Study of Quality Control and Uncertainty in Multiresidue Pesticide Estimation in Packaged Drinking Water Using GC- MS/MS

 

S. D. Garway, D. G.Garway, Y. T.Gaikwad, R.M.R. Azad, G.H. Pandya*

Research and Development Division, Anacon Laboratories Pvt. Ltd., F-34,35, MIDC, Butibori, Nagpur 441122, India

*Corresponding Author E-mail: pandyagh@rediffmail.com

 

 

 

1. INTRODUCTION:

Consumption of packaged drinking water for drinking purposes has been increasing considerably in the country. Apart from water shortages at times, real and perceived needs to safeguard health has also contributed to an escalating trade in package drinking water at the national and international level. Considering the consumer’s health and safety it has become imperative to ensure that the package water offer for sale is safe and free from harmful organisms. One of the requirement under Indian Standard for Packaged Drinking water(IS 14543:2004) ¹ is that the pesticide residue  limits considered individually should not be more than 0.0001 mg/L.The total pesticide residue should not be more than 0.0005 mg/L. Such stringent limits require development and quality control of analytical method for accreditation by internationally established organization such as NABL under ISO/IEC 17025. The paper reports the study carried out for development of quality control in the analysis of package drinking water for pesticide residues. Consideration has also been given in the Prevention of Food Adulteration Act, 1954 and the Rules framed there under regarding the presence of pesticide in samples. Many pesticides remain as residues in foodstuffs after their application, and they can be widespread in the environment (soils, surface and underground waters) ^{2, 3, 4}.

 

Methods to determine pesticide residues include the extraction of the analytes from the matrix, appropriate cleanup of the raw extracts and subsequent determination by gas chromatography (GC) or liquid chromatography (LC).Globalization of commodities market and concerns for the consumer has put pressure on regulatory agencies to increase pesticide monitoring programs in terms of specificity of analysis and number of samples analyzed. These demands have caused the development of methods to reliably and rapidly detect as many pesticides as possible from a single extraction. Many of the methods cover a limited number of pesticides, analyzed by multiple gas chromatographic detectors ECD, NPD etc., thus requiring multiple injections of each sample.

 

Multi residue GC/MS methods come close to meeting the needs of regulatory agencies since they allow for the determination of a broad spectrum of pesticides and metabolites in a variety of food extracts. Specificity is provided by the combination of chemical structure information and retention time on an analytical GC column. These techniques are also sensitive, precise and sufficiently accurate to be useful for regulatory purposes, while being cost effective and rapid.

 

In this work, GC was coupled to a triple quadrupole mass spectrometer. The application of selected reaction monitoring (SRM) in the analyzer allows one to obtain a sensitivity and selectivity gain over single ion monitoring (SIM) because the fragmentation reaction implies two different characteristics of the target compound. The Triple Quadropole analyzer also assists the development of fast chromatographic methods because it makes it possible to monitor a high number of co-eluted compounds. The aim of this study was to develop and validate a GC/-MS/MS for determining 15 pesticide residues in packaged drinking water. The target compounds were determined in less than 21 min after extraction. The number of transitions per segment was increased and the dwell time reduced, in order to achieve the lowest possible running time for the simultaneous analysis of the target compounds. The use of GC- MS/MS makes the chromatographic analysis faster.  In consequence, this method is suitable for application in routine analysis of packaged drinking water samples where a high sample throughput is required.

 

EXPERIMENTAL:

Chemicals and Standards:

Pesticide reference standards were obtained from Dr. Ehrenstorfer GmbH,Augsburg, Germany. Pesticide quality chromatography grade solvents (hexane, acetonitrile, diethyl ether and methanol) were purchased from J.T.Baker, USA and Merck, India respectively. Special grade anhydrous sodium sulfate was heated to 700ºC for 8 hours, cooled and then used for analysis. Purified RO grade water with a conductivity of 0.5 µSi/cm and Florisil: (60/100 mesh) heated at 130ºC for 8 hours then cooled slowly in a desiccators was utilized during cleanup.

 

Apparatus:

A Thermo Trace GC Ultra gas chromatograph (Thermo Fisher Scientific Instruments, San Jose, CA95134, USA) with electronic flow control (EFC) was used. A Thermo Fisher Scientific TSQ Quantum GC triple quadrupole mass spectrometer was coupled to the gas chromatograph. Samples were injected with a 1 ul syringe, into a split/splitless septum-equipped injector .A fused-silica analytical capillary column TR 5 – MS, 30m x 0.25mm was used .The mass spectrometer was operated in electron ionization (EI) mode at 70 eV. NIST library was applied for identification of pesticides. The mass spectrometer mass scale was calibrated with perfluorotributylamine every 3 days. Helium (99.999%) at a flow rate of 1 mL/min was used as carrier gas; argon (99.999%) at a pressure of 2 mTorr was used as collision gas. Thermo Workstation software XCaliber was used for instrument control and data analysis.

 

The GC operating conditions were : Injector temperature was 240^oC, column temperature was  60º(hold for 1 min), increased at  15 ⁰C/min to 180ºC(hold for 1 min), then to 240 C(hold for 1 min ) at 8ºC/min.  MS with EI Ionization voltage: 70eV.The mass spectrometer was operated in the Full scan and MS/MS (SRM) mode. The ionization current was 25μA.The ionization voltage was 70 eV. The temperatures of the transfer line, manifold and ionization source were set at 280, 40 and 220^oC, respectively. The analysis was performed with a filament-multiplier delay of 4 min in order to prevent instrument damage. The scan time was 0.01s in segments 1–10. Rotary vacuum evaporator Model Evator, from Medica Instruments, India was used for concentration of the samples.

 

Sample Preparation:

The packaged drinking water samples were obtained from the local markets. 50 g of sodium chloride was added to 1L of water sample. 60 ml of 15% Diethyl ether in n-Hexane was added to the sample bottle and shaken for 10 minutes. The hexane layer was separated. The extraction was repeated twice, combining the hexane layers, dehydrating with anhydrous sodium sulfate, filtering and concentrating (reducing) the hexane solution to 5ml with rotary vacuum evaporator. The extract was cleaned using Florisil column and further concentrating this solution to 1ml by blowing nitrogen gas across the surface of the solution. 1µl of the extract was used for injection into GC-MS/MS for analysis. A blank sample was prepared as an analytical control by using the same procedure as described above.

 

Stock standard solution was(1.00 ug/ul) prepared from pure certified standard reference material by accurately weighing  0.01 g of pure material on a 5 decimal place analytical balance. The material was dissolved in n-Hexane and volume made up to 10 ml in certified volumetric flask. The standards were stored at low temperature in a freezer. The calibration standard were at five  concentration levels for each compound by adding appropriate volume of one or more stock standards to a volumetric flask and diluting to volume with n- Hexane. While preparing working standards, a record was kept of the identity and amount of all solutions and solvents employed. The standards were labeled indelibly, allocated an expiry date, and stored at low temperature in the dark in containers that prevent any loss of solvent and entry of water.

 

In order to carry out the quantitative analysis of the samples with GC-MS/MS, electronic ionization (EI) mode was used due to its applicability for all the target compounds. The MS/MS conditions were fixed for each compound, trying to select as precursor ion the one with highest m/z ratio (greater selectivity) and abundance (greater sensitivity) (Table 1).The same approach was applied to choose the product ions used in the quantitation. In this work, one transition per compound was monitored for the majority of the pesticides. It is essential that the scan time should be fast enough to register all the programmed transitions or reactions in the minimum chromatographic separation time. The scan time was set at 0.010 s for all the segments. Identification and confirmation of the target compounds was based on the use of retention time of the chromatographic peak of the analyte. The RTs were established for all the pesticides under study. Confirmation of the analytes was performed by comparing the MS/MS sample product ion spectrum with a reference spectrum obtained under the same experimental conditions.

 

 

 

TABLE 1 MS/MS CONDITIONS FOR ORGANOCHLORINE PESTICIDES

 

 

Fig. 1  Calibration curve for   α –HCH                                    ( A)

 

RESULTS AND DISCUSSION:

In order to carry out multiresidue pesticide analysis, it was necessary to develop an in-house quality control program for ongoing analysis of spiked samples. Ongoing data quality checks were compared with established performance criteria to meet the performance characteristics of the method. The multi-residue analysis of pesticides in water samples require validation of all procedures (steps) that were undertaken in the method. This required assessment of linearity, recovery (as a measure of trueness or bias) and precision.

 

Linearity was studied in the range 10–200 μg/L with five calibration points by matrix-matched standard calibration. Calibration curves for all the 15 pesticides were developed in the 10–200 μg/L range. Figure 1 illustrates the calibration of α –HCH. The resulting chromatogram and mass spectra of α –HCH. is summarized in Figure 2.  Linear calibration graphs were constructed by least-squares regression of concentration versus relative peak area of the calibration standards. Linearity values, calculated as determination of correlation coefficient (r²), were in the range 0.9814–0.9999. The deviation of the individual points from the calibration curve was lower than 20%.

 

 

(A)

 

(B)

Fig. 2.    Chromatogram (A) and Mass Spectra (B)  of  α –HCH

 

 

Accuracy was evaluated in terms of recovery by spiking blank samples of packaged drinking water with the corresponding volume of the multi compound pesticide working standard solution. Total of seven samples, one on each day, were spiked with a concentration of 0.09 ug/L. The samples were than   processed for analysis by GC-MS. The results of day to day analyses are summarized in Table 2. Recoveries between 97 and 109 % were found in water samples. The intraday precision was expressed as percent relative standard deviation for each pesticide analysed. The minimum RSD was 4.70 (β- HCH) and the maximum 14.24 (2, 4’ DDD). Therefore, these results meet the requirement criteria of trueness or mean recovery for quality control ⁵

 

Method detection limits (MDL) were also determined for all the pesticides under this study. It provides a useful mechanism for illustrating the capability of the analytical method. MDLs were calculated for the pesticides as follows:

 

The sample standard deviation is multiplied by the correct Student's t-value from the statistical Tables

 

In the present study seven replicates were taken, hence six degrees of freedom was considered. t is found to be 3.143. The MDL was calculated for a compound like α – HCH is as follows:

MDL= (s)(t-value)= 0.0093 x 3.143= 0.0293 ug/L. Rounding to the correct number of significant figures, the calculated MDL becomes 0.029 ug/L.

 

Similarly, LOQs were subsequently established as 10 times the Standard Deviation of the recovered pesticide. The limit of quantitation was also calculated as :

 

LOQ= 10 x (s)= 10 x 0.0093 = 0.0934 ug/L

 

The MDL and LOQ were thus calculated for all the pesticide under study and are summarized in Table 2.

 

Attempt was also made to estimate the uncertainty associated with the multiresidue analytical method in water matrix by applying a bottom-up approach. All data appearing in this study complies with NABL 17025 requirements. It was implemented in our laboratory as a pesticide residue analysis routine method and our laboratory was accredited. The uncertainty of each step was estimated identifying which of them are relevant in the global uncertainty analysis are illustrated by a cause and effect diagram as shown in Figure 3.The parameters of the measure and are represented by the main branches in the diagram. Further factors are added to the diagram, considering each step in analytical procedure. The values and uncertainties are shown in Table 3.

The relevant uncertainty sources in the pesticide residue The standard uncertainties associated with each step is quantified by estimating analyte concentration from the calibration curve, calculating recovery of the sample extract. After obtaining the standard uncertainty (u(x)), expressed as a standard deviation, and combined standard uncertainty were determined.

 

The different aspects explained above for estimating the combined uncertainties have been applied to the multiresidue of 15 pesticides in water.

 

Table 3.summarises the relevant information for calculating uncertainties associated with the preparation of primary standard solutions, volumetric materials, and analytical balance.

 

 

TABLE 2  RECOVERY STUDY FOR ORGANOCHLORO PESTICIDES IN PACKAGED DRINKING WATER

Name of Pesticide	Day 1 Conc. μg/L	Da y 2 Conc. μg/L	Day 3 Conc g/L	Day 4 Conc. μg/L	Day 5 Conc. μg/L	Day 6 Conc. μg/L	Day 7 Conc. μg/L	SD	RSD	Mean	Recovery  %	MDL   μg/L	LOQ μg/L	Combined uncertainty Uc
α HCH	0.076	0.086	0.100	0.103	0.086	0.095	0.088	0.0093	10.31	0.0905	100.63	0.0293	0.093	0.0291
β-HCH	0.092	0.100	0.098	0.101	0.098	0.089	0.100	0.0046	4.70	0.0968	107.61	0.0143	0.045	0.0292
γ-HCH	0.101	0.075	0.091	0.095	0.09	0.093	0.100	0.0086	9.38	0.0921	102.38	0.0271	0.086	0.0291
γ  HCH	0.08	0.094	0.097	0.107	0.095	0.097	0.102	0.0084	8.71	0.096	106.66	0.0262	0.083	0.0292
Aldrin	0.089	0.088	0.086	0.08	0.091	0.093	0.085	0.0043	4.89	0.0874	97.142	0.0134	0.042	0.0291
2,4’-DDE	0.095	0.081	0.100	0.105	0.098	0.098	0.105	0.0096	9.70	0.0992	108.25	0.0302	0.096	0.0292
α-Endosulfan	0.077	0.087	0.103	0.106	0.095	0.102	0.096	0.0102	10.70	0.0951	105.71	0.0319	0.101	0.0292
4,4’- DDE	0.092	0.076	0.087	0.110	0.101	0.110	0.094	0.0124	12.91	0.0957	106.34	0.0388	0.123	0.0292
Dieldrin	0.098	0.074	0.098	0.108	0.092	0.110	0.092	0.012	12.50	0.096	106.66	0.0376	0.120	0.0291
2,4’- DDD	0.085	0.073	0.110	0.110	0.096	0.089	0.092	0.0133	14.24	0.0935	103.96	0.0418	0.133	0.0291
β-Endosulfan	0.102	0.086	0.097	0.11	0.1	0.099	0.091	0.0072	7.24	0.0994	108.73	0.0226	0.072	0.0292
4,4’-DDD	0.087	0.091	0.107	0.077	0.104	0.101	0.097	0.0106	11.13	0.0948	105.39	0.0331	0.105	0.0291
2,4’- DDT	0.087	0.091	0.091	0.078	0.1	0.112	0.097	0.0107	11.45	0.0937	104.12	0.0337	0.1073	0.0292
Endosulfan sulfate	0.076	0.085	0.104	0.107	0.1	0.102	0.096	0.0112	11.74	0.0957	106.35	0.0353	0.112	0.0292
4,4’-DDT	0.09	0.11	0.095	0.099	0.101	0.102	0.094	0.0065	6.35	0.1015	109.68	0.0202	0.064	0.0292

 

 

 

Fig. 3   Cause and effect diagram for pesticide analysis

 

TABLE 3 VALUES AND UNCERTAINITIES IN PESTICIDES MEASUREMENTS

Description	Value x	u(x)	u(x)/x
Mass of the pesticide(ug)	10000	8.165x 10-2	8.165 x10-5
Volume of flask(mL)	10	0.1079	1.079x 10-2
Volume of pipette(mL)	1	2.3 x10-2	2.3 x10-2
Volume of pipette(mL)	0.2	2.3x10-2	1.1x 10-2
Repeatability	Standard Deviation/√3	4.85x10-3	7.020x10-4
Calibration	0.9916	3.99x10-3	4.023x10-3

 

 

In some cases, it is feasible to use relative uncertainties which represent the value of the uncertainty normalized. It is obtained as the quotient between the standard uncertainty u(x) and the value of x:

 

Urel(x) =        or  urel(x) =   

The uncertainty estimation was carried as per the following steps:

(1) Specifying the measurand. This involved making a clear statement of what is being measured, including the relationship between the measurand and the input quantities (measured quantities, constants and calibration standard values. (2) Identifying uncertainty sources i.e listing the possible sources of uncertainty, usually specified in the above step.(3)  Quantifying uncertainty components i.e. estimating the uncertainty component associated with each potential source of uncertainty identified. The different contributions to the overall uncertainty is expressed as standard deviation which is calculated depending on the data available  from a standard deviation value( this value is directly used); from a coefficient of variation; from the standard deviation of experimental data sets; from a declared purity and  uncertainty value(which is given in a certificate of calibration for reference materials) and  from a correlation coefficient of calibration curves etc.(4)  Calculate combined uncertainty by combining different contributions to the overall uncertainty according to the appropriate rules. The combined standard uncertainty u(f) is calculated as

 

u(f ) =   [c²(x)u²(x) + c²(y)u²(y)+· · ] ^½

 

Where c is a sensitivity coefficient associated to each one of variables, given by the partial derivative of the function: c(x) = ∂f/∂x. and (5)Expanded uncertainty by applying the appropriate coverage factor.

 

The combined uncertainty and expanded uncertainty were calculate for all the 15 pesticides under study.

For sake of illustration, calculation of uncertainty in analyte concentration for a pesticide like  α HCH is as follows :

 

Where,  u(P) is the uncertainty in purity of the pesticide  as quoted in the suppliers certificate, u(m) is the uncertainty in the mass of the pesticide in the certified reference standard solution, u(V_flask ) is the uncertainty in volumetric measurements estimated by considering the influences of calibration, repeatability and temperature effects and determined by measurement of uncertainty in internal volumes and variation in filling volumetric flask to the mark, u(v_pip1) and u(V_pip2) are the uncertainties in volumetric measurements using  pipettes , u(rep) is the uncertainty in repeatability estimation considering the standard deviation of replicate divided by √3. U(calib) is the uncertainty in calibration determined from the coefficient of correlation obtained for the calibration curve. U (recov) is the uncertainty in recovery of spiked samples involving precision and the homogeneity of the pesticide sample. The precision was determined by measuring the standard deviation of a set of spiked samples, which were extracted and analysed each day.

 

u_c(c α HCH) = c _{α HCH}  { 8.677 x 10^-6 +6.667 x 10^-6 +1.164 x10^-4  + 5.409 x 10^-4 + 1.38 x 10^-2 + 4.928 x 10^-7 + 1.619 x 10^-5 + 4.85 x 10 ^-3  +4.147 x 10^-2} ^1/2  

                     = c _{α HCH}  x 0.3241

 

The expanded uncertainty U(C _{α HCH}) was subsequently determined to develop an interval within which the value of the measurand may lie. A  factor of  2  was thus used for obtaining  a confidence level of  95%.

 

U(C _{α HCH}) = 2 u_C (c _{α HCH} ) = 0.6482 (c _{α HCH} )

Where c _{α HCH} is the concentration of analyte such as α HCH, expressed in ug/L.

 

The developed method was applied to real samples of packaged drinking water with several internal quality controls to ensure that the measurement process is under statistical control. Each batch of samples was processed together with a reagent blank, composed of only solvent. The reagent blank was obtained by performing the whole process without a sample. The majority of recoveries were in the range 70–110%. Calibration curves were prepared daily and the determination coefficients must be higher than 0.98 and the deviation curve in the first and second calibration points must be lower than 20%.

 

The developed method was validated in order to ensure the feasibility of the method for its application in routine pesticide analysis of packaged drinking water. Parameters such as linearity, recovery, precision and confirmation parameters such as MDLs, LOQs and uncertainty were studied.

 

REFERENCES:

1.       IS 14543:2004, Indian Standard “Packaged Drinking Water (Other than Packaged Natural Mineral Water), Bureau of Indian Standards, New Delhi, 2004

2.       UNEP : Global report on regionally based assessment of persistent toxic substances, ENEP Chemicals, Geneva, Switzerland(2002).

3.       Oxynos,K.,J.Schmitzer and A.Kettrup: Guidelines for environmental specimen banking in the Federal Republic of Germany, Federal Environmental Agency, Berlin(1989).

4.       WHO: Public Health Impact of pesticide used in agriculture, WHO in collaboration with Unep, Geneva(1990).

5.       SANCO: Quality control procedures for pesticide residue analysis. European Commission, Directorate General Health and Consumer Protection. Document No SANCO/10232/ 2006, March 2006.

 

 

Received on 27.01.2012        Modified on 25.02.2012

Accepted on 18.03.2012         © AJRC All right reserved