternating quadrilinear decomposition algorithm for the kinetic analysis of four-way room-temperature phosphorescence data
Chemom. Intell. Lab. Syst.
(2014) - et al.
A novel fourth-order calibration method based on alternating quinquelinear decomposition algorithm for processing high performance liquid chromatography–diode array detection– kinetic-pH data of naptalam hydrolysis
Anal. Chim. Acta
(2015) - et al.
Unfolded partial least-squares with residual quadrilinearization: a new multivariate algorithm for processing five-way data achieving the second-order advantage. Application to fourth-order excitation–emission–kinetic–pH fluorescence analytical data
Chemom. Intell. Lab. Syst.
(2011) - et al.
Simultaneous determination of imidacloprid and carbendazim in water samples by ion chromatography with fluorescence detector and post-column photochemical reactor
Talanta
(2013) Recent developments in electrochemical flow detections—a review: part II. Liquid chromatography
Anal. Chim. Acta
(2011)- et al.
A review of post-column photochemical reaction systems coupled to electrochemical detection in HPLC
Anal. Chim. Acta
(2010) - et al.
Second-and third-order multivariate calibration: data, algorithms and applications
Trends Anal. Chem.
(2007) - et al.
Multi-way chemometric methodologies and applications: a central summary of our research work
Anal. Chim. Acta
(2009) - et al.
A review on second- and third-order multivariate calibration applied to chromatographic data
J. Chromatogr. B
(2012)
MVC3: a MATLAB graphical interface toolbox for third-order multivariate calibration
Chemom. Intell. Lab. Syst.
Multidimensional voltammetry: Four-way multivariate calibration with third-order differential pulse voltammetric data for multi-analyte quantification in the presence of uncalibrated interferences
Chemom. Intell. Lab. Syst.
Simultaneous determination of tyrosine and levodopa in human plasma using enzyme-induced excitation-emission-kinetic third-order calibration method
Chemom. Intell. Lab. Syst.
Exploiting third-order advantage using four-way calibration method for direct quantitative analysis of active ingredients of Schisandra chinensis in DMEM by processing four-way excitation–emission-solvent fluorescence data
Chemom. Intell. Lab. Syst.
Excitation-emission-kinetic fluorescence coupled with third-order calibration for quantifying carbaryl and investigating the hydrolysis in effluent water
Talanta
Fluorescence quantification of tetracycline in the presence of quenching matrix effect by means of a four-way model
Talanta
Four-way modeling of 4.2 K time-resolved excitation emission fluorescence data for the quantitation of polycyclic aromatic hydrocarbons in soil samples
Talanta
An alternative quadrilinear decomposition algorithm for four-way calibration with application to analysis of four-way fluorescence excitation-emission-pH data array
Anal. Chim. Acta
Trilinear least-squares and unfolded-PLS coupled to residual trilinearization: new chemometric tools for the analysis of four-way instrumental data
Chemom. Intell. Lab. Syst.
Efficient way to estimate the optimum number of factors for trilinear decomposition
Anal. Chim. Acta
Intra-laboratory testing of method accuracy from recovery assays
Talanta
The challenge of micropollutants in aquatic systems
Science
Pesticides in U.S. streams and rivers: occurrence and trends during 1992–2011
Environ. Sci. Technol.
Streams in the urban landscape
Annu. Rev. Ecol. Syst.
The role of indirect photochemical degradation inthe environmental fate of pesticides: a review
Environ. Sci. Process. Impacts
Pesticides: an important but underused model for the environmental health sciences
Environ. Health Perspect.
Cited by (16)
High sensitivity on-site early warning system monitoring of pesticides by photo-induced fluorescence
2023, Analytica Chimica ActaThe monitoring of chemical pesticides pollution on ecological environment by GIS
2021, Environmental Technology and InnovationRecent applications of multiway calibration methods in environmental analytical chemistry: A review
2020, Microchemical JournalCitation Excerpt :Qing et al. extended PARAFAC and ATLD to five-way and proposed five-way PARAFAC and alternating quinquelinear qecomposition (AQQLD) for handling five-way data array [18]. Besides, AFWRQQLD has been proposed to analyze the five-way data of emission-excitation-UV irradiation time-volume-sample for accurate quantification of imidacloprid in water samples [38]. Currently, there are many commonly used software packages and toolboxes with graphical user interfaces (GUIs) that are available for multiway calibration, including N-way Toolbox [42], MVC2 GUI [43], MVC3 GUI [44,45], GUIPRO GUI [46], MCR-ALS GUI [47,48] and MCRC software [49], they are freely available on the internet.
Recent advances in chemical multi-way calibration with second-order or higher-order advantages: Multilinear models, algorithms, related issues and applications
2020, TrAC - Trends in Analytical ChemistryCitation Excerpt :AQQLD was first developed by Qing et al. for the processing of fourth-order data, which were obtained by recording the HPLC-DAD data of a sample set as function of the reaction time and pH. It also has the advantage of fast convergence rate and being insensitive to the excessive component number [30]. Then AFWRQQLD as a variant of AQQLD was used for the analysis of imidacloprid in environmental waters to deal with the five-way data, which were obtained by recording EEM at different UV irradiation times for different volumes of sample [87]. CAMLD as an extension of CAQLD to M-way data array was proposed by Kang and coworkers, the related iterative formulas were also introduced [70].
Inner filter effect in fluorescence spectroscopy: As a problem and as a solution
2019, Journal of Photochemistry and Photobiology C: Photochemistry ReviewsCitation Excerpt :The corrected EEMF is used for advanced multi-linear analysis; the authors propose CDA to be a relevant pre-treatment step in PARAFAC analysis of highly absorbing samples. Nie and group have proposed a third-order calibration algorithm known as “Alternating Fitting Weighted Residue Quadrilinear Decomposition (AFWRQLD)”, for analysis of fourth order “excitation-emission-UV irradiation time-volume” data array [73]. Satisfactory quantification of spiked pesticides in river water samples had been achieved using AFWRQLD method, even in presence of IFE and other matrix interferences.