SciELO - Scientific Electronic Library Online

 
vol.56 número3SYNTHESIS, CHARACTERIZATION AND BIOLOGICAL EVALUATION OF TRIAZOLE AND FUSED TRIAZOLE DERIVATIVESSYNTHESIS AND ANTITUBERCULAR ACTIVITY OF PYRIDAZINONE DERIVATIVES índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Journal of the Chilean Chemical Society

versión On-line ISSN 0717-9707

J. Chil. Chem. Soc. vol.56 no.3 Concepción  2011

http://dx.doi.org/10.4067/S0717-97072011000300012 

J. Chil. Chem. Soc., 56, N° 3 (2011), págs.: 774-777.

 

PARTIAL LEAST SQUARES FOR SIMULTANEOUS DETERMINATION OF FE (III) AND HG(II) IN WATER AND PHARMACEUTICAL PREPARATIONS BY RP-HPLC USING 1,4-BIS-(4-PYRIDIL)-2,3-DIAZO-1,3-BUTADIENE CHELATING AGENT

 

MOHAMMADREZA KHANMOHAMMADI *, MAJID SOLEIMANI, MAJID AFSHAR, AMIR BAGHERI GARMARUDI

Chemistry Department, Faculty of Science, IKIU, Qazvin, Iran. e-mail: mrkhanmohammadi@gmail.com


ABSTRACT

A method has been developed for simultaneous determination of Fe(III) and Hg(II) by RP-HPLC, utilizing partial least squares technique. Vertex C18, with UV-visible detector was applied in reversed phase high performance liquid chromatography, while 1,4-bis-(4-Pyridil)-2,3-diazo-1,3-butadiene (4-bpdb) was added to the mobile phase as the chelating agent. Several effective parameters e.g. mobile phase make up, buffers, concentration of 4-bpdp and pH salts were investigated in order to optimize the quantification conditions. The optimum condition was achieved at 25°C, 10% methanol containing mobile phase, 30 mmol L1 acetate buffer, 0.05 mmol L1 4-bpdb and pH=4.5. Partial least squares chemometric technique was applied. The root mean square error (RMSE) of independent test set in the optimized model was 0.024 and 0.393 for Fe(III) and Hg(II) respectively.

Keywords: Determination; Complexation; 1,4-bis-(4-Pyridil)-2,3-diazo-1,3-butadiene; Mercury; Iron ; RP-HPLC, PLS.


INTRODUCTION

Quantitative and qualitative analysis of heavy metals e.g. mercury in different media is of particular importance. The toxicity of mercury is known to be highly dependent on its chemical form. In many of the reported methods, a pre-concentration step should be included in the analysis procedure in order to achieve a final concentration level matching the detection limits accessible by applied detection technique1. Mercury has been also considered as a human health hazard. In the other hand, iron is of great interest to different fields of science. It is strongly involved in global biogeochemical cycling2. Several efforts have been taken for development of methods which provide accurate and precise separation and determination of these metal ions in biological, pharmaceutical, food and environmental samples during the last decades. There are several analytical methods described in the literature for the determination of mercury and iron at low concentrations in environmental and biological samples3. Although the most widely used determination method for the speciation of metals e.g. mercury in environmental samples is gas chromatography with capture electrode or with other atomic element specific detectors4,5, the use of HPLC for mercury speciation has the advantage of simplified sample preparation. In GC analysis, it is essential to form volatile, thermally stable derivatives, whereas this is not necessary for HPLC1. During the last decades high-performance liquid chromatography (HPLC) in combination with various detection techniques has been extensively applied to the speciation of a number of metal ions. Many chromatographic techniques have been applied for the analysis of Hg(II) and Fe(III), e.g. thin layer chromatography, ion exchange chromatography, ion pair chromatography and micro HPLC. Combination of chemometrics with analytical chemistry can improve selectivity of determination, optimize the experimental conditions, raise the analytical operation efficiency and provide scientific information. Hence, it is rapidly attracting analyst attention and is used for simultaneous determination of multi-components in recent years6. Among the various chemometric approaches applied to multi-component analysis, partial least squares (PLS) has been successfully adopted in many quantitative assays. PLS is a factor analysis method which allows establishing a relationship between matrices of chemical data7. One of the clearest explanations of this method was given by Haaland and Thomas8. In PLS, the response data decomposition is weighted to the concentration. The aim of this work was to establish a simple, selective, sensitive, and robust RP-HPLC method for simultaneous determination of Fe(III) and Hg(II) in real samples, applying 1,4-bis-(4-Pyridil)-2,3-diazo-1,3-butadiene (4-bpdb) (Figure 1) as a new ligand. The obtained data were processed by PLS technique.


EXPERIMENTAL

1.Apparatus and Materials

The HPLC (KNAUER, Germany) instrument was equipped with a Mod.7125 Rheodyne injector with a 20 ì! external loop and a K2500 UV-Visible detector at 228 nm wavelength. Vertex (KNAUER Co.) reversed-phase C18 analytical column (250mmx4.6mm i.d., 5ìç1 particle size) was applied at 25 °C. Isocratic elution was performed (flow-rate: 1 ml min1). A B2000 pH meter equipped with a GCFC 11 combination glass electrode was used for pH measurements. The 4-bpdb chelating agent was prepared by the method reported previously9. HPLC grade methanol was from Merck (Germany). All solutions were prepared by HPLC grade water, being filtered by a 0.45ì PTFE filter. Other reagents were of analytical grade unless stated otherwise.

2.General Analysis Procedure

A 0.05 mmol L-1 solution of 4-bpdb was prepared by methanol:water (10:90) as the mobile phase. The pH was adjusted at 4.5 using 30 mmol L1 of sodium acetate-acetic acid buffer. The capacity factors were calculated using k = (tr-t0)/t0, where tr is the average retention time of sample solute and to is that of a non-absorbed substance (nitrate ion here).

3.Partial Least Squares (PLS)

PLS regression is an important multivariate calibration tool based on the use of a large number of variables, which permits to evaluate the concentration of interesting analytes. The PLS method is an important multivariate calibration tool that has been growing in importance for the last years and has been incorporated in new analytical chemistry textbooks10. PLS regression is based on the resolution of two initial multivariate matrices, R (response matrix) and C (concentration matrix), by projection onto smaller matrices T and U (or R and C score matrices, respectively). They contain the coordinates of the objects on the new axes or PLS components, with orthogonal columns, and relates the information in the response matrix R to the concentration matrix C, through correlation between R and C covariance matrices. In this work, R represents the independent variables (the original chromatographic data of the calibration set), while C represents the dependent variables (concentration of analytes in the calibration set). The determination of a significant number of model dimensions (number of PLS principal components) was made by cross-validation. The PLS method was employed using chromatographic data.11.

RESULTS AND DISCUSSIONS

1. Investigating Experimental Chromatography Conditions

Standard solution of each analyte and a binary mixture solution were prepared by dissolving Hg(CH3COO)2 and Fe(NO3)3.H2O in the mobile phase (methanol:water, 10:90). As observed in figure 2, Hg(II) demonstrates 2 chromatographic signals according to formation of [M(4-bpdb)] and [M(4-bpdb)2] complexes but Fe(III) does not form any complex and thus appears at t0 In RP-HPLC, the behavior of analytes is estimated according to their polarity, because the retention is mainly dependent on the partition between non-polar stationary phase and polar mobile phase.


According to preliminary experiments, acetate buffer was selected as the suitable reagent for the experimental. Phosphate, citrate, formiate and acetate buffers (30 mmol L1) were separately added to water:methanol (90:10) containing 0.1 mmol L-1 of 4-bpdb chelating agent. Phosphate causes Hg to sediment, citrate would demonstrate long time of analysis (more than 20 min) and the chromatographic signals of Hg (II) and Fe (III) would highly overlap using formiate as the buffer. In the next step, the effect of pH on separation of Fe(III) and Hg(II) ions was studied in the 2.5-7.5 range, using acetate buffer solution (30 mmol L-1), HCl (1 mol L-1) and NaOH (1 mol L-1). It was observed that gradually increment of eluent's pH from 2.5 to 5.5 would increase the resolution of analytes' signals but higher pH is no more effective (Figure 3).


In addition, buffer concentrations over 40 mmol L1 are not influential on the resolution of analytes while concentrations below 20 mmol L1 would decrease it (Figure 4). Thus pH=4.5 and buffer concentration of 30 mmol L1 were set as the optimum condition. Effect of eluent composition on resolution was studied changing the water:methanol ratio from 90:10 to 50:50 v/v. (containing 0.05 mmol L1 (4-bpdb) at pH=4.5). Increasing the methanol content would decrease the peak widths and resolution. A perfect separation (Rs=3) was obtained by 10% methanol content (Figure 5). Concentration of 4-bpdb was varied (0.0063-0.3000 mmol L1) in the mobile phase. In 0.05 mmol L1 4-bpdb ligand concentration, analysis time and peak widths are optimum. The increasing of (4-bpdb) concentration over 0.05 mmol L-1 would decrease the resolution and sensitivity (according to high matrix absorbance) (Figure 6).




2. Classical RP-HPLC Determination of Hg (II) and Fe (III)

Figure 2-c shows 3 main signals around 2 and 11 min; The first signal is related to Fe(III) and others of due to Hg(II). Classical calibration models were constructed, applying binary standard solutions of Fe(III) (0.2-1.0 mmol L1) and Hg(II) (3.00-15.00 mmol L1). Both of the Hg(II) chromatographic signals were observed to be linearly related to the analyte's concentration. As the interference of Fe(III) and Hg(II) chromatographic signals at 1.50-3.50 min region would cause some errors in classical determination, the 1st and 3rd chromatographic signals (Figure 2-c) were used for quantitative determination of Fe(III) and Hg(II) ions respectively. The obtained quantification equations for analytes according to classical method are as:

where h is the signal height and C is the analyte's concentration. R2 and RSD (3 times analysis) for Fe (III) and Hg (II) were (0.993, 0.935%) and (0.992, 1.257%) respectively. The LOD obtained by CLOD = 3 Sb where Sb is standard deviation of blank for 10 replicate determination and m is the slop in the regression equations. The LOD was 4.11x103־ mmol L1 and 9.14x102־ mmol L1 for Fe(III) and Hg(II), respectively.

3.Partial Least Squares Technique

The main advantage of multivariate analysis using Partial Least Squares (PLS) calibration is its ability in fast determination of the components in mixtures especially with signal overlapping. Among chemometrics methods, partial least squares technique (PLS) has been used frequently according to quality of the obtained calibration models, ease of its implementation and software availability12. Thus it was very useful to take the advantage of chemometrics in our research, improving the idea by its benefits.

4.Calibration Model for RP-HPLC Analysis, Utilizing PLS

A set of 20 binary standard solutions were provided (0.20-1.00 mmol L-1 of Fe and 3.00-15.00 mmol L-1 of Hg). Twelve standard solutions were used to set up the calibration model and the remaining 8 solutions were used as the validation set to evaluate the calibration models. The investigated retention time region was 1.80-4.30 min. Chromatograms were obtained according to the procedure detailed before. The root mean square error (RMSE) is an indication of average error for each of components in analysis and thus was calculated. Tables 1 and 2 show the actual and predicted concentration of analytes in addition with R2 and RMSE for both calibration and validation samples. The constructed calibration model was optimized by varying the number of PLS factors from 1 to 6. Cross validation method, leaving out one sample at a time, was used to select the optimum number of factors in the PLS algorithm. Cross validation makes the calibration set as large and representative as possible. The prediction error sum of squares (PRESS) was calculated each time a new factor was added. The optimum number of factors would yield the minimum PRESS. In the finalized model, optimum number of factors was 4 in the model for both of analytes. The PLS algorithm was also applied to an independent data set consisting of 6 samples (Fe(III): 0.30-0.80 mmol L-1 and Hg(II): 3.00-15.00 mmol L-1) in the proposed method. Root mean square error of test (RMSET) for independent data set was 0.024 and 0.393 for Fe(III) and Hg(II) respectively.



5.Analysis of Real Samples and Recovery Study

In order to evaluate the ability of the proposed method for quantification of Fe(III) and Hg(II) in real analysis, variety of samples consisting of mineral water and pharmaceuticals (tablet and syrup) were analyzed, applying standard addition and recovery study. RSD of analytical procedures was determined by 5 times of replication in each experiment. Results are detailed in tables 3 and 4.



CONCLUSIONS

A new method was introduced for separation and simultaneous determination of Fe(III) and Hg(II), which produced reliable and reproducible results. Different chemical matrices were also analyzed by the proposed method, obtaining acceptable data. PLS chemometric technique provides the ability in performing the determination process in the first 5 minutes of elution times, resolving the problem of interfering chromatographic signals. The analytes can be sensitively determined without any matrix ion effect. The PLS benefited RP-HPLC method can be applied for simultaneous determination of Fe(III) and Hg(II) in different environmental samples.

REFERENCES

1.Morita M.; Yoshinaga, J.; Edmonds, J.S. Pure Appl. Chem. 1998, 70, 1585.         [ Links ]

2.de Baar H.J.W., de Jong J.T.M., in The Biogeochemistry of Fe in Seawater (IUPAC Book Series on Analytical and Physical Chemistry of Environmental Systems), edited by Turner, D.; and Hunter, K., JohnWiley & Sons, Chichester, 2001.         [ Links ]

3.Marco, S.G.; Cremonini, M.A.; Esteban, P.; Yunta, F.; Apaolaza, L.H.; Placucci, G.; Lucena, J.J. J. Chromatogr. A 2005, 1064, 57.         [ Links ]

4.Johanson, M.; Emteborg, H.; Glad, B.; Reinholdsson, F.; Baxter, D.C. Fresenius J. Anal. Chem. 1995, 351, 461.         [ Links ]

5.Hadad, G.M.; El-Gindy, E.; Mahmoud, W.M.M. Spectrochim Acta A 2008, 70, 655.         [ Links ]

6.Fang, G.; Liu, N. Anal. Chim. Acta 2001, 445, 245.         [ Links ]

7.Ragno, G.; Ioele G.; Risoli, A. Anal. Chim. Acta 2004, 512, 173.         [ Links ]

8.Haaland, D.M.; and E.V. Thomas, Anal. Chem. 1988, 60, 1193.         [ Links ]

9.Kesslen, E.C.; Euler, W.B. Tetrahedron Lett., 1995, 36, 4725.         [ Links ]

10.Kellner, R.; Mermet, J.M.; Otto, M.; Valcarcel, M.; Widmer, H.M. Analytical Chemistry, (Wiley-VCH, Weinheim, 2004).         [ Links ]

11.Pylypiw, H.M.; Grether, M.T. J. Chromatogr. A 2000, 883, 299.         [ Links ]

12.Goicoechea H.C.; Olivieri, A.C. Analyst 2001, 126, 1105.         [ Links ]


Running Head: RP-HPLC determination of Fe(III) and Hg(II) by 4-bpdb chelating.

(Received: January 3, 2011 - Accepted: April 12, 2011).