Micro-Extraction and Detection/Quantification of Trace Pesticides in Various Matrices

Micro-extraction and detection/quantification of trace pesticides in various matrices

JOSEPH MOSOTHO GEORGE

THESIS

Submitted in fulfilment of the requirements for the degree

DOCTOR OF PHILOSOPHY

CHEMISTRY

in the

FACULTY OF SCIENCE

of the

UNIVERSITY OF JOHANNESBURG

Supervisor: Prof DBG Williams Co-supervisor: Dr L Marjanovic

June 2012

AFFIDAVIT: MASTER’S AND DOCTORAL STUDENTS TO WHOM IT MAY CONCERN

This serves to confirm that I______Full Name(s) and Surname

ID Number______

Student number______enrolled for the

Qualification______

Faculty of Science______Herewith declare that my academic work is in line with the Plagiarism Policy of the University of Johannesburg which I am familiar.

I further declare that the work presented in the ______(minor dissertation/dissertation/thesis) is authentic and original unless clearly indicated otherwise and in such instances full reference to the source is acknowledged and I do not pretend to receive any credit for such acknowledged quotations, and that there is no copyright infringement in my work. I declare that no unethical research practices were used or material gained through dishonesty. I understand that plagiarism is a serious offence and that should I contravene the Plagiarism Policy notwithstanding signing this affidavit, I may be found guilty of a serious criminal offence (perjury) that would amongst other consequences compel the UJ to inform all other tertiary institutions of the offence and to issue a corresponding certificate of reprehensible academic conduct to whomever request such a certificate from the institution.

Signed at Johannesburg______on this ______of ______2012

Signature______Print name______

STAMP COMMISSIONER OF OATHS Affidavit certified by a Commissioner of Oaths This affidavit conforms with the requirements of the JUSTICES OF THE PEACE AND COMMISSIONERS OF OATHS ACT 16 OF 1963 and the applicable Regulations published in the GG GNR 1258 of 21 July 1972; GN 903 of 10 July 1998; GN 109 of 2 February 2001 as amended.

Table of Contents List of Figures...... viii

List of Tables ...... xiii

Acknowledgements ...... xv

List of abbreviations ...... xvi

Synopsis ...... xix

Chapter 1: Introduction and literature review ...... 1

1.1 Introduction to the thesis ...... 1

1.2 Applications of chemicals, their fate and social aspects in everyday life ...... 2 1.2.1 Applications of chemicals in agriculture ...... 2 1.2.2 Fate of pesticides in plants or the environment and their effects thereon ...... 3 1.2.3 Effects of pesticides on human health ...... 3 1.2.4 Legal, political and social aspects of pesticide use and misuse ...... 4

1.3 Analysis of pesticides and other organic compounds ...... 5 1.3.1 Analytical method development and validation ...... 6 1.3.2 Chromatographic methods of analysis ...... 8 1.3.2.1 Column chromatography ...... 8 1.3.2.2 Gas chromatography...... 9 1.3.3 Detectors for chromatography ...... 10 1.3.3.1 Mass spectrometry as a detector for chromatographic analysis ...... 12 1.3.4 Hyphenated/coupled instrumental methods ...... 12 1.3.4.1 Hyphenation of HPLC and MS ...... 13 1.3.4.2 Hyphenation of GC with MS ...... 13 1.3.5 Quantitative analysis in chromatography ...... 14 1.3.6 Recent developments in chromatography ...... 15

1.4 Sample preparation techniques for chromatography analysis ...... 16 1.4.1 Liquid-based sample preparation techniques ...... 17 1.4.1.1 Soxhlet extraction ...... 17 1.4.1.2 Supercritical extraction techniques ...... 18

1.4.1.3 Solvent micro-extraction ...... 20 1.4.2 Solid-based sample preparation techniques ...... 25 1.4.2.1 Solid-phase extraction (SPE)...... 25 1.4.2.2 Solid-phase micro-extraction SPME ...... 27 1.4.3 Dispersive methods ...... 28 1.4.3.1 Dispersive solvent micro-extraction ...... 28 1.4.3.2 Dispersive solid-phase extraction (D-SPE) ...... 29 1.4.4 Sample preparation involving chemical reactions ...... 30 1.4.4.1 Derivatisation-coupled sampling ...... 30 1.4.4.2 Chelation and complex formation ...... 32

1.5 Mass-transfer in solvent microextraction ...... 33

1.6 Objectives of the study ...... 35

1.7 References ...... 37

Chapter 2: Application of the bubble-in-drop single drop micro-extraction

(BID-SDME) method for monitoring of herbicides from farm areas ...... 47

2.1 Introduction ...... 47

2.2 Experimental ...... 49 2.2.1 Reagents and chemicals ...... 49 2.2.2 Standard solutions ...... 49 2.2.3 Preparation of soil samples ...... 50 2.2.4 Instrumentation ...... 50 2.2.5 Micro-extraction procedure ...... 51 2.2.6 Performance of the analytical method for metolachlor ...... 51 2.2.7 Sampling of the soil samples ...... 51 2.2.8 Evaluation of extraction recovery ...... 52

2.3 Results and discussions ...... 52 2.3.1 Development and validation of BID-SDME method for triazines analysis ...... 53 2.3.2 Validation of the BID-SDME method for accuracy using atraton ...... 55

iii

2.3.3 Validation of the BID-SDME method for reproducibility ...... 57

2. 4 Application of the BID-SDME method in the analysis of metolachlor and atrazine in agricultural soil samples ...... 59 2.4.1 Extraction and analysis for monitoring of metolachlor and atrazine from farm soil. 59 2.4.2 Extraction efficiency of metolachlor using BID-SDME...... 62 2.4.3 Linear dynamic range and detection limits for metolachlor ...... 62 2.4.4 Application of the BID-SDME method in monitoring of metolachlor and atrazine from agricultural soils ...... 64 2.4.5 Exploration of more robust extraction methods for soil samples combined with BID- SDME as a clean-up step ...... 69 2.4.5.1 Hot water extraction (HWE) ...... 70 2.4.5.2 Superheated water extraction ...... 72 2.4.6 Analysis of breakdown products for both atrazine and metolachlor ...... 73

2.5 Recovery of herbicides from the soil samples for quantitative analysis ...... 76 2.5.1 Evaluation of spiking protocol ...... 78 2.5.1.1 The effect of spiking using aqueous surrogate (trace-less spiking) ...... 82 2.5.1.2 Recovery of the atrazine breakdown products (desisopropyl atrazine – DIPA, desethylatrazine – DEA) ...... 84 2.5.1.3 The effect of period post spraying on recovery ...... 85

2.6 Quantitation of the amounts of the herbicides in the soil ...... 87

2.7 Exploration of samples whose spraying history was unknown ...... 89 2.7.1 Analysis of soil samples from a field with unknown history ...... 89 2.7.2 Analysis of stream water samples ...... 93 2.7.2.1 Analysis of water samples from stream 1 ...... 93 2.7.2.2 Analysis of water samples from stream 2 ...... 96

2.8 General discussions and conclusions ...... 99

2.9 References ...... 100

Chapter 3: Application of BID-SDME in the analysis of organochlorine pesticides in water ...... 102

3.1 Introduction ...... 102

3.2 Experimental ...... 104 3.2.1 Chemicals ...... 104 3.2.2 Apparatus ...... 105 3.2.3 Instrumentation ...... 105 3.2.4 Extraction procedures...... 105

3.3 Experimental results and discussions ...... 106 3.3.1 Chromatographic separation and identification of the compounds ...... 106 3.3.2 Optimisation of the extraction method...... 108 3.3.2.1 Solvent choice – effect of various solvents on extraction efficiency ...... 108 3.3.2.2 The effect of increasing ionic strength ...... 109 3.3.2.3 The effect of pH of the solution on extraction ...... 110 3.3.2.4 Extraction – time profiles using chloroform and dichloroethane ...... 111 3.3.2.5 The effect of stirring on extraction efficiency ...... 112 3.3.2.6 The effect of temperature on extraction using dichloroethane ...... 114 3.3.2.7 Optimised condition and enrichment factors ...... 115 3.3.3 Validation of the method ...... 116 3.3.3.1 Linearity, limits of detection and quantification ...... 116 3.3.3.2 Accuracy validation ...... 121 3.3.3.3 Reproducibility and repeatability validation ...... 124 3.3.4 Application of the method to environmental samples ...... 126

3.4 General discussions ...... 129

3.5 Conclusions and recommendations ...... 130 3.6 References…………………………………..……………………………………………………………………….. 132

Chapter 4: Development and application of mixed-solvent BID-SDME for determination of some growth hormones from bovine urine...... 133

4.1 Introduction ...... 133

4.2 Experimental ...... 135 4.2.1 Chemicals and standards used ...... 135 4.2.2. Apparatus ...... 135 4.2.3 Chromatography and mass spectrometry instrumentation ...... 136 4.2.4 Micro-extraction procedure ...... 136 4.2.5 Evaluation of extraction recovery from other biological matrices ...... 138

4.3 Results and discussion ...... 138 4.3.1 Chromatographic method development ...... 138 4.3.2 GC-MS analysis of the hormones ...... 139 4.3.3 The effect of solvent on extraction of hormones ...... 141 4.3.4 Effect of bubble size on both individual and mixed solvent systems ...... 143 4.3.5 Effect of NaCl on the extraction efficiency using mixed-solvent system ...... 145 4.3.6 The effect of sodium chloride and air-bubble size extraction efficiency ...... 146 4.3.7 Extraction-time profile using different salt and solvent composition ...... 147 4.3.8 The effect of temperature on air bubble and solvent mixture on extraction efficiency ...... 149 4.3.9 Characterisation of the urine for particulate/solutes matter...... 150 4.3.10 Effects of NaCl at ng/mL level using the matrix-matched samples ...... 151 4.3.11 The effect of pH on extraction efficiency ...... 153 4.3.12 Determination of enrichment factor at optimum conditions...... 154 4.3.13 Determination of linearity, limits of detection and quantification ...... 156 4.3.14 Evaluation of reproducibility ...... 159 4.3.15 Evaluation of the accuracy of the method...... 161

4.4 Application of the method to other biological matrices ...... 163

4.5 General discussions and conclusions ...... 164

4.6 References ...... 166

Chapter 5: Development of a dispersed solvent-assisted headspace ...... 168

5.1 Introduction ...... 168

5.2 Experimental ...... 169 5.2.1 Chemicals and standards used ...... 169 5.2.2 Apparatus ...... 170 5.2.3 Chromatography and mass spectrometry instrumentation ...... 170 5.2.4 Extraction procedure ...... 171

5.3 Results and discussions ...... 171 5.3.1 Development of chromatography and identification of compounds ...... 171 5.3.2 Demonstration of the proof of concept ...... 172 5.3.3 Choice of solvent ...... 174 5.3.4 Optimisation of the added solvent volume ...... 176 5.3.5 The effect of ionic strength of the aqueous solution ...... 178 5.3.6 Extraction as function of time ...... 179 5.3.7 Exploration of smaller volume of samples...... 180 5.3.8 Evaluation of different fibre phases ...... 181 5.3.9 Comparison of different solvent for use as dispersants ...... 181

5.4 General discussions and conclusions ...... 182

5.5 References ...... 183

Chapter 6: General Conclusions ...... 184

vii

List of Figures

Figure 1 Photograph of the ‘bubble-in-drop’ single-drop micro-extraction (BID-SDME) arrangement ...... 53 Figure 2 Calibration curves for some representative triazines from the US-EPA TP-619 herbicide mixture ...... 54 Figure 3 A standard curve for the accuracy validation ...... 57 Figure 4 A chart showing reproducibility of extraction over four days ...... 58 Figure 5 A TIC chromatogram of metolachlor standard (1 µg/mL) ...... 59 Figure 6 The expanded SIM chromatogram showing both reference and qualifying ions for metolachlor ...... 60 Figure 7 TIC chromatogram of 9-week post-spraying sample from farms ...... 60 Figure 8 An extracted ion chromatogram for diphenylamine (IS) and metolachlor ...... 61 Figure 9 The SIM chromatogram of the TP-619 mixture using the m/z values from the scanning mode ...... 61 Figure 10 Standard curves for metolachlor and atrazine following BID-SDME extraction..... 63 Figure 11 A map showing the places where the samples were collected ...... 64 Figure 12 An expanded picture showing the field that was sampled ...... 65 Figure 13 Moisture content of the soil as a function of number of weeks post spraying ...... 66 Figure 14 A TIC chromatogram for the extract of the soil sample collected on week 1 post spraying ...... 67 Figure 15 Dissipation-time profile for metolachlor in the sprayed field ...... 68 Figure 16 Evaluation of different extraction methods for soil sample (22 Jan 2010) ...... 70 Figure 17 Evaluation of hot-water and hot-water-sonication assisted extraction modes ..... 71 Figure 18 Dissipation-time profile from soil samples extracted with hot-water extraction .. 73 Figure 19 A TIC chromatogram of desethyl-atrazine ...... 74 Figure 20 The mass spectrum of desethyl-atrazine ...... 74 Figure 21 An EIM chromatogram of the TP-619 mixture including metolachlor and desethylatrazine ...... 75 Figure 22 Dissipation-time profiles for atrazine and DEA from the soil following HWE ...... 75 Figure 23 A chart showing the influence of spiking metolachlor into the soil ...... 76

viii

Figure 24 A comparison of solvent extraction with hot-water coupled to BID-SDME ...... 78 Figure 25 The effect of solvent composition of the spiking solution on extraction ...... 81 2 2 Figure 26 The standard curves for H5-atrazine and H6-metolachlor respectively ...... 87 Figure 27 A photograph of the farm whose spraying history was unknown ...... 89 Figure 28 A TIC chromatogram of the soil sample post BID-SDME extraction ...... 90 Figure 29 The extracted ion chromatogram of the ions of interest from the soil sample ..... 90 Figure 30 An MS spectrum of the peak 200 on the extracted ion chromatogram from soil sample...... 91 Figure 31 An MS spectrum of peak 214 on the extracted ion chromatogram from soil sample ...... 91 Figure 32 An MS spectrum of the peak 162 from the extracted ion chromatogram of the soil sample...... 92 Figure 33 Analysis of the herbicides present in the field whose history is unknown ...... 92 Figure 34 A photograph of stream 1 showing where water sample was collected ...... 93 Figure 35 A TIC chromatogram of stream-water (stream 1 – collected 14 Feb 2010) sample94 Figure 36 An EIM chromatogram of 169, 172, 200 (m/z values for DIPA, DEA and atrazine) 94 Figure 37 The MS spectrum of the 2nd peak (blue trace) ...... 95 Figure 38 The MS spectrum of the 3rd compound ...... 95 Figure 39 Dissipation profiles of atrazine and herbicides in the run-off stream water – stream 1 ...... 96 Figure 40 A map picture showing stream 1 where water samples were collected ...... 96 Figure 41 A TIC chromatogram of the water from stream 2 (14 Jan 2010) ...... 97 Figure 42 An expanded EIM chromatogram using ions of interest ...... 97 Figure 43 The MS spectrum of the peak at 9.8 minutes with 84% similarity with NIST 2007 library for atrazine ...... 98 Figure 44 A TIC chromatogram for water from stream 2 collected on week 11 post spraying (22 Jan 2010) ...... 98 Figure 45 A EIM chromatogram of water from stream 2 (22 Jan 2010) with an internal standard ...... 98 Figure 46 A total ion chromatogram of the OCP 625 mixture ...... 106 Figure 47 An EIM chromatogram of mixture showing m/z values ...... 106

Figure 48 The mass spectra of the various organochlorine compounds and their chemical formulae ...... 107 Figure 49 The extraction efficiency of different solvents on extraction ...... 108 Figure 50 Effect of sodium chloride content on extraction efficiency using 1 µL of DCE ..... 109 Figure 51 The effect of pH on extraction efficiency of organochlorines ...... 110 Figure 52 Extraction-time profile of the organochlorines using dichloroethane ...... 111 Figure 53 The extraction-time profile using chloroform as an extracting solvent ...... 112 Figure 54 Effect of stirring on BID-SDME extraction using dichloroethane ...... 113 Figure 55 Effect of stirring on extraction efficiency using simple SDME ...... 114 Figure 56 The effect of temperature on extraction efficiency of organochlorines ...... 115 Figure 57 Standard curves for the organochlorine pesticides following BID-SDME extraction ...... 117 Figure 58 Mass spectrum of typical multi-organochlorinated compound (aldrin) ...... 119 Figure 59 Mass spectrum of a single chlorinated compound (propazine) ...... 119 Figure 60 Individual chromatograms for a 1 ng/mL standard post extraction with BID-SDME ...... 121 Figure 61 A TIC chromatogram of the CRM mixture ...... 122 Figure 62 An extracted ions chromatogram of the CRM mixture ...... 122 Figure 63 Mass spectra of DDD and DDT from the NIST library (version 07) ...... 122 Figure 64 mass spectrum of DDT ...... 123 Figure 65 Mass spectrum of DDD ...... 123 Figure 66 Reproducibility of the extraction method over 5 days using 5 different samples 125 Figure 67 A satellite picture of the place from which the dam water was collected ...... 127 Figure 68 A TIC (black) together with an EIM chromatogram (pink) with m/z 243, the base 2 peak for H8-DDT ...... 127 2 Figure 69 The mass spectra of the two peaks from H8-DDT standard (100 µg/mL) ...... 128 2 Figure 70 Mass spectra of the two H8-DDT peaks after extraction from spiked water ...... 128 Figure 71 The chromatogram of the water extract from Limpopo dam ...... 128 Figure 72 The GC-FID chromatogram of the HEX and DES mixture, showing an additional peak identified as dienestrol (DNS) ...... 139 Figure 73 An EIM GC-MS chromatogram for the mixture of HEX (135) and DES (268) ...... 140 Figure 74 Mass spectrum of hexestrol (95% similarity with NIST 05 library) ...... 140

Figure 75 Mass spectrum of peaks 1 and 2 (88% and 92% similarities to DES respectively) 141 Figure 76 The effect of solvent of extraction efficiency from aqueous samples ...... 142 Figure 77 Effect of mixed-solvent composition on extraction efficiency ...... 143 Figure 78 The effect of bubble size and solvent composition ...... 144 Figure 79 Effect of NaCl content on extraction efficiency using 75% chloroform mixture .. 146 Figure 80 Effect of NaCl content and bubble size on the extraction of 5 µg/mL hormones solution ...... 147 Figure 81 The extraction-time profiles for BID-SDME using different solvent and NaCl compositions ...... 148 Figure 82 An expanded picture showing the extraction-time profile with 10% w/v sodium chloride ...... 149 Figure 83 the effect of temperature and air bubbles on extraction efficiency...... 150 Figure 84 A SIM chromatogram of the urine extracted without spiking ...... 152 Figure 85 Effect of NaCl content on extraction efficiency at 50 ng/mL hormones spiking level ...... 152 Figure 86 The effect of pH on extraction of the hormones ...... 153 Figure 87 An expanded chromatogram of a 50 ng/mL solution of hormones in methanol . 155 Figure 88 An expanded chromatogram of 50 ng/mL extracted using the optimized conditions for BID-SDME ...... 156 Figure 89 Concentration versus relative response for the hormones ...... 156 Figure 90 A SIM chromatogram of a 0.05 ng/mL mixture ...... 158 Figure 92 An expanded SIM chromatogram for the 0.05 ng/mL mixture following extraction ...... 158 Figure 93 Reproducibility of the extractions using four solutions extracted over four days 160 Figure 94 Relative reproducibility of four solutions extracted over four days ...... 161 Figure 95 A TIC chromatogram of the mixture of 5 chlorophenols ...... 171 Figure 96 The SIM chromatogram showing the 5 components and their corresponding m/z values ...... 172 Figure 97 The effect of temperature on headspace extraction of aqueous and methanol solutions ...... 173 Figure 98 Headspace extraction of aqueous and MeOH solutions after diethylether spiking ...... 174

Figure 99 Effect of different solvents on organic solvent-assisted headspace extraction ... 175 Figure 100 Effect of the organic solvent volume on efficiency ...... 176 Figure 101 Repeatability of the extraction following addition of 100 µL of diethylether .... 178 Figure 102 Effect of ionic strength on extraction ...... 179 Figure 103 The extraction-time profile for solvent assisted headspace sampling ...... 179 Figure 104 The effect of organic solvent volume on 0.5 mL solution ...... 180 Figure 105 Extraction as a function of a fibre phase ...... 181

xii

List of Tables

Table 1 Clinical symptoms of some pesticides on human exposure ...... 4 Table 2 A table showing different groups on individual triazines and their CAS numbers .... 53 Table 3 Analytical data from the calibration curves (0.05 - 0.5 ng/mL spiking) ...... 55 Table 4 Relative responses for atraton standard and CRM extractions using 100 ng/mL DPA internal standard ...... 56 Table 5 Relative responses for atraton and CRM extractions using 5 ng/mL DPA internal standard ...... 56 Table 6 Data for accuracy validation using standard curve approach and 0.5 ng/mL CRM ... 56 Table 7 Combined relative extraction results for reproducibility over four days ...... 58 Table 8 Assessment of extraction efficiency using metolachlor standard ...... 62 Table 9 Some calibration data from the standard curves ...... 63 Table 10 A table showing presence of herbicides as a function of soil depth ...... 73 Table 11 The effect of spiking solution on extraction of the analytes ...... 79 Table 12 Recovery data of the surrogate and analytes from different solutions ...... 80 Table 14 Recoveries of the analytes and surrogates from different samples spiked using aqueous standard solution obtained by direct comparison ...... 83 Table 14 Normalised recoveries of the surrogate herbicides with and without suspension . 84 Table 15 Recoveries of the atrazine breakdown products following spiking ...... 85 Table 16 Recoveries of deuterated standards for soils collected at different times ...... 86 Table 17 Obtained concentrations (ng/mL) from spiked soil samples collected at different periods ...... 87 Table 18 Quantitative data for different samples obtained from the calibration curves using recovery data ...... 88 Table 20 The enrichment factors for the organochlorines using the optimum extraction conditions ...... 116 Table 21 Analytical parameters obtained from the calibration curves of the organochlorines mixture ...... 117 Table 21 S/N ratios for 1 ng/mL organochlorines standard mixture with the corresponding relative response ...... 120

xiii

Table 22 Calculated concentrations from the standard curves compared to the CRM ...... 124 Table 25 Relative responses obtained during reproducibility studies of the organochlorines extraction over four days using three different solutions ...... 126 2 Table 24 Recoveries of from dam water from Limpopo using set at around H8-DDT standard ...... 129 Table 27 The amount of particulate matter in the urine following air drying ...... 151 Table 26 Enrichment factors of hormones from various conditions using 50 ng/mL solution ...... 155 Table 27 Table showing the analytical data obtained from the calibration curves ...... 157 Table 28 A table showing some hormones and the matrix from which they were determined from ...... 159 Table 29 Reproducibility of the extractions using 1 ng/mL hormones solutions: individual and relative extractions ...... 160 Table 30 Relative responses for extractions of the CRM and the corresponding calculated concentrations ...... 161 Table 31 The comparison of the calculated concentrations to the CRM ...... 162 Table 32 Different samples and their pH values...... 163 Table 33 Comparison of extraction efficiencies of the hormones from water, urine and milk samples ...... 164

xiv

Acknowledgements I owe deep debt of gratitude to a number of people without whose support and encouragement this would have been much harder to achieve.

Firstly my wife Camilla, and our two beloved kids Nomsa and Themba for their immeasurable support and always keeping me in their prayers. Even in physical absence their spiritual presence was a constant reminder to keep me going!

My supervisor and academic father Prof. D. Bradley G. Williams without whose advice and patient guidance this would not have been possible. The financial support he afforded me is also gratefully appreciated.

My co-supervisor Dr. Ljiljana Marjanovic, thanks to her patience and sacrifices she had to make to accommodate me. It has not been the easiest road, but she remained steadfast. Her invaluable insight in Analytical Chemistry is highly appreciated.

Prof. Luke Chimuka of the University of Witwatersrand for extending a welcoming hand to use his facilities for the carrying out some of the analysis.

Colleagues, staff and students in the Departments of Chemistry and Biochemistry; those silly jokes, the words of encouragement, and sometimes the hurting truth, kept the zeal. Hope God gives you strength to face all challenges the academic life puts ahead of you. May you all receive the Lord’s blessings aplenty!

The University of Johannesburg is further acknowledged for the much needed funding for the project and bursary through the Research Office. The National University of Lesotho for affording me the opportunity to pursue these studies and financing my dependants’ allowance.

Above all, The Almighty God for the strength and guidance to carry on despite the challenges and hardships throughout this journey!

List of abbreviations

BID-SDME Bubble-in-Drop Single-Drop Micro-Extraction EIM Extracted Ion Chromatogram EPA Environmental Protection Agency FID Flame Ionisation Detector GC Gas Chromatography GC-MS Gas Chromatography Mass Spectrometry HPLC High Performance Liquid Chromatography LOD Limit of Detection MS Mass Spectrometry NIST National Institute of Standards and Technology RSD Relative Standard Deviation S/N Signal-to-Noise Ratio SDME Single-Drop Micro-Extraction SIM Selected Ion Monitoring SME Solvent Micro-Extraction SPE Solid-phase Extraction SPME Solid-phase Micro-Extraction

xvi

A fact is a simple statement that everyone believes. It is innocent, unless found guilty.

A hypothesis is a novel suggestion that no one wants to believe. It is guilty, until found effective.

~~ Edward Teller ~~

xvii

Dedication

This work is dedicated to my late Mother ‘Machaka Julia George

who passed on during the last stages of this thesis

xviii

Synopsis

A lot of chemicals are used in agriculture to increase production per cost. Unfortunately most of these chemicals find their way into the final agricultural product or get washed off into other systems where they may pose health and environmental concerns. As such monitoring of these compounds where they are not needed is necessary to avoid unwarranted pollution and the effects associated with them. This requires the use of analytical techniques and instrumentation.

Analytical chemistry, especially in the area of sample preparation and clean-up, has shifted focus mainly on greener methods that produce only minute quantities of waste without sacrificing efficiency. This has led to the conception of a term QuEChERS (acronym for Quick, Easy, Cheap, Effective, Robust and Safe) to reflect this trend in sample preparation. Efforts have been made in this regard to reduce the amount of chemicals used; as a result miniaturised techniques have evolved. One such technique that is easiest and cheapest is solvent micro-extraction especially the single-drop format, wherein only a few micro-litres of an organic solvent are used to sample the aqueous solution with the same volume being transferred into the instrument for analysis.

This study reports the advances made in development of a modified single-drop micro- extraction (SDME) technique through the deliberate introduction of an air-bubble to facilitate mass transfer. This is termed bubble-in-drop single-drop micro-extraction (BID- SDME). The method has been reported for the first time in our earlier work. This study started off with the validation of the method using triazine mixture (TP 619) as model herbicides. The method was validated for linearity (0.05 – 5 ng/mL), reproducibility and repeatability (%RSD < 10%), matrix effect, limits of detection (pg/mL range) and quantification as well as accuracy.

Following the validation of the method and preliminary application in laboratory-based work, the method was applied in monitoring and determination of metolachlor and atrazine from agricultural soils and water samples. Contributions along the way included the use of

xix hot-water extraction coupled to this pre-concentration method for extraction of soil samples and the importance of surrogate composition in spiking experiment. Hot-water extraction improved the extraction of these herbicides from the soil samples 4-fold in the case of metolachlor compared to room temperature sonication. This has led to the ability to monitor qualitatively the dissipation of the herbicides in the farm soil from the time of spraying through harvesting.

Most researchers use an analyte surrogate in the solvent in which they are commercially obtained without much effort to examine the effect such solvents have on the actual analyte in the matrix. The present study has revealed that the use of organic solvents dissolved surrogates result in increased recovery of the analytes which leads to an overstatement of the recovery and hence a report of higher levels of the analytes than are actually present in the soil. A traceless spiking method was thus sought and developed. The recoveries observed were in the range 70 – 80% (atrazine) and 80 - 100% (metolachlor), with the use of traceless aqueous surrogate spiking. The method demonstrated not only qualitative capability but also quantitative capability in that these herbicides were quantified at levels in the pg/mL range (0.03 ng/mL and 0.9 ng/mL for atrazine and metolachlor, respectively) which were almost at the detection limits of the method, determined as 0.024 ng/mL (metolachlor) and 0.013 ng/mL (atrazine), respectively.

The BID-SDME method was further applied to the analysis of organochlorines compounds where it demonstrated flexibility. Here only a change of the extracting solvent was necessary to render the method applicable to organochlorines. The method yielded slightly higher (0.1 ng/mL range) detection limits compared to those prescribed by the US-EPA (0.01 ng/mL range) for the same compounds possibly due to the use of electron impact mass spectrometry detector instead of the negative ionisation mass spectrometry used contemporarily. Other than the slightly higher detection limits, other parameters were satisfactory: linearity (R2 ≥ 0.9991 in the range 0.05 to 5.0 ng/mL), repeatability and reproducibility (%RSD < 10%) and accuracy. The method was then applied in analysis of water samples originating from the areas in which the use of organochlorines pesticides is suspected. Although no compounds were detected in the samples, the recoveries of the surrogates of these compounds from the said matrices were satisfactory (> 90%) at ng/mL

xx level suggesting that if these compounds were present at this level, the method would have detected them.

The method was further modified through the use of mixed-solvents (toluene and chloroform at the ratio of 3:1 chloroform to toluene) and applied in the analysis of synthetic growth hormones (hexestrol and diethylstilbestrol) from bovine urine without much negative matrix effects being observed. The method achieved comparable LODs (< 1 ng/mL) and LOQs to those published in the literature. Other parameters were still satisfactory: linearity (R2 ≥ 0.9991 in the range 0.5 ng/mL to 10 ng/mL), repeatability and reproducibility (RSD < 10%). The method was validated for accuracy with the use of a certified reference material in a bovine urine matrix. The advantages such as speed of extraction, low use of consumables and low waste production were demonstrated for this application. This further demonstrates the robustness of the method and its flexibility to adapt to different analytes and matrices.

Generally, the need for solvents more dense than water has been demonstrated in the study. BID-SDME is not possible with less dense solvents since the solvent is lost through wicking due to higher buoyancy aggravated by the presence of the air bubble. The average gain ratio of BID-SDME to simple SDME has been consistently 1.5, indicating a distinct benefit of introducing the air bubble in facilitating mass transfer.

The last part of the study explored the potential of using the modified dispersive liquid- liquid micro-extraction for headspace sampling. The concept was shown to have potential although it was not fully developed as a technique. It involves the spiking of diethyl ether into the aqueous solution and warming up the solution to about 50 °C after which headspace sampling is performed. This led to increased extraction of the chlorophenols used as model compounds compared to standard headspace analysis at the same temperature. Despite the improved efficiency, there was substantial fibre swelling associated with presence of organic solvents preventing the method to be fully pursued. However, the concept holds potential to be further explored particularly with the view to identifying swell-resistant polymers for their use in this assisted headspace sampling and developing a full understanding of the dynamics involved.

xxi

Chapter 1: Introduction and literature review

1.1 Introduction to the thesis In this introduction, the reader is introduced to the thesis, which is composed of three main sections: 1) introduction and literature review, 2) results and discussion (four chapters), 3) general conclusion. The first chapter sets the scene with a brief discussion on the discovery, history, global trends in usage, fate and effects of pesticides and other hazardous chemicals with regards to the environmental health and societal impacts. Various analytical methods especially those with relevance to the study are discussed.

The second section of this chapter presents a literature review that serves as a foundation for the study and leads to the formulation of the study objectives. The main focus of this chapter is to explore the origin and recent developments in the area of sample preparation relating to pesticides and similar chemicals with special relevance to environment.

Chapters 2, 3, 4 and 5 present and discuss the results relating to the development and applications of several aspects of sample preparation for analysis of some pesticides from the environment and from various matrices. Each of these chapters is set up as an independent section chapter with a brief introduction, experimental procedures, results and discussion and rounded up with the relevant conclusions and recommendations followed by the references cited in that chapter. Specific results will be preceded by brief outline of the methods for ease of conceptualisation. Chapter 5 is more of a concept development and initial study which breaks ground for future work. As such it will become clear that it is presented as proof of concept study only.

Chapter 6 presents a general discussion and conclusions summarising the preceding results chapters and relates them to the objectives of the project as a whole. General conclusions and contribution to the existing knowledge or understanding of the sample preparation methods shall also be outlined. This chapter shall also present some recommendations for future work and or other aspects that maybe relevant for successful application of the developed techniques.

1.2 Applications of chemicals, their fate and social aspects in everyday life Everyday life revolves around the use of thousands of chemicals and/or chemical processes, most of which consumers are not even cognisant. The use of chemicals varies from domestic products like those used in controlling bugs, cosmetics, and also pharmaceutical products. In South Africa the skin lightening products business reportedly hit US$12.8 million in 1976 wherein these products were used by black people in efforts towards improving their social standing (taking away the blackness).1 The deleterious effect of these products led to their ban for over-the-counter sales in many countries.2,3 Other uses of chemicals are in textile industries for purposes such as dyeing and optical whitening,4 recreation and sports as appearance/performance enhancers5,6 or to maintain the “green” in golf courses and stadia.7,8

The large scale handling of chemicals is found in the areas of pharmaceuticals, agriculture, industrial activities and in warfare where most of the toxic chemicals are used as raw materials, produced either as products or by-products or generated as waste. Although most of these chemicals are safe, except the warfare, they can cause some degree of damage to the environment and the health of workers and consumers with prolonged exposure directly or through their by-products.9 Their fate and that of their breakdown products is of global concern as they always find their way into the environment.10

1.2.1 Applications of chemicals in agriculture As demand for food security and other economic pressures increases, there is greater use of pesticides at the expense of the environment and public health. In Europe 854 pesticide products were in use for domestic use in gardens and homes in 2003.16 Although recent statistics could not be obtained, there has been a considerable investment in pesticides use, a total of US$14,118bn was spent on pesticides in 2001 worldwide, while in the United States of America alone, a total of US$6,410bn was spent in the same period,17 and in South Africa the expenditure in total on agrochemicals grew by almost 100% between 1985 and 1990 from US$93 million to US$177 million.18

1.2.2 Fate of pesticides in plants or the environment and their effects thereon Pesticide residues in the soil and water form a major focus of many studies on environmental pollution as essentially all pesticides end up in such systems. An understanding of the degradation processes and the products thereof is important as the transformation products may also be toxic, sometimes even more toxic than parent compounds,19,20 although regulation often focuses more on the parent compounds than the degradation products.21 The degradation processes can either be biotic or abiotic.22 For example, Mandelbaum et al.23 isolated a bacterium species Pseudomonas sp. that is capable of degrading triazines to generate its own nitrogen source from a herbicide spillage site. Lányi et al.24 have studied photo-degradation of triazines in the environment. This complexity makes it even more difficult to regulate against the by-products, which may themselves show toxicity effects, as noted.

Contamination of the environment by pesticides, especially the persistent types like chlorinated hydrocarbons, has been the subject of global concern for many years, leading to the banning of DDT in many countries.25 This is due to the fact that most of these pesticides have a tendency to accumulate in fat tissues of the animals and hence build up higher in the food chain.26 They could also pose the danger of causing mutations through DNA interactions leading to more resistant strains of the pests in addition to changing the natural flora and fauna.27 For example, the Anopheles gambiae mosquito has developed resistance to pyrethroids due to excessive applications.28 One of the most commonly used herbicides, namely atrazine, has been reported to have serious teratogenicity wherein it has led to hermaphroditism of male frogs.29

1.2.3 Effects of pesticides on human health The inherent toxicity of individual pesticides may be additive, antagonistic or synergistic.7,30 This toxicity is assessed through the Lethal Dose (LD) phenomenon which represents the dose in milligrams that kills half the sample population per kilogram of the animal's body weight. Pesticide toxicity varies with the mode and duration of exposure.

Table 1 below illustrates the effects of a few routinely used domestic pesticides. The effects range from minor irritations to eventual death.

Table 1 Clinical symptoms of some pesticides on human exposure

Chemical Family Action on human Symptoms of internal Symptoms of external Symptoms of chronic (and examples) Type of pesticide System exposure exposure exposure Organo- phosphates Insecticides, Inhibits Headache, dizziness, Minimal rashes but readily Loss of appetite & weight, diazinonC acaricides acetylcholinesterase in weakness, shaking, nausea, absorbed through the skin weakness, SpectracideT the tissues stomach cramps, diarrhoea, and general feeling of sickness sweating Carbamates Insecticides, Reversible changes in Headache, dizziness, Minimal rashes but readily Loss of appetite weakness, carbaryl C acaricides acetylcholinesterase weakness, shaking, nausea, absorbed through the skin weight loss, & feeling of Sevin T enzyme of tissues stomach cramps, diarrhoea, sickness sweating Organo-chlorines Insecticides, Disrupt function of Headache, dizziness, Minimal rashes but readily Build-up in the fat tissues. methoxychlor C acaricides nervous system, weakness, shaking, nausea, absorbed through the skin May cause nervousness, MarlateT (HCB is a fungicide) mainly the brain excitability, disorientation weakness, and shaking Pentachlorophenol Herbicides, Toxic to liver, kidneys, Headache, weakness, Highly irritating to skin, Weight loss, weakness, penta C molluscicides, and nervous system nausea, excessive sweating eyes, nose, and throat anaemia T Pentacon germicides, fungicides, wood preservatives Chloro-phenoxy Herbicides Irritant to lung, Prompt vomiting, burning Moderately irritating to Do not remain in body; passed Pesticides stomach and intestinal sensation in stomach, eyes, skin, and lungs out within hours or days 2,4-DC Weed-B-GonT linings. Injure liver, diarrhoea, muscle twitching kidney, and nervous system Thiocarbamates & Fungicides Low human toxicity Nausea, vomiting, diarrhoea, Irritates skin, eyes, nose, Dithiocarbamates weak-ness, and nasal and throat ZinebT stuffiness Pyrethrins & Very low human Slight toxic reaction Pyrethroids toxicity PrentoxT EctibanT Triazines Herbicides Irritant Mildly irritating to skin, atrazineC AAtrexT eyes, nose & throat

C = Common name, T = Trade name (Table extracted from: PMEP23)

1.2.4 Legal, political and social aspects of pesticide use and misuse There is a need to regulate toxic commercial products introduced to the market.31 Globally there are a number of institutions, some of which are country specific, charged with regulation of production, use, disposal and treatment of waste by registered companies. Environmental data for parent compounds and their metabolites, degradation processes and transformation products are normally sought before placing a new chemical on the market.

Cooperation between policy makers, scientists, environmentalists, development workers, consumers and farmers towards awareness campaigns and eradication of the misuse of these important chemicals, and to enforce the legal instruments against this threat both at

4 community, national, regional and global levels is absolutely critical.32 For example, the US EPA through the Pesticide Programs and the Office of Compliance has been reported to anticipate spending of $515 000 annually between 2010 and 2014 on educational programs for senior managers, senior scientists, and supervisors and managers of pesticide regulatory enforcement and compliance assurance programs on issues such as water quality, worker safety, endangered species, and applicator certification and training, pesticide compliance and enforcement laboratory issues, new regulatory programs and any emerging issues.33

In sub-Saharan countries, where resources are meagre, although the regulation policies are aligned with international conventions that dictate the maximum allowable residual limits of chemicals, especially in potable water, they only serve as policy documents without much enforcement.34 Examples of such conventions are The Stockholm, Basel, Rotterdam, and Agenda 21 of The Rio Summit protocols, each of which deals with only a few regulatory aspects. For example, persistent organic pollutants such as aldrin, chlordane, dieldrin, DDT, endrin, heptachlor, polychlorinated benzenes, dioxin and furan products are included in one document but others are not mentioned.35

1.3 Analysis of pesticides and other organic compounds It is virtually impossible to develop only a few common methods for the analysis of all types of pesticides given their structural diversity.36 The chemical class of a compound often determines the method of analysis. These methods vary from ordinary wet chemistry (titrometry) to the most highly sophisticated instrument based protocols. The current project is focussed on chromatographic techniques; hence the focus of the following sections shall be on chromatographic and separation techniques, notwithstanding the importance of other instrumental techniques.

Wet chemistry methods are of the oldest and least reported recently for analysing pesticides, yet it still offers the most reliable methods for analysing pesticide formulations for applicable classes of compounds. These methods are relatively simple, including, for example, situations where the sample is titrated to a set end-point that can either be detected visually or electronically. Common domestic products may be analysed by these 5 techniques, but they (the techniques) are limited in scope by virtue of their high limits of detection.

Chromatographic methods are important in that they allow the separation of mixtures of several pesticides from the matrix before being detected. Matrices are generally complex, for example, plant material, soil, water, processed food, etc. as such play a big role in the applicability of a specific technique. Instrumental methods are sophisticated and demand some degree of competence before the results generated thereof may be deemed reliable and fit for the purpose. The advent of automation facilities eliminates much human error and in the process often improves the precision of the results;37 however this still comes at a financial price. Chromatographic methods can be coupled to a number of techniques like mass spectrometry, infra-red, ultraviolet and NMR spectroscopy that are also applicable in identifying the analytes and their metabolites, thus offering a larger benefit over other methods.38

A lot of work has been done to analyse pesticide residues in foodstuffs and in the environment. For example, Hamlet et al.39 analysed traces of ethyl carbamate in bread in the order of parts per billion (using positive ion chemical ionisation MS technique) in wheat harvested from fumigated areas, while Garcia et al.40 used positive and negative ion chemical ionisation LC-MS analysis noting the same analytical sensitivity. Other techniques that are useful in pesticide analysis include immunochemical methods and offer a high degree of analyte specificity.41,42

1.3.1 Analytical method development and validation New analytical methods and results obtained thence need to be validated as an assurance to the analyst and the end user of such results with regards to accuracy and precision of such a method and its intended application.43 For any newly developed method intended for quality assurance and control, the quality plan detailing all analytical steps in the process and the exact experimental conditions such as temperature, concentration and instrument tuning are a pre-requisite.44 The validated method is able to distinguish between false negative/positive responses from the true negative/positive appropriately,

6 thereby ensuring higher precision and accuracy of the results.45 Other aspects of method validation include specificity, stability, ruggedness, robustness, limits of detection and quantification as well as linearity (dynamic range).46 This process may be, and is often, carried out by inter-laboratory co-operation.47

The precision of the results from a method is expressed in terms of standard deviations or as relative standard deviations (RSD) often expressed in percentage. The other important parameter in trace analysis is the limit of detection (LOD), which is understood as the lowest possible quantity that can be detected. Due to controversy posed by this definition, the LOD is expressed as the amount of analyte that gives a signal that is 3 times the standard deviation of the method blank.48

LOD = SB + 3σB

The other definition that closely resembles the one above expresses LOD as the ratio of 3 times standard deviation of the analytical blank to calibration slope/sensitivity.49,50

In the case of chromatography where the blank cannot be measured, the term SB is dropped and the LOD is estimated as three times the standard deviation of the lowest concentration used in the calibration curve.51 Other methods use the 3 times standard error of the calibration curve divided by the slope.52 Ribani et al.53 have assessed and reported several ways of determining the LOD from chromatography based method including the use of the standard error of calibration and the error of the intercept with the slope. Their results demonstrate that the LOD obtained with the use of error of intercept was lower than that obtained using the standard error of calibration. However, both determinations are heavily dependent on the slope.

The other commonly used technique is the S/N ratio approach where LOD is estimated as that concentration which gives the S/N = 3, although this is highly subjective.53 Due to this controversy, it is important for the analytical community to debate and agree on the best method that would be widely accepted, other than leaving it to individual scientists to use the ones that suit their results the best.

Validation for accuracy (trueness of the results/method) requires the use of certified reference materials, whose certified values are compared against those obtained in the new method.54 If the CRM is not available, the accuracy of the method could be evaluated by using a validated method and comparisons made using the results thereof.55,56

1.3.2 Chromatographic methods of analysis Chromatography has been well-developed into several classes depending on the physical states of both the mobile and the stationary phases, the two major classes of which are liquid chromatography and gas chromatography. The extent to which analyte separation occurs depends on the strength and nature of the interactions between the analyte and the mobile and stationary phases. These interactions could be due to polarity differences based on respective dipole moments or dielectric constants or the weaker van der Waals forces.57 To achieve stationary phase stability, the stationary material is packed in a column (column chromatography) or applied on a solid surface (thin-layer chromatography) or immobilised on a solid surface in the case of liquid stationary phase.

There are a number of official standard methods that have been developed and published by a number of institutions, which are used for environmental monitoring. These methods are typically ISO 17 025 accredited and as such they are widely acceptable. Methods such as EN 1528, NEN 6408, EPA 8260, DIN 38407, EPA 8270, EN ISO 6468 are commonly followed to identify and determine the levels of pesticides in environmental, food and other samples, mostly using state-of-the art equipment such as GC-MS.58

1.3.2.1 Column chromatography Column packing is one of the parameters that can adversely affect efficiency of a column chromatographic method. Sanchez59 presented in great detail a number of aspects relating to silica-based packing materials as sorbents for sample preparation for HPLC, supercritical chromatography and SPE. The development of packing materials with very good chemical and physical properties, have arisen consequent to intensive research on column packing materials.60 The Sol-Gel technique results in the formation of regular porous spheres providing mechanically and chemically stable structures with predictable pore size, pore distribution and surface area.61 In the polyethoxysiloxane technique on the other hand,

8 the particles are spongy, porous and have larger pore size and lower mechanical durability than those obtained from the Sol-Gel approach.62

To change the applicability of the stationary phase, the surface layer of the bed may be modified resulting in reverse phase chromatography.63 This is achieved by reacting the surface silanols with organic silanes preferably mono-functional silanes, resulting in packing materials with varying polarities and applications that can withstand different pH ranges.64 A lot of research is taking place in the area of column chemistry resulting in much 65 smaller particle size leading to higher efficiency.

The chromatographic elution may be either isocratic or by gradient depending on the need. In the former, the conditions are constant and in the latter they are altered as elution progresses. Gradient elution may be achieved through programming relating to and affecting the mobile phase (solvent) mixture and flow rate.66 Compared to isocratic elution, gradient elution provides narrower peak widths and faster elution of the sample.

1.3.2.2 Gas chromatography Gas chromatography has been a technique of choice for the determination of pesticides because of the favourable combination of very high selectivity, resolution, accuracy and precision, wide dynamic range and high sensitivity to thermo-stable and yet volatile compounds.67 GC is also a cleaner method as it uses less organic solvents (for injections only) than HPLC which uses them for mobile phase. A specific drawback of GC methods is that many compounds of interest do not satisfy the thermo-stability and volatility requirements applicable to a GC technique.68 This issue has been counteracted by the introduction of derivatisation techniques, which serve to render the compound of interest more volatile than the parent analyte. 69

The ability of GC to be coupled to various detectors, including MS and infra-red spectrometry, broadens the scope of application of the technique. The GC column may run to tens of metres as opposed to the much shorter HPLC columns, which are only a few centimetres long. GC columns of 60 metres length have been used in various applications.70,71 The requirements for such long columns can be easily understood

9 considering the high speed of gas molecules (determined by the formula v = [3kT/m]1/2 or [3RT/M]1/2 where k is the Boltzmann constant, T temperature, m and M are mass and molar mass of the gas, and R is the molar gas constant) even at room temperature. The extra time required when using a long column may be compensated by the use of fast GC techniques.72

GC columns can be either packed or capillary columns.22 Packed columns are normally made of steel with an internal diameter ranging between 1.6–9.5 mm and packed with the stationary phase. However, these types of column are no longer commonly used due to vast growth of the counterpart capillary columns except in preparative HPLC. Capillary columns, on the other hand, have narrower diameters *≤ 1 mm], and are normally made up of fused silica glass, which has a higher capacity to cross-link with the silicon–oxygen

[Si-O2] matrix than ordinary glass. Two major capillary column types include micro packed and open tubular types. The open tubular type is divided into wall-coated open tubular and porous-layer open tubular. The internal surface of the wall-coated open tubular column is coated with a stationary phase, while the porous-layer open tubular column the stationary phase is packed in the column.

The liquid polymer used to generate the coating wets the surface of the fused SiO2 resulting in a film covering the wall, hence the term “wall coated”. The stationary polymer could also be bonded onto the surface, which reduces its vapour pressure and reduces column bleed. A variety of functional groups can be introduced to change the polarity and selectivity of the column. These factors (polarity and column selectivity) are important to achieve a speedy and reliable separation of the sample mixture. Wall-coated open tubular columns are preferred over packed columns since they offer shorter retention times, greater inertness, long life, lower bleed, higher efficiencies and greater reproducibility.

1.3.3 Detectors for chromatography As previously stated, the choice of detector depends heavily on the chemical properties of the compounds being analysed and most importantly the compatibility of the mobile phase with the detector of choice. A detailed discussion of available detectors for chromatography using both GC and HPLC can be obtained from various sources.73 For

10 efficient functioning of any detector, the volume of the detector should be small to avoid mixing of compounds post separation.74 The most commonly used detector for routine HPLC is UV-Visible spectrophotometry which may be configured to run in either in fixed wavelength, variable wavelength and scanning wavelength (photodiode array – PDA) modes. PDA detectors can also play a vital role in the identification of components with different chromophores that may co-elute.75 Other detectors used are refractometric, fluorescence and electrochemical detectors.

The most commonly used GC detector is the flame ionisation detector (FID). This detector measures the current from ionisable compounds as they burn in the flame. A limitation of the FID method is the quenching effect of some functional groups on the compound due to poorer ionisability. These includes most natural gas samples and those without carbon 16 or those saturated with oxygen, e.g. CO, CO2, CS2, NH3, and halogenated SiCl4.

Other detectors include the thermionic emission detector76 which uses almost the same principles as the FID except that in this case nitrogen or phosphorus compounds can be detected, and it is thus sometimes called a nitrogen-phosphorus detector (NPD).16, This detector uses a small plasma poor in hydrogen fuel resulting in a low temperature that suppresses ionisation of compounds which do not contain nitrogen and phosphorus atom. This detector is sufficiently sensitive to be used in trace analysis.

The electron capture detector (ECD) is one of the mostly used detectors for halogenated compounds, most of which are pesticides. The eluent passes between two electrodes, one of which has a radioisotope that emits high-energy β-particles as it decays. These particles bombard the carrier gas resulting in the formation of plasma of positive ions, radicals and thermal electrons through a series of collisions. The application of a potential to the electron capture cell allows collection of thermal electrons that constitute the detector baseline current. As the electron-absorbing species from the sample collect these electrons, a decrease in the baseline current results, which produces a signal. This detector is ideal for analysis of electrophilic compounds like organo-chlorines77,78 giving detection limits comparable to those of the nitrogen-phosphorus detector.

Several other detectors including thermal conductivity, flame photometry, photo ionisation and electron conductivity detectors are available but not commonly used because they are very analyte specific.

1.3.3.1 Mass spectrometry as a detector for chromatographic analysis MS is the most widely used and generally applicable method for qualitative and quantitative analyses of numerous compounds to determine both molecular and atomic compositions following chromatographic separation. It produces charged particles consisting of ion fragments of the original parent molecule. These particles are arranged according to their mass-to-charge (m/z) ratio, which is sometimes is referred to loosely as mass. The mass spectrum produced by the MS is characteristic of the specific compound in question including isomers79 and isotopes.80

Comparative studies for several chromatographic detectors for pesticides analysis proved that none could match the efficiency of MS for trace analysis.81 High resolution MS can give elemental information on the composition of molecular and fragment ions. Fragmentation patterns are consistent for identical compounds under identical ionisation conditions, allowing a measure of predictability. The versatility of MS as an analytical technique facilitated the detector to find its way into modern physical and biological sciences ranging from bio-molecular and clinical studies to material sciences and extra- terrestrial materials.82

1.3.4 Hyphenated/coupled instrumental methods Hyphenation is defined as an on-line coupling of a separation/chromatography technique and an analytical spectroscopic method that can provide detection and possibly also structural information. The coupled methods complement each other in that one may be dedicated to separation, while the other for detection and analysis. The two methods are joined by an interface that allows the preceding to be able to feed directly into the subsequent without compromising performance of one another. The usual primary consideration of coupling methods is compatibility at the interface.83,84,85

1.3.4.1 Hyphenation of HPLC and MS Most of the recent literature relating to HPLC concentrates on the hyphenation of HPLC with MS due to the flexibility of the MS as a detector and analytical instrument, especially its applicability in sample identification86 and the suitability of HPLC-MS to the analysis of polar and non-volatile compounds.87 The major problem associated with this coupling is the relative incompatibility of the two as they operate under completely different conditions. However, some progress has been made to by-pass the incompatibility limitation through the development and application of electro-spray, thermo-spray and atmospheric pressure interfaces that have widened applications of HPLC-MS to those now having relatively high sensitivities.88

The principles of the atmospheric pressure interface used with chemical ionisation involve nebulisation and vaporisation of the mobile phase followed by ionisation of the sample by some electrostatic discharge. The electro-spray involves the formation of charged liquid droplets and their reduction in volume by evaporation and disintegration into ionised species that are directly transferable to the MS. In electro-spray ionisation there is no degradation of thermo-labile compounds. This technique is suitable for polar and non- polar samples and the resulting mass spectrum contains mainly an M+ signal with minimum fragmentation.53 The polarity of the samples and the negative or positive ion character of the ions in the chromatographic conditions employed govern the choice of an appropriate interface.

1.3.4.2 Hyphenation of GC with MS The applications of the hyphenated GC-MS technique are well-documented and so well- developed that commercial analyses of samples running into hundreds of traceable compounds are being carried out world-wide. GC-MS methods have taken centre stage in trace analyses due to the improved sensitivity associated with the MS detection.

There are a vast number of publications in the area of GC-MS, too many to meaningfully summarise here. These tend to focus on specific sample matrices and the extraction methods applicable thereto. Depending on the extraction method used, the injected sample may contain large amounts of unavoidable co-extractants, and these are

13 responsible for matrix effects occurring on the injector, column and/or detector. In order to decrease the matrix effects, efficient sample preparation techniques are important. Alternatively, the fast GC method can be used without a loss of separation efficiency when utilising narrow-bore capillary columns.89,90

The ruggedness of fast GC-MS analysis for pesticides has been studied by Kirchner et al.91 who established that a PTV inlet in the cold split-less mode under optimised conditions provided sample vaporisation and sample transfer into the column with excellent repeatability (RSD < 6.6% at the ng/mL level). In addition, fast GC-MS using a capillary column with an internal diameter less than 0.15 mm provided good ruggedness for analyses of polychlorinated biphenyls92 and organic volatiles from alcoholic beverages.93

1.3.5 Quantitative analysis in chromatography It is not sufficient to perform a qualitative analysis for the detection of pesticide residues, without ultimately resorting to quantitative analysis. Chromatographic detectors produce a signal that is proportional to the quantity of the solute. Therefore the presence of an analyte is detected by the presence of a signal, but the amount of the analyte present may be obtained only from precise accurate data.

Quantitative information can be obtained through the integration of the peaks on the chromatogram trace. The traditional peak area integration methods are gradually being replaced by computer-based methods. Both integration techniques do not give absolute, but rather relative, quantities of analyte present in the sample. Compounds with known concentration and behaviour under the prevailing conditions could be used as either external or internal standards.94,95 For external standards, there is a need to use various concentrations of the standard and set up a calibration curve from which the quantity of the unknown may be determined. For an internal standard, the sample is spiked with a known quantity of the standard and the measurement is made and comparisons effected using the data generated from that sample alone. Usually these internal standards are the isotopic labelled analogues of the analytes.96 However, Liang et al.97 reported suppression of ionisation due to the co-eluting 2H-labelled analogues when analysing drugs on electro- spray while the converse was true for atomic pressure chemical ionisation. 14

Another method that can be used is normalisation. This is useful when absolute values are not required, but rather when information regarding the composition of the mixture is required. Peak heights or areas from different samples are compared and thereafter the respective ratios are taken to establish the percentage composition. This method does not require the use of standards.

1.3.6 Recent developments in chromatography Research in chromatography, either HPLC or GC, is focused more on developing more efficient methods yet with high resolution in the shortest possible time. The most important parameter to consider in achieving this is the reduction of theoretical plate height thereby increasing the number of theoretical plates per given length of the column. Different ways have been developed in response to this challenge. One way of dealing with this is reducing the size of stationary particles; however, this results in increased back pressure which subsequently resulted in the development of ultra-high pressure liquid chromatography (UPLC).98 To overcome the increased back pressure resulting from smaller particle size, elevation of operating temperatures to achieve the same chromatography efficiency with much lower back pressure has been introduced. This is referred to in different terms: rapid resolution high throughput columns (RRHTC, hence also RRHT chromatography)99,100 and ultra-fast liquid chromatography (UFLC) and high temperature liquid chromatography (HTLC).101

Another development has been introduction of the same particle size typical of a normal HPLC column (5 µm) with an inner core thereby restricting the analyte-stationary phase interactions to the outer 1.5 µm coated on the core.102 This apparently reduces the back pressure that would be generated if the analyte were to traverse the whole cross-section of the particle.

Recently a “single fibre-in-capillary annular” GC column where in a single optical fibre of outer diameter of 0.12 mm pre-coated with stationary phase was inserted into the capillary has been reported.103 The length of 6 m for this column reduced the duration time required for the separation of a 9-component mixture (n-propanol, isobutanol,

15 isoamylol, isooctane, butylacetate, amylacetate, 1-hexanol, n-decane, and n-undecane) to 3.5 minutes compared to 5.3 minutes for a length of 12 m. This type of column could also offer flexibility in that it can allow the use of different stationary phases. Research is not only restricted to hardware, but also on the software manipulations. For example, the use of computers to resolve non-resolved peaks from polybrominated diphenyl ether congener has been reported with success.104

Chromatography on its own cannot be sufficient for analysis, more so in trace analysis applications where concentrations of the analytes may well be outside the detection limits of the instrument. In such cases sample preparation is mandatory, not only to clean up the samples but also to pre-concentrate the analytes to levels detectable by the method.

1.4 Sample preparation techniques for chromatography analysis Sample preparation is one of the most highly explored phenomena relating to chromatography besides the column chemistry and detectors. The most important benefit to a good sample preparation method is the effective transfer of analytes out of the bulk matrix into/onto another phase before subsequently being subjected to the instrumental method, thus improving the detection limits of the overall method. Sample preparation involves mainly two processes: extraction and clean-up.

Focus in this area has been initially towards exhaustive methods without much exclusivity/specificity while lately it has shifted to not only more selective methods that are easy to couple with the instrumental methods, but also to pre-concentration of the analytes which is vital in trace analysis. These can be broadly classified as solid-based and liquid-based methods based on the physical state of the sampling phase. Sample preparation goes a long way in determining the extent to which the method can be applied and the sensitivity thereof. A good sample preparation could lower the detection limits of the method quite significantly.

Some of the important properties that the extracting materials should have include an ability to selectively extract the analyte, be reasonably quick and safe, pose minimal losses through volatility or by unwanted adsorption of the analyte on the walls of the container

16 being used and it should not introduce contaminants from the materials/solvents being used.105 The simplest concentration method for solutions is evaporation of the solvent. However, this presumes that no losses take place due to volatility of the analyte. Accordingly, this concentration method is applicable only to specific scenarios.

This section focuses on sample preparation techniques for chromatographic separation and analysis as this thesis deals with chromatographic techniques with an emphasis on GC. All the sample preparation methods discussed here can be applied both in direct immersion or headspace sampling unless where specifically stated otherwise. Direct immersion mode is where the sampling phase (hereunder referred as acceptor) in immersed into the aqueous solution (hereunder referred as donor solution). There is a direct interface between the two phases. In headspace sampling the two phases are separated by some air volume. As such this situation presents a three phase system with two interfaces: donor-air and air-acceptor interfaces.

1.4.1 Liquid-based sample preparation techniques Liquid-based sample preparation methods are the oldest and most commonly used methods in pesticide analysis. The oldest forms of liquid-based sample preparation are solvent extraction and Soxhlet extraction. In the former, a sample which is dissolved in water is distributed into the water immiscible organic solvent that is brought into contact with the aqueous solution. This method is not commonly used in pesticide analysis. Soxhlet extraction is discussed below.

1.4.1.1 Soxhlet extraction Soxhlet extraction is the most widely used liquid extraction and pre-concentration method in environmental and trace analysis. In this method, the sample is brought into contact with the solvent based on reflux distillation to constantly deliver fresh solvent to the sample matrix, and as such avoids saturation of the compound in the solvent during the extraction procedure. It is, however, important when extracting heat sensitive analytes to ensure their structural integrity. A specific drawback of this method is that organic solvents often lack efficacy when applied to humate matrices like soils and sediments, and also pose some health hazards.106 17

There have been several developments to improve liquid extraction, such as sonication- assisted extraction,107,108 microwave-assisted extraction, microwave-assisted soxhlet extraction109,110 and pressurised liquid extraction.111 These are mainly used to enhance the availability of the analytes from the matrix for other extraction techniques like solvent or solid based extractions without necessarily being the main method. Microwave heating involves ionic conductance and dipolar rotation in the molecule.112 Since polar solvents can absorb microwave energy, they are preferred over non-polar solvents.113 The improved conditions and shorter times used in these techniques limits degradation of the analytes. The precision, accuracy and reproducibility with microwave-assisted extraction are better than those observed for standard liquid extraction methods.114

1.4.1.2 Supercritical extraction techniques Supercritical techniques are those techniques that do not use the normal standard conditions for physical states of the extractant nor the analyte. Supercritical conditions are achieved by changing the temperature and pressure of the sample solution. The variations of these conditions have led to two key variants for water extraction, pressurised hot water and superheated water extraction. These approaches are closely related, while being mindful that heating a liquid beyond its boiling point should obviously require some additional pressure to prevent the liquid from boiling. However, for pressurised water extraction, external pressure can be applied to the system without necessarily heating it. These conditions are referred to as superheated, subcritical and/or pressurised water extraction. Reportedly, when water is heated beyond its boiling point, its chemical properties change to those of organic solvents, such that by temperatures above 200 °C, it demonstrates the same polarity as methanol.115 The other supercritical extraction mode uses carbon dioxide at its critical pressure and temperature conditions. These techniques demonstrate “greener” properties than other extraction techniques in that only water and

CO2 are used and generated as waste.

1.4.1.2.1 Superheated and pressurised water extraction Hot water has been widely used for the extraction of volatile compounds in biological systems, as in steam distillation. The major advantage of using superheated water extraction is the relatively lower pressure required for the extraction, as compared to the

18 supercritical fluid extraction method, and the fact that both polar and non-polar compounds can be extracted over a wide range of temperatures up to 400 °C. Superheated water extraction functions due to decreasing polarity under superheated conditions (solubility of organic compounds increases about 20 000 times in the range 25 – 200 °C).116 The method is sufficiently robust to be used for extraction of sequestered pesticides from aged soils and similar matrices.117,118 Superheated water proved better than conventional water distillative extraction of volatile oils from plants.119

Superheated water can also be used as a chromatographic technique offering “greener” and much cheaper analysis replacing the use of copious amounts of hazardous organic solvents. However, due to the considerations about analyte and/or stationary phase stability at elevated temperatures, together with a lack of ready access to the specialised instrumentation required, this technique has not seen considerable growth.120

On the other hand, pressurised water extraction uses water under elevated pressures. The temperatures used for this technique range from superheated temperatures to sub-boiling temperatures (below 100 °C).121,122 The advent of “green chemistry” has led to an increased interest in such techniques123 and the use of superheated water as a chromatographic mobile phase without organic solvents has been explored.124

1.4.1.2.2 Supercritical fluid extraction

Supercritical fluid extraction using CO2 under supercritical conditions is the most commonly used method relying on such types of fluids. CO2 has a moderate critical temperature (32 °C) and it is chemical inert, as such it is recommended and frequently applied for extraction of thermally labile compounds under supercritical conditions. Its applicability in routine analysis of carbamates and organophosphates extracted from soil samples shows similar efficiencies when compared with conventional soxhlet extraction.125 Supercritical fluids show low viscosity and high diffusion coefficients (intermediate between gases and liquids) and these are good solvent properties. Supercritical solvent properties may be easily modulated by pressure adjustments and the analyte is recovered by simple isothermal depressurisation.126

As a chromatographic method, supercritical fluid chromatography shows slightly higher selectivity ranges than HPLC. The lower operating temperature than in GC allows it to be used with thermally labile compounds. Another specific advantage is that the CO2 can be easily recycled. This technique may be coupled to other techniques such as MS for on-line detection.127 The main limitations of supercritical fluid chromatography is the highly sophisticated apparatus needed, making it expensive for routine analysis. Another drawback is the limited solubility of polar and large compounds in pure CO2.

1.4.1.3 Solvent micro-extraction Miniaturisation of the liquid-liquid extraction technique results in a dramatic reduction of the volumes of organic solvents used. This aspect has thus led a lot of attention being focused in this area, leading to the development of two major techniques, namely liquid- phase micro-extraction128,129 and drop-based liquid micro extraction techniques.130,131 In the latter method, a micro-drop of the organic solvent is directly suspended in the aqueous solution or in the headspace of such a solution, depending on the properties of the analyte being extracted. After equilibration has been established, the drop is introduced directly to the appropriate instrument for analysis. The drop-based liquid micro-extraction method is the most cost-effective method since the apparatus involved is no more than an ordinary micro-syringe used in the GC introductions.132

The dynamics and merits of the solvent micro-extraction have been explored extensively and compared against the well-developed and highly rated counterpart solid-phase micro extraction, which will be discussed in later sections. Solvent micro-extraction proves to be an equally good or better candidate than its counterpart, solid-phase micro-extraction (SPME).133,134 However, its mass transfer rate surpasses that of SPME as equilibrium is achieved much faster due to higher diffusion rates in liquids than in solids. It is worth mentioning that both techniques are more inclined towards sample clean-up and pre- concentration than extraction as they require that the sample be already present in solution form. This is not always the case as in, for example, soil or biological samples.

Solvent micro-extraction is a fast evolving technique due to its affordability, since it requires only a few micro-litres of solvent.135,136 This technique is one of the cheapest and

20 safest developed so far. Most of the extraction methods discussed thus far are far more expensive, capital intensive and/or not as environmentally friendly and sustainable as solvent micro-extraction. A number of modifications that reportedly increase efficiency quite considerably have been published and will be discussed in the following sections. The simplest of these modifications is a syringe modification where a syringe is cut horizontally to increase its adhesion surface area thus improving the micro-drop stability,137 which together with problems posed by solvent volatility and water miscibility are major set-backs associated with these drop-based methods.

1.4.1.3.1 Single-drop micro-extraction The drop-based solvent micro-extraction was first reported by Liu and Dasgupta in 1995 where a drop of water was used to extract NH3 and SO2 from gaseous samples in a sequential analysis system.130 Jeannot et al.138 were the first to report the drop-based SME in the format that is widely used to date, namely single-drop micro-extraction (SDME). This method has seen an explosive growth in both evolution and application. Ho et al.139 have demonstrated that the micro-extraction techniques offer higher enrichment of compounds with relatively large partition coefficients (Korg/d > 500), but lacks selectivity for those with lower distribution coefficients, than conventional liquid-liquid extraction posing a potential limitation to this technique.

A major set-back of this method is the difficulty of automation due to mechanical instability of the droplet, volatility and miscibility with water of the solvents used. Solvent volatility limits its applications in headspace sampling. However, it is seen as a relatively “green” method as it only requires a few micro-litres for rinsing of the syringe and sampling. This rinsing also assures no sample carry-over between samplings. Recent studies on the employment of ionic liquids to counteract the effect of solvent volatility have also been reported.140,141 However, this comes at a considerably higher price and poses some other complications, especially for GC applications; ionic liquids are not sufficiently volatile for GC applications and they are relatively more expensive than common organic solvents. Their application requires modification of the injection port so that the remaining liquid can be disposed of.142,143

SDME has been applied for various analytes in myriad matrices both in headspace and direct immersion modes.144 Jeannot’s group145 has done a lot of work in developing the dynamics and theoretical background of the method. Below are some of the modifications that reportedly result in an increase in extraction efficiency or improved detection limits over classical simple SDME.

 Direct drop suspension:146 an organic drop is introduced as a freely floating (drifting) droplet into the mixture that is being agitated and eventually allowed to settle as in classical solvent extraction so that it can be sucked by the syringe to the instrument. The additional surface area engendered by the free droplet (as opposed to one attached to the needle of a micro-syringe) increases mass transfer. However, that could also result in an accelerated solvent loss due to increased surface area and of course the kinetic energy from stirring.

 Another variant that is related to the direct drop suspension147 has recently been reviewed and is termed solid drop-based liquid-phase micro-extraction.148 In this technique an organic solvent with melting point around the room temperature is introduced to the aqueous solution in its liquid form. After equilibration is reached the solution is cooled and the solidified solvent is isolated and thereafter melted by heating and injected into the instrument. The method has been used by various groups for analysis of organic analytes in water149 and biological samples,150 as well as in extraction of metals following chelation to render such metals soluble in organic solvents.151,152

 Funnel-form SDME derivative: in an effort to improve stability, Qian et al.153 used a funnel to protect the drop from the effect of stirring. This modification improved the precision of the method. The same observation was made by another group that used a conical device to protect the drop as in the funnel-form derivative.154

 Drop-to-drop solvent micro-extraction:155 a micro-drop of the organic solvent is used to extract a drop of the aqueous solution. The difference in this method from the other drop-based micro-extractions is that the extracted solution is only drop-

sized as opposed to the larger volumes (≥ 1 mL) of the aqueous solution used in other methods. The properties of this method resemble the directly suspended drop method discussed above.

 Continuous-flow micro-extraction:156 the drop is in contact with a continuously moving sample; as a result the sample is always fresh, leading to much improved mass transfer.157 Sample flow rate should ensure an effective micro-extraction of analytes without drop dislodgement or bubble formation.158

 Dynamic solvent microfilm159 uses a microfilm of the organic solvent being introduced through dynamic motion of the syringe plunger; the equilibration is between the organic film and the aqueous solution. The analyte then diffuses to the bulk of the droplet as the plunger is depressed. This process requires high precision of the frequency and period of the cycles to maintain precision of the results. This is the only non-supported SME method where automation has been reported successfully resulting in detection limits between 0.18 and 0.35 ng/mL for a BTEX mixture.

1.4.1.3.2 Supported solvent micro-extraction This is commonly referred to as liquid-phase micro-extraction, LPME. The liquid is supported inside a membrane to enhance stability. The membrane may also function to minimise matrix effects by providing a barrier for larger molecules than its pore size.

 Membrane-based/assisted techniques160,161 include hollow fibre liquid micro-extraction (HF-LPME)162 and solvent bar micro-extraction.163 Solvent bar extraction and or micro- extraction resembles the stir bar sorption discussed under solid-based methods in the proceeding sections. The organic solvent is trapped inside a small membrane often in the form of a small bar which is suspended in a stirred or static solution as required. The solvent is eventually collected using a syringe and analysed. The hollow fibre methods such as this, in which the hollow membrane fibre encloses the organic solvent, have a physical appearance that looks similar to SPME. This membrane could be selective towards some compounds and hence brings about some degree of specificity.

In other instances the membrane may simply be a permeable membrane, in which case it is used solely for droplet protection. This method has been reported for various analytes in different media.164,165 Lee et al.166 reviewed the applications of the HF-LPME in environmental and biological samples.

 Membrane-assisted solvent micro-extraction167 and the extracting syringe derivative168 thereof have also been reported. In the former, a small membrane sheet is used for the extraction and it is directly introduced into the injector port of the GC. The extracting syringe derivative uses an extraction “card” prepared with two symmetrical polypropylene pieces. The inner surface of the card has a groove that can contain a known volume of the organic solvent. A small membrane is pressed in between the sheets and it creates two channels, one for stagnant organic solvent and the other for the mobile aqueous solution. This configuration is assembled such that organic solvent can be easily injected into the GC injector port once the extraction is complete.

Membrane-assisted methods can be easily used in a back extraction format (three phase system), in which case the membrane is impregnated with an organic solvent while a central cavity is filled with an acceptor solution which is different to that of the bulk (donor) solution in some way. For example, for acidic organic compounds, the aqueous (donor) solution may be acidified to reduce ionisation of the analytes and thereby increase their solubility in intermediate organic solvent. Meanwhile, a basic aqueous (acceptor) solution is filled into the membrane cavity. As the molecular analytes dissolve in the organic solvent-impregnated membrane they come into contact with the base on the acceptor phase inside the membrane and they are ionised. This ionisation increases the gradient for the analytes to dissolve into the acceptor solvent. This mode works very well for ionisable compounds like amines,169,170 hydroxylated aromatic compounds171 and phenoxy acids172,173 where enrichment factors of 400-fold are observed in the latter in dynamic HF-LPME.174

Recently, LPME was reported using a direct-current electrical potential difference across the supported liquid membrane as the driving force for rapid mass transfer into the donor phase based on electrokinetic migration, reaching equilibrium within five minutes.175 New

24 composite hollow-fibres membrane made up of polysulfone and accurel popypropylene with sulfonated poly(ether ether ketone), PEEK, have been reportedly used to improve the stability of the membrane for up to 5 months of continuous use.176

1.4.2 Solid-based sample preparation techniques Efforts to reduce the quantities of solvents used in extractions have led to the development of solid-phase extraction (SPE) techniques. This technique restricts the use of solvents to the desorption step of the process where such a step is used for desorption of the sample from the solid-phase cartridge or bar. Solid-based techniques follow the same principles of extraction as discussed in the preceding section in terms of partition of analytes between the solid and liquid phases. These techniques reduce/eliminate the use of organic solvents and hence reduce interferences from the solvent and impurities thereof in the chromatogram.

1.4.2.1 Solid-phase extraction (SPE) The origin of solid-phase extraction is attributed to the successful use of solid stationary phases in liquid chromatography, more specifically adsorption chromatography. An aqueous solution of the analytes is passed through the cartridge where the analytes become adsorbed and retained by the cartridge. Thereafter the analytes are desorbed by the use of an organic solvent and introduced to the chromatograph.

Stir-bar sorptive extraction is another solid based sample preparation method first described by Baltussen et al. in 1999.177 In this technique the stationary phase was thinly coated onto a magnetic stirrer bar and introduced into the sample. Subsequently the bar was removed and the analytes desorbed thermally or through dissolution into the organic solvent. Polyurethane foams have been reported as “new generation polymeric phases” for the extraction of triazine herbicides mixture, showing better efficiency than the polydimethylsiloxane solid-phase that is commonly used as a standard sorbent.178 This method has also received quite considerable attention with commercial bars being introduced to the market.179

Efforts to reduce the particle size for the sorbent material have resulted in the preparation of nano-fibres of the sorbent material. The smaller particle size results in improved surface area necessary for faster extraction kinetics. The most common nano-fibres are prepared through electro spinning.180 The other commonly used nano-fibres are carbon nano- tubes.181 Just like in the other SPE methods, these nano-fibres may also be molecularly imprinted by grafting or co-synthesis for improved selectivity. Most nano-fibre sorbents are applied in extraction of metal ions.182

1.4.2.1.1 Naturally occurring material as sorbents for sample preparation In efforts to advance “green chemistry” most researchers have explored the possibility of using naturally occurring sorbents as they tend to be the least toxic of all available. Such sorbents may also be replenishable, depending on their source, and obviates the need for chemical synthesis thereby improving their “green” standing. Most naturally occurring sorbents have been used for the extraction of metals.183,184 For example, orange peel materials have been used for the extraction of dyes185,186 while pine cone-derived substances have been used in the extraction of Cu [II] ions.187 Polymers such as lignin and cellulose are isolated from these materials and used. For example, lignin both in its pure and amidated form has been used to extract organic compounds.188 Despite the growth in the application of these sorbents, miniaturised derivatives have not yet been reported.

1.4.2.1.2 Molecularly imprinted polymers Molecularly imprinted polymers (MIPs) mimic immunochemical phenomena in that some complimentary recognition sites are created in a polymer network to complement those in the analyte molecule resembling antigen-antibody interactions.191,192 These sites increase the level of interaction between the analyte and the sorbent at the expense of other compounds in the mixture, allowing some degree of selectivity to be achieved. These sites are created using a template that is subsequently removed through dissolution or thermal desorption.193 They can also be grafted onto a solid support such as silica.194

Caro et al.196 reviewed the application of MIPs to solid-phase extraction of compounds from environmental and biological samples. Highly selective MIP-SPME has also been reported recently for ultra-trace analysis of folic acid (LOD of 0.0036 ng/mL) in human

26 serum.197 MIPs have found applications not only in sample preparation but in chemo- sensing as well, for example in DNA detection systems in gel electrophoresis.199 Grafting of naturally occurring polymers has also been reported in analysis of (N-phenyl-1- 200 naphthylamine) with LOD of 1.38 μM (ng/mL levels).

1.4.2.2 Solid-phase micro-extraction SPME Recent advancements in this area have resulted in the development of a miniaturised version of SPE, namely solid-phase micro-extraction SPME, whereby micro quantities of the solid-phase material are employed. SPME surpasses its predecessor SPE due to the fact that it does not require further treatment but rather can be coupled on-line with the analysis instrument, the analyte then becomes thermally desorbed and analysed. This coupling does not only reduce costs and time of analysis but also reduces risks of contamination associated with various multi-stepped off-line sample preparation/manipulation techniques.

In gas chromatography, the sample desorption from the stationary phase to the carrier gas is effected at elevated temperatures for SPME.201 The time lapse for desorption is said to be less than one second for most compounds,202 thereby facilitating avoidance of tailing peaks resulting from slow desorption of analytes. Many commercial suppliers provide several stationary phases of varying characteristics for SPE and SPME.203 SPME may also use a solid-phase impregnated into a semi-permeable hollow membrane as a stationary and the product is configured the same way as the SPME for inlet systems.

As in the case of SDME, the SPME extraction can be performed by direct immersion or by insertion into a headspace, avoiding the collection of solids in the latter method. To facilitate extraction using headspace SPME, the vapour pressure of the sample can be increased by gently heating the sample matrix and adding water or surface-active materials to the matrix.205,206 With the direct immersion protocol, a suspension is prepared in water, especially if the sample particles do not bond too strongly to the matrix.207

Some problems associated with SPME include the fragility of the fibre, the need to desorb the sample at temperatures below that of the lowest boiling sample, the maximum 27 operating temperature of the stationary phase itself, formation of air bubbles in and on the fibre, persistent adsorption of some samples (especially organo-chlorines) onto the fibre and irreversible adsorption of proteins onto the fibre thereby changing its properties. The problem of irreversible adsorption can be prevented by protecting the fibre with semi- permeable membranes to reduce its accessibility.208 Further improvements, especially as they relate to the detection limit, may be realised by manipulating the pH, ionic strength and temperature of the system during extraction.209

Efforts have been made recently to develop cheaper and “home-made” types of SPME derivatives. Materials such as pencil lead,210 silicon glue coated stainless steel,211 graphite rods212 polymer- and ionic liquid-coated materials,213,214 metal oxides such as PbO plated on metal wire215 have been reported for the extraction various analytes. Cao et al.217 oxidised titanium using hydrogen peroxide to produce titanium oxide nanostructured fibres. Gierek et al.218 reported preparation of SPME fibres by pyrolysis of methylene chloride on quartz fibres and epoxide-acrylic polymer and used such fibres for the extraction and analysis of volatiles.

1.4.3 Dispersive methods Dispersive SME or SPE involves the dispersion of the extracting phase throughout the solution. This dispersion creates tiny individualised interfaces where interactions and mass transfer can occur; as such the surface area is expected to be tremendously increased. This technique can also be available in dispersive solid-phase extraction and dispersive liquid- liquid extraction. Both can also be used at micro-extraction techniques.

1.4.3.1 Dispersive solvent micro-extraction Dispersive solvent micro-extraction (also known as dispersive liquid-liquid micro- extraction) has seen some quite explosive exploration since coming to the scene in 2006 following the work of Rezaee et al.219 This technique is based on a ternary systems where a few micro-litres of an organic solvent with high density and a disperser solvent which is miscible in both extractant and aqueous phases are rapidly mixed resulting in cloudy mixtures with micro droplets of the organic extractant solvent dispersed in the aqueous

28 solution. Due to these tiny droplets, the total surface area is greatly increased, offering rapid mass transfer.

Following equilibration which typically takes a few seconds, the solution is centrifuged and the organic droplets aggregate into one drop, which it is eventually recovered with a syringe and introduced to the relevant instrument for analysis. Recently Rezaee et al.220 have reported a new modification whereby a less dense 1-undecanol floating droplet was used in extraction of an aluminium-morin complex in water samples instead of the usually used denser solvent that settles at the bottom of the solution upon centrifugation following sampling.

This method has already produced hundreds of papers over the last few years wherein remarkable successes have been reported.221,222 It has been applied for analytes such as chlorophenols223 and triazines224 wherein the speed of the method has been further demonstrated. It has been applied successfully in the extraction of metals, possibly due to the fact that the other liquid-based methods are not compatible with the considerably higher volumes required in most analytical techniques amenable to metal trace analysis, such as ICP-MS.225 However, a pre-requisite is that the form in which the metal is present should be soluble in an organic medium.

1.4.3.2 Dispersive solid-phase extraction (D-SPE) As opposed to the liquid-based counterpart, the dispersed sorbent in the D-SPE is further dispersed in organic solvent to dissolve the analytes prior to the injection into the instrument. This extra step poses threats of contamination and/or loss due to dilution. D- SPE has been applied for extraction of organophosphate pesticides in milk;226 synthetic growth promoters in bovine milk;227 tetracycline antibiotics in water and milk;228 organo- phosphorus pesticides in milk, fruits and vegetables.229 The method can also be used for further clean-up following conventional LLE as demonstrated by Nguyen et al.230 who applied it by extracting 95 pesticides in soybean with recoveries of 80 to 114%, while recoveries ranged from 93 to 105% and from 96 to 111% for DLLME and DMSPE, respectively, were obtained from extraction of quinolones in swine meat using HPLC with LODs ranging from 5.6 to 23.8 μg/kg and from 7.5 to 26.3 μg/kg respectively.231

Dispersive SPE has also been reported to have been used to remove the matrix as opposed to extracting the analytes out of the matrix making use of an approach which has been termed QuEChERS (Quick, Easy, Cheap, Effective, Rugged, and Safe).232 This approach has been adopted by many laboratories wherein a number of QuEChERS kits have been produced for various anlytes and matrices.233 This QuEChERS method has already been adopted as First Action by the Association of Official Analytical Chemists (AOAC) International Official Method 2007.01.234

1.4.4 Sample preparation involving chemical reactions There are cases when extraction is not so straightforward. It is sometimes necessary to consider transforming the analytes into the form that is compatible with the extraction mode as well as with the analytical method. In the following sections, some transformations that do not require further steps in the extraction of both solid and liquid based analytes are discussed. The simplest of the reactions could be suppression or amplification of ionisation/dissociation through pH variation.235 Mohammadi et al.236 recently reported the use of “ligandless-dispersive liquid-liquid micro-extraction” for analysis of Cu (II) ions in which the extraction was enhanced by simple altering of the pH of the sample solution.

1.4.4.1 Derivatisation-coupled sampling Derivatisation is important and is usually used in transforming compounds as part of sample preparation for compounds that are not amenable to the extraction methods and/or the subsequent analytical method. There are a few reports where online coupled derivatisation either in-drop or on-fibre derivatisation has been reported to enable the extraction and analysis of a range of compounds. This approach improves both the extraction efficiency because solubility increases, and therefore also the sensitivity of the analysis. For example, a 100-fold higher sensitivity for the extraction of Bisphenol A has been achieved upon derivatisation with acetic anhydride, when compared with extraction of the free phenol.237 In some instances, the derivatisation agent is spiked into the sample and the resultant solution is subjected to extraction (in-solution/on-matrix derivatisation).238 On-column derivatisation has also been reported wherein the

30 derivatisation agent is injected immediately after the sample and the reaction occurs in the injection port. 239,240

1.4.3.1.1 Liquid-based sampling with derivatisation There are a few publications where the principle of in-drop derivatisation has been investigated. In this method the derivatising agent is spiked into the extraction solvent so that the reaction can occur simultaneously with extraction. As the analyte comes into contact with the derivatisation agent at the interface of the extracting solvent, a reaction occurs and the product is extracted into the extracting solvent. Due to this chemical reaction, Le Chatellier’s principle is obeyed, causing enrichment of the product in the extracting solvent so long as the derivatisation reagent is not depleted and providing the concentration of product in the extracting solvent does not exceed its solubility.

Stalikas et al.241,242 have worked on the in-drop derivatisation method as applied to the headspace micro-extraction of carbonyl compounds and some phenolic compounds that were derivatised using ethyl chloroformate as well as some ion-pairing agents.243 Hollow fibre liquid micro-extraction has also been successfully applied to protect the moisture sensitive derivatising reagents.244,245 In situ derivatisation has also been reported in continuous flow micro-extraction in the analysis of lead by graphite furnace atomic absorption spectrometry.246

1.4.3.1.2 Solid-based sampling with derivatisation In situ or online derivatisation is applicable to solid-based techniques as well. This derivatisation could be either before extraction if the derivatisation agent is spiked into the sample solution,247 during the extraction, in which case the derivatisation agent is coated on to the solid extractant (on-fibre)248 and/or post extraction where the derivatisation agent is added during thermal desorption of the analyte/s from the fibre.249 Applications of stir bar sorptive techniques with in situ derivatisation have recently reviewed by Prieto et al.250 and they include the work detailing the derivatisation of analytes containing phenolic compounds and hydroxy-PAHs.251,252 In situ derivatisation with SPME has received a lot of attention for in-solution,253 pre-exposure254,255 and post extraction derivatisation.256

1.4.4.2 Chelation and complex formation As has already been pointed out, chelation and ligation can be used as a step towards making the metal ions soluble in an organic solvent from which they may be analysed. Metal analysis using gas chromatography is restricted to those metals that can form volatile organometallic derivatives, such as mercury, tin and lead.257 Further developments have advanced this field to the stage where headspace-solid-phase micro-extraction sampling for these metals may be effected.258,259 This approach may also be performed to make metal ions amenable to liquid chromatography. For example, analysis of tris(2,2’- bipyridine) chelated ruthenium(II) ions using liquid chromatography has been reported.260 Metal chelation can also be used for rendering the metal ions more interactive with solid sorbents. Extraction of metal ions using solid-phase extraction has been reported in work where chelated Pb(II) ions were extracted with diphenylthiocarbazone-anchored polymeric micro-beads.261

Metal chelation with a view to use liquid-liquid micro-extraction has seen a substantial interest since the advent of dispersive liquid-liquid micro-extraction technique. Its applications has been reported for various metal ions ranging from the Al,220 heavy metals,262 to selective extraction of Au in the presence of Fe, Co, Ni, Cu, Pd and Pt ions in acidic medium.263 Dadfarnia et al.264 recently reviewed solvent micro-extraction techniques for trace analysis of metals.

Ion pairing is one form of chelation wherein an ion-pairing agent, the function of which is to reduce the polarity of the ionisable compounds, is added to the sample solution.265,266 This promotes ionisation of the sample, but because of the bulkiness of counter-ion, the net properties of the molecule resemble covalent compounds, imparting improved solubility of the ion pair in organic solvents. This method is used predominantly in liquid chromatography.267,268 Bianchi et al.269 investigated the role of ion-pair in liquid chromatography for analysis of heterocyclic amines. Chienthavorn et al.270 reported the online coupling of superheated water extraction with ion-pairing derivatisation for the analysis of chlorophenoxy acid pesticides.

1.5 Mass-transfer in solvent microextraction Mass transfer in classical solvent extraction can be defined mathematically as follows:

D = Co/Ca

= no × Va/Vo × na where D, Co, no, Ca, and na represent distribution coefficient, and the concentrations and number of moles of the analyte in question in organic and aqueous media, respectively. This rule holds where the extraction is not exhaustive, but is rather equilibrium-based, and provided mass transfer is restricted to diffusion without any energy requirement except the chemical potential of the analyte.

In the case of micro-extraction, taking into consideration that Va>>Vo and that the concentration of the analyte in the aqueous medium remains almost constant throughout the extraction, the above equation can be rearranged as follows:

ŋo = DCaVo

Co,eq = DCa = D Ca,i /(1 + DVo/Va) where the “ŋ” represent number of moles of analyte, subscripts “a” and “o” designate the relevant quantities in the aqueous and organic phases, and where “i” and “eq” designate initial and equilibrium conditions.

Jeannot et al.144 derived a number of equations governing mass transfer for solvent micro- extraction that eventually led to following equation:

dCt/dt = Ai × β / Vo(D × Ca-Ct)

where Ai is the interfacial surface area and β is the overall transfer coefficient which takes into consideration the transfer between the aqueous and organic media represented as follows:

1 / β = 1 / βo + D / βa. The interface flux is related to concentration and mass transfer coefficients as follows:

Fi = 1/A ×dŋ/dt = βa(Ca – Ca,i) = βo(Co,i – Co)

The subscript “i” represents the volume closest to the interface.

The rate of extraction follows a first order kinetics

-λt Ct = Co,eq(1 – e ) where λ is a rate constant given by

83,91 λ = Aiβ/Vo(DVo/Va + 1) or Aiβ(D/Va + 1/Vo)

From the above equations, it can be deduced that β is an empirical parameter relating mass transfer/flux to concentration. It also serves as a driving force for the extraction, as long as stirring rate and temperature are kept constant, regardless of the solvent used. 271 Bagheri and Khalilian related the extraction efficiency denoted Eo to the distribution coefficient as follows: -1 Eo = [Vo/Va + 1/D] This equation shows that a considerable amount of the analyte may be extracted provided the distribution coefficient is reasonably high, as has been pointed out earlier.

In the case of SPME, there is a maximum number of active sites on the fibre that governs the maximum extractable amounts.272 This maximum extractable amount is related mathematically to the number of active sites on the fibre as follows:

ŋf,e = Cf,e Vf

= KCaVaVf (Cf,max – Cf,e) / [Va + DVf (Cf,max – Cf,e) where the subscripts: a, e, and f represent aqueous, equilibrium and fibre, respectively, and K is the adsorption coefficient. The term “f,max” denotes the active sites parameter, which is an indication of a finite number of active sites on the fibre.

Mass transfer in a three-phase system is more complicated than in the two-phase system discussed above. This is understandable in that equilibrium processes occur across all available interfaces, resulting in a more complex mathematical relationship. Headspace

34 analysis is considered a three-phase system as well due to presence of more than one interface, namely liquid-gas and gas-sampler.145,273

1.6 Objectives of the study Having discussed the analytical and other requirements for analytical methods, the main objective of this study can be summarised as the development of affordable and greener yet effective sample preparation methods for chromatographic analysis of trace pesticides and other environmentally exogenous chemicals. Once the method has been developed and validated it will be applied to various systems to assess its applicability and robustness. Specifically, environmental samples shall be used for this purpose where the samples are to be sourced from a farming area where known pesticides are being sprayed as crop protection chemicals. Most of the literature reports focus on the detection of residues but typically do not trace given pesticides following their application. A significant part of this project will be dedicated to monitoring the pesticides from application onto farm fields to harvesting of the crops.

The desired method is to be based on solvent micro-extraction as a technique that has proven sufficiently sensitive for trace analysis applications and does not require significant investment in infrastructure and manpower development, beyond what many labs already have at their disposal. Once the new method has been fully developed and validated, it will be applied to various matrices especially environmental matrices, water and soil samples collected from exposed areas.

In summary, therefore, this project aims to work with single-drop micro-extraction as a pre-concentration method. The author has already worked in this area and the method discussed here has formed part of a previous study. In the present case, the method shall be more fully investigated to establish a protocol which can be validated against certified reference materials. The method is to be applied to the analysis of samples gathered from a farming area where certain types of herbicides are known to be used. These herbicides include atrazine and metolachlor. In so doing, the method shall be not only pursued from a developmental perspective, but also from an application point of view. The application of

35 the method to the analysis of metolachlor shall be fully investigated since the method was originally developed for the pre-concentration of triazines.

The method, which has been developed for aqueous samples, shall be modified to allow its application to soil samples. For this purpose, the soil samples shall be subjected to water-based extraction methods and the aqueous samples so generated shall be subjected to pre-concentration extraction making use of the single-drop micro-extraction method. Relevant spiking methods for soil samples shall also be investigated.

This method shall also be applied for organochlorines specifically DDT that is still a problem in South Africa due to occurrence of malaria some of the areas of the country. Being highly resistant to degradation, DDT finds its way into water and soil bodies.

South Africa also has a large beef farming industry and as such it is imperative that the beef products produced in the country meet the international stipulations for doping. Growth hormones are used for accelerated growth of the animals for high conversion of grains into body weight gain. Most of the methods in use and literature for the pre- concentration of such substances are solid-based being either SPE or SPME. The application of the single-drop micro-extraction method shall be assessed on synthetic growth hormones with the view to expand the scope of application of this technique.

Finally efforts will be made to develop a technique combining both solvent micro- extraction and solid-phase micro-extraction for headspace sampling. The solvent will be used to aide mass transfer for SPME extraction.

1.7 References

1. Bentley-Phillips, B.; Bayles, M.A.H. South African Medical Journal, 1975, 49, 1391-1395 2. Burke, P.; Maibach, H. Journal of Dermatological Treatment, 1997, 8, 21-26 3. Shankar, P. R.; Subish, P. Journal of Pakistan Association of Dermatologists, 2007, 17, 100-104. 4. Tutak, M.; Demiryürek, O.; Bulut, Ş.; Haroğlu, D. Textile Research Journal, 2011, 81, 58- 66 5. Dodge, T.L.; Jaccard, J.J. Journal of Adolescent Health, 2006, 39367-39373 6. Hildebrandt, T.; Alfano, L.; Langenbucher, J.W. Journal of Psychiatric Research, 2010, 44, 841-846 7. Cox, C. Journal of Pesticide Reform, 1991, 11, 2-4 8. Wan, H.B.; Wong, M.K.; Mok, C. Y. Bulletin of Environmental Contamination and Toxicology, 1996, 56, 205-209 9. Blount, E.; Erikson, B.; R. Green, Strategic Approach to International Chemicals Management (SAICM) Statement, 2007 UNEP, 10. Goldman, L.R.; Koduru, S. Environmental Health Perspectives, 2000, 108, 443-448 16. Buffin, D.; Diamand, E.; McKendry, R., Wright, L. Breaking the pesticide chain - Report 2003, Pesticide Action Network UK & Friends of the Earth 17. Kiely, T.; Donaldson D.; Grube, A. Pesticide Industry Sales and Usage, 2004, 2000-2001 Market Estimates, USEPA. Washington USA. 18. London, L. Pesticides News, 1995, 27, 6-7 19. Andreu, V.; Pico, Y. Trends in Analytical Chemistry, 2004, 23(10-11), 772-789 20. Lo, H.H.; Brown P.I.; Rankin, G.O. Toxicology, 1990, 63(2), 215-231 21. Sinclair, C.J.; Boxall, A.B.A. Environmental Sciences and Technology, 2003, 37, 4617- 4625 22. Pesticides Management Education Program, 1995. University of Cornell, USA. 23. Mandelbaum, R.T.; Allan, D.L.; Wackett, L.P. Applied and Environmental Microbiology, 1995, 61, 1451-1457 24. Lányi, K.; Dinya, Z. Microchemical Journal, 2005, 80, 79-87 25. Turusov, V.; Rakitsky, V.; Tomatis, L. Environmental Health Perspective, 2002, 110, 125-128 26. Albanis, A.; Hela, D.; Papakostas, G.; Goutner V. Science of The Total Environment, 1996 182(1-3), 11-19 27. Tennant, A.H.; Peng, B.; Kligerman, A.D. Mutation Research/Genetic Toxicology and Environmental Mutagenesis, 2001 493, 1-10 28. Pesticides Action Network Pesticides and Alternatives, 2000, 12 29. Hayes, T.; Haston, K.; Tsui, M.; Hoang, A.; Haeffele C.; Vonk, A. Environmental Health Perspective, 2003, 111, 568-575 30. Yang, Z. The Application Notebook. 2004. Varian Inc

31. US EPA -Office of Enforcement. Pesticide Production Laboratory - Procedure Manual, 1980, EPA-330/9-79-001. National Enforcement Investigations Center. Denver, Colorado 32. Scheringer, M.; Bergman, Å.; Fiedler, H.; Holoubek, I.; Zetzsch, N.C. Initiative for an International Panel on Chemical Pollution (IPCP), 2003 33. PESPWire April 2009, United States Environmental Protection Agency 34. London, L.; Dalvie, M.A.; Nowick, A.; Cairncross, E. Water SA, 2005, 31, 53-59 35. Pesticide Action Network [PAN]; Pesticides and Alternatives, 1999, 9 36. La Guardia, M.J.; Hale, R.; Harvey, E.; Chen, D. Environmental Science and Technology 2010, 44, 4658–4664 37. Risticevic, S.; Niri, V.H.; Vuckovic, D.; Pawliszyn, J. Analytical and Bioanalytical Chemistry, 2009, 393, 781-795 38. Wilson, I.D.; Brinkman, U.A.Th. Journal of Chromatography A, 2003, 1000, 325-356 39. Hamlet, C.G.; Jayeratne, S.M.; Morrison, C. Rapid Communications in Mass Spectrometry, 2005, 19(16), 2235-2243 40. Juan-Garcia, A.; Manés, J.; Font, G.; Picό, Y. Journal of Chromatography A, 2004, 1050, 119-127 41. Garcés-Garcia, M.; Brun, E.M.; Puchades, R.; Maquieira, Á. Analytica Chimica Acta, 2006, 556, 347-354 42. Gabaldón, J.A.; Maquieira, A.; Puchades R. Talanta, 2007, 71, 1001-1010 43. Green, M.J. Analytical Chemistry, 1996, 68, 305-309 44. Fournier, J.B.; El’Hourch, M.; Fournier, J. Analusis 1999, 27, 726-734 45. Ellison, S.L.R.; Fearn, T. Trends in Analytical Chemistry, 2005, 24, 468-476 46. Rozet, E., Marini, R.D., Ziemons,E., Boulanger, B., Hubert, Ph. Journal of Pharmaceutical and Biomedical Analysis, 2011, 55, 848-858, 47. El Mrabet, K.; Poitevin, M.; Vial, J.; Pichon, V.; Amarouche, S.; Hervouet, G.; Lalere, B. Journal of Chromatography A. 2006, 1134, 151-161 48. Thomsen, V., Schatzlein, D., Mercuro, D. Spectroscopy, 2003, 18(12), 112-114 49. Voigtman, E. Spectrochimica Acta B, 2008, 63, 154-165 50. Longerich, H.P., Jackson, S.E., Günther, D. Journal of Analytical Atomic Spectrometry, 1996, 11, 899-904 51. Tor, A.; Aydin, M.E.; Özcan, S. Analytica Chimica Acta, 2006, 559, 173-180 52. Eide, M.; Martin, L.; Motley, K.; Plummer, T.; Potter, W. Method Evaluation Guidelines, 1993, Occupational Safety and Health Administration, USA 53. Ribani, M.; Collins, C.H.; Bottoli, C.B.G. Journal of Chromatography A, 2007, 1157, 201- 205 54. González, A.G., Herrador, M.Á., Asuero, A.G. Talanta, 2010, 82, 1995-1998 55. Swartz, M.; Krul, I. LCGC North America, 2005, 23, 46 -52 56. García, I., Ortiz, M.C.,Sarabia, L., Vilches, C., Gredilla, E. Journal of chromatography A, 2003, 992, 11-27

57. Farrington K., Regan, F. Talanta, 2009, 78, 653-659 58. European Pharmacopoeia 2002, Chap 2.8.13 59. Sanchez, D. Analusis Magazine, 1998, 26(7), 33-37 60. Guo, Y., Colόn, L.A. Analytical Chemistry, 1995, 67, 2511-2516 61. Willard, H.H.; Merritt Jr, L.L.; Dean, J.A.; Settle Jr F. A., Instrumental Methods of Analysis 7th Ed.; Wadsworth Pub. Co.; Belmont, California, USA. 1988; p514 62. Baklanov M. R. Journal of Vacuum Science and Technology B, 2000, 18(3), 1385-1391 63. Majors, R.E. Column Match, 2004, Agilent Technologies, Pittcon Conference. 64. Waters Technical Bulletin, The Column, Sep. 2005 65. Nováková, L., Matysová, L. Solich, P. Talanta 2006, 68, 908–918 66. Frenich, A.G.; Vidal, J.L.M.; Galera, M.M. Analytical Chemistry, 1999, 71, 4844-4850 67. Delgado, M.J.S.; Barroso, S.R.; Fernandez-Tostado, G.T.; Polo-Diez, L.M. Journal of Chromatography A, 2001, 921, 287-296 68. Soler, C., Manes, J., Pico, Y. Journal of Chromatography A, 2004, 1048, 41-49 69. Zhang, J., Lee, H.K. Journal of Chromatography A, 2006, 1117, 31-37 70. Ragsdale, J.D. Application Note 9168. Thermo Finnigan, Thermo Electron Business 71. Thermo Scientific Technical note 20073. Thermo Fisher Scientific Inc. 72. Kirchner, M.; Matisova, E.; Hrouzkova, S.; de Zeeuw J. Journal of Chromatography A, 2005, 1090, 126-132 73. Scott, R.P.W. Chrom Ed. Series, Library 4 Science. www.chromatography-online.org 74. Bayer, E. Albert, K., Nieder, M., Grom. E. Analytical Chemistry, 1982, 54, 1747-1750 75. Frenich, G.A.; Vidal, J.L.M.; Galera, M.M. Analytical Chemistry, 1999, 71, 4844-4850 76. Lambopoulou, D.A.; Albanis, T.A. Journal of chromatography A, 2005, 1072, 55-61 77. Rajeswaran, J.; Merlinkamala, I.; Chandrasenkaran, S.; Jayakumar, R.; Kuttalam, S. Food, Agriculture and Environment 2004, 2 (2), 276-277 78. Schmidt,T.C.; Haas, R.; Steinbach, K.; von Löw, E. Fresenius Journal of Analytical Chemistry, 1997, 357, 909-914 79. Mechref, Y., Kang, P., Novotny, M.V. Rapid Communications in Mass Spectrometry, 2006, 20, 1381-1389 80. Saad, O.M., Leary, J.A. Journal of American Society for Mass Spectrometry, 2004, 15, 1274-86 81. Eiert,R.; Levsen, K.; Wuensch, G. International Journal of Environmental Analytical Chemistry, 1995, 58, 103-120 82. Zafman, D.; Rudich, Y.; Sagi, I.; Strasser, D.; Savin, D.W.; Goldberg,S.; Rappaport, M.; Heber, O. Journal of Mass Spectrometry, 2003, 229, 55-60 83. Begnaud, F., Chaintreau, A. Journal of Chromatography A, 2005, 1071, 13-20 84. Wilson, I.D., Brinkman, U.A.Th. Trends in Analytical Chemistry, 2007, 26(9), 847-854 85. Yanes, E.G., Miller-Ihli, N.J. Spectrochimica Acta B, 2004, 59, 883-890 86. Jeannot, R.; Sauvard, E. Analusis, 1999, 27, 271-80

87. Hernandez, F.; Pozo, O.J.,;Sancho, J.V.; Lopez, F.; Marin, J.M.; Ibanez, M. Trends in Analytical Chemistry, 2005, 24 (7), 596-616 88. D’ascenzo, G.; Gentili, A.; Marchese, S.; Marino, A.; Perret, D. Analusis, 1998, 26, 251- 255 89. Matisová, E.; Dömötörová, M. Journal of Chromatography A, 2003, 1000, 199-221 90. Korytár, H.; Janssen, G.; Matisová, E.; Brinkman, U.A.Th. Trends Analytical Chemistry, 2002, 21, 558-572 91. Kirchner, M.; Matisová, E.; Otrekal, R.; Hercegová, A.; de Zeeuw J. Journal of Chromatography, 2005, 1084, 63-70 92. Covaci, A. Schepens, P. Journal of Chromatography A, 2001, 923, 287-293 93. Namara, K.M., Leardi, R., Sabuneti, A. Analytica Chimica Acta, 2005, 542, 260-267 94. Ng, L.L. Validation of Chromatographic Methods, Center for Drug Evaluation and Research, US Food and Drug Authority. 1994 95. Rohrback, B.G.; Ramos, L.S. Aligning Chromatograms, 2003, Infometrix, Inc. Gulf Coast Conference 96. Häubl, G.; Berthiller, F.; Krska, R.; Schuhmacher, R. Analytical and Bioanalytical Chemistry, 2006, 384,692-696 , 97. Liang, H. R.; Foltz, R. L.; Meng, M.; Bennett, P. Rapid Communications in Mass Spectrometry, 2003, 17, 2815-2821 98. Swartz, M. E. Separation Science Redefined 2005, 8-14 99. Yoshida, T. ; Majors, R.E. Journal of Separation Science, 2006, 29, 2421-2432 100. Agilent, 2003, ZORBAX, Rapid Resolution High Throughput Columns (cat. # 5988- 9617EN) 101. Nováková, L.; Vlčková, H. Analytica Chimica Acta, 2009, 656, 8-35 102. Oláh, E.; Fekete, S.; Fekete, J.; Ganzler, K. Journal of Chromatography A, 2010, 1217, 3642-3653 103. Li, P.; Xu, Z.; Yang, X.; Bi, W.; Xiao, D.; Choi, M.M.F. Journal of Chromatography A, 2009, 1216, 3343-3348 104. Mydlová, J.; Krupčík, J.; Korytár, P.; Sandra P. Journal of Chromatography A, 2007, 1147, 95-104 105. Prosen, H.; Zupančič-Kralj, L. Trends in Analytical Chemistry, 1999, 18 , 272-282 106. Zhang, Z.; Xiong, G.; Lie, G., He, X. Analytical Sciences, 2000, 16, 221-224 107. Sánchez-Brunete, C.; Miguel, E.; Tadeo, J.L. Journal of Chromatography A, 2002, 976, 319-327 108. Gonçalves, C.; Alpendurada, M.F. Talanta, 2005, 65, 1179-1189 109. de Sabando, O.L.; de Balugera, Z.M.; Goicolea, M.A.; Rodriquez, E.; Sampero, M.C.; Barrio, R.J. Chromatographia, 2002, 55, 667-671 110. Barriada-Pereira, M.; González-Castro, M.J.; Muniategui-Lorenzo, S.; López-Mahía, P.; Prada-Rodríguez, D.; Fernádez-Fernández, E. Journal of Chromatography A, 2004, 1061, 133-139

111. Lüthje, K.T.; Hyotylainen, T.; Rautianen-Rama, M.; Riekkola, M.-L. The Analyst, 2005, 130, 52-58 112. Ericsson, M. Ph.D Thesis, 2003, Stockholm University. Sweden 113. Shah, S.; Richter, R.C.; Kingston, H.M.S. LCGC North America, 2002, 20, 280-286 114. Saifuddin, N.; Chua, K.H. Malaysian Journal of Chemistry, 2003, 5, 30-33 115. Smith, R. M. Analytical and Bioanalytical Chemistry, 2006, 385, 419-421 116. Smith, R.M. Journal of Chromatography, 2002, 975, 31-46 117. Crescenzi, C.; Di Corcia, A.; Nazzari, M.; Samperi, R. Analytical Chemistry, 2000, 72, 3050-3055 118. Tajuddin, R.; Smith, R. M. Journal of Chromatography A, 2005, 1084, 194-200 119. Özel, M.Z.; Göğüs, F.; Lewis, A.C. Analytica Chimica Acta, 2006, 566, 172-177 120. Smith R. M. Journal of Chromatography A, 2008, 1184, 441-455 121. Pardo, O., Yusà, V., León, N., Pastor, A. Journal of Chromatography A, 2007, 1154, 287-294 122. Marcic, C.; Lespes, G.; Potin-Gautier, M. Analytical and Bioanalytical Chemistry, 2005, 382, 1574-1583 123. Ridgway, K.; Lalljie, S. P.D.; Smith, R. M. Journal of Chromatography A, 2007, 1153, 36-53 124. Smith, R. M., Journal of Chromatography A, 2008, 1184, 441-455 125. Kreunzig, R.; Koinecke, A.; Bahadir M., Journal of Biochemical & Biophysical Methods, 2000, 43, 403-409 126. Jusforgues, P.; Shaimi, P. Analusis Magazine, 1998, 26, 55-59 127. Sakamoto, T.; Yamamoto, A.; Owari, M.; Nihei, Y. Analytical Sciences, 2003, 19, 853- 857 128. Rasmussen, K.E.; Pedersen-Bjergaard, S.; Krogh, M.; Ugland, H.G.; Gronhaung, T. Journal of Chromatography A, 2000, 873, 3-11 129. Rasmussen, K.; Pedersen-Bjergaard, S. Trends in Analytical Chemistry, 2004, 23, 1-10 130. Liu, H.; Dasgupta, P.K. Analytical Chemistry, 1995, 67, 2042-2049 131. Liu, H.; Dasgupta, P.K. Analytical Chemistry, 1996, 68, 1817-1821 132. Psillakis, E.; Kalogerakis, N. Trends in Analytical Chemistry, 2002, 21, 53-63 133. López-Blanco, M.C.; Blanco-Cid, S.; Cancho-Grande, B.; Simal-Gándara, J. Journal of Chromatography A, 2003, 984, 245-252 134. Das, P.; Gupta, M.; Jain, A.; Verma, K.V. Journal of Chromatography A, 2004, 1023, 33- 39 135. Dalvie, M.A.; Sinanovic, E.; London, L.; Cairncross, E.; Solomon, A.; Adam, H. Environmental Research, 2005, 98, 143-150 136. Psillakis, E.; Kalogerakis, N. Journal of Chromatography A, 2001, 938, 113-120 137. Ahmadi, F.; Assadi, Y.; Milani, S.M.R.; Rezaee, M. Journal of Chromatography A, 2006, 1101, 307-312 138. Jeannot, M.J.; Cantwell, F.F. Analytical Chemistry, 1996, 68, 2236-2240

139. Ho, S.T.; Pedersen-Bjergaard, S.; Rasmussen, K.E. Journal of Chromatography A, 2002, 963, 3-17 140. Zhou, Q. ; Ye, C. Microchimia Acta, 2008, 162, 153-159 141. Zhou, Q. X. ; Xie, G. H. ; Pang, L. Chinese Chemical Letters, 2008, 19, 89-91 142. Aguilera-Herrador, E.; Lucena, R.; Cárdenas, S.; Valcárcel, M. Journal of Chromatography A, 2008, 1209, 76-82 143. Chisvert, A.; Román, I. P.; Vidal, L.; Canals, A. Journal of Chromatography A, 2009, 1216, 1290-1295 144. Jeannot, M.A.; Przyjazny, A.; Kokosa, J. M. Journal of Chromatography A, 2010, 1217, 2326-2336. 145. Schnobrich, C.R.; Jeannot, M.A. Journal of Chromatography A, 2008, 1215, 30-36 146. Yangcheng, L.; Quan,L.; Guangsheng, L.; Youyuan, D. Analytica Chimica Acta, 2006, 566, 259-264 147. Zanjani, M.R.K.; Yamini, Y.; Shariati, S.; Jonsson, J. Å. Analytica Chimica Acta, 2007, 585, 286-293 148. Ganjali, M.R.; Sobhi, H.; Farahani, H.; Norouzi, P.; Dinarvand, R.; Kashtiaray, A. Journal of Chromatography A, 2010, 1217, 2337-2341 149. Farahani, H.; Ganjali, M.R.; Dinarvand, R.; Norouzi, P. Talanta, 2008, 76, 718-723 150. Sobhi, H. R.; Yamini, Y.; Esrafili, A.; Adib, M. Journal of Pharmaceutical and Biomedical Analysis, 2008, 48, 1059-1063 151. Bidabadi, M. S.; Dadfarnia, S.; Shabani, A. M. H. Journal of Hazardous Materials, 2009, 166, 291-296 152. Dadfarnia, S.; Shabani, A.M. H.; Kamranzadeh, E. Talanta , 2009, 79, 1061-1065 153. Qian,L.; He, Y. Journal of Chromatography A, 2006, 1134, 32-37 154. Ye, C.; Zhou, Q.; Wang, X. Journal of Chromatography A, 2007, 1139, 7-13 155. Wu, H.F.; Yen, J.H.; Chin, C. Analytical Chemistry, 2006, 78, 1707-1712 156. Liu, W.; Lee, H.K. Analytical Chemistry, 2000, 72, 4462-4467 157. Liu, Y.; Hashi, Y.; Lin, J.-M. Analytica Chimica Acta, 2007, 585, 294-299 158. Pena-Pereira, F.; Lavilla, I.; Bendicho, C. Spectrochimica Acta B, 2009, 64, 1-15 159. Mohammadi, A.; Alizadeh, N. Journal of Chromatography A, 2006, 1107, 19-28 160. Jönsson, J. Å.; Mathiasson, L. Trends in Analytical Chemistry, 1999, 18(5), 318-325 161. Ndungu, K.; Mathiasson, L. Analytica Chimica Acta, 2000, 404, 319-328 162. Petersen-Bjergaard, S.; Rasmussen, K.E. Analytical Chemistry, 1999, 71, 2650-2656 163. Jiang, X.; Lee, H.K. Analytical Chemistry, 2004, 76, 5591-5596 164. Basheer, C;Lee, K. H. Journal of Chromatography A, 2004, 1047, 189-194 165. Rasmussen, K.E; Petersen-Bjergaard, S. Trends in Analytical Chemistry, 2004, 23, 1-10 166. Lee, J.; Lee, H.K.; Rasmussen, K. E.; Pedersen-Bjergaard, S. Analytica Chimica Acta, 2008, 624, 253-268 167. Schellin, M.; Popp, P. Journal of Chromatography A, 2003, 1020, 153-160

168. Barri, T.; Bergström, S.; Hussein, A.; Norberg, J.; Jönsson, J. Journal of Chromatography A, 2006, 1111, 11-20 169. Zhao, L.; Zhu, L.; Lee, H.K. Journal of Chromatography A, 2002, 963, 239-248 170. Hou, L.; Lee H.K., Analytical Chemistry, 2003, 75, 2784-89 171. Melwanki, M.B.; Huang, S.D. Journal of Separation Science, 2006, 29, 2078-2084 172. Wu, J.; Ee, K.H.; Lee, H.K. Journal of Chromatography A, 2005, 1082, 121-127 173. Chen, C.C.; Melwanki, M.B.; Huang, S.D. Journal of Chromatography A, 2006, 1104, 33-39 174. Jiang, X.; Oh, S.Y.; Lee, H.K. Analytical Chemistry, 2005, 77, 1689-1695 175. Pedersen-Bjergaard, S.; Rasmussen, K.E. Journal of Chromatography A, 2006, 1109, 183-190 176. He, T.; Versteeg, L.A.M.; Mulder, M.H.V.; Wessling, M. Journal of Membrane Science 2004, 234, 1-10 177. Baltussen, E.; Sandra, P.; David, F.; Cramers, C. Journal of Microcolumn Separation, 1999, 11, 737-747 178. Portugal, F. C.M.; Pinto, M. L.; Nogueira, J.M.F. Talanta, 2008, 77, 765-773 179. Prieto, A.; Basauri, O.; Rodil, R.; Usobiaga, A.; Fernández, L.A.; Etxebarria, N.; Zuloaga, O. Journal of Chromatography A, 2010, 2642-2666 180. Kang, X.; Pan, C.; Xu, Q.; Yao, Y.; Wang, Y.; Qi, D.; Gu,Z. Analytica Chimica Acta, 2007, 587, 75-81 181. Zhou, Q.; Xiao, J.; Wang, W.; Liu, G.; Shi, Q.; Wang, J. Talanta, 2006, 68, 1309-1315 182. Wang, H.; Ding, J.; Lee, B.; Wang, X.; Lin,T. Journal of Membrane Science, 2007, 303, 119-125 183. Memon, J. R.; Memon, S. Q.; Bhanger, M. I.; El-Turki, A.; Hallam, K. R.; Allen, G. C. Colloids and Surfaces B: Biointerfaces 2009, 70, 232-237 184. Kumar U. Scientific Research and Essay, 2006, 1, 33-37 185. Namasivayam, C.; Muniasamy, N.; Gayatri, K.; Rani, M.; Ranganathan K., Bioresource Technology, 1996, 57, 37-43 186. Benaïssa, H. Ninth International Water Technology Conference, IWTC9, 2005, Sharm El-Sheikh, Egypt, p1175-1187 187. Ofomaja, A.E.; Naidoo, E.B.; Modise, S.J. Journal of Hazardous Materials, 2009, 168, 909–917 188. Dizhbite, T.; Zakis, G.; Kizima, A.; Lazareva, E.; Rossinskaya, G.; Jurkjane, V.; Telysheva, G.; Viesturs, U. Bioresource Technology, 1999, 67, 221-228 191. Remcho, V. T.; Tan, Z. J. Analytical Chemistry News and Features, 1999, 248A-255A 192. Fernández-González, A.; Guardia, L.; Badía-Laínõ, R.; Díaz-García, M. E. Trends in Analytical Chemistry, 2006, 25, 949-957 193. Boyd, J. W.; Cobb, P. G.; Southard, G. E.; Murray, G. M. Johns Hopkins APL Technical Digest, 2004, 25, 44-49

194. Lu, Q.; Chen, X.; Nie, L.; Luo, J.; Jiang, H.; Chen, L.; Hu, Q. ; Du, S. ; Zhang, Z. Talanta, 2010, 81, 959-966 196. Caro, E.; Marcé, R.M.; Borrull, F.; Cormack, P.A.G.; Sherrington, D.C. Trends in Analytical Chemistry, 2006, 25, 143-154 197. Prasad, B. B.; Tiwari, M. P.; Madhuri, R.; Sharma, P. S. Analytica Chimica Acta 2010, 662, 14-22 199. Ogiso, M.; Minoura, N.; Shinbo, T.; Shimizu, T. Biosensors and Bioelectronics, 2007, 22, 1974-1981 200. Ng S. M.; Narayanaswamy, R. Sensors and Actuators B: Chemical, 2009, 139, 156-165 201. Arthur, C.L.; Pawliszyn, J. Analytical Chemistry, 1990, 62, 2145-2148 202. Prosen H.; Zupancic-Kralj, L. Trends in Analytical Chemistry, 1999, 18, 272-282 203. Krutz, L.J.; Senseman, S.A.; Sciumbato, A.S. Journal of Chromatography A, 2003, 999, 103-121 205. Zhang, Z.; Pawliszyn, J. Analytical Chemistry 1993, 65, 1843-1852 206. Fromberg, A.; Nilsson, T.; Larsen, B. R.; Montanarella, L.; Facchetti, S.; Madson, J. Journal of Chromatography A,1996,746, 71-81 207. Boyd-Boland, A. A.; Pawliszyn, J.B. Journal of Chromatography A, 1995, 704, 163-172 208. Zhang, Z.; Poerschmann, J.; Pawliszyn, J. Analytical Communications, 1996, 33, 219- 221 209. Lesellier, A. Analusis, 1999, 27, 363-368 210. Djozan, Dj.; Baheri, T.; Farshbaf, R.; Azhari, Sh. Analytica Chimica Acta, 2005, 554, 197-201 211. Panavaitė, D.; Padarauskas, A.; Vičkačkaitė, V. Analytica Chimica Acta, 2006, 571, 45- 50 212. Luo, F.; Wu, Z.; Tao, P.; Cong, Y. Analytica Chimica Acta, 2009, 631, 62-68 213. Mohammadi, A.; Ameli, A.; Alizadeh, N. Talanta, 2009, 78, 1107-1114 214. Huang, K.-P.; Wang, G.-R.; Huang, B.-Y.; Liu, C.-Y. Analytica Chimica Acta, 2009, 645, 42-47 215. Mehdinia, A.; Mousavi, M. F.; Shamsipur, M. Journal of Chromatography A, 2006, 1134, 24-31 217. Cao, D.; Lü, J.; Liu, J.; Jiang, G. Analytica Chimica Acta, 2008, 611, 56-61 218. Gierak, A.; Seredych, M.; Bartnicki, A. Talanta, 2006, 69, 1079-1087 219. Rezaee, M.; Assadi, Y.; Hosseini, M.-R. M.; Aghaee, E.; Ahmadi, F.; Berijani, S. Journal of Chromatography A,2006, 1116, 1-9 220. Rezaee, M.; Yamini, Y.; Khanchi, A.; Faraji, M.; Saleh, A. Journal of Hazardous Materials, 2010, 178, 766-770 221. Rezaee, M.; Yamini, Y.; Faraji, M. Journal of Chromatography A, 2010, 1217, 2342- 2357 222. Zang, X.-H.; Wu, Q.-H.; Zhang, M.-Y.; Xi, G.-H.; Wang, Z. Chinese Journal of Analytical Chemistry, 2009, 37(2), 161-168

223. Lin, C.-Y.; Huang, S.-D. Journal of Chromatography A, 2008, 1193, 79-84 224. Nagaraju, D.; Huang, S.-D. Journal of Chromatography A, 2007, 1161, 89-97 225. Shamsipur, M.; Ramezani, M. Talanta, 2008, 75, 294-300 226. Di Muccio, A.; Pelosi, P.; Camoni, I.; Barbini, D. A.; Dommarco, R.; Generali, T.; Ausili, A. Journal of Chromatography A, 1996, 754, 497-506 227. Malone, E.M.; Elliott, C.T.; Kennedy, D.G.; Regan, L. Journal of Chromatography B, 2010, 878, 1077-1084 228. Tsai, W-H. ; Huang, T-C. ; Huang, J-J. ; Hsue, Y-H. ; Chuang, H-Y. Journal of Chromatography A, 2009, 1216, 2263-2269 229. Liu, M.; Yuki, H.; Song, Y.; Lin, J. Chinese Journal of Analytical Chemistry, 2006, 34( 7), 941-945 230. Nguyen, T. D.; Lee, M.H.; Lee, G. H. Microchemical Journal, 2010, 95, 113-119 231. Tsai, W-H.; Chuang, H-Y.; Chen, H-H.; Huang, C.J-J.; Chen, H-C.; Cheng, S-H.; Huang, T- C. Analytica Chimica Acta, 2009, 656, 56-62 232. Anastassiades, M.; Lehotay, S.J.; Stajnbaher, D.; Schenck, F.J. Journal of AOAC International, 2003, 86, 412-31 233. Restec Application Notes, QuEChERS Methodology:AOAC Approach: Q-sep™ Q150, cat.# 26214 234. Majors, R. E. LCGC North America, 2007 235. Pena-Pereira, F.; Lavilla, I.; Bendicho, C. Spectrochimica Acta B, 2009, 64, 1-15 236. Mohammadi, S. Z.; Afzali, D.; Baghelani, Y. M. Analytica Chimica Acta, 2009, 653, 173- 177 237. Kawaguchi, M.; Inoue, K.; Yoshimura, M.; Ito, R., Sakui, N., Okanouchi, N., Nakazawa, H. Journal of Chromatography B, 2004, 805, 41-48 238. Llompart, M.P.; Lorenzo, R.A.; Cela, R.; Paré, J.R. J.; Bélanger,J. M.R.; Li, K. Journal of Chromatography A, 1997, 757, 153-164 239. Amijee M.; Cheung, J.; Wells,R. J. Journal of Chromatography A, 1996, 738, 43-55 240. Basheer, C..; Lee, H.K. Journal of Chromatography A, 2004, 1057, 163-169 241. Fiamegos, Y. C.; Stalikas, C. D. Analytica Chimica Acta, 2007, 599, 76-83 242. Fiamegos, Y. C.; Stalikas, C. D. Analytica Chimica Acta, 2008, 609, 175-183 243. Fiamegos, Y. C.; Stalikas, C. D. Analytica Chimica Acta 2007, 597, 32-40 244. Kramer, K.E.; Andrews, A.R.J. Journal of Chromatography B, 2001, 760, 27-36 245. Lee, H.S.N.; Sng, M.T.; Basheer, C.; Lee, H.K. Journal of Chromatography A, 2007, 1148, 8-15 246. Cao, J.; Liang, P.; Liu R. Journal of Hazardous Materials, 2008, 152, 910-914 247. Itoh, N.; Tao, H.; Ibusuki, T. Analytica Chimica Acta, 2005, 535, 243-250 248. Poli, D.; Goldoni, M.; Corradi, M.; Acampa,O.; Carbognani, P.; Internullo, E.; Casalini, A.; Mutti, A. Journal of Chromatography B, 2010, 878, 2643-2651 249. Montero, L.; Conradi, S.; Weiss, H.; Popp, P. Journal of Chromatography A, 2005, 1071,163-169

250. Prietoa, A.; Basauria, O.; Rodilb, R.; Usobiagaa, A.; Fernándeza, L.A.; Etxebarriaa, N.; Zuloagaa, O. Journal of Chromatography A, 2010, 1217, 2642-2666 251. Kawaguchi, M.; Sakui, N.; Okanouchi, N.; Ito, R.; Saito, K.; Nakazawa, H. Journal of Chromatography A, 2005,1062, 23-29 252. Desmet, K.; Tienpont, B.; Sandra, P. Chromatographia, 2003, 57, 681-685 253. Saison, D.; De Schutter, D.P. ; Delvaux, F.; Delvaux, F.R. Journal of Chromatography A, 2009, 1216, 5061-5068 254. Mateo-Vivaracho, L.; Ferreira, V.; Cacho, J. Journal of Chromatography A, 2006, 1121, 1-9 255. Wang, Q.; O’Reilly, J.; Pawliszyn, J. Journal of Chromatography A, 2005, 1071, 147- 154 256. Pizarro, C.; Pérez-del-Notario, N.; Gonźalez-Śaiz, J.M. Journal of Chromatography A, 2007, 1143, 26-35 257. Reuther, R.; Jaeger, L.; Allard, B. Analytica Chimica Acta, 1999, 394, 259-269 258. Beceiro-González, E.; Guimaraes, A.; Alpendurada, M.F. Journal of Chromatography A, 2009, 1216, 5563-5569 259. Centineo, G.; González, E. B.; Sanz-Medel, A. Journal of Chromatography A, 2004, 1034, 191-197 260. Tsukagoshi, K.; Miyamoto, K.; Nakajima, R.; Ouchiyam, N. Journal of Chromatography A, 2001, 919, 331-337 261. Salih, B. Spectrochimica Acta B, 2000, 55, 1117-1127 262. Yamini, Y.; Rezaee, M.; Khanchi, A.; Faraji, M.; Saleh, A. Journal of Chromatography A, 2010, 1217, 2358-2364 263. Kagaya, S.; Takata, D.; Yoshimori, T.; Kanbara, T.; Tohda, K. Talanta, 2010, 80 1364- 1370 264. Dadfarnia, S.; Shabani, A. M. H. Analytica Chimica Acta, 2010, 658, 107-119 265. Rodrıguez, I.; Llompart, M.P.; Cela, R. Journal of Chromatography A, 2000, 885, 291- 304 266. Xu, L.; Gong, X. Y.; Lee, H. K.; Hauser, P.C. Journal of Chromatography A, 2008, 1205, 158-162 267. Palmans, R.; Claes, S.; Vanatta, L.E., Coleman, D.E. Journal of Chromatography A, 2005, 1085, 147-154 268. Wang, S.-P.; Liao, C.-S. Journal of Chromatography A, 2004, 1051, 213-219 269. Bianchi, F.; Careri, M.; Corradini, C.; Elviri, L.; Mangia, A.; Zagnoni, I. Journal of Chromatography B, 2005, 825, 193-200 270. Chienthavorn, O.; Pengpumkiat, S.; Noomhorm, A.; Smith,R. M. Journal of Chromatography A, 2007, 1152, 268-273 271. Bagheri, H.; Khalilian, F. Analytica Chimica Acta, 2005, 537, 81-87 272. Lord, H.; Pawliszyn, J. Journal of Chromatography A, 2000, 902, 17-63 273. Lambropoulou, D.A.; Konstantinou, I.K.; Albanis, T.A. Journal of Chromatography A, 2007, 1152, 70–96

Chapter 2: Application of the bubble-in-drop single drop micro-extraction (BID- SDME) method for monitoring of herbicides from farm areas

2.1 Introduction

Herbicides are widely used in South Africa for control of weeds and improved crop production. Metolachlor is a pre-emergence herbicide sprayed immediately after ploughing for control of narrow-leafed weeds and grasses.1 However, it has been classified as a human carcinogen by the WHO in 1993.2 On the other hand, atrazine is used against broad-leaved weeds and it is applied as a post-emergent herbicide. The ill-effects of some of these important herbicides in biodiversity raises big concerns, for example, atrazine is believed to cause hermaphrotidism in male frogs hence a threat to biodiversity.3

Rivard4 has studied and documented the environmental fate of metolachlor; the main transformation processes are the replacement of the chlorine atom with the OH group and sulphonation following dechlorination. Hartzler5 studied the binding parameters for atrazine and metolachlor in soils and demonstrated that metolachlor has a binding constant of 200 while that of atrazine is 100. This means that metolachlor binds more strongly to soils than atrazine. In a study by Accinelli et al.6 on degradation of atrazine and metolachlor in sub-soils, it was observed that degradation depends on the biological composition of the soil. Both herbicides showed varying decomposition rates: metolachlor degradation occurred only in non-sterile surface soil, with a half-life of 37 days, while under anaerobic conditions, atrazine showed half-life of 124 and 407 days under non-sterile surface and sub-soil respectively. However, in the same study it was reported that under sterile conditions, the half-lives for surface and sub-soils were not significantly different.

In laboratory simulation experiments, Singh et al.7 showed that movement of metolachlor and terbutylazine (a member of triazine herbicides family to which atrazine belongs) is much higher on the soils with lower organic matter content, and that metolachlor binds less strongly than terbutylazine, which is contrary to the work by Hartzler.5 In a study by Kochany and Maguire8 on photodegradation of metolachlor, it was demonstrated that metolachlor is quite stable under poor sunlight resulting in 205 days half-life in winter as opposed to 22 days

in summer. However, this was the case when no organic matter was present; the stability increased 2-3 fold with some organic matter introduction.

Degradation of metolachlor in soils is reportedly more biological in origin than chemical and it is considered co-metabolic.9 Different biological agents including fungi10,11 and bacteria12 have been reported to aid in the biodegradation of both chloro-acetamide derived herbicides, such as alachlor and metolachlor. The mostly reported bio-transformation and degradation products from metolachlor whether biological or chemical are dechlorination, dehydroxylation, dehydrochlorination, N-dealkylation and some degree of morpholine moiety formation.13,14

Pesticides monitoring has received a lot of attention with researchers reporting various methods ranging from electrochemical,15,16 immunochemical17 and chromatography,18 just to mention but a few. Some of the sample preparation methods used in monitoring and analysis of pesticides in environmental samples in combination with chromatography include solid- phase extraction (SPE),19 solid-phase micro-extraction (SPME),20 and solvent micro-extraction in the form of single-drop micro-extraction SDME.21 Pinto et al.22 have reviewed the use of liquid-based miroextraction methods as miniaturisation endeavour to reduce the hazards associated with classical solvent extraction. SDME has recently been introduced into the pool of sample preparation methods wherein it has shown similar efficiencies23 at a fraction of cost to its counterpart SPME.24

In the present work, validation of a recently developed BID-SDME25 method of pre- concentration for triazines and its application in the analysis of the triazines and metolachlor herbicides in field samples from the farming areas is reported. The samples in question include soil samples from sprayed fields and the surface run-off water from the fields with no prior knowledge of their spraying history. This work also demonstrates the potential of using BID-SDME as a sample clean-up coupled with other extraction methods for monitoring the dissipation not only of these herbicides (metolachlor and atrazine) but their breakdown products as well from time of spraying through harvesting.

2.2 Experimental

2.2.1 Reagents and chemicals

The pesticides standards (TP-619 mixture containing atraton, prometron, atrazine, propazine, simazine, terbutylazine, prometryn, symetryn, ametryn and terbutryn each at concentration of 100 µg/mL), 100 µg/mL diphenylamine (DPA), desethylatrazine (DEA) and desisopropylatrazine (DIPA) were obtained from Chem Service (Pennsylvania, USA), metolachlor (Met), atrazine (Atrz) and their deuterated analogues (100 µg/mL in 1 mL ampules each) were obtained Dr Ehrenstorfer GmbH (Augsburg, Germany), while chloroform, methanol and water (all HPLC grade) were purchased from Riedel-de Haën (Seelze, Germany). Sodium chloride (AR grade) was obtained from Sigma-Aldrich (Seelze, Germany). The agricultural soil samples were collected from a farming area in the North West Province near the small town called Derby about 150 km West of Johannesburg.

2.2.2 Standard solutions

The H-standards of 100 µg/mL atrazine (2-chloro-4-(ethylamino)-6-(isopropylamino)-s- triazine) and metolachlor (2-chloro-N-(2-ethyl-6-methyl-phenyl)-N-(1-methoxypropan-2-yl) acetamide) respectively were mixed and diluted to 1 µg/mL in MeOH and kept in a refrigerator set at around 2 °C. The deuterated standards (2H-stds) for the same herbicides were prepared and diluted in an identical manner. The breakdown products DEA and DIPA were also mixed with atrazine and metolachlor to prepare 1 µg/mL standard in MeOH and was kept in the freezer. The working solutions were diluted appropriately when required to make the organic and aqueous solutions, respectively (1 mL in GC vials). For the extraction procedure, 0.1 g of NaCl were added to the aqueous solutions to make 10% NaCl aqueous solutions.

The 100 ng/mL diphenylamine solution was prepared by dilution from the 100 µg/mL solution with chloroform, the mixture of which was used as the extracting solvent for the BID-SDME method and the diphenylamine used as an internal standard to control the droplet loss through evaporation and incomplete retraction of the plunger. The choice of DPA was based on the preliminary work25 that was performed and showed that it did not show significant leaching into the aqueous solution compared to acetanilide (same group as metolachlor). It 49

was a challenge to get the isotopically labeled standards, by the time they were accessed a bulk of the work had already been carried out.

2.2.3 Preparation of soil samples The soil samples were air-dried to a constant mass before being stored in a freezer. The soil suspensions were prepared by adding 2 g of air dried soil to 2 mL of water (1 g/mL). This suspension was treated by vortexing it at 30 Hz using a Zx3 vortexer (Velp Scientifica Italy), sonication for 15 minutes (FungiLab, Barcelona Spain), water was heated to boiling temperature (97 °C) using a hot plate (Labotec, South Africa), and thereafter centrifuged (Eppendorf centrifuge 5415D – Hamburg, Germany) to settle the particulate matter. Given portions of the clear supernatant liquid were transferred into a gas chromatograph (GC) vial containing 0.1 g NaCl, followed by BID-SDME extraction.

2.2.4 Instrumentation The 10 μL calibrated gas-tight Hamilton GC syringe (Seelze, Germany) was used for sampling and injections. Analyses were carried out using a Shimadzu (Kyoto, Japan) QP2010 gas chromatograph coupled to mass spectrometry (GC-MS) equipped with a Zebron 35MS column with 30 m x 0.25 mm x 0.25 μm dimensions. Pure helium (99.999%, Afrox, South Africa) at a constant flow rate of 1 mL/min was used as the carrier gas. Injections (1 μL) were carried out in the splitless mode with sampling times of 2 minutes; thereafter the split ratio of 1:10 was used throughout the runs for trace levels analyses. The injector and transfer line were maintained at 250 °C. The oven temperature programme included an: an initial temperature of 100 °C (held for 4 minutes), followed by 50 °C/min ramp to 200 °C, then ramped by 10 °C/min to 280 °C and held for 5 minutes. The total time for one GC run was 17 minutes.

For qualitative analyses of soil with high levels of herbicides content (those samples collected between November and January), the injection was carried out using split ratio of 1:50.

The MS (EI 70 eV and 200 °C ion source temperature) was set up on the scanning mode with mass range 50 to 350 mass units with detector potential set at 1.5 kV for the monitoring experiments. The ions of interest were extracted out of the total ion chromatogram as

follows: 162 (166), 169, 172, 187, 200 (205) for Met, DPA, DEA, DIPA and Atrz, respectively; (the values in the parentheses represent the deuterated analogues). For quantitative analysis at trace level (lower ng/mL range) the MS was set on the selected ion monitoring SIM mode using the same ions as listed above, otherwise it the MS was always used on scanning mode using the extracted ion monitoring (EIM) facility. In this mode, the ratios of the intensities of the SIM (reference) ion to the qualifying ion were determined to complement the NIST Library incorporated in the instrument as a secondary tool used to identify the analyte. For qualitative analyses with higher levels of herbicides content, the detector voltage was reduced to 0.7 kV to avoid saturation.

2.2.5 Micro-extraction procedure The set-up for the micro-extraction protocol is reported in details in literature and in our earlier work.25 1 µL of the extracting solution was drawn into the syringe, followed by 0.5 µL air, these contents were introduced into the aqueous solution through by gentle depression of the plunger, causing and the air volume to form an air bubble contained within the micro- droplets in the picture below. Following a static extraction period of 20 minutes, the total volume was carefully retracted back into the syringe carefully to avoid loss of the droplet through wicking, and injected into the GC-MS.

2.2.6 Performance of the analytical method for metolachlor Since the method was already developed and validated, only a limited amount of development and validation work was carried out. The main thrust included the assessment of the method for performance with metolachlor. The extraction efficiency was evaluated using three 10% aqueous solutions at concentrations 1, 50 and 100 ng/mL for each pesticide together with the corresponding organic solutions. Linearity and LOD were evaluated using aqueous mixture of atrazine and metolachlor in the concentration range of (0.05 - 1.0 ng/mL).

2.2.7 Sampling of the soil samples Individual grab samples were obtained from the field in triplicate. The samples were air-dried to a constant mass and the moisture content determined from the difference before and after drying of the samples. Each of the soil samples were analysed in triplicates.

2.2.8 Evaluation of extraction recovery 2 g of the air-dried soil samples, ground and sieved through a 0.0035 inch (0.089 mm) mesh, were suspended in HPLC grade water and spiked using the d-standard solutions to prepare concentrations of 25 ng/g soil samples. The suspension was allowed to air dry thereafter, and the samples were extracted with hot water (2 mL aliquots), and the water extracts were then subjected to the BID-SDME as for the other soil samples described above. The effect of organic solvents in the spiking mixture was also evaluated using different spiking solutions, including 100% aqueous, 25%, 50%, 75% and 100% MeOH, respectively. The extraction values especially those from the existing herbicides (H-metolachlor and H-atrazine) from these solutions were compared with the HPLC grade water treated the same way as the soil samples. All soil samples were analysed in triplicates, so every analysis has been carried out for n = 6 minimum.

2.3 Results and discussions Below shows the structure of the two herbicides monitored in the study.

Atrazine Metolachlor

In the case of the other triazines studied, the Cl atom and the alkyl groups attached on the exo-cyclic N atoms of the atrazine molecule above are replaced with different groups as tabulated in Table 2 below.

Table 2 A table showing different groups on individual triazines and their CAS numbers

X R1 R2 CAS number

Prometron OCH3 CH(CH3)2 CH(CH3)2 1610-18-0

Atraton OCH3 CH2CH3 CH(CH3)2 1610-17-9

Propazine Cl CH(CH3)2 CH(CH3)2 139-40-2

Atrazine Cl CH2CH3 CH(CH3)2 1912-24-9

Simazine Cl CH2CH3 CH2CH3 122-34-9

Terbutylazine Cl CH2CH3 C(CH3)3 5915-41-3

Prometryn SCH3 CH(CH3)2 CH(CH3)2 7287-19-6

Ametryn SCH3 CH2CH3 CH(CH3)2 834-12-8

Simetryn SCH3 CH2CH3 CH2CH3 1014-70-6

Terbutryn SCH3 CH2CH3 C(CH3)3 886-50-0

2.3.1 Development and validation of BID-SDME method for triazines analysis The development of this method has already been reported in the author’s previous work25 and the reader is as such referred to the earlier work appropriately. For ease of understanding the summary of the method is provided in the following paragraphs. The method involves use of a micro-droplet hanging on the tip of a micro-syringe and this droplet is in contact with the aqueous solution being sampled. A variant to this method was developed whereby an air-bubble is introduced into the droplet (see Figure 1 below).

Figure 1 Photograph of the ‘bubble-in-drop’ single-drop micro-extraction (BID-SDME) arrangement (solvent containing a blue dye was used for clarity and contrast).

The two pictures illustrate the ideal (A) and non-ideal (B) configurations of the drop-bubble assembly at the tip of the syringe. In picture A the droplet and air bubble are in ideal size ratio of volume of solvent to bubble size (2:1 for a 1 µL droplet) while in picture B, the air-bubble is overly large and is almost floating away from the opening of the syringe. This makes it difficult to suck the droplet back in the syringe for injection; as such it results in not only lower extraction efficiency but also erratic results due to variable amounts being drawn back into the syringe at the end of the extraction.

The linearity of the method was determined at the optimum conditions: 1 mL of a 10% aqueous NaCl solution extracted for 20 minutes with a 1 µL chloroform containing diphenylamine as an internal standard, and a 0.5 µL air bubble with no stirring. The results presented in Table 1 and Figures 2 below demonstrate sufficient linearity of the method at low concentrations approaching the detection limits (0.05 - 0.5 ng/mL). This also showed sufficiently low detection limits.

Prom Atro Prpz Atrz Simz 6

2 relative response relative 1

0 0 0.1 0.2 0.3 0.4 0.5

concentration (ng/mL) Figure 2 Calibration curves for some representative triazines from the US-EPA TP-619 herbicide mixture

The standard curves in Figure 2 above represent only half of the total number of components in the mixture to avoid overcrowding the figure. The calibration data extracted from these curves and the data corresponding to the other herbicides is shown Table 3 below.

Table 3 Analytical data from the calibration curves (0.05 - 0.5 ng/mL spiking) Prom Atro Prpz Atrz Simz Tbtz Prmy Amty Simy Tbty Intercept 0.0744 0.1387 0.7266 0.0263 0.0174 -0.035 -0.006 0.10283 0.1662 0.0486 Slope 3.1534 5.9144 8.5841 4.7578 2.683 11.299 9.2732 12.3702 8.2947 8.1705

SDIntercept 0.0128 0.1337 0.0094 0.0177 0.0135 0.0387 0.1028 0.18432 0.1176 0.1773 R2 0.987 0.99 0.993 0.986 0.983 0.982 0.991 0.993 0.991 0.991 Calculated LOD (ng/mL)a 0.01 0.07 0.003 0.01 0.02 0.01 0.03 0.05 0.04 0.07 Calculated LOD (ng/mL)b 0.0479 0.0152 0.0350 0.0317 0.0112 0.0077 0.0344 0.0374 0.0210 0.0571 Prescribed LOD (ng/mL)c 0.041 0.17 0.014 0.15 0.14 0.10d 0.024 0.2 0.035 0.031 a Limit of detection (LOD) calculated using 3  standard deviationintercept/Slope b LOD calculated using 3  standard deviation (SD) /slope; using SD of lowest concentration26,27

c,dTo comply with US EPA method 507 detection limits,28 except for Tbtz which is taken from WHO guidelines for drinking-water quality.29

From Table 3 above, it is clear that different ways of calculating the LODs result in different

values. For example the values obtained using the formula LOD = 3 x CL/slope where CL represents the standard deviation of the lowest concentration are lower LOD than those obtained using the 3 x SD/Slope.

Ribani et al.30 have demonstrated several ways of determining the detection limits from the chromatographically obtained results. The other approach uses the standard error of the calibration although in most cases it results in the higher LODs than the one determined from the standard error of intercept as used in Table 3 above. Vial and Jardy31 reported that the best approach is the use of residual standard deviation of a weighted regression as these values are more reliable and compare to those obtained with signal-to-noise ratio approach. No clear guidelines regarding which method is recommended have been established by the scientific community. In the case of the present project, the standard deviation of the intercept to slope was used since it takes into consideration the linearity of the method.

2.3.2 Validation of the BID-SDME method for accuracy using atraton The method was validated for accuracy using a certified reference material (0.5 ng/mL ± 5% atraton in water – Chem Service USA) and the corresponding standard solution. To validate

the method, a standard solution of atraton with a concentration of 0.5 ng/mL was prepared in water. 0.1 g of sodium chloride was added to both the resulting solution and into the certified reference material (CRM) solution to make 10% NaCl aqueous solutions respectively. These solutions were then extracted with BID-SDME as described above. The results are presented in Table 4 below for n = 9.

Table 4 Relative responses for atraton standard and CRM extractions using 100 ng/mL DPA internal standard

Standard solution (5 ng/mL) CRM (5 ng/mL) Int Std Atro Std Data Int Std Atro Std Data 0.5 ng/mL std 1485406 10254 7.21 CRM 1843648 12692 7.18 SD 199754 1683 0.6 SD 148030 2105 0.6

RSD 13 16 8.5 RSD 8 17 8.8

The concentration of the internal standard was reduced to be in the same order as the CRM and the corresponding standard. The results are presented in the table below for n=8.

Table 5 Relative responses for atraton and CRM extractions using 5 ng/mL DPA internal standard 5 ng/mL standard 5 ng/mL certified reference solution (CRM) Int Std Atro Std Data Int Std Atro Std Data 0.5 ng/mL std 96796 13369 138.02 CRM 107164 14861 138.60 SD 14263 2081 3.26 SD 10059 1491 1.20 RSD 15 16 2.4 RSD 9 10 0.9

The results from the two tables show that the method is sufficiently consistent and accurate since both the standard and the CRM give values that are within the experimental error of one another. Following this comparison, it was decided to determine the accuracy using a calibration curve method. Table 5 below presents the data used to achieve this feat.

Table 6 Data for accuracy validation using standard curve approach and 0.5 ng/mL CRM

Concentration (ng/mL) 0.10 0.25 0.50 0.75 1.00 relative response 0.0666 0.1850 0.3315 0.5260 0.6960 CRM relative response 0.349

When these data are plotted graphically, the standard curve below is achieved together with the calibration equation that is use dto calculate the concentration of the CRM.

0.8 y = 0.6951x - 0.0004 R² = 0.9984 0.6

0.4

0.2 relative responserelative

0 0 0.2 0.4 0.6 0.8 1 concentration (ppb)

Figure 3 A standard curve for the accuracy validation

The calculated concentration of the CRM from the calibration curve using the regression equation in Figure 3 above is 0.50 ng/mL ± 0.8% which is in agreement with the CRM concentration given as 0.50 ng/mL ± 5%. Interestingly the precision of the calculated CRM concentration is better than the reported precision on the CRM value.

2.3.3 Validation of the BID-SDME method for reproducibility Reproducibility gives a measure of how reproducible the method is over a period of several days. A set of standard solutions with concentration of ng/mL TP-619 mixture was prepared. Each day fresh solutions were prepared and extracted with the BID-SMDE. This was performed over four day. The results demonstrated below represent the averages (3n for n = 3 solutions; resulting in total of 9 replicates for each day) for each day using fresh solutions appropriately.

Table 7 Combined relative extraction results for reproducibility over four days Prom Atro Prpz Atrz Simz Tbtz Prmy Amty Simy Tbty Day-1 6.5 12.7 7.5 8.8 16.7 11.5 16 9.3 16.3 17.3 Day 2 5.9 13.2 7.6 9 17 11.4 16.2 9.3 16.5 17.5 Day 3 6.5 13.2 7.7 9.2 16.5 12.1 16.4 9.3 16.8 17.5 Day 4 7.1 13.1 7.7 9.2 16 11.9 15.7 9.0 15.7 16.5 Average 6.5 13.1 7.6 9.1 16.6 11.7 16.1 9.2 16.3 17.2 SD 0.490 0.238 0.096 0.191 0.420 0.330 0.299 0.150 0.465 0.476 % RSD 7.5 1.8 1.3 2.1 2.5 2.8 1.9 1.6 2.8 2.8

Figure 4 below shows the same data as in Table 7 above, pictorially.

Figure 4 A chart showing reproducibility of extraction over four days

Both presentations (Figure 4 and Table 7) demonstrate a high degree of reproducibility since the responses from different solutions over various days are highly consistent. The error bars in Figure 4 represent the standard deviations of three solutions for each day representing the values tabulated above, where the standard deviations shown reflect the inter-day repeatability.

2. 4 Application of the BID-SDME method in the analysis of metolachlor and atrazine in agricultural soil samples Having developed the method for the triazines determination based on the triazines mixture 619 from the US-EPA, it was discovered that metolachlor is commonly applied together with atrazine (a member of the triazine herbicides) for control of the other classes of plants that atrazine would be inactive against. Hence it became important to assess the applicability of the developed method in so far as the metolachlor is concerned. A few principles were assessed and the results are presented in the following sections, most of which are presented with those of atrazine as a co-analyte.

2.4.1 Extraction and analysis for monitoring of metolachlor and atrazine from farm soil Figure 5 below represent a total ion chromatogram for a metolachlor standard run on a scanning mode set between 60 and 350 amu.

(x1,000,000) 2.5 TIC

2.0

1.5

1.0

0.5

0.0

5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 Figure 5 A TIC chromatogram of metolachlor standard (1 µg/mL)

Metolachlor was positively identified from its MS spectrum using the NIST 2007 Library incorporated into the instrument with the similarity of 95% as well as the ratio of reference ion to qualifying ion (Ref Ion/Qual Ion).32,33 From the mass spectrum, the mass of 162 was used as a reference ions while 238 (M+) was used as a qualifying ion for the single ion monitoring (SIM) experiments necessary for trace analysis (where the ratio in the present case was 1.54:1 based on the full scan mass spectrum for the given eluted peak, alternatively

expressed as where the qualifying ion provides an intensity of 65% of that of the reference ion). The SIM chromatogram showing both the reference (162) and qualifying (238) ions is shown below. The black trace represents the m/z 162 while the yellow represents the m/z 238. From the SIM chromatogram below (Figure 6), the ratio of the reference to the qualifying ion is 1.54.

(x1,000) 162.00 (1.00) 238.00 (1.00) 5.0

2.5

0.0 11.00 11.25 11.50 11.75 12.00 12.25 12.50 12.75

Figure 6 The expanded SIM chromatogram showing both reference and qualifying ions for metolachlor

Previously, the experiment was set-up on true SIM while in this case the full scan was still maintained but the specific ions were selected to be shown on the monitor. This manipulation results in the same clean MS chromatogram (with good S/N ratio) without compromising the power of MS full scan necessary for identification even at low concentrations. This latter SIM/full scan hybrid was termed “extracted ion chromatogram” EIM and it was used for the monitoring experiments. Figure 7 shows the total ion chromatogram of the soil extracted with sonication followed by BID-SDME with the DPA used as an internal standard.

(x100,000) 7.5 TIC DPA

5.0 Met

2.5

7.0 8.0 9.0 10.0 11.0 12.0 13.0 Figure 7 TIC chromatogram of 9-week post-spraying sample from farms

When the base peak m/z values are extracted from the total ion count chromatogram a cleaner chromatogram is obtained as in Figure 8 below.

(x100,000) 169.00 (1.00) 162.00 (1.29) DPA 5.0 Met

2.5

0.0

7.0 8.0 9.0 10.0 11.0 12.0 13.0 Figure 8 An extracted ion chromatogram for diphenylamine (IS) and metolachlor

It is notable the extent to which the background is reduced by the use of extracted ion chromatogram, yet all important details for structural determination is still preserved.

Figure 9 below shows a chromatogram showing the 10 components of the TP-619 US-EPA triazine mixture, diphenylamine (internal standard) and metolachlor separated using the ZB- 35ms column. The identities are shown as the m/z values used in the SIM analysis. As it can be seen, metolachlor (Met) fits very well with the triazines without compromising resolution of other peaks through peak overlapping.

(x10,000) (214) (241) 5.0 DPA Atrz Met (169) (200) (226) (162) 4.0 (227)

3.0 (213) (210) (211) (201) 2.0 1.0 0.0

8.5 9.0 9.5 10.0 10.5 11.0 11.5 12.0

Figure 9 The SIM chromatogram of the TP-619 mixture using the m/z values from the scanning mode The m/z values in Figure 8 represent 162 (metolachlor), 169 (diphenylamine – internal standard), 200 (atrazine), 201 (simazine), 210 (atraton), 211 (prometron), 213 (simazine), 214 (propazine and terbutylazine), 226 (ametryn), 227 (terbutryn) and 241 (prometryn), respectively. 61

2.4.2 Extraction efficiency of metolachlor using BID-SDME Extraction efficiency is defined as the ratio of concentration after sample preparation to concentration in the original solution before/without sample preparation; E = Cextracted/Coriginal. For ease of interpretation extraction efficiency can be expressed in percentage.

To assess this, six solutions of metolachlor were prepared, three organic and the other three aqueous at concentrations of 1, 50 and 100 ng/mL, respectively. The organic solutions were injected directly into the GC-MS while the aqueous solutions were pre-extracted with BID- SDME as developed previously. The responses obtained when injecting the extracted aqueous solutions were taken as ratios with reference to the corresponding organic solutions resulting in the enrichment factors presented in the fourth column of Table 7 below.

Table 8 Assessment of extraction efficiency using metolachlor standard Response for Concentration Response for organic extracted 10% NaCl (ng/mL) solutions aqueous solutions Enrichment Factor 1 1 175 140 50 11 1728 154 100 23 4008 178

The extraction efficiency seems to drop with decreasing concentration, the decrease is linear with the R² = 0.9764. These data show that metolachlor is enriched 139 times the original concentration following BID-SDME method from aqueous solutions prepared using HPLC grade water.

2.4.3 Linear dynamic range and detection limits for metolachlor Linear dynamic range was determined from 10% NaCl aqueous solutions with concentrations in the range of 0.01 – 1 ng/mL, extracted with the BID-SDME method (n = 3) as previously developed for triazines using a mixture of atrazine and metolachlor standards. The extraction was carried out using the optimised conditions from the triazine mixture as follows: 1 mL of a 10% aqueous NaCl solution extracted for 20 minutes with a 1 µL chloroform containing diphenylamine as an internal standard, and a 0.5 µL air bubble with no stirring.

Figure 10 Standard curves for metolachlor and atrazine following BID-SDME extraction

The results presented in Figure 9 demonstrate linearity in the range used which is at the expected the detection limits based on atrazine. Table 9 presents some data derived from the standard curves above and the use of such data for calculation of detection limits of the two compounds respectively. In this case the LODs were calculated using standard error of the intercept approach.

Table 9 Some calibration data from the standard curves Parameter Metolachlor Atrazine Intercept 3.8069 0.2193 Slope 47.966 56.147 R2 0.9988 0.9997 LOD (ng/mL)* 0.024 0.013

* LOD determined using 3 × SDintercept / slope

The obtained LODs are better than the method reporting limits obtained with the US FDA method 302 for metolachor and atrazine (0.010 µg/mL)34 while the LOD cited in the EPA Method 507 from water samples are 0.19 ng/mL and 0.015 ng/mL for metolachlor and atrazine, respectively.35

2.4.4 Application of the BID-SDME method in monitoring of metolachlor and atrazine from agricultural soils Since the method has already been developed for triazines, matrix effect experiments were not independently developed for metolachlor, but rather extended to cover metolachlor as well, since the method is intended for analysis of both herbicides simultaneously as they are applied together in the farms.

The soil samples were collected from a farm field in North West Province near a small town called Derby about 150 km away from Johannesburg (see the picture below, Figure 11), over a period of 6 months commencing a week after spraying and stored in a cold room or freezer, respectively. They were then subjected to the sonication extraction protocol (see section 2.4.4.1.1) followed by the optimised BID-SDME and subsequently analysed by a GC-MS. Sonication at ambient temperature is the method commonly used to leach the analytes from soil matrices. This method was modified during the course of the present study, details of which are given below (see section 2.4.5.2).

Stream 1

Stream 2

Sampling field

Field with unknown history

Figure 11 A map showing the places where the samples were collected

Below is an expanded picture to show the field where the soil samples were collected for monitoring of the herbicides used mainly in this project. This is the area labelled “sampling field” in Figure 11 above.

Sampling site Farm 1

Figure 12 An expanded picture showing the field that was sampled

The area as can be seen in these two figures is flat and as a result does not have large surface run-off, thus a lot of herbicides loss should be more by downward leaching, normal breakdown and bio-transformation than by surface wash-off.

3.2.1 Determination of water moisture for samples collected at different times Water moisture plays an important role in mobility of herbicides in the soil depending on their relative solubility and adsorption onto the soil. The soil’s moisture levels were determined by weighing the samples before and after air-drying to a constant mass. Figure 12 below shows the moisture content of the soil samples obtained from week 1 to week 32 after spraying analysed in triplicates. It also shows the moisture profile as a function of soil depth (depicted as negative values relative to surface designated as 0 cm).

0 cm - 5 cm - 10 cm - 15 cm - 25 cm 18

15 12 9 6 relative moisture (%) moisture relative 3 0 5 10 15 20 25 30 35 number of weeks post spraying Figure 13 Moisture content of the soil as a function of number of weeks post spraying

Figure 13 above shows that there was a lot of moisture in the period between 10th to 15th week arising from heavy rains that were experienced from about week 7. The figure also indicates a remarkable water holding capacity in that the soil was not soaked despite heavy rains, yet still managed to retain sufficient moisture content during the dry part of the sampling period (weeks 1 -5 and 20 onwards). It is evident that the water retention seems to increase slightly from 0 cm (surface) to 5 cm then decrease downwards. The higher values for 5 cm could be attributed to evaporation on the surface, while the other decrease below 10 cm could be attributed to the soil structure. It was observed that the lower the depth, the higher silt content in the soil.

2.4.4.1.1 Treatment of soil samples Soil samples were weighed out without prior drying to prepare suspensions of 1 g/mL (soil/water) to give about 1 mL supernatant liquid. The suspensions were sonicated for 15 minutes since sonication has been used a simple method for extraction of pesticides from soils.36 Ozcan et al.37 recently reported the development and application of a miniaturised ultrasonic extraction method for organochlorine pesticides. The suspension was centrifuged to obtain a clear supernatant liquid which was transferred (1 mL) into a GC vial containing 0.1 g NaCl (to prepare a 10% NaCl solution). The resulting solution was then subjected to BID- SDME extraction. A TIC chromatogram of the samples collected a week after spraying was as shown in Figure 14 below. 66

(x1,000,000) 1.75 TIC 1.50 Met

1.25

1.00

0.75

0.50

0.25

5.0 7.5 10.0 12.5 15.0 Figure 14 A TIC chromatogram for the extract of the soil sample collected on week 1 post spraying

Only metolachlor, labelled Met, was obtained and identified from its retention time, comparison of its mass spectrum with that in the 2005 NIST Library as well as the ratio of the intensities of the reference (SIM) ion to the qualifying ion (1.47) compared to 1.54 obtained for a standard (see Figure 6 under section 2.4.1).

From this chromatogram, atrazine was not detected. Judging by the absence of its peak at its expected retention time (9.4 minutes) it can be seen that there is no atrazine in the soil, or at least the levels are below detection levels of the method as applied. Since the levels were expected to be relatively higher than trace levels, the instrument configuration was set up for higher levels. A split ratio of 1:50 was used on the GC injector together with the MS detector voltage of 0.7 kV. The idea of diluting the samples after extraction as a means to lower the concentration of the herbicides to fit in the range of the instrument sensitivity was ruled out since dilution it would have reduced the matrix effect as well. Therefore, this instrument sensitivity was the plausible choice.

2.4.4.1.2 Monitoring of dissipation-time profile The method was applied to analyse for the presence of metolachlor and degradation thereof in the farm field where it was sprayed. Figure 15 illustrates the depletion of metolachlor from the field soil over a period of 12 weeks. The depletion approximates the expected first order -kt decay (Ct = Co ) kinetics.

Figure 15 Dissipation-time profile for metolachlor in the sprayed field

The decay reaches its half-life in less than 14 days, thereafter it levels off as expected. The removal of herbicides from a specific environment (including farming soil) is usually attributed to various environmental factors, biological breakdown, leaching, as discussed in the introduction.

Interestingly neither atrazine nor its breakdown products were detected in these samples, despite this herbicide having been sprayed onto the soils in question about four weeks after the spraying of metolachlor.

Attempts to detect the presence of both herbicides at different soil depths showed that metolachlor was more prevalent in the surface soil as the sub-surface soil layers (from a depth of 15 cm downwards) did not show any detectable amounts. This contradicts earlier reports which classified metolachlor as “transient” in soils with poor biological activity.38 Analysis of metolachlor at different soils depths showed that the herbicide was more bound to the surface soil as analyses of different depths did not show any detectable amounts. No atrazine was detected in any of the samples at both surface and sub-surface levels.

The absence of triazines in the soil could be attributed any of the following:  To high mobility of triazines in the soil as reported by Harzler5 more so when the triazines were sprayed onto soil that had metolachlor already bound to it. This would be true if 68

both compounds compete for similar binding sites onto which they can bind to the soil. This point is further supported by the fact that atrazine was detected in the flood/run-off streams that were flowing around the farming area. This is despite atrazine not being detected in the farming soil sampled.

 Another possible cause of this observation which is contrary to the one above could be the stronger adsorption of the atrazine onto the soil despite having metolachlor already bound. The extraction was performed using sonication with water at room temperature which may not sufficiently release these analytes for BID-SDME to pick them up for subsequent analyses.

 Heavy rains that were prevailing over that period; the farmer from whose field the samples were collected anecdotally explained that such heavy rains were not experienced in the last 13 years of his farming.

 Low spraying level: the farmer alluded to the fact that he had sprayed atrazine at lower levels than prescribed due to the fear of its persistent presence in the soil lasting to the following year, thereby affecting his next crop (sun flower, a broad-leaved crop that would be inhibited by presence of atrazine).

Interestingly, no broad-leafed plants were seen growing near the run-off streams (where atrazine was detected) which are indicative of presence of triazines in the running water.

2.4.5 Exploration of more robust extraction methods for soil samples combined with BID- SDME as a clean-up step A number of more robust extraction methods have been documented and discussed in the preceding sections of this thesis. Unfortunately most of these methods require further clean- up or work-up stage at times leading to use and generation of organic solvent waste, losses and/or contamination of the analytes. Some of the techniques used in the extraction of soil samples are more robust than simple sonication and include microwave-assisted extraction39 and capital and technologically intensive supercritical extraction methods.40

The failure to detect atrazine and its breakdown products implied that it was prudent to think of alternative and more stringent methods to extract the soil samples before subjecting the resultant solution to BID-SDME. Having been using water at room temperature with vortexing and sonication for the prior work, it was decided to explore hot/boiling and superheated water extractions. As discussed in the literature review chapter, superheated water can have the same solvent strength as organic solvents at sufficiently high temperatures. A benefit of this technique is that the hazards associated with organic solvents are avoided when using water. However, thermal stability of the analytes still needs to be considered to prevent loss from thermal degradation. The results of these experiments are presented below.

2.4.5.1 Hot water extraction (HWE) This method was attempted in two versions, namely, using hot-water to extract cold soil and heating the soil-water suspension to near boiling. In the former case, boiling water was introduced to the sample and vortexed for about half a minute before being allowed to settle and the routine procedure followed from this point onwards (centrifugation, addition of salt, etc.). In the other version, the soil-water was vortexed to mix it up before being heated to near boiling to avoid loss of solvent (water), the mixture was allowed to cool and centrifuged as routinely. The results demonstrated in Figure 16 below compare the three different extraction techniques: room temperature sonication, hot-water extraction and super-heated water extraction using the soil samples collected on 22 January (12 weeks after metolachlor spraying).

Figure 16 Evaluation of different extraction methods for soil sample (22 Jan 2010) 70

It was noted that sample extractions with hot water extraction suffer from pressure differences in the pipette leading to variable volumes being transferred, accounting for the relatively higher standard deviations as observed by the error bars in Figure 16 above, compared to the ambient temperature conditions.

It is worth noting that both cases of hot or boiling water have their own inherent draw-backs; adding boiling water poses a problem in measuring the volume accurately as a lot of vapour gets sucked in to the pipette and renders volume readings a bit problematic in addition to the fact that the water temperature gets reduced due to cold soil sample. In the case of boiling water, there is potential loss of water due to evaporation during the heating process.

2.4.5.2 Evaluation of sonication on hot water extraction efficiency It was decided to assess the effect of sonication on hot water extraction following the realisation that hot-water does have an influence on dissolution of the herbicides especially atrazine from the soil particles. 2 x 3 samples (labelled HWE and HWE + Sonic with numbers 1, 2 and 3, respectively) were each collected and extracted n = 3 resulting in Figure 17 below.

Figure 17 Evaluation of hot-water and hot-water-sonication assisted extraction modes

From Figure 17 above, it seems that sonication improves the extraction efficiency of hot water, especially in the case of atrazine thus supporting the idea that atrazine binds more strongly to the soil than metolachlor, since the extraction of metolachlor increased by only about 10% relative to the hot-water extraction. This therefore suggests that monitoring of 71

atrazine would require hot-water sonication while that of metolachlor would not benefit that much from the extra heating step.

2.4.5.2 Superheated water extraction To evaluate the superheating format, the soil suspension was heated in a sealed stainless steel reactor and heated to 140 °C (oil bath temperature) and held for about 5 minutes, followed by cooling to about 50 °C; the suspension was then transferred to 2 mL Eppendorf tubes for centrifugation to obtain clear supernatant liquid for BID-SDME.

The results as shown in Figure 17 above comparing the three techniques show a dramatic increase in efficiency when using superheated water extraction (SHWE) compared to hot- water extraction (HWE). However, it must be pointed out that despite its benefits the superheated water extraction method still suffers from being quite time consuming given the heating and cooling as well as the closing and opening processes involved.

2.4.6 Dissipation of the herbicides extractions using hot water extraction Having shown quite good improvements to the solubilisation protocols when employing hot water extraction, it was decided to revisit the soil samples with the view to 1) improving sensitivity for the detection of metolachlor in the latter part of the sampling period and 2) reassessing the samples for the presence of atrazine. Figure 18 below shows the merged dissipation profiles for the two extraction methods: room temperature sonication (blue trace) and hot water (green and purple traces for atrazine and metolachlor respectively) extraction. It must still be borne in mind that these levels were still relatively high for trace analyses, as such the instrument was still not set-up for trace analyses.

As can be seen from Figure 18 below, hot water extraction allows the tracking of dissipation of atrazine as well. The dissipation of atrazine at this level did not show the exponential decay demonstrated by metolachlor, possibly due to the fact that the earlier time segment was missed during weeks 5 to 8. However, the trend between weeks 9 and 11 still shows some degree of exponential behaviour.

Figure 18 Dissipation-time profile from soil samples extracted with hot-water extraction

2.4.6 Dissipation into the soil layers using the hot-water extraction Dissipation into different soil layers gives variable results. Table below shows results obtained using samples obtained from about 25 cm deep. Seemingly the herbicides did not leach down substantially into the deeper layers of the soil profile. They were only detected at the depth of 25 cm after week 11 after spraying.

Table 10 A table showing presence of herbicides as a function of soil depth Metolachlor Atrazine Week 9 n.d. n.d. Week 10 n.d. n.d. Week 11 4 2 Week 16 4 1 Week 28 2 0 Key: n.d. means not detected

2.4.6 Analysis of breakdown products for both atrazine and metolachlor The major chemical breakdown product of atrazine is desethyl-atrazine (DEA with m/z 187) and the metabolic breakdown product is 1-hydroxyl-atrazine (m/z 197). The analysis for these breakdown products was carried-out on the soil samples. The results showed that no considerable concentration of these compounds was observed. The triazines sample was

spiked with metolachlor and desethyl-atrazine and sampled to determine the retention time of desethyl-atrazine. The relevant portion of chromatogram is attached in Figure 19 below.

(x10,000,000) TIC

4.0 DEA 3.0

2.0

1.0

0.0

6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 Figure 19 A TIC chromatogram of desethyl-atrazine

The mass spectrum of the desethyl-atrazine is presented in Figure 20 below for reference.

100.0 172

75.0

50.0 68 187 25.0 104 145 11094 189 53 79 130 0.0 159 207 219 232 240253 269 281 297 50.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 300.0 Figure 20 The mass spectrum of desethyl-atrazine

Desethyl-atrazine gives the same m/z value (172) with a reference-to-qualifying ion ratio (Ref Ion/Qual Ion) of 3.3. Terbutylazine on the other hand, gives m/z 214, with a fragment at 173. Atrazine gives m/z 200 (100%), followed by molecular ion peak 215, and a small peak at m/z 173. Terbutylazine is confirmed by both the MS spectrum and the retention time (9,92 min). Desethyl-atrazine has a retention time of 9.30 minutes and it was identified from the injection of a pure standard as well as using the MS library.

An extracted ion monitoring (EIM) chromatogram of the TP-619 mixture used in the triazine method development including metolachlor and desethyl-atrazine is shown below. The new compound posed no serious interference issues.

(x100,000) 169 187 (DEA) 1.00 162 214 0.75 227 200 241 213 226 210 211 0.50 201 0.25

0.00 8.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 12.0 12.5 Figure 21 An EIM chromatogram of the TP-619 mixture including metolachlor and desethylatrazine Peak order starting with the pink dominant trace at about 8.9 minutes: diphenylamine (169), desethylatrazine (187), prometron (210), atraton (211), propazine (214), atrazine (200), simazine (201), terbutylazine (214), prometryn (241), ametryn (226), simetryn (213), terbutryn (227) and metolachlor (162) around 12 minutes.

Applying the HWE solubilisation technique to the soil samples also allowed desethylatrazine (DEA) to be traced (Figure 22 below) over time together with atrazine. The results show that the relative concentration of DEA (pink trace) is higher than the parent Atrz (blue) at week 5 post-application (remembering that the application of atrazine was performed at week four following metolachlor spraying). The concentration of DEA diminishes essentially as a mirror of the concentration of the parent atrazine. No breakdown products for metolachlor could be identified in the extracts and were thus not monitored over the time.

Figure 22 Dissipation-time profiles for atrazine and DEA from the soil following HWE

It must be emphasised that the levels used in the dissipation-time profile above were still higher than trace analysis level.

2.5 Recovery of herbicides from the soil samples for quantitative analysis

Since no information had been gathered on the extent of recoveries of the analytes from the soil samples, it was decided to pursue experiments that would assist in answering that question. To this end, it was decided to use soil sample that was free of the analytes of interest and perform the recovery experiment on the soil collected thereof (to cater for similarity of soil) and thereafter to apply those results to the soil samples of interest (that was sprayed with the herbicides). All the experiments from this section onwards were performed with the instrument configured for trace analysis (detector voltage set from 0.7 kV to 1.5 kV and the split-splitless injector set at splitless). The conditions were set the same as those used in the optimisation and calibration curves construction for trace analyses. Accordingly, a soil sample A soil sample (BA-1) collected from a field that had been sprayed in the previous year, was weighed appropriately in duplicate. One sample was spiked with 10 μL of 100 ng/mL metolachlor solution and agitated to mix and then allowed to settle overnight while the other remained without spiking. The following morning, both samples were extracted with boiling water and thereafter treated the same way soil samples had been treated in this study. The results are as follows.

Figure 23 A chart showing the influence of spiking metolachlor into the soil

These results demonstrate that the sample already was not yet free of metolachlor, which may pose a problem for this part of the study. It was then decided to use the actual samples containing both atrazine and metolachlor to the levels already pre-determined and use the deuterated standards in place of the analytes that would be used had the samples been free

of metolachlor. Further experiments were conducted using the sample collected in May (5 and 7 months post spraying – atrazine and metolachlor respectively).

The use of deuterated standards is important as these standards resemble the analytes for all chemical properties except for their molecular mass. The molecular mass difference allows for mass spectroscopic identification hence they would provide a good basis for recovery experiments of the deuterated standards. The recovery experiments would now lead to quantitative information regarding the trend of pesticides dissipation demonstrated earlier.

2 2 2 A 2H6-metolachlor (propyl- H6) and H5-atrazine (ethylamino- H5) at concentrations of 100 µg/mL in acetone each, were mixed and diluted with methanol and water respectively to prepare the standard solutions with concentration of 1 µg/mL each. These solutions were used in the subsequent spiking experiments for recovery. The possibility of 2H – 1H exchange was investigated by measuring the isotope content of the standard, made up as described above, over several days. During this period, no change in the ratio of the protonated/deuterated analyte could be detected, pointing towards the integrity of the 2H- standards over this period of time when made up in a protonated solvent.

A preliminary experiment was set up where in 2 x 2 g of soil sample were subjected to solvent extraction and the other to hot-water extraction followed by BID-SDME. For solvent extraction, the sample was heated under reflux in 2 mL of methanol for an hour followed by centrifugation; thereafter the supernatant liquid was directly injected into the GC-MS. In the latter case, 2 mL of boiling water was introduced to a 2 g soil sample which was left to stand 15 minutes. The supernatant liquid following centrifugation was extracted with BID-SDME as described above.

The results presented in Figure 24 show the expected results that indeed HWE/BID-SDME method out-performed the standard solvent extraction approach. This is understandably due to the pre-concentration nature of the BID-SDME as a technique.

Figure 24 A comparison of solvent extraction with hot-water coupled to BID-SDME

2.5.1 Evaluation of spiking protocol While the data generated above provide good relative information, absolute data based on recovery experiments are important for the quantification of analytes. To assess the level of recovery, labelled analogues of the analytes of interest that are readily discriminated and quantified by detectors such as mass spectrometers are usually doped into the sample.41 Given their essentially identical chemical and physical characteristics to the analyte in question, this approach is widely used to establish quantitative recovery of analytes from samples.42 The spiking solution is usually added with the organic analyte dissolved in an appropriate organic solvent such as methanol43,44 and the influence of the spiking solvent is ignored. In other instances, aqueous spiking solutions are used.45,46

Spiking and equilibration are important in assessing the recovery of the spiked analytes. If equilibration is not sufficient this results in relatively high recovery at the expense of accuracy of the results and may lead to an over-estimation of the level of the target analytes in such matrices. As such a series of experiments was set-up to evaluate this concept. The most fundamental principle is to attempt to ensure homogeneity of the deuterated standard across the entire volume of the sample which is then left until all the solvent has evaporated in an effort to mimic the analyte behaviour. This in itself is a challenge since the solvent chosen should be able to distribute in the soil and mix with the water moisture residues in the soil. As such methanol and acetone were selected and used in the experiments. 78

Two soil samples were weighed (2.00 g) and one was suspended in acetone (1.5 mL) overnight (treated), while the other was just left in open polytop (reference soil). Both samples were then subjected to the same extraction protocol as all soil samples. Results were as shown in Table 11 below.

Table 11 The effect of spiking solution on extraction of the analytes Metolachlor 2H-Metolachlor Atrazine 2H-Atrazine Reference Soila 16 (13) 0 2 (12) 0 Treated Soilb 50 (6) 0 18 (6) 0 a Reference soil – non-treated soil b Treated soil was suspended in acetone until dryness before both being extracted with HWE and BID-SDME; values in parentheses are the %RSD values

Interestingly, the results demonstrate a phenomenon that was not anticipated. It was observed that the treatment of the soil sample with acetone somehow increased the extractability (3 times and 9 times for metolachlor and atrazine, respectively) of the H- analytes from the control to the spiked.

This phenomenon was further observed on the spiking experiment detailed below. A series of solutions were prepared as follows: Solution 1. Water – 1 ml of water spiked with 50 µL of 1 µg/mL solution of d-standards and evaporated overnight, then extracted with 2 mL of hot-water and BID-SDME appropriately.

Solution 2. Control – 2 g of field soil sample was extracted with hot-water without treatment nor spiking. This soil was virgin from the field and had not in any way been treated. This is the control for the H-pesticides.

Solution 3. Spiked control – 2 g of the field soil sample spiked with 50 μL of the 1 µg/mL 2H- pesticides; vortexed and allowed to stand, thereafter subjected to hot-water extraction.

Solution 4. Spiked-treated soil – 2 g of the field soil sample, with 50 μL of the 1 µg/mL d- pesticides solution; then 1.0 mL of MeOH was added and the sample vortexed for 2 minutes. The additional methanol was added in an effort to ensure that the surrogate was evenly

distributed over the entire sample. The solution was left for the MeOH to evaporate overnight and thereafter subjected to HWE.

Solution 5. Spiked-treated and oven dried (expressed as spiked treated x 2) – 2.0 g of the field soil sample was spiked 50 µL of the 1 µg/mL d-pesticides solution and suspended in 1 mL of MeOH. Following an overnight drying at room temperature the sample was further heated in the oven at 60 °C for 4 hours. The solution was allowed to cool to room temperature before being subjected to the HWE as other soil samples described above. The results are as shown in Table 12 below.

Table 12 Recovery data of the surrogate and analytes from different solutions Relative responses from extraction % Recovery 2 2 2 2 Met H6-Met Atr H5-Atrz Met H6-Met Atr H5-Atrz Water 1 (12) 21 (12) 0 34 (16) 0 100 0 100 Control 13 (8) 0 3 (6) 0 100 0 100 0 Spiked control 17 (10) 28 (11) 6 (6) 35 (10) 131 133 187 100 Spiked treated 49 (9) 28 (9) 22 (17) 33 (17) 371 133 684 97 Spiked treated (oven dried) 50 (4) 27 (4) 23 (5) 31 (3) 377 127 730 90 Values in parentheses represent the %RSD values; green code represents normal analytes relative to control soil sample; yellow colour represents d-labelled (surrogate) values with reference to water (aqueous solution)

The recovery of the 2H-standards ranged between 97% and 133% (atrazine and metolachlor, respectively) relative to the treated water (see the bold entries). The recovery data were measured for n = 7 and the %RSD is as shown on the table (9 - 17%). It seems oven drying of the sample does not have a marked effect on the recovery of the 2H-standards; which remained almost constant (90 – 100%) and 127 – 133%) within experimental error.

One of the most important observations, though, was that the extraction of the analytes already present in the sample) was considerably higher (6.84 and 7.30 times respectively) in all cases of spiking, relative to unspiked samples. This indicated some form of effect resulting from the spiking procedure, which should ideally not influence extractability of the analytes. The surrogate solution was prepared in an organic solvent – methanol. This solvent is commonly used to deliver the spiking analytes.

Could this additional recovery be as a result of the added methanol? It was thought possible that the methanol treatment could act to improve the extraction of the soil-bound analytes and hence result in improved extraction. The aqueous leaching of the analytes from the soil would possibly not leach all of the analyte in question from the soil, resulting in some of the analyte remaining on the soil post-leaching with HWE. The Table shows that the 2H-standard gave essentially identical recoveries in all instances. This indicated that the action of the solvent (used for spiking and wetting the soil) was on the metolachlor and atrazine that pre- existed on the soil, not on the spiked 2H-standard. The action in question could be to solubilise the analytes, thereby possibly accounting for the improved recoveries.

To respond to the question above, experiments were conducted using the 1 µg/mL d- standards mixtures. Four soil samples were prepared and suspended in 1 mL of solvents with the following water-methanol compositions: 100%, 75%, 50% and 0% MeOH, the remainder being water. These solutions were spiked with the 2H-analytes as before. The results are given in Figure 25 below.

Figure 25 The effect of solvent composition of the spiking solution on extraction

What is interesting from the results (Figure 25) is that the metolachlor and atrazine analytes (pre-existing in the soil samples from the farming operations) both showed significantly higher relative amounts upon analysis using the BID-SDME technique. The amount of the analytes being extracted increased as the organic solvent concentration of the suspension solution in increased. The deuterated spiking surrogate materials showed little changes to 81

their levels of extraction (with the exception of the 100% MeOH suspension sample) as was suspected from the previous results (see Table 13). The implication of this exercise is that the suspension of the soils in methanol-containing solutions enhances the extractability of the analytes in question, despite the samples being air dried and essentially solvent-free at the time of analysis. Practically, this implies that the method of choice for the application of doping solutions to dry soil samples is to suspend the soil samples in water, to add the spiking solution to the suspension, to vortex the suspension to assist in intimate mixing and dispersion of the dopant and to allow the samples to air dry.

This organic solvent effect is consistent with the results obtained by Merini et al.47 who reported recoveries of 49% when using an aqueous surrogate and 81% when using a surrogate for 2,4-dichlorophenoxyacetic acid which was prepared in methanol. However no explanation was made about this effect. The authors simply noted it. This observation is very important since most of the spiking reported in literature is carried out using an organic solvent in which the surrogates are already dissolved in.

2.5.1.1 The effect of spiking using aqueous surrogate (trace-less spiking) Having determined the influence of the spiking solution and the suspension solvent on the extractability of the analytes, it was decided to use strictly aqueous solutions even for spiking so that no extra effect on the sample would be exerted during spiking. For this purpose, a 1 µg/mL aqueous spiking solution containing the surrogate and the atrazine break down products (DEA and DIPA) was prepared by dilution of a methanolic 100 µg/mL solution with HPLC grade water. This solution was used (50 µL of the 1 µg/mL aqueous surrogate) to spike the soil samples (2 g) which were suspended in water and then air dried. The results in Table 13 demonstrate the recovery of the both the analyte and surrogate herbicides from differently treated soils and water used as a reference solution.

Table 13 Recoveries of the analytes and surrogates from different samples spiked using aqueous standard solution obtained by direct comparison

DIPA DEA Atrz Met D-Atrz D-Met Water (reference) 5.0 22.5 0 0.9 11.8 40.0 Control soila 0.8 2.2 3.0 62.5 0.0 0.0 Control treated soilb 0.8 1.7 2.8 61.6 0.0 0.0 Spiked treated soil 5.6 22.9 3.0 63.5 8.3 42.6 % Recoveryc 96 92 - - 70 106 a used without spiking nor suspension in water b suspended in water to mimic spiking and suspension without the use of surrogate solution c The % recovery was calculated as response ratio of spiked soil to water (reference) x 100%

These results show that indeed water did not provide any effects since control, treated control, spiked solutions gave the same extractability for the H-analytes (average 2.9 and 63 atrazine and metolachlor respectively – lighter shade). The recoveries obtained for the surrogate herbicides are 70 and 106% respectively (darker shade). This result was critical since it was important to establish a traceless spiking protocol, from which quantitative data could be obtained. These data are important in the translation of the qualitative data discussed earlier in this chapter into quantitative data. Additionally, these recoveries demonstrate that atrazine binds more strongly to the soil than metolachlor since it is recovered to a lesser extent. The results also show that both atrazine breakdown products DEA and DIPA demonstrate good recoveries (see also section 2.5.1.2, below).

To evaluate the effect of distribution of the surrogate evenly throughout the soil, two spiked samples were compared. In one case the soil sample was spiked using aqueous surrogate, mixed and then allowed to evaporate without suspending the sample in any liquid. Recall that this was done to effect even distribution of the surrogate. In the other case, the soil sample was spiked and suspended in water as usual to distribute the spiked analytes over the soil sample. The extractions were also compared with control sample that was not spiked with the surrogate. The results from the extraction of these samples are presented in Table 14 below as percentages of the recovery in HPLC grade water (entry 1).

Table 14 Normalised recoveries of the surrogate herbicides with and without suspension Entry Sample treatment 2H-Atrazine 2H-Metolachlor 1 Water (reference) 100 100 2 Control 0.0 0.0 3 Treated Control 0.0 0.0 4 Spiked 314 188 5 Treated Spiked 70 106

Interestingly, the recovery in the case of non-treated spiked sample (entry 4) was higher than that of the analogous sample (entry 5) where water had been added to the sample to distribute the spiked analytes. This observation probably arises since the surrogates were not distributed throughout the soil, but rather over a small volume covered by the 50 µL of the surrogate solution. However, in the pre-suspended sample, the surrogate distributed over a larger amount of sample and hence bound more strongly. The latter provides a more realistic situation than the former since the surrogate is allowed to distribute over the entire sample volume and therefore more accurately reflects the situation that would be found in the field. A comparison of the treated spiked solution to the reference yields recovery values of 70% and 107% for 2H-atrazine and 2H-metolachlor, respectively, whilst that of the non-treated spiked solution results in 314% and 187%, respectively, very clearly demonstrating the importance (at least in this type of scenario) of even distribution of the spiking analyte.

2.5.1.2 Recovery of the atrazine breakdown products (desisopropyl atrazine – DIPA, desethylatrazine – DEA) It was also important to perform a similar set of experiments but with the atrazine breakdown products DIPA and DEA as the subjects of the study. The treatment of the soil sample and the spiking solutions was identical to that detailed above. Table 15 below illustrates the data for breakdown products for the H-analogues of the herbicides that were already in the soils since the soil was not free of these herbicides, along with data on the 2H- standards for ease of comparison.

Table 15 Recoveries of the atrazine breakdown products following spiking Entry Sample treatment DIPA DEA 1 Water (reference) 5.0 22.5 2 Control 0.8 2.2 3 Control treated 0.8 1.7 4 Spiked control 10.3 30.8 5 Spiked treated 5.6 22.9 6 % Recovery* 96 92 DIPA – desisopropylatrazine, DEA – desethylatrazine, * value obtained by dividing the difference between the values obtained for spiked and treated control by water x 100%

Both breakdown products DIPA and DEA demonstrate good recovery values (96% and 92%, respectively). It is interesting to note the low recovery of atrazine (70% - see Table 15 above) relative to its breakdown products and metolachlor. This could be due to the solubility difference between the parent atrazine and the daughter products. These daughter products, having lost the alkyl groups, would be relatively more polar, and thus more soluble in water. This relatively high recovery could also explain the higher levels of the DEA than the parent atrazine in the time-dissipation profile in Figure 22 above.

2.5.1.3 The effect of period post spraying on recovery It is important to consider the possibility that recovery of the analytes may vary with time due to the ever changing conditions which the field is exposed, including, organic matter content, plants growth and heat/sun exposure as mentioned in the introduction. It would be ideal to determine the levels of recovery for each soil sample to obtain a perfect picture. Moisture content of the samples used was ranging between 10% and 15% (w/w) and was not expected to exert a strong influence on the recovery experiments.

The levels of recovery of the 2H-standards were determined by making use of the spiking/suspending method described above. Soil samples collected on 08 Dec, 14 Jan and 12 May were used. Each of the samples was spiked with the d-standards while one set was used as control. Table 16 below shows the recoveries for samples collected at different times whose recoveries were determined by direct comparison method. The levels of recovery were calculated based on the use of the spiked HPLC grade water as the reference consistent with 85

the earlier sections. Accordingly, the recovery data presented up to this point are with reference to the HPLC grade water sample spiked at the same levels as the soil samples with the 2H-standards.

Table 16 Recoveries of deuterated standards for soils collected at different times Date of collection Recovery (%) 2 2 H5-atrazine H6-metolachlor Dec_1 58 81 Dec_2 50 75 Dec_3 56 82 Jan_1 55 80 Jan_2 63 89 Jan_3 60 86 May_1 78 85 May_2 63 79 May_3 65 82

These data show that recoveries do not differ that much between the months and that in future work it may be accepted that the levels of recovery for a given soil sample would not vary much over a period of months.

2.5.1.3.1 Determination of the recoveries from the calibration curve Recoveries determined above were obtained through direct comparison, i.e. computation of the recovery levels obtained from the soils against that obtained with spiked HPLC grade water. These recovery levels can also be determined from a standard curve using the response vs concentration of the 2H-standards. The concentration is determined from the response observed with the respective analytes from reference and the spiked samples and this is compared to the known levels of spiking.

2 2 Four standard solutions of H5-atrazine and H6-metolachlor were prepared in the concentrations 1, 5, 10, 25 ng/mL with which to set up a calibration curve. Following the extractions of all the samples, a standard curve was plotted for each of the two compounds from which the recoveries of the standards from the soil samples were determined.

D5-atrazine D6-metolachlor 60 y = 2.1095x + 1.1157 50 R² = 0.9986 40

relative response relative y = 1.488x - 0.3324 10 R² = 0.9995 0 0 5 10 15 20 25 concentration (ppb) 2 2 Figure 26 The standard curves for H5-atrazine and H6-metolachlor respectively

The values obtained from the standard curves (see Table 17 below) are comparable to those obtained by direct comparison. Recovery at the beginning of the growing season (December) is slightly lower than towards harvesting, when most of the plant material has begun to die and rot in the soil thus providing a lot of humic/fulvic acids.

Table 17 Obtained concentrations (ng/mL) from spiked soil samples collected at different periods 2H-Atrazine (ng/mL) 2H-Metolachlor (ng/mL December spiked 21 (58) 41 (77) January spiked 23 (63) 47 (89) May spiked 29 (77) 46 (85) Values in parentheses represent the percentage recovery using the known levels of spiking as the basis for the calculations.

2.6 Quantitation of the amounts of the herbicides in the soil Having determined the recoveries of the analytes from the soil samples, it was then decided to go back to the soil samples and determine the quantitative levels of these herbicides. It must be highlighted that the dissipation-time profiles presented earlier are based on high levels of the herbicides (higher µg/mL levels) not trace levels. As stated under Section 2.4.4.1.1, the herbicides levels were too high (higher than µg/mL or µg/g) for the first few weeks and the GC-MS was too sensitive as such the sensitivity of the MS was reduced through 87

reduction of the detector voltage as well as increasing the split ratio, as stated. The working range of the MS detector is the low µg/mL to ng/mL without pre-concentration.

The configuration was changed when the recovery experiments were started in order to increase the sensitivity of the instrument for application in trace analyses (concentrations in the low ng/mL to pg/mL range). The soil samples collected from March (week 18) through June (week 30) were used for these analyses. Table 18 below presents the relative responses and the corresponding concentrations calculated from the calibration curves obtained in Section 2.4.3.

Table 18 Quantitative data for different samples obtained from the calibration curves using recovery data

Week number Metolachlor Atrazine Relative responses obtained from the extractionsa 18 79.8 (5.7)b 8.3 (4.5) 22 65.9 (1.9) 5.1 (5.7) 26 62.9 (0.7) 2.9 (3.5) 30 48.4 (6.3) 1.9 (3.6) Calculated concentrations (ng/g) c 18 1.585 (6.2) 0.144 (2.9) 22 1.295 (2.0) 0.087 (8.4) 26 1.231 (0.8 0.048 (3.5) 30 0.930 (6.8) 0.031 (7.3) Adjusted concentration (ng/g) using the recovery data 18 1.617 ± 0.131 (98%)e 0.206 ± 0.007 (70%) 22 1.523 ± 0.044 (85%) 0.124 ± 0.012 (70%) 26 1.231 ± 0.016 (100%) 0.069 ± 0.003 (77%) 30 1.094 ± 0.107 (85%) 0.044 ± 0.004 (70%) a Average values for n = 3 replicates b The values in parentheses depict the %RSD C Calculated using the calibration equations: ymet = 47.966x + 3.8069 and yatrz= 56.147x + 0.2193 d Recovery values used: 100% (metolachlor) and 70% (atrazine) e The values in parentheses are the percentage recoveries

2.7 Exploration of samples whose spraying history was unknown It was then decided to test the method developed here on the samples which the author had no prior knowledge of pesticides used in the past in that area. To this effect, a soil sample was collected from a field where the history of the application was not known. The field seemed to be still active since there were small plants growing on it; hence it was justifiable to think that some herbicides could have been applied. The map of that field is shown in Figure 27 below.

Collection site Field with unknown history

Figure 27 A photograph of the farm whose spraying history was unknown

2.7.1 Analysis of soil samples from a field with unknown history

The relevant soil samples were collected (08 December 2009) and screened for presence of any herbicides. Figure 28 shows the total ion chromatogram resulting from the extraction of the soil sample with hot water and BID-SDME pre-concentration method as described above.

(x1,000,000) TIC

1.5

1.0

0.5

5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 Figure 28 A TIC chromatogram of the soil sample post BID-SDME extraction

Figure 28 below shows the extraction ion chromatogram of the corresponding TIC above. It shows the presence of atrazine (9.8 minutes), metolachlor (black trace - 12 minutes) and terbutylazine (9.9 minutes). The retention times of the herbicides coincided with those previously determined as part of the study.

(x10,000) 5.0 DPA Tbtz 4.0 169 214 3.0 Atrz Met 2.0 200 162 1.0

0.0

5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0

Figure 29 The extracted ion chromatogram of the ions of interest from the soil sample DPA - diphenylamine (internal standard); Atrz – atrazine, Tbtz – terbutylazine, Met - metolachlor

The intensity of the DPA (internal standard) peak (169) has been reduced by a factor of 2 to match the intensities of the other compounds. As has been iterated above, the use of extracted ion monitoring clearly extracts the information needed out of the otherwise highly busy chromatogram. The observed m/z value of 172 (9.3 minutes) was attributed to the desethylatrazine based on a combination of the retention time, ratio of reference to qualifying ions (172/187) of 3.2 compared to the standard with a ratio of 3.3, and by comparison with the NIST 2007 library. Desethylterbutylazine would produce the same fragment but would show a different retention time and a different ratio of the two ions.

The signal at the retention time 9.8 minutes was assigned to atrazine, by comparison of its retention time, mass spectrum (Figure 30) compared to that of a genuine sample and the NIST 2007 library as well as the ratio of peak intensities for peak 215 and 200 (1.8).

% 100.0 200

75.0

91 215 50.0 155 173 202 104 25.0 132 184 68 145 256 297 76 122 239 284 0.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 300.0

Figure 30 An MS spectrum of the peak 200 on the extracted ion chromatogram from soil sample This spectrum matches that of atrazine from the NIST 2007 library with similarity of 65%. % 125.0

100.0 214

75.0

173 50.0 229 25.0 68 132 158 175 71 104 145 91 197 210 240 256 297 0.0 284 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 300.0 Figure 31 An MS spectrum of peak 214 on the extracted ion chromatogram from soil sample This spectrum matches that of terbutylazine from the NIST 2007 library with similarity of 80%.

The signal at retention time 9.9 minutes was assigned to terbutylazine. This assignment was done using the same approach as for the other compounds above. The retention time coincides with that of the terbutylazine as determined from that constituent in the TP-619 mixture on the same GC column.

The elution peak at 12 minutes was confidently assigned to metolachlor, again based on the same principles.

100.0 162

75.0

238 50.0

25.0 146 194 91 132 185 211 256 0.0 52 165 284 50.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 Figure 32 An MS spectrum of the peak 162 from the extracted ion chromatogram of the soil sample This spectrum matches that of metolachlor from the NIST 2007 library with similarity of 75%.

Besides the similarities in the generated mass spectra with those in the NIST library, these compounds were also identified on the bases of their retention times and the peak intensity ratios of their base peaks and the qualifying peaks as mentioned above. Their abundances relative to the internal standard (DPA at 100 ng/mL) are presented below. The actual concentrations were not determined since the recoveries of these herbicides on this soil were not determined. Hence the concentration would be stated with a high degree of uncertainty.

Figure 33 Analysis of the herbicides present in the field whose history is unknown

The data above show that not only atrazine but also terbutylazine has been applied in the soil. Since access to the field was limited, monitoring over time was not possible. However this experiment still demonstrates that the BID-SDME method coupled to hot-water

extraction can be applied in environmental applications where no prior information is available about nature and abundance of this type of herbicides.

2.7.2 Analysis of stream water samples Water samples were collected from run-off streams (see Figure 11) down-stream from some farms whose history of the use of these herbicides was not known. The water samples were centrifuged to remove any particulate matter. Thereafter, 0.1 g of NaCl was added to 1 mL aliquots followed by the BID-SDME extraction.

2.7.2.1 Analysis of water samples from stream 1

A map of the collection area is shown in Figure 34 below. The area was a bit swampy with a shallow furrow which gave the water some direction of flow down the gradient to the right of the road (yellow line cutting from top to bottom). The water was collected on the edge of the road as there was a fence securing the fields on either side of the road, and entrance was not permitted to these fields as their owners could not be contacted.

Figure 34 A photograph of stream 1 showing where water sample was collected

The TIC chromatogram (Figure 34) of the water sample (collected on January 22, 2010) following the extraction procedure summarised above is presented below. The

chromatogram generated using the extracted ions of the compounds identifiable from their corresponding MS spectra and the retention times is also shown (Figure 35).

(x1,000,000) TIC

1.25

1.00

0.75

0.50

0.25

0.00 5.0 7.5 10.0 12.5 15.0 Figure 35 A TIC chromatogram of stream-water (stream 1 – collected 14 Feb 2010) sample

(x1,000) 162 6.0 200 214 5.0

4.0 DPA 169 3.0

2.0

1.0

0.0 7.5 10.0 12.5 15.0 Figure 36 An EIM chromatogram of 169, 172, 200 (m/z values for DIPA, DEA and atrazine)

Analysis of this water for metolachlor showed no detectable levels thereof (absence of peak 162 around 12 minutes). However, high levels of atrazine were detected in both streams. The retention time of the elution peak in question at 9.8 minutes coupled to its mass spectrum (see Figure 36) and a comparison thereof with the NIST 2007 library and other spectra generated by authentic samples of atrazine led to this conclusion. This was also compared by the ratio of the intensities SIM ion (200) to the qualifying ion (215) giving ratio of 1.53 compared to the pure atrazine standard (1.67) with a difference of about 8.4%. 94

Due to high rainfalls experienced in the area at the time, the levels of atrazine in these water bodies was attributed to high surface run-off from the farms laying up-stream of the rivulet sample site.

% 100.0 200 58 75.0 215

173 50.0 71

122 132 158 25.0 104 92 217 180 265 281 79 295 0.0 253 50.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 300.0 Figure 37 The MS spectrum of the 2nd peak (blue trace)

The signal at retention time 9.9 minutes was assigned to terbutylazine based on its retention time, peaks (214/229) intensity ratio of 2.5 compared to 2.59 from authentic sample, and obtained mass spectrum (Figure 38 below) compared to that of the authentic sample and the one in the NIST 2007 library. No breakdown compounds (i.e. des-alkyl-) were detected.

100.0 214

173 75.0 57 71

50.0 96 132 229 68 97 111 175 25.0 158 207 295 197 256 267 0.0 50.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 Figure 38 The MS spectrum of the 3rd compound

Time profile for the herbicides in the stream water (stream 1) The streams were not running full-time so determination of the dissipation over time was erratic and the results did not show any trend as presented in Figure 38 below. Beyond the 6th week the streams had completely stopped flowing.

Figure 39 Dissipation profiles of atrazine and herbicides in the run-off stream water – stream 1

2.7.2.2 Analysis of water samples from stream 2

Water sample was collected from stream 2 indicated on the picture below. The samples were collected alongside the road as there was a fence securing the fields along which the stream was flowing, with a flow from left to right. This stream had to diversions of two streams, one directly leading to the water reservoir on the right while the other was running along the road into the area labelled “sampling site” and gets through the culvert just by the start of the trees at the bottom of the picture. The water sample was analysed in the same way as the sample from stream 1.

Sampling site

Collection site

Figure 40 A map picture showing stream 1 where water samples were collected

The results are presented below.

(x1,000,000) TIC 5.0

4.0

3.0

2.0

1.0

5.0 7.5 10.0 12.5 15.0 Figure 41 A TIC chromatogram of the water from stream 2 (14 Jan 2010)

As can be seen on the chromatogram (Figure 41) there are many peaks that could not be linked to the herbicides under investigation. This could be associated with deteriorating column showing excessive column bleed. The regular slightly increasing period between the peaks often result from different fractions of the polysiloxanes that form the column bed. An extracted ion chromatogram of the SIM ions is presented below (Figure 42), and it revealed presence of atrazine in the sample. This assignment was based on the retention time as well as other factors discussed in the above sections.

(x1,000,000) TIC 4.0 172.00 (50.00) 214.00 (50.00) 3.0

2.0 Atrz 200 1.0

0.0

8.5 9.0 9.5 10.0 10.5 11.0 11.5 Figure 42 An expanded EIM chromatogram using ions of interest

% 150.0

125.0

100.0 200

75.0 215

50.0 68 173 202 69 93 25.0 122 138 96 158 105 187 281 0.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 Figure 43 The MS spectrum of the peak at 9.8 minutes with 84% similarity with NIST 2007 library for atrazine

Another sample of water from the same stream was collected on the 22 Jan 2010. This sample was also centrifuged and subjected to BID-SDME extraction, the extract of which was injected into the GC-MS for analysis (Figure 44 below).

(x1,000,000) TIC 7.0

6.0

5.0

4.0

3.0

2.0

1.0

7.0 8.0 9.0 10.0 11.0 12.0 13.0 Figure 44 A TIC chromatogram for water from stream 2 collected on week 11 post spraying (22 Jan 2010)

(x10,000) 172.00 (1.00) 4.5 169.00 (0.50) 214.00 (1.00) 4.0 200.00 (1.00) 173.00 (1.00) DPA 3.5 3.0 2.5 DEA 2.0 1.5 1.0 0.5 0.0 -0.5 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 Figure 45 A EIM chromatogram of water from stream 2 (22 Jan 2010) with an internal standard The ions in the figure represent internal standard (m/z 169), desethylatrazine (m/z 172),

It can be noted that only DEA, an atrazine breakdown product, could be identified in this water sample based on the same principles as in the above sections. Neither atrazine not metolachlor were detected.

2.8 General discussions and conclusions BID-SDME has been demonstrated as an alternative method for the determination and monitoring of metolachlor and triazine herbicides from farming areas. The method is efficient and cheaper compared to its reported counter-parts, as already mentioned in the introduction parts. The importance of temperature of the extracting water is critical as has been demonstrated in this study. Extraction of these herbicides using hot water improved four fold to the extraction efficiency of room temperature sonication.

The recovery of herbicides from soil samples is not an easy task as evidenced above. This work has also demonstrated the importance of the composition of the spiking solution in being able to effect traceless spiking of the soil sample. It was not possible to detect metolachlor in the deeper layers of the soil profile, nor was it detected in the run-off streams. These results may be treated with confidence since it was shown that the recoveries for metolachlor are about 100% at the 50 ng/mL level, with the use of hot water extraction coupled to BID-SDME. Atrazine could be detected in the soil samples and in the streams that were sampled, and its disappearance from the soil could be reliably tracked. The BID-SDME method coupled to hot water extraction has been demonstrated to be sufficiently effective to determine both qualitatively and quantitatively analyses of these herbicides from period of spraying where the concentrations are high to the time of harvesting where levels are approaching detection limits, 0.930 - 1.094 ng/mL (metolachlor) and 0.044 - 0.038 ng/mL (atrazine), compared to their LODs (0.024 ng/mL (metolachlor) and 0.013 ng/mL (atrazine).

The dissipation in the run-off streams cannot be modelled since the streams ran only immediately after rains. However, the method seems to work satisfactorily for the run-off water as well.

2.9 References

1. Liu, H.; Huang, R.; Xie, F.; Zhang, S.; Shi, J. Journal of Hazardous Materials, 2012, 217-218, 330-337 2. WHO, Guidelines for Drinking-Water Quality, vol. 1. 2nd ed. (Chap3). 1993 WHO Geneva 3. Hayes, T.; Haston, K.; Tsui, M.; Hoang, A.; Haeffele, C.; Vonk, A. Environmental Health Perspectives, 2003, 111, 568-575 4. Rivard, L. Environmental Monitoring Branch, Department of Pesticide Regulation, 2003, Sacramento, CA 95812 5. Hartzler, B. Department of Agronomy, 2002, Iowa State University, 6. Accinelli, C.; Dinelli, G.; Vicari, A.; Catizone, P. Biology of Fertility of Soils, 2001, 33, 495–500 7. Singh, N.; Kloeppel, H.; Klein, W. Chemosphere, 2002, 47, 409-415 8. Kochany, J.; Maguire, R.J. Journal of Agricultural and Food Chemistry, 1994, 42, 406-412 9. Barker, A.V.; Bryson, G.M. The Scientific World Journal, 2002, 2, 407-420 10. Krause, A.; Hancock W.G.; Minard, R.D.; Freyer, A.J.; Honeycutt, R.C.; Lebaron, H.M.; Paulson, D.L.; Liu, S.Y.; Bollag, J.M. Environmental Toxicology and Water Quality, 1985, 33, 584–589 11. Liu, D.; Maguire R.J.; Pacepavicius, G.J.; Aoyama, I.; Okamura, H. Environmental Toxicology and Water Quality, 1995, 10(4), 249–258 12. Munoz, A.; Koskinen W.C.; Cox, L.; Sadowsky, M.J. Journal of Agricultural and Food Chemistry, 2011, 59, 619-627 13. Mathew, R.; Khan, S.U. Journal of Food and Agriculture, 1996, 44, 3996-4000 14. Zheng, H.-H.; Ye, C-M. Journal of Environmental Sciences, 2003, 15, 783-790 15. Songa, E.A.; Arotiba, O.A.; Owino, J.H.O.; Jahed, N.; Baker, P.G.L.; Iwuoha, E.I. Bioelectrochemistry, 2009, 75(2), 117-123 16. Chouteau, C.; Dzyadevych, S.; Durrieu, C.; Chovelon, J.-M. Biosensors and Bioelectronics, 2005, 21, 273–281 17. Reder, S.; Dieterle, F.; Jansen, H.; Alcock, S.; Gauglitz, G. Biosensors and Bioelectronics, 2003, 19(5), 447-455 18. Hogenboom, A.C.; van Leerdam, J.A.; de Voogt, P. Journal of Chromatography A, 2009, 1216, 510-519 19. Gaynor, J.D.; MacTavish, D.C.; Labaj, A.B. Chemosphere, 1998, 36, 3199-3210 20. Frías, S.; Rodríquez, M.A.; Conde, J.E.; Pérez-Trujillo, J.P. Journal of Chromatography A, 2003, 1007, 127-135 21. Bagheri, H.; Khalilian, F. Analytica Chimica Acta, 2005, 537, 81-87 22. Pinto, M.I.; Sontag, G.; Bernardino, R.J.; Noronha, J.P. Michrochemical Journal, 2010, 96, 225-237 23. Jeannot, M.J.; Cantwell, F.F Analytical Chemistry, 1996, 68, 2236-2240 24. London, L.; Dalvie, M.A.; Nowicki, A.; Cairncross, E. Water SA, 2005, 1(31), 53-59 25. George, M.J. MSc Thesis, 2007, University of Johannesburg, South Africa 100

26. Eide, M.; Martin, L.; Motley, K.; Plummer, T.; Potter, W. Method Evaluation Guidelines - Occupational Safety and Health Administration, USA (1993). 27. Tor, A.; Aydin, M.E.; Özcan, S. Analytica Chimica Acta, 2006, 559, 173-180 28. Munch, J. Method 507, 1995, United States Environmental Protection Agency 29. Chambon, P.; Lund, U.; Galal-Gorchev, H.; Ohanian, E. Guidelines for drinking-water quality, 2nd Ed., 2, WHO, Geneva. Switzerland (2003). 30. Ribani, M.; Collins, C.H.; Bottoli, C.B.G. Journal of Chromatography A, 2007, 1157, 201-205 31. Vial, J.; Jardy, A. Analytical Chemistry, 1999, 71, 2672-2677. 32. Pitarch, E., Medina, C., Portolés, T., Hernández, F. Analytica Chimica Acta, 2007, 583, 246- 258 33. Portolés, T.; Pitarch, E.; López, F.J.; Hernández, F. Journal of Chromatography A, 2011, 1218, 303–315 34. Peterson, D. Analytical Report to eXfuze, LLP, 2009, Environmental Micro Analysis Inc 35. Munch, J. Method 507, United States Environmental Protection Agency, 1995, 1-31 36. Gonçalves, C.; Alpendurada, M.F. Talanta, 2005, 65, 1179-1189 37 . Ozcan, S.; Tor, A.;Aydin, M.M. Analytica Chimica Acta, 2009, 640, 52-57 38. Caracciolo, A. B.; Giuliano, G.; Grenni, P.; Guzzella, L.; Pozzoni, F.; Bottoni, P.; Fava, L.; Crobe, A.; Orrù, M.; Funari, E. Environmental Pollution, 2005, 134, 525–534 39. Shen, G.; Lee, H.K., Journal of Chromatography A, 2003, 985, 167-174 40. Smith, R.M. Analytical and Bioanalytical Chemistry, 2006, 385, 419–421 41. van de Merbel, N.C. Trends in Analytical Chemistry, 2008, 27, 924-933 42. Morgan, L.; Dike, L.; Dandeneau, A.; Crespi,C. L.; David, M. Stresser BD Biosciences – Discovery Labware, Woburn, MA 43. D’Agostino, P.A.; Hancock, J.R.; Provost L. R. Journal of Chromatography A, 2001, 912, 291-299 44. Saifuddin, N.; Chua, K.H. Malaysian Journal of Chemistry, 2003, 5, 30-33 45. Psillakis, E.; Naxakis, G.; Kalogerakis, N. Global Nest: the Int. J. 2000, 2, 227-236 46. D’Agostino, P.A.; Hancock, J.R.; Chenier, C.L. Journal of Chromatography A, 2004, 1058, 97–105 47. Merini, L.J.; Cuadrado, V.; Giulietti, A.M. Chemosphere, 2008, 71, 2168–2172

101

Chapter 3: Application of BID-SDME in the analysis of organochlorine pesticides in water

3.1 Introduction

Organochlorines (OCPs) are a class of pesticides used for control of insects. One specific candidate in this class is 1,1,1-trichloro-2,2-di(4-chlorophenyl)ethane commonly known as dichlorodiphenyltrichloroethane (abbreviated DDT) which was developed as a synthetic insecticide in the 1940s and was subsequently used against insect-borne diseases by the military in the Second World War and civilian applications.1 However, despite its efficiency, DDT has been the subject of many discussions worldwide2 due to its non-biodegradability and consequent build-up in the foodchain due to its high lipophilicity.3

Their endocrine disruption effects range from poor semen quality, testicular cancer, menstrual cycle abnormalities, spontaneous abortions, and cause prolonged pregnancy, reduced birth weight, skewed sex ratio, and altered age or sexual development.4 The use of DDT has been restricted by the Stockholm convention5 in 2001 and the WHO6 to countries where malaria is rampant. However, it is spread and detected even in places that it is not applied through means such as wind, water and the migration of animals and birds from contaminated areas.7 It is not only DDT but most of the organochlorine compounds that are used as pesticides including endosulfan, lindane (γ-HCH), aldrin, just to mention a few,8 that are persistent in the environment.

Vaclavic et al.9 attempted to develop a predictive model for lipophilic accumulation of DDT in Danish women; and they showed that adipose tissue concentrations of DDE and PCBs were consistently positively associated with age and the consumption of fish with a high fat content. Omori-Inoue et al.10 reported the determination of some organochlorines including DDT in umbilical cords and studied the effects thereof on genetic expression, although their results were inconclusive. Siddiqui et al.11 studied the relationship between breast cancer and levels of organochlorines. Despite high levels of OCPs their study could not conclude a direct correlation.

102 In South Africa DDT use is controlled and restricted to regions where malaria still persists. These areas include Limpopo province and some parts of the Mpumalanga and KwaZulu-Natal Provinces.12 It has been reported that the loss in revenue in Africa as a whole related directly or indirectly to malaria exceeded US$12 billion annually as at 2004.13 These costs include control of vectors, treatment of malaria patients, loss of productivity and eventual deaths. This demonstrates an aspect of the controversy of the use of DDT. Should its use be stopped, the repercussions could be severe. However, there are also fears of tolerance towards DDT by the mosquito vectors it is intended to eradicate, resulting from careless application and consequent constant exposure.14 It is reported that about 800 000 people die of malaria annually in sub-Saharan Africa.15

Most of the methods employed in the analysis of DDT and other organochlorines use gas chromatography coupled to an electron capture detector16 and mass spectrometry.17 Having a high electron affinity due to the chlorine atoms, organochlorines and other halogenated compounds are normally analysed using the negative ionisation (also known as electron capture negative ionisation) mass spectrometry.18 Due to the abundance of chlorine isotopes in the organochlorines, negative chemical ionisation (NCI) mode is preferred over electron impact ionisation.19 Due to their mild nature, the molecular ion mass is maintained in NCI.20 Chromatography requires good sample clean-up to reduce the matrix effects and pre- concentration of the analytes is required so that they can be easily detected. Jain and Verma have reviewed a number of liquid-based sample clean-up techniques reported for analyses of organic compounds including organochlorines.21 Solid-phase techniques including the micro- extraction formats are also reported.22

SDME making use of mixed solvents (p-xylene and acetone, 8:2 v/v) has been applied for GC- MS analysis of organochlorines from various vegetables with the determination limits at an S/N ratio of 3 ranging from 0.05 ng/mL for α, β, γ-HCH (hexachlorocyclohexane) to 0.2 ng/mL for dicofol, dieldrin or p,p-DDT.23

This study presents the development and application of the bubble-in-drop single-drop micro- extraction (BID-SDME) method on organochlorines and their subsequent analysis using gas chromatography and mass spectrometry detection. The method was validated for analysis of

103 DDT and applied to real water samples collected from a dam in Limpopo (the northern province of South Africa) where DDT is still applied to control mosquito that transmits malaria. This is the first report of this novel technique being applied on organochlorines. It is known to the author that EI-MS is not an ideal detector for quantitative analysis of organochlorines at ultra-trace levels especially using SIM mode, despite its importance in identification in monitoring analysis. The better quantitative devices, ECD and NCI-MS were not available hence the study was performed using the EI-MS detection which could compromise the limits of detection and quantification, but would allow for requisite principles to be demonstrated.

3.2 Experimental 3.2.1 Chemicals The US-EPA organochlorine mixture (PM 625 JM) containing 100 µg/mL each of aldrin (1,2,3,4,10,10-Hexachloro-1,4,4a,5,8,8a-hexahydro-1,4:5,8-dimethanonaphthalene), lindane (hexachlorocyclohexane), endosulfan (6,7,8,9,10,10-Hexachloro-1,5,5a,6,9,9a-hexahydro- 6,9- methano-2,4,3-benzodioxathiepine-3-oxide) - α and β isomers, 4,4’-DDE (1,1-bis-(4- chlorophenyl)-2,2-dichloroethene) and 4,4’-DDT (1,1,1-trichloro-2,2-di(4- 2 chlorophenyl)ethane) was obtained from Chem Service (Pennsylvania, USA). H8-DDT was obtained from Dr Erhenstorfer GmbH (Augsburg, Germany), 1,3,5 trichlorobenzene (TCB) from SigmaAldrich, (Seelze, Germany), toluene (pestanal grade) from Riedel-de Haën (Seelze, Germany), chloroform, methanol (HPLC grade), dichloroethane (DCE) and dichloromethane were obtained from Labscan (Sowioskiego, Poland), while formic acid and ammonium hydroxide were obtained from Merck (Johannesburg, South Africa).

Standard solutions (1 µg/mL) were prepared by serial dilution of a standard stock solution (100 µg/mL) using HPLC grade methanol and stored in a freezer. From this solution, the working solutions were prepared by dilution as described below before being used. The extracting solvents were prepared by spiking the organic solvent in question with TCB to make up the internal standard with a concentration of 50 ng/mL.

104 3.2.2 Apparatus A Hamilton 10 µL graduated syringe was used for injections, and a PIERCE multi-vial thermo bath with stirring facility (PIERCE, Rockford, Illinois - USA) was used to control temperature. Sub-mm length magnetic fleas from Sigma-Aldrich (Steinheim- Germany) were used for stirring solutions on the thermo bath. pH was measured using a Hanna pH meter (Hanna Instruments, Bulgaria).

3.2.3 Instrumentation A Shimadzu QP2010 GC-MS (Shimadzu, Kyoto - Japan) was fitted with a Zebron -1ms column with dimensions 30 m × 0.25 mm × 0.25 µm (Phenomenex, Torrance, California – USA). Helium was used as a carrier gas at a flow rate of 1 mL/min. The instrumental settings were as follow, the split-splitless injector at 260 °C was operated using split-less (2 minutes) followed by split ratio of 1:10 with the linear velocity and a column flow of 1 mL/min. The column temperature was set at 80 °C held for 4 min; ramped at 40 °C/min to 250 °C; thereafter ramped by 15 °C/min to 280 °C and held for 5.75 min for a total runtime of 16 min.

The mass spectrometer was set as below. The acquisition mode was set to scanning mode with m/z range of 100 to 450 at a scan speed 769 and time event (0.5 sec) resulting in scan rate of 1538 scans/sec for the identification of the compounds. The ionisation was carried out with the electron impact mode with energy of 70 eV. The interface temperature was set at 250 °C, the ion source set at 200 °C with the solvent cut-off time of 4.5 minutes. The detector voltage was set at 1.5 kV. The following m/z values were used for selected ion monitoring: 180 (internal standard), 181 (219), 195 (237), 235 (165), 246 (318) and 263 (293), respectively where the values in parentheses indicate the qualifying ion m/z values selected on the basis of their relative intensities compared to the SIM ions.

3.2.4 Extraction procedures The extraction procedure has been outlined in the earlier chapter and it was applied without change unless otherwise stated in the section preceding each set of results for ease of reference.

105 3.3 Experimental results and discussions

3.3.1 Chromatographic separation and identification of the compounds

Chromatography is used to separate compounds of interest and to their retention times to be accurately measured under specified conditions. The total ion chromatogram of the PM 625 JM organochlorine mixture is shown in Figure 46 below. All six components are well-resolved and the chromatography is achieved within reasonably short time.

(x10,000,000) TIC 4.0 DDE DDT

3.0 Aldrin Endos β Endos α Lindane 2.0

1.0

0.0 7.0 8.0 9.0 10.0 11.0 Figure 46 A total ion chromatogram of the OCP 625 mixture

The corresponding extracted ion chromatogram (EIM) is shown below in Figure 47. Note the improvement in the baseline and the ability to select a compound (m/z value) at will. As before, it was anticipated that this mode would improve the detection limit of the instrument. Whereas previously EIM was also useful to assess in the determination of overlapping peaks, in this case the peaks are well-resolved.

(x1,000,000) 2.0 246 235 1.5 181 263 195 1.0 195

0.5

0.0

7.0 8.0 9.0 10.0 11.0 Figure 47 An EIM chromatogram of mixture showing m/z values

106 The mass spectra of the compounds as appearing in the TIC chromatogram above are shown together with their molecular structures in Figure 48.

Lindane (m/z 181) Aldrin (m/z 263)

Endosulfan A and B (m/z 195) DDE (m/z 246)

DDT (m/z 235) Trichlorobenze (m/z 180) (internal standard - IS)

Figure 48 The mass spectra of the various organochlorine compounds and their chemical formulae

As was expected, the mass spectra in Figure 48 above show the various isotopic masses characteristic of organochlorine compounds. This is due to the presence and abundance of the two chlorine isotopes – 35Cl and 37Cl, which also stymies mass spectrometer which, in SIM mode, must be set to every one isotopic mass. The following reference/qualifying ion ratios were obtained from the standards: lindane – 1.013, aldrin – 2.469, endosulfan A and B – 1.653, DDE – 1.429 and DDT – 2.026 respectively.

107 3.3.2 Optimisation of the extraction method The preliminary optimisation process used a 25 ng/mL standard solution of the PM 625 organochlorines mixture extracted using an extracting solvent spiked with trichlorobenzene (TCB) to the 50 ng/mL level. There are a number of parameters that needed to be optimised before the method could be validated and applied to real systems. Amongst these parameters, solvent choice in terms of solubility, volatility, density and viscosity are important, temperature, ionic strength, pH and stirring are amongst the most often studied parameters. In addition to this, the effect of the air-bubble has also been introduced in the BID-SDME method.

3.3.2.1 Solvent choice – effect of various solvents on extraction efficiency The choice of solvent in solvent micro-extraction is key since it determines the efficiency of extraction. This is also very important in BID-SDME where density of the solvent is vital in holding the assembly in place at the tip of the needle despite the buoyancy and wicking forces pulling the droplet and the air-bubble upwards along the needle of the syringe.

Since the BID-SDME method has already been described, all the experiments unless otherwise stated were carried out using the 1 µL droplet with 0.5 µL air bubble to form the BID-SDME arrangement. Figure 49 below present the extraction efficiency achieved when employing a simple SDME arrangement for different organic solvents. The extraction time was 20 minutes throughout.

Figure 49 The extraction efficiency of different solvents on extraction

108 The overall extraction efficiency is similar for dichloroethane (DCE) and chloroform. Importantly, DDT and its metabolite DDE extract slightly better into DCE than chloroform, which could make it a better solvent given its slightly higher boiling point than chloroform which may render it more conducive to prolonged BID-SDME for improved extraction efficiency. It was found, as part of this study that chloroform could be used up to 20 minutes of extraction time, whereafter the solvent would evaporate somewhat causing growth of the bubble. This led to destabilisation of the BID-SDME set-up and erratic results.

3.3.2.2 The effect of increasing ionic strength

Ionic strength is an important phenomenon in extraction due to potential salting out effects of organic compounds from the aqueous solution. Water molecules solvate the ions which adds more structure to the solution causing it to become more polar, rendering organic solutes less soluble. Figure 50 below illustrates the results obtained in response to increases in ionic strength (0 – 10% w/v NaCl content) resulting from use of sodium chloride, which is commonly used for ionic strength experiments.

Figure 50 Effect of sodium chloride content on extraction efficiency using 1 µL of DCE

From Figure 50, the extraction efficiency seems to be negatively affected by the increase in ionic strength. This is contrary to what has been observed earlier in this study as it pertained to the triazines. However, Raposo and Ré-Poppi24 reported the same behaviour for the same compounds using the SPME pre-concentration method. The same reduction in extraction efficiency has been reported in literature, although the explanation given is inconclusive. It is

109 only reported that the dissolved salt changes the physical properties of the Nernst diffusion film thereby lowering extraction efficiency,25,26 but the explanation is limited in its extent and was not fully explored.

3.3.2.3 The effect of pH of the solution on extraction Three solutions were prepared at different pH levels, being acidic, neutral and basic. These solutions were extracted for 20 minutes using the BID SDME extraction method. The aqueous solutions were acidified by the addition of a few drops of formic acid, while in the case of basic medium. A few drops of ammonium hydroxide were added. Due to the limited volume of solution, the actual pH was not determined each time. Instead the corresponding volume of acid or base was calculated, to give pH 4 and 10 for acid and base, respectively.

10 7 4 500

400

300

200

relative relative response 100

0 Lindane Endosulfan A Endosulfan B DDT DDE Aldrin compound Figure 51 The effect of pH on extraction efficiency of organochlorines

The endosulfans seemed to be less extractable in basic medium. This is difficult to explain since both compounds do not show potential acidic sites which may be neutralised by the added base. The other compounds showed little response to the pH, judged by their relative extractabilities, which implies that overall these compounds are adequately extracted across a range of pH values.

110 3.3.2.4 Extraction – time profiles using chloroform and dichloroethane Extraction-time profiles were set-up for the analytes using BID-SDME with both chloroform and 1,2-dichloroethane (DCE). The extent of extraction shows a steady increase up to 30 minutes, and then it flattens beyond this point. The error bars in this figure represent the standard deviation of 3 replicates.

Lindane Endosulfan A Endosulfan B DDT DDE Aldrin

400

300

200 relative response relative

100

0 5 10 15 20 25 30 35 time (min)

Figure 52 Extraction-time profile of the organochlorines using dichloroethane

The extraction-time profile using chloroform as an extracting solvent

Given that both chloroform and DCE gave almost the same extraction efficiency, it was decided to explore the time-profile for chloroform. Figure 53 below shows the results of this experiment.

111 Lindane Endosulfan A Endosulfan B DDT DDE Aldrin

400

300

200

100

0 relative resonse relative 10 15 20 25 30

time (min) Figure 53 The extraction-time profile using chloroform as an extracting solvent

As it has been expected from the boiling point, chloroform plateaued a little earlier than DCE, possibly due to its lower boiling point. Despite the earlier maximum in the case of chloroform, similar overall efficiency was observed. It was decided to use DCE for further experiments due its higher boiling point.

Unlike in the case of SPME, extraction increases to a maximum followed by a dip in efficiency as the droplet dissolves into the aqueous solution or evaporates into the air-bubble. SPME is characterised by a constant efficiency once saturation has been achieved.

3.3.2.5 The effect of stirring on extraction efficiency

The acidified aqueous solution was used in this case to assess the effect of dynamic extraction (with stirring) on efficiency. Earlier in this study, the negative effects of stirring on BID-SDME using chloroform as a solvent were noted. The effect of stirring on BID-SDME using DCE was investigated for extraction time of 20 minutes and the results are shown in Figure 54 below.

112

Figure 54 Effect of stirring on BID-SDME extraction using dichloroethane

As expected, dynamic extraction is overall more efficient than static extraction due to higher mass transfer. However, this was not always the case for all of the compounds as the extraction efficiency of DDE and aldrin decreased to a small extent. It is also reported that stirring prevents formation of a Nernst diffusion layer around the droplet that reduces the diffusion gradient into droplet.27 However, this must be done with care since at higher stirring rate droplet used in the BID-SDME set-up may get dislodged, a phenomenon noted several times here. It is also possible to lose the solvent through increased dissolution/mixing thereof into the aqueous solution.

One of the interesting observations was the loss of solvent during some of the dynamic extractions. In the case of 4 × stirring the volume of the droplet following 20 minutes extractions was between 1 and 1.2 µL as opposed to the initial 1.5 µL (droplet and bubble together) used at the beginning of the extraction. This observation, together with the dislodgment of the droplet, indicated away from stirring, or at least towards its minimal application.

To investigate if this behaviour would still hold with the simple SDME and in order to evaluate the efficiency thereof, simple SDME was carried out using the same conditions as with the BID-SDME above. The results are presented in Figure 55 below.

113

Figure 55 Effect of stirring on extraction efficiency using simple SDME

From the results above it is evident that stirring significantly affects simple SDME more than the BID-SDME approach, since there is a notable increase in extraction efficiency with stirring in the case of simple SDME as opposed to the slightest effect observed with the BID-SDME. The observed drop with higher stirring rate could be explained in terms of loss of solvent due to increased energy leading to dissolution of the droplet or some wicking. The latter was evidenced by rotation of the droplet in the direction of the stirring. This would affect effective sucking of the solvent into the syringe for injection. However, the system was quite stable for milder stirring rates.

3.3.2.6 The effect of temperature on extraction using dichloroethane Increased temperature, just like stirring, provides higher kinetic energy and as a result increases extraction efficiency. However, this comes at a cost in respect of water solubility, reduced viscosity and volatility of the solvent, all of which contribute to solvent loss.

114 Lindane Endosulfan A Endosulfan B DDT DDE Aldrin 600

500

400

300

200 relative relative response 100

0 20 25 30 35 temperature (°C)

Figure 56 The effect of temperature on extraction efficiency of organochlorines

From the results in Figure 56 it is evident that increasing the temperature does not result in much improvement in the extraction efficiency. However, the precision is compromised as evidenced by the larger error bars depicting standard deviations in the figure. Therefore, it was decided to use room temperature for further work given the good compromise between extraction efficiency and precision.

3.3.2.7 Optimised condition and enrichment factors The optimised conditions were used to determine the enrichment factors for each of the analytes. The optimum conditions thus far determined are summarised as follows: 1 µL of dichloroethane (DCE) with 0.5 µL of air-bubble, stirring rate of 2 × stirring (mild stirring), neutral to slightly acidic without any addition of salt, temperature 25 °C and an extraction time of 25 minutes.

The method shows good enrichment factors of the original concentrations ranging from about 1600% (16 × enrichment) to about 5000% (50 × enrichment). These values are satisfactory considering fact that SDME is equilibrium based and moreover the sampling volume is also very limited. Importantly, the enrichment factors for DDT and DDE are the highest (average of around 5000%). This was particularly useful because this method is aimed at determination of these compounds from the environmental matrices.

115 Table 19 The enrichment factors for the organochlorines using the optimum extraction conditions

Lindane Endosulfan A Endosulfan B DDE Aldrin DDT 41.80 9.79 11.07 5.70 Standard (3.3)a (13.8) (12.2) 11.74 (9.5) (4.1) 3.58 (10.8) 906.55 104.26 137.47 407.44 110.41 111.54 Simple SDME (8.7) (12.8) (10.6) (4.5) (11.9) (8.3) 1257.51 157.04 196.52 587.66 197.30 177.34 BID-SDME (3.5) (7.2) (4.5) (9.9) (11.4) (3.1)

EFb (SDME)c 2169 1065 1242 3469 1938 3116 EF (BID-SDME)d 3008 1604 1775 5004 3464 4954 BEFe 1.39 1.51 1.43 1.44 1.79 1.59 a the values in parentheses represent the %RSD from n = 3 b EF represents extraction factor c ratio of SDME extraction to the original standard d ratio of BID-SDME extraction to the original standard e the ratio of BID-SDME to simple SDME (bubble enrichment factor)

Another important point to stress is the fact that, consistent with the earlier chapter, the bubble enrichment factor averaged about 1.5 times the extraction without the air-bubble. This cements the point that the enhancement noted earlier (with the triazines) with the air- bubble do not necessarily depend on the nature of compounds.

3.3.3 Validation of the method The method was validated for accuracy, reproducibility, limits of detection and quantitation and the results thereof are presented in the next sections. Validation is important since it pits the new method against tried and tested techniques.

3.3.3.1 Linearity, limits of detection and quantification The linearity of the method was evaluated in the range of 0.05 to 25.0 ng/mL. Figure 57 below shows the relative response versus concentration of the solutions being extracted. The standard curves shown in Figure 57 below show sufficient linearity in the range used.

116 Endosulafan A Endosulfan B DDE Aldrin DDT 250

200

150

100

relative response relative 50

0 0 5 10 15 20 25 concentration (ng/mL)

Figure 57 Standard curves for the organochlorine pesticides following BID-SDME extraction

The analytical parameters resulting from these curves are presented in Table 20 below. The standard error of the intercept was used to calculate the LOD of the individual compounds as a three times ratio of the slope (LOD = 3 × SEintercept/slope).

Table 20 Analytical parameters obtained from the calibration curves of the organochlorines mixture Lindane Endosulfan A Endosulfan B DDT DDE Aldrin R2 0.9998 0.9991 0.9999 0.9998 0.9996 0.9996 slope 36.979 6.1108 6.2363 3.8967 7.7056 4.2867 intercept 29.2431 4.6677 3.23 2.0794 7.6411 2.7702 Standard error of calibration 0.4546 0.1048 0.0313 0.027 0.0734 0.0459 Standard error of intercept 6.2284 1.2847 0.3497 0.3018 0.8215 0.5629 LODa (ng/mL) 0.5623 0.6307 0.1682 0.2324 0.3198 0.394 Prescribed LODb (ng/mL) 0.025 0.015 0.024 0.01 0.075 0.06 LOQc (ng/mL) 1.87 2.10 0.56 0.77 1.07 1.31 a LOD calculated as 3 × standard error of calibration/slope b Prescribed by the US-EPA method 508 for organochlorine pesticides c LOQ calculated as 10 × standard error of calibration/slope

Interestingly, the obtained LODs are higher than the lowest standard used in the standard curves. This suggests that the lowest standard might have fallen out the linearity range of the

117 method where the signal was not very reliable. However, the linearity is quite sufficient with the average R2 = 0.9995.

These LOD values are one order of magnitude higher than those prescribed by the EPA method 508. This could be attributed to two principle differences: the extraction techniques and the detector. In the EPA method, the extraction was carried out using the SPE technique which is exhaustive as opposed to the single-drop method which is equilibrium based. The prescribed detector for organochlorines and other halogenated compounds is the ECD detector due to its sensitivity and specificity for this type of compounds. For MS-based detection, negative electron capture and negative chemical ionisation are useful for organochlorines as stated in the introductory section. However, these facilities were not available when the research was conducted. It was decided to nevertheless pursue the present work, despite the lack of sensitivity that would be achievable, making use of EI-MS. This would allow the proof of concept to be generated, namely the application of the BID- SDME method of the present work to organochlorine analytes. A small amount of additional work in the future, making use of ECD or negative chemical ionisation detection would be able to provide improved LOD and sensitivity data.

Despite the positive EI-MS being the most suitable for a wide range of trace analyses, it suffers in the analysis of organochlorines especially when the selected ion monitoring technique is employed. This is due to the abundance of the two isotopes 35Cl and 37Cl. For a compound with only one Cl atom, two M+ ions are observed with the ratio of 1:3 and this leads to a drop in the intensity of the signal. The more the Cl atoms are present, the greater the loss in sensitivity, due to the ever increasing isotopic masses observed. This phenomenon is shown with the two mass spectra below. Figure 58 represents a mass spectrum of aldrin, a compound with several Cl atoms, while Figure 59 represents a monochlorinated compound, namely atrazine.

118 %

100.0 263

75.0 101

50.0 293 298 25.0 103 186 220 152 127 223 250 329 167 269 364 0.0 100 150 200 250 300 350 Figure 58 Mass spectrum of typical multi-organochlorinated compound (aldrin)

% 125.0

100.0 214

75.0

173 50.0 229 25.0 68 132 158 175 71 104 145 91 197 210 240 256 297 0.0 284 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 300.0 Figure 59 Mass spectrum of a single chlorinated compound (propazine)

From the two mass spectra, it can be seen how the complexity grows to the mass spectrum of a multi-chlorinated compound relative to a singly chlorinated compound (atrazine) where the base peak value has only 2 peaks (214 and 215) with relative intensities of 1:3. However, in the case of multi-chlorinated compound (aldrin) the base peak (263) is surrounded on either side by a number of peaks, each representing an identical molecular formula but different isotope ratios. It is no leap of logic to understand why mass spectrometer in the SIM or EIM mode suffers severe sensitivity losses when such compounds are targeted. Eljarrat et al.28 reported higher LOD values (0.53 – 32.09 pg) when using EI-MS compared to the NCI (30 fg – 1.72 pg) from a solution of polybrominated diphenyl ethers, reflecting some of the improvements that may be obtainable to the LOD/LOQ of the present method, once it is applied with NCI detection.

Nevertheless, the data generated for the organochlorines compounds show distinctly the promise for application to such analytes. It is almost certain that the LOD values will be improved if an ECD device is used in place of a mass spectrometer.

119 Farahani et al.29 reported competitive LOD (7 - 19 ng/L) determined as 3 times S/N ratio to the EPA method in their work using the solidified organic droplet and GC-FID. However, most literature reports LOD determination using S/N ratio other than statistically calculated values. For example, Fernandez-Alvarez et al.30 obtained LOD below 0.02 ng/mL except for DDT that was at 0.19 ng/mL; Huang and Huang31 obtained LODs in the range 0.0361 to 0.059 for OCPs except for DDE which was at 0.164 ng/mL using hollow fibre liquid-phase micro-extraction; Li et al.32 obtained much lower LODs (0.5–10ng/L) using thiophene coated SPME.

Ribani et al.33 demonstrated that different methods of calculating LODs result in different values. The S/N approach yields much lower values but can suffer from being subjective. In the other two approaches, namely using 3 × ratio of either standard error of intercept or a regression to slope of calibration, a greater degree of objectivity is built in, improving the confidence levels. Vial and Jardy34 stated that the LOD obtained from the regression line is better since there is less subjectivity in its determination. This further illustrates the importance of engagements of the scientific community in determining the most suitable approach to reduce the subjectivity of the reported results. Table 21 below demonstrates the standard S/N ratio of the responses from the 0.1 ng/mL standard (n = 5) auto-generated by the GC-MS solution incorporated in the GC-MS software (GCMS solution®).

Table 21 S/N ratios for 1 ng/mL organochlorines standard mixture with the corresponding relative response

Lindane Endosulfan A Endosulfan B DDT DDE Aldrin Average reading 35837 5917 5683 7267 16078 3958 Average S/N 54.5 19.2 9.3 72.8 80.3 27.6

It is clear from the table that if the LODs were determined by the S/N ratio approach they would be much lower than those reported in Table 21 above. The corresponding chromatograms are presented in Figure 60 below.

120

Figure 60 Individual chromatograms for a 1 ng/mL standard post extraction with BID-SDME

Visual inspection would have resulted in a much higher S/N ratio for endosulfan B than the value recorded. The lower value has resulted due to the ever increasing baseline around that peak, which gives an artificially low S/N ratio. Ribani et al.33 used the baseline of the blank solution for calculation of the noise level when using the S/N ratio method to calculate the LOD values. That approach has its own flaws, the main one being the fact that each injection is accompanied by some degree of error. Moreover, this approach carries relatively higher level of operator intervention as opposed to using the software auto-generated S/N ratios.

3.3.3.2 Accuracy validation The accuracy of the method was validated using the certified reference mixture in water, containing DDT (6.78 ± 0.744 ng/mL), DDE (5.21 ± 0.538 ng/mL) and aldrin (8.78 ± 0.799 ng/mL) together with DDD, dieldrin, heptachlor and heptachlor epoxide that were not contained in the PM 625 JM mixture used for the method development. Below is the total ion chromatogram of the CRM mixture, showing that by and large there is good peak separation.

121 (x10,000,000) TIC 1.00

0.75

0.50

0.25

0.00 8.75 9.00 9.25 9.50 9.75 10.00 10.25 10.50 Figure 61 A TIC chromatogram of the CRM mixture

Below is the view of the same chromatogram with only target m/z values. The small peaks at 8.9 minutes (heptachlor), 9.35 minutes (heptachlor epoxide) and 9.9 minutes (dieldrin) represent the named compounds whose mass spectra are attached in the appendix (last page before references in this report).

(x100,000) 7.5 181.00 (1.00) 195.00 (1.00) 246 235 5.0 263

2.5

0.0 8.50 8.75 9.00 9.25 9.50 9.75 10.00 10.25 10.50 Figure 62 An extracted ions chromatogram of the CRM mixture

Figure 63 Mass spectra of DDD and DDT from the NIST library (version 07)

122 % 125.0

100.0 235

75.0 165 50.0

25.0 199 75 88 176 63 136 248 281 354325 429 0.0 50 100 150 200 250 300 350 400 Figure 64 mass spectrum of DDT

100.0 235

75.0

50.0 165

25.0 199 75 88 178 212 63 101123 136 151 248 320 0.0 264 283 341 50 100 150 200 250 300 350 Figure 65 Mass spectrum of DDD

The accuracy of the method was assessed using the certified reference standard containing aldrin (8.78 ± 0.799 ng/mL), DDE (5.21 ± 0.538 ng/mL) and DDT (6.78 ± 0.744 ng/mL) in addition to other compounds (DDD, heptachlor and heptachlor epoxide and dieldrin) that were not present in the standard PM 625 JM mixture used in this part of the project. Table 22 below illustrates the obtained values from the extraction together with the corresponding concentrations calculated from the standard curves.

123 Table 22 Calculated concentrations from the standard curves compared to the CRM Relative responses Calculated concentrations (ng/mL) DDE Aldrin DDT DDE Aldrin DDT Run 1 47.7 41.3 30.9 5.204 8.997 7.071 Run 2 47.8 43 32.3 5.216 9.379 7.066 Run 3 43.5 38.6 31.9 4.660 8.359 7.146 Run 4 46.8 42.2 31.5 5.078 9.202 7.561 Run 5 46.7 40.9 32.9 5.064 8.903 7.422 Run 6 47.6 40.2 30.2 5.192 8.740 7.207 Run 7 47.2 40.1 27.1 5.132 8.718 6.021 Run8 47.5 36.7 29.5 5.176 7.911 7.083 Run 9 52.6 36 30.7 5.834 7.759 7.354 Average 46.6 39.9 29.7 5.173 8.663 7.103 SD 1.5 1.4 2.6 0.301 0.554 0.442

Certified concentration of CRM (ng/mL) 5.21 ± 0.538 8.78 ± 0.799 6.78 ± 0.744 Calculated concentration of CRM (ng/mL) 5.17 ± 0.231 8.66 ± 0.425 7.10 ± 0.339 t-value 0.667 0.650 2.123

The confidence interval was calculated as CI = t × s/√(n) where t denotes the statistical t- parameter and s denotes standard deviation whiles n denotes the number of replicates.

The method shows good accuracy with the calculated concentrations within the limits of the CRM concentrations for the three standards. The comparison using the student t-test is given below using the equation

t = ( x - µ)√n/s where represents the average concentration determined based on the method, µ denotes the CRM value, n denotes the number of replicates, and s denotes standard deviation. The student t-test supports the results obtained with the direct comparison since the value of tcric,

0.95 = 2.306 which is higher than the t-values from the experimental determination.

3.3.3.3 Reproducibility and repeatability validation Reproducibility of the method is normally assessed over a series of days to evaluate the variation between days. This is some kind of extension to the repeatability which looks at the

124 variation between individual measurements for a given set of conditions. The method was validated for both reproducibility and repeatability. However, it must be emphasised that repeatability has already been demonstrated by the results already demonstrated in the earlier sections relating to triazines since all the measurements were performed using the minimum of three replicates where the %RSD was higher 10%. The results in Figure 66 and Table 23 represent the data obtained from extractions performed over four days using three different solutions each day. Since each of the solutions was extracted n = 3 times the total number of replicates is 12 × 3 = 36 replicates.

Figure 66 Reproducibility of the extraction method over 5 days using 5 different samples The letter “S” in the figure legend represents the sample number

From the results, the method clearly demonstrates sufficient reproducibility as well as repeatability. Most of the published literature does not report reproducibility. There are therefore no data to compare the results. However, it has been shown that the method demonstrates sufficient reproducibility with small deviations that are generally within %RSD of 10%, except only a few.

125 Table 23 Relative responses obtained during reproducibility studies of the organochlorines extraction over four days using three different solutions

Lindane Endosulfan A Endosulfan B DDE Aldrin DDT Day 1 S1a 151.8 (4.2)b 21.7 (8.7) 23.0 (4.8) 145.0 (11.1) 37.4 (10.4) 41.6 (4.1) Day 1 S2 152.4 (6.1) 20.9 (4.0) 24.6 (9.0) 118.7 (6.3) 35.2 (11.4) 42.5 (3.0) Day 1 S3 140.2 (9.2) 19.1 (13.8) 23.4 (1.80) 119.3 (2.0) 35.3 (2.5) 43.9 (3.0) Day 2 S1 160.1 (6.4) 22.9 (7.9) 24.3 (4.4) 139.0 (6.3) 39.4 (6.9) 43.9 (4.8) Day 2 S2 159.6 (7.8) 21.9 (8.1) 25.7 (5.6) 124.3 (2.0) 36.9 (1.9) 44.5 (7.7) Day 2 S3 142.1 (5.1) 21.1 (12.8) 23.7 (4.4) 120.9 (7.6) 35.8 (8.6) 39.4 (5.0) Day 3 S1 153.6 (8.6) 22.0 (5.3) 23.3 (13.7) 127.6 (4.1) 37.8 (0.8) 42.1 (3.7) Day 3 S2 172.3 (8.4) 20.6 (9.8) 27.8 (6.9) 129.1 (4.5) 39.8 (2.7) 48.0 (4.7) Day 3 S3 143.5 (7.1) 19.6 (7.3) 23.9 (5.7) 122.0 (3.5) 36.1 (2.1) 39.8 (3.6) Day 4 S1 145.6 (3.7) 20.8 (0.6) 22.1 (8.5) 135.5 (8.2) 35.9 (6.3) 39.9 (8.9) Day 4 S2 153.2 (1.0) 21.0 (0.7) 24.7 (2.7) 123.3 (10.2) 35.6 (3.9) 42.7 (5.1) Day 4 S3 156.2 (4.7) 21.3 (11.7) 26.0 (1.8) 132.8 (6.0) 39.3 (2.5) 43.3 (3.3) Average 152.5 (5.9) 21.1 (4.9) 24.4 (6.4) 128.1 (7.7) 37.0 (6.6) 42.6 (4.6) a The letter “S” represents the sample number b Values in parentheses represent the %RSD values for n = 3

3.3.4 Application of the method to environmental samples

A sample of water was collected from a reservoir (dam) in Limpopo where DDT is still sprayed for the control of vector mosquitoes responsible for transfection of malaria parasites. The other water samples were collected from the North West Province. Both water samples were donated by the Department of Zoology of the University of Johannesburg where they were going to be used for the other types of pollution. The author is grateful for the kind donation of the samples.

A map of the area in Limpopo Province where the sample was collected is attached below.

126

Figure 67 A satellite picture of the place from which the dam water was collected (Downloaded from googlemaps.com 05 September 2011)

Recovery is one of the important parameters that are often overlooked whenever an application of new method is assessed for water. To assess the recovery of the analytes from 2 the water sample, the deuterium labelled DDT standard ( H8) was used and spiked into the water. Thereafter it was extracted and compared against the extraction of correspondingly 2 spiked HPLC grade water sample. Figure 68 shows the chromatogram of the H8-DDT and the 2 corresponding mass spectra. These data evidence what may be two isomers of the H8-DDT, but what may also be a sample containing DDT and DDD. It will be noted that the mass spectra relating to the two peaks are similar but not identical (See Figure 69), pointing to two different compounds.

(x1,000,000) TIC 1.5 243.00 (5.00)

1.0

0.5

0.0

9.75 10.00 10.25 10.50 10.75 2 Figure 68 A TIC (black) together with an EIM chromatogram (pink) with m/z 243, the base peak for H8-DDT

127 The mass spectra of these two peaks are presented in Figure 69 below.

% %

100 243 100 243

50 173 129 50 173 104 112 174 206 101 220 109 140 256 328 183140 255 290 362327 0 293 0 100 200 300 100 200 300 2 Figure 69 The mass spectra of the two peaks from H8-DDT standard (100 µg/mL)

According to the mass spectra and taking the normally expected elution pattern of DDD and DDT, it seems like the first peak is DDD due to the presence of the m/z values 328, 243 and 173 which correspond to +8Da with 320, 325 and 165, respectively, the m/z values expected for DDD. The second peak is likely to be DDT since the m/z value of 362 corresponds to the M+ 2 peak for H8-DDT, i.e. m/z 354 for H-DDT plus 8Da.

% %

100 243 100 243

50 50 173 129 173 112101 341 206 327 207 281 143104 186 256 140 174 254 0 192 0 100 200 300 100 200 300 2 Figure 70 Mass spectra of the two H8-DDT peaks after extraction from spiked water

(x1,000,000) (x1,000,000) 4.0 5.0TIC TIC 243.00 (25.00) 243.00 (25.00) 4.0 3.0

3.0 2.0

2.0 1.0

1.0 0.0

0.0 9.75 10.00 10.25 5.0 7.5 10.0 12.5

Figure 71 The chromatogram of the water extract from Limpopo dam

128 2 The water samples from different sources were spiked with the H8-DDT and extracted using the BID-SDME method with DCE as the solvent, and subsequently analysed using GC-MS. Table 24 below demonstrates the extraction efficiency of the DDT from the water samples.

2 Table 24 Recoveries of from dam water from Limpopo using H8-DDT standard

Reference water Limpopo 1 Limpopo 2 Dam 1a Hartebeespoort

2 DDT- H8 64.0 57.0 60.2 63.1 59.4 (89)b (94) (99) (93) a Dam close to where the soils were collected for atrazine analysis b Values in parentheses show recovery (%) relative to reference solutions prepared using HPLC grade water.

These results demonstrate that there is negligible matrix effect since the recoveries are almost 100% relative to the reference solution.

Extraction of the water samples did not show the presence of DDT or any of its breakdown compounds. There was no prior knowledge of either presence or absence of the DDT and its breakdown products in the water samples except that the water was collected in a suspected DDT use area in the case of the Limpopo samples. It is possible that there was DDT present in these samples but at levels below the detection limit of the GC-MS. Future efforts will include the use of the ECD detector as well as ECNI-MS and NCI-MS to improve the detection limits.

3.4 General discussions The BID-SDME method has been satisfactorily applied to the analysis of organochlorinated pesticides. The flexibility of the method has been demonstrated by changing from chloroform to DCE as the extracting solvent. This study further demonstrated the importance of a balance between the optimised conditions as they are not always mutually beneficial when combined. For example, despite dynamic extraction being efficient, it cannot be combined increased temperature and vice versa. However, the optimised method shows good linearity and accuracy for the determination for DDT and DDE as a breakdown product, as well as of aldrin.

Despite the non-detectability of DDT and its breakdown products from water samples from a dam in Limpopo where the use is suspected, the efficacy of the method cannot be disputed

129 owing to the accuracy demonstrated by the accuracy validation using statistical t-test, falling within the limits of accuracy for the 95% confidence level.

3.5 Conclusions and recommendations

The BID-SDME method for organochlorine pesticides has been successfully developed and validated for analysis of these pesticides in water samples. It demonstrated sufficient linearity, reproducibility, repeatability and accuracy. The only thing that is a bit of a worry is the slightly higher detection limits to those prescribed and obtained by the EPA method 508. This limitation is attributed to the use of a positive EI-MS detector instead of a negative EI or NCI-MS or even an ECD which are better suited for organochlorines since the EI-MS has a dilution effect due to abundance of the isotopes as discussed in the preceding section. It is likely that much-improved limits of detection will be obtainable once an appropriate detector is becomes available.

Importantly, BID-SDME showed consistent bubble enrichment factors over simple SDME, the improvement averaging about 1.5 times.

The method has been applied to dam water samples water with good recoveries being noted throughout. While DDT was not detected in any of the water samples, it remains possible that DDT and/or its breakdown products are present at levels below the detection limits of the current method but still at levels that would raise concern. Once the appropriate facilities become available, application of this work thereto should clarify the matter with confidence.

130 3.6 References

1. Russell, E.P. Technology and Culture, 1999, 40, 770-796 2. Roberts, D. R.; Laughlin, L. L.; Hsheih, P.; Legter, L. J. Perspectives, 1997, 3, 295-302 3. Noakes, D. N.; Benfield, C. A. Journal of the Science of Food and Agriculture, 1965, 16, 693– 697 4. Toft, G.; Hagmar, L.; Giwercman, A.; Bonde, J. P. Reproductive Toxicology, 2004, 19, 5–26 5. Stockholm Convention on Persistent Organic Pollutants (POPs), 2001, Stolkholm Switzerland 6. Global Malaria Programme, World Health Organization, 2007, Geneva, Switzerland 7. Ritter, R.; Scheringer, M.; MacLeod, M.; Hungerbühler; K. Environmental Health Perspectives 119 (2011) 707-712 8. Wong, M.H.; Leung, A.O.W.; Chan, J.K.Y.; Choi, M.P.K. Chemosphere, 2005, 60, 740-752 9. Vaclavik, E.; Tjonneland, A.; Stripp, C.; Overvad, K.; Weber, J. P.; Raaschou-Nielsen, O. Environmental Research, 2006, 100, 362–370 10. Omori-Inoue, M.; Fukata, H.; Komiyama, M.; Todaka, E.; Osada, H.; Aburatani, H.; Mori, C. Reproductive Toxicology, 2007, 23, 283–289 11. Siddiqui, M.K.J.; Anand, M.; Mehrotra, P.K.; Sarangi, R.; Mathur, N. Environmental Research, 2005, 98, 250–257 12. Wells, M.; Leonard, L. GroundWork, The International POPs Elimination Project (IPEP), 2006, South Africa 13. Kapp, C. The Lancet, 2004, 364, 1113-1114 14. African Network for Vector Resistance (ANVR) and WHO Regional Office for Africa, 2005, Harare 15. Genthe, B. Science Scope, CSIR, 2009, South Africa 16. Scott, R.P.W. Chrom-Ed Series, 2003, Book 4 17. Bedendo, G.C.; Carasek, E. Journal of Chromatography A, 2010, 1217, 7-13 18. La Guardia, M.J.; Hale, R.; Harvey, E.; Chen, D. Environmental Science and Technology 2010, 44, 4658–4664 19. Lacorte, S.; Guillamon, M. Chemosphere, 2008, 73, 70-75 20. Rothweiller, B.; Berset, J.-D. Chemosphere, 1999, 38, 1517-1532 21. Jain, A.; Verma, K.K. Analytica Chimica Acta, 2011, 706, 37-65 22. Picό, Y.; Fernandez, M.; Ruiz, M.J.; Font, G. Journal of Biochemical and Biophysical Methods, 2007, 70, 117-131 23. Zhang, M.; Huang, J.; Wei, C.; Yu, B.; Yang, X.; Chen, X. Talanta, 2008, 74, 599–604 24. Raposo Júnior, J. L.; Ré-Poppi, N. Talanta, 2007, 72, 1833-1841 25. Bagheri, H.; Khalilian, F. Analytica Chimica Acta, 2005, 537, 81-87 26. Lin, C.-Y.; Huang, S.-D. Journal of Chromatography A 2008, 1193, 79-84 27. Jeannot, M. A.; Cantwell, F. F. Analytical Chemistry, 1997, 69, 235-239 28. Eljarrat, E.; Lacorte, S.; Barcelό, D. Journal of Mass Spectrometry, 2001, 17, 76-84

131

29. Farahani, H.; Yamini, Y.; Shariati, S.; Khalili-Zanjani, M.R.; Mansour-Baghahi, S. Analytica Chimica Acta, 2008, 626, 166-173 30. Fernandez-Alvarez, M.; Llompart, M.; Lamas, J.P.; Lores, M.; Garcia-Jares, C.; Cela, R.; Dagnac, T. Journal of Chromatography. A, 2008, 1188, 154-63 31. Huang, S.-P.; Huang, S.-D. Journal of Chromatography. A, 2006, 1135, 6-11 32. Li, X.; Li, C.; Chen, J.; Li, C.; Sun, C. Journal of Chromatography. A, 2008, 1198-1199, 7-13 33. Ribani, M.; Collins, C.H.; Bottoli, C.B.G. Journal of Chromatography. A, 2007, 1156, 201-5 34. Vial, J.; Jardy, A. Analytical Chemistry, 1999, 71, 2672-2677

132 Chapter 4: Development and application of mixed-solvent BID-SDME for determination of some growth hormones from bovine urine

4.1 Introduction

Hormones are generally defined as chemical messengers transporting signals between and among cells in multi-cellular organisms. Their range of message transporting extends from cell to cell, cell to tissue/organ, to the extreme being from one organism to another without the two organisms being in physical contact.1 Although hormones are naturally occurring, some are synthetic and are used in a variety of applications, for example, in medical and pharmaceutics,2,3 pest control,4 and in animal husbandry,5 just to mention a few. Such hormones may function to replace or complement those of the natural analogues. Despite their benefits, hormones also possess ill-effects such as being carcinogenic and sometimes causing over-expression of the desired effects, for example, obesity, impotence and sterility which are unhealthy despite the intent to obtain positive results.6,7

A review covering the use, regulation and monitoring of hormones in edible matrices has been published by Noppe et al.8 The uses mentioned in this review are those with good intentions in clinical applications. Examples include growth hormones which may be prescribed to children with poor growth records,9 and in animal husbandry10 for fattening and improving fertility, amongst others. The history of the use/abuse of hormones other than for the functions intended includes leisure as in body building and performance enhancement in sports.11,12

Common matrices used in hormones and drugs studies include saliva, urine, blood plasma,13 hair,14,15 sweat16 and faeces.17 Other interesting biological matrices reported include human breath18,19 and skin secretions.20 Analysis of animal tissue and urine sometimes requires enzymatic hydrolysis prior to hormonal analysis.21,22 Analysis for hormones in urine is seen as a non-invasive approach since the animal is not stressed as may be the case when blood is drawn for analysis.23 Germano et al.24 used ELISA to analyse sex hormones from frogs’ urine to assess their reproduction. Akre et al. analysed 17β-estradiol in bovine urine using OASIS™ HLB SPE cartridge with LOD of 170 pg/mL.25 The importance of screening for these

133 hormones before the meat products are sent to the markets is important to avoid loss of income by the farmers following slaughtering of the suspected animals; as such urine testing may be more important than analysis of meat products.

Sándor Görög26 has reviewed the analysis of steroid hormone drugs in papers published between 2004 and 2010, including in matrices from pharmaceutical products and samples from the environment. Amongst the chromatographic methods predominantly featured in this review are HPLC which is used as a method of choice by the US pharmacopeia with UV and MS detectors, thin layer chromatography coupled to densitometry, gas chromatography, micellar electrokinetic chromatography and microemulsion electrokinetic chromatography, ELISA. Capillary electrophoresis has been reported to be rarely used due to the neutral nature of most hormones. Izumu et al.27 developed a quantitative analysis method for plant hormones (auxins, cytokinins, abscisic acid and gibberellins) from tobacco dry seeds using nanoflow liquid chromatography–electrospray ionisation-ion trap mass spectrometry. ELISA has also been reported in determination of hexestrol from spiked pork and liver tissue samples, and to compare with liquid chromatography and tandem mass spectrometry, in which an internal standard was used.28

Sample preparation techniques reported for hormonal analysis using chromatography include a column switching system with restricted access pre-column packing for coupled integrated sample cleanup for liquid chromatography. Here, C4, C8 and C18 pre-column cartridges were used for the analysis of alkylphenolic compounds and steroid sex hormones from sediments at ng/mL levels29 along with solid-phase extraction using different commercially available phases and molecularly imprinted polymers, etc.30 Kataoka has reviewed developments and applications of micro-extraction techniques in drug analysis.31 However, there are quite few articles where solvent micro-extraction has been applied either for hormones or biological matrices which are not necessarily captured in these reviews. Vidal et al. 32 reported the use of single-drop micro-extraction of free benzophenone-3 in human urine samples based on an ionic liquid while Xie et al.33 analysed medroxyprogesterone in water samples using dispersive liquid-liquid micro-extraction.

134 In the present study, the development and application of the bubble-in-drop single-drop micro-extraction (BID-SDME) developed in this laboratory34 is reported for the determination of growth hormones from bovine urine. As stated above there are a few publications using solvent micro-extraction techniques in this type of application. Hexestrol and diethylstilbestrol, representatives of the group of stilbene estrogens, have been used as model hormones in the current study.

4.2 Experimental 4.2.1 Chemicals and standards used All solvents used in this study were of HPLC grade or higher: methanol, toluene, chloroform, dodecane, 2-octanol, xylene, butylacetate and water were obtained from Riedel-de Haën (Seelze, Germany); 4,4’-dihroxybiphenyl, formic acid, ammonia, sodium chloride AR grade were obtained from Sigma-Aldrich (Seelze, Germany). The hormone standards, hexestrol (4,4'-(1,2-Diethylethylene)diphenol -CAS 84-16-2) and diethylstilbestrol (4,4'-hex-3-ene-3,4- diyldiphenol- CAS 56-53-1) and the CRM (diethylstilbestrol in lyophilised bovine urine) were obtained from Dr Ehrenstorfer GmbH (Augsburg, Germany). The bovine urine sample was obtained from the Agricultural Research Council in Pretoria, South Africa.

Standard solutions of the two hormones were prepared by dissolving 0.0101g of diethylstilbestrol and 0.0103 g hexestrol in HPLC grade MeOH (1 mL) resulting in concentrations of 10 mg/mL of each hormone. This solution was diluted to produce 100 µg/mL and both solutions were kept in a freezer as stock solutions. Working solutions were prepared as required by serial dilution of the stock solutions with methanol kept in a fridge when not in use. The appropriate mass of sodium chloride was weighed and dissolved in the aqueous solution post spiking and equilibration to assess the effect of ionic strength.

4.2.2. Apparatus A multi-vial compartmental water bath with magnetic stirrer facility (Pierce, Rockford Illinois USA) was used for thermal equilibrium; a centrifuge (Eppendorf 5415D, Hamburg, Germany) was applied for settling the particulate matter; a Zx3 vortexer (Velp Scientifica, Italy) was used for mixing; an ultrasonic bath (FungiLab, Barcelona Spain) for the solubilisation of samples; a pH meter (Hanna Instruments, Romania) was used for measuring the pH of

135 samples; and a 10 µL graduated Hamilton GC syringe (Seelze, Germany) was employed for extractions and injection into the gas chromatograph for analysis.

4.2.3 Chromatography and mass spectrometry instrumentation Analyses were carried out using a Shimadzu 17A gas chromatograph (GC-FID) for the preliminary experiment at µg/mL (ppm) level, and a Shimadzu QP2010 gas chromatograph coupled to mass spectrometry (GC-MS) (Kyoto, Japan) for lower concentration levels (ng/mL). Both of these instruments were equipped with a Zebron 35ms column with 30 m × 0.25 mm × 0.25 μm dimensions. The carrier gases used were nitrogen and helium (99.999%, Afrox, South Africa), respectively, passing through the GC column at a constant column flow rate of 1 mL/min. A split-splitless injector operated on splitless for 2 minutes followed by split ratio of 1:10 was set at 260 °C. The oven temperature was set at 150 °C (4 min) then ramped at 25 °C/min to 280 °C (8.8 min) total time 18 minutes. The FID was set at 350 °C

The MS settings included EI 70 eV with 1.5 kV detector voltage with the ion source temperature of 200 °C, the interface temperature at 240 °C. Acquisition was made between 6 and 18 minutes using scan mode in the range 50 – 350. For SIM analyses, the MS was set on m/z values 135 (107), 186 and 268 (239) for HEX, DHB and DES respectively. The values in parentheses represent the qualifying.

4.2.4 Micro-extraction procedure The set-up for micro-extraction is reported in detail in the literature and in the preceding chapters of this thesis. 1 μL of the extracting solvent was drawn into the syringe and transferred to the vial containing the aqueous solution where it was introduced into the solution in a form of a droplet hanging on the tip of the syringe. After the equilibration time had elapsed, the droplet was sucked back into the syringe and injected into the GC for analysis whether by flame ionisation or mass spectrometry detection, depending on the concentration range employed.

For the BID-SDME, following the sucking of the appropriate volume, a further volume of air is drawn in equivalent to the required volume, thereafter the entire contents are gently introduced into the aqueous solution through depression of the plunger and the air volume

136 forming the air bubble droplet terms BID-SDME. Following the equilibration, the entire volume was retracted back into the syringe carefully to avoid loss of the droplet through wicking, and subsequently injected into the GC for analysis.

4,4’-Dihydroxylbiphenyl was used as an internal standard and was dissolved in the extracting solvent before the extraction. The same internal standard was spiked into the organic solutions before the injection to standardise the signal intensities obtained from direct injections of the organic solution before calculating the extraction efficiency.

4.2.5 Optimisation and validation of the extraction method Various parameters amenable to solvent micro-extraction, particularly SDME and more recently BID-SDME, were evaluated. These included: choice of solvent, single and mixed- solvent system, effect of salt (ionic strength), extraction time, the effect of air bubble (BID- SDME vs simple SDME), the effect of temperature and the effect of pH. Some of these parameters were investigated individually or paired to assess their interdependence if any.

Following optimisation, the extraction efficiency was determined as a ratio of the relative intensity of the signal after the extraction to that before the extraction (original intensity), using two solutions (aqueous and organic respectively) at three different concentrations (1, 50 and 100 ng/mL). Matrix-matched samples were prepared by spiking the urine samples with the hormones and evaluating the extraction from these solutions compared to the spiked aqueous solutions.

The linearity of the optimised method was determined using the standards prepared in the urine (matrix matched solutions); thereafter the limits of detection and quantification were determined from the calibration curve. The reproducibility was assessed using four solutions analysed over four days and evaluating their standard deviations to assess their differences. Repeatability was explored for each parameter assessed since all of the extractions were performed a minimum of three times and the standard deviation assessed for all the measurements.

137 Accuracy was evaluated using a certified reference material (diethylstilbestrol in urine – 12.8 ± 2.5 ng/mL). This solution was treated the same way as the other urine samples except that it was not spiked, but was extracted in an identical fashion to the other samples. Individual concentrations were calculated using the calibration data and the average calculated together with the confidence interval. The student t test was used to evaluate the accuracy of the method.

4.2.5 Evaluation of extraction recovery from other biological matrices The recovery of these hormones was evaluated on the other biological samples (blood and milk) collected from the same source (the same cow). These samples were extracted to assess the base levels, and thereafter spiked (10 ng/mL) with the hormones and subjected to the same treatment as the urine samples. The recovery was calculated as a percentage of the extraction in urine, which therefore uses the urine analyses as a reference. The same procedure was done using HPLC grade water to assess the matrix effect arising from urine when HPLC grade water is used as reference matrix.

4.3 Results and discussion 4.3.1 Chromatographic method development A chromatographic separation method was developed for the two standards, hexestrol (HEX) and diethylstilbestrol (DES) whose chemical structures are represented below. Figure 72 shows the chromatogram of the 25 µg/mL mixture of the hormones analysed using the GC-FID.

HO HO OH OH

Hexestrol trans-Diethylstilbestol

138 (x1,000,000) TIC 6.5 6.0 HEX 5.5 5.0 4.5 4.0 3.5 DES 3.0 DNS 2.5 2.0 1.5 1.0 0.5

5 12.00 12.25 12.50 12.75 13 13.25 13.50 13.75 14.00 14.25 14.50 14.75 15.00 15.25

Figure 72 The GC-FID chromatogram of the HEX and DES mixture, showing an additional peak identified as dienestrol (DNS)

The two compounds of interest could not be fully separated as can be seen from the broad elution band between 12.5 and 13.5 minutes, despite much effort. The peak at 14.1 minutes was later identified as dienestrol (DNS) from its mass spectrum and the NIST 05 Library which accompanies the instrument software. When injecting individual samples of the two hormones it was shown that dienestrol is present in both DES and HEX standards. Nevertheless, the main compound in each sample constitutes 80% (DES) and 70% (HEX) and the remainder being constituted by the dienestrol and the cis-isomer of DES (a small shoulder around 12.75 minutes in Figure 72 above).

Despite the presence of this extra component, the DES/HEX mixtures could still be used for the purposes of the present study.

HO OH

cis-Diethylstilbestrol Dienestrol

4.3.2 GC-MS analysis of the hormones

For GC-MS analysis, the individual compounds (1 µg/mL solutions in methanol) were injected into the GC to determine their retention times and the mass spectra so that their characteristic m/z values could be identified for SIM analysis. Having determined the m/z

139 values for SIM, the standards were mixed and the mixture injected producing the extracted ion chromatogram (EIM) shown in Figure 73 below. The figure shows that the compounds could not be separated using the particular stationary phase (5% phenyl, 95% methyl siloxane phase), but this problem was overcome using the SIM for quantitative analysis.

(x1,000,000) 135.00 (1.00) 5.0 268.00 (5.00) HEX 4.0 DES 3.0 2.0 1.0 0.0 11.2 11.4 11.5 11.6 11.7 11.8 11.9 12.0 12.1 12.2 12.3 12.4 12.5

Figure 73 An EIM GC-MS chromatogram for the mixture of HEX (135) and DES (268)

A full scan mass spectrum of the m/z 135 is presented below (Figure 74) with the NIST 05 library hit at 95% for hexestrol. This thus informed the decision to use the m/z 135 in the SIM mode for quantitative analysis and m/z 107 as a qualifying ion resulting in a reference/qualifying ion ratio of 1.667.

Figure 74 Mass spectrum of hexestrol (95% similarity with NIST 05 library)

140 As can be seen from the chromatogram in Figure 73, DES gave two peaks whose mass spectra were virtually identical (Figure 74) with a slight difference in their retention times. The same phenomenon was observed when using with GC-FID. These compounds are probably the cis- and trans- isomers of DES. The observed reference to qualifying ion ratios for (1.961 and 1.963) respectively supports this assignment. The literature from the suppliers indicated the possibility of the cis-isomer being present in the product as supplied, although the specific content was not certified.

Figure 75 Mass spectrum of peaks 1 and 2 (88% and 92% similarities to DES respectively)

4.3.3 The effect of solvent on extraction of hormones

The most important parameter in solvent micro-extraction is the solvent choice. This is governed by mainly three parameters:

 Nature of analytes - this dictates what kind of solvent to use.

 Water miscibility - the biggest dilemma comes when the analytes are highly polar. It may be the case that the most suitable solvent for the analytes is miscible with water, so a compromise has to be reached, usually with the use of a more polar solvent to avoid excessive miscibility which leads to mixing of the solvent with water.

 Volatility – solvent loss is the greatest concern especially with prolonged extraction times. The solvent may evaporate into air bubbles.

141  Surface tension and cohesion forces between the droplet and the surface of the syringe are also important as lower surface tension lessens the attachment of the droplet onto the syringe, while the poorer cohesion forces ease the detachment of the droplet from the remainder of the solvent in the needle of the syringe.

Since the two hormones could not be chromatographically separated, it was decided to develop the method and conditions using only one sample (HEX) in the case of GC-FID, and thereafter to transfer the method to the other sample (DES) at a later stage when the MS was to be used. With MS use, resolution would be achieved using the SIM mode of analysis. Accordingly, a variety of solvents was used with which to extract an aqueous sample of HEX (5 µg/mL). The relative extraction efficiencies are shown in Figure 76.

Figure 76 The effect of solvent of extraction efficiency from aqueous samples

It may be noted from Figure 76 that chloroform and toluene afford the best level of extraction of HEX. It was decided to try the combination of chloroform with toluene to evaluate the efficiency of a combined solvent mixture. This was thought to be important as efforts to introduce the air bubble in the case of toluene were futile (the bubble-in-drop set- up was too buoyant and floated up the needle). Toluene was selected given its aromatic nature (a like substance to the two hormones in question). Surprisingly, the mixed solvent mixture resulted in higher efficiency of extraction than each of the individual solvents (see Figure 77 below).

142

Figure 77 Effect of mixed-solvent composition on extraction efficiency

The use of combined solvents was reported in the 1975 by Medir and Mackay35 who discovered that ester-alcohol mixtures provided better distribution coefficients than either pure solvent. Samaratunga et al. 36 reported the use of between 0.1 and 0.6 mole fraction of t-BuOH in toluene for selective extraction of Cr(IV) from aqueous media in the presence of other divalent ions. At the micro-extraction level, specifically drop-based, mixed solvents have been reported by few researchers: Zhang et al.37 reported a slight increase in efficiency when the organic solvent (p-xylene) was spiked with acetone (20%) in extraction of organochlorines; Qian and He38 reported the use of n-hexan/ethylacetate (2:1 – v/v) for extraction of organochlorines with better efficiency than either solvent. However, explanations for these improvements in efficiency have been scant, with the authors typically simply reporting the observed improvement.

4.3.4 Effect of bubble size on both individual and mixed solvent systems

Figure 78 depicts the effect of solvent composition together with the bubble size on extraction efficiency. It is worth noting that without the air-bubble, 50% chloroform seems to be the best extracting solvent. However, when a bubble is introduced into the droplet, the 75% chloroform mixture becomes the most favourable.

143

Figure 78 The effect of bubble size and solvent composition

Another important aspect that is demonstrated in Figure 78 is that for both 50% and 75% chloroform mixtures an air-bubble size of 1 µL led to a stable bubble-in-drop configuration, whilst in the case of pure chloroform the droplet became unstable and was buoyant. It will be recalled from the work on triazines and organochlorines earlier, that similar observation was made. What is interesting in the present instance is that the chloroform/toluene mixtures tolerated the larger bubble size without suffering this instability although there is a drop in extraction efficiency. This could be due to lower evaporation of the solvent into the droplet due to the presence of the higher boiling toluene in the solvent mixture.

In the case of chloroform, the observed drop in efficiency probably relates to the growth in the bubble size owing to its lower boiling temperature. This evaporation would add extra volume to the introduced air bubble volume and as such the bubble grows beyond the effective volume where it starts to add negative contribution. As discussed in the previous sections, too large bubbles result in reduced efficiency.

In the case of pure toluene, the presence of air bubbles, even as small as 0.5 µL, led to the system becoming unstable, probably due to the low density of the solvent giving rise to buoyancy effects overcoming the adhesion and cohesion forces leading to wicking of the organic droplet. This made it difficult to extract with this solvent and no results could be observed (no entries in the Figure 78 corresponding to this solvent).

144 Another observation is that the introduction of an air-bubble in the cases of 25% and 50% chloroform/toluene mixtures seems to decrease the efficiency of the extraction, contrary to the realised benefit of the BID-SDME method as reported in the earlier chapters. Closer physical inspection of the set-up revealed a possible reason for this problem, namely that the sizes of the bubble in these instances grew with time, also leading to additional buoyancy as described above. In extreme instances, the buoyancy of the bubble was so high that the bubble broke away from the liquid and floated off, resembling the case of pure toluene. This presumably accounts for the lower extraction efficiencies and points towards the fact that denser solvents are favoured for the BID-SDME set-up.

4.3.5 Effect of NaCl on the extraction efficiency using mixed-solvent system The salting out effect is one of the mostly studied phenomena in the extraction of organic compounds from aqueous solutions since it decreases further the solubility of such compounds in water. However, this phenomenon has resulted in conflicting reports where some researchers reported increases in efficiency while others report the opposite.39,40 In the previous part of the study, an increase in the efficiency of extraction with the addition of NaCl was noted for triazines, followed by a decrease after a maximum was reached.

Importantly, addition of salt (10% w/v) was observed to stabilise the BID-SDME arrangement. However, negative effects were noted with the organochlorines when NaCl was added. It became important to assess the influence of the addition of NaCl in this case, given the instability of the bubble-in-drop set-up as noted above, when toluene or its mixtures with chloroform are used as the extracting solvent. Accordingly, various samples were prepared from which the extraction of the hormones was performed (Figure 79).

145

Figure 79 Effect of NaCl content on extraction efficiency using 75% chloroform mixture

Quite surprisingly, the results demonstrate a decrease in extractability with an increase in salt content, similarly to the DDT results. Moreover, urine by its nature already contains salts of various descriptions in varying concentrations. These salts may already be exerting a given influence on the extractability of the hormones via a matrix effect.

4.3.6 The effect of sodium chloride and air-bubble size extraction efficiency It has been established in our earlier report34 that the presence of sodium chloride stabilises the air-bubble thereby assisting with the observed efficiency of the BID-SDME set-up. This observation posed a dilemma in the present instance since increasing concentrations of NaCl led to diminished efficiency of extraction of the hormones, yet the presence of NaCl has been proven desirable to stabilise the droplet. Despite the issues with extraction efficiency noted above, it was nevertheless decided to investigate in a cursory fashion the influence of bubble size on the extraction efficiency when employing 10% NaCl solutions.

146

Figure 80 Effect of NaCl content and bubble size on the extraction of 5 µg/mL hormones solution

Figure 80 above represents the results for the effect of air bubble in salt-less and 10% NaCl aqueous solutions, respectively. Values, 0.0, 0.5 and 1.0 represents various air-bubble volumes in microlitres. Evidently, the presence of sodium chloride does stabilise the BID- SDME, allowing the achievement even of a 1 µL air bubble, thereby supplying additional evidence for the stabilising effect of the NaCl on the BID step-up. Nevertheless, the efficiency of extraction in this instance was lower than that of a 0.5 µL air bubble. In this case, some buoyancy effects were noted, which may account for the lower extraction efficiency. In the absence of NaCl, only a bubble size of 0.5 µL could be reached. Thereafter, the set-up was unstable and could not be employed at all.

4.3.7 Extraction-time profile using different salt and solvent composition The effect of NaCl and BID-SDME over the simple SDME (noted as basic in Figure 81) were evaluated over various extraction times. To carry this out, a series of solutions was prepared and extracted using 50% and 75% chloroform/toluene mixtures resulting in the trends presented in Figure 81 below.

147 50%+NaCl 50%BID+NaCl 50%Basic 50%Basic-BiD 75%Basic 75%basic-Bid 75%Basic+NaCl 75%-Bid-NaCl 250

200

150

100

50 Response(x1000)

0 5 10 15 20 25 30 35 Extraction time (min)

Figure 81 The extraction-time profiles for BID-SDME using different solvent and NaCl compositions BID represents the BID-SDME, basic BID represent the BID-SDME without any sodium chloride.

From Figure 81 above, it can be seen that indeed the presence of sodium chloride reduces extraction efficiency. Comparison of 50% and 75% chloroform compositions, the same results observed earlier are still observed, namely the 75% chloroform in toluene yields higher extraction efficiency. The observed shift in maximum efficiency towards longer times indicates a more stable droplet arrangement even if the actual efficiency is lower than in the other case. Comparison of 75% chloroform simple SDME and BID-SDME demonstrates that the air-bubble improves the mass transfer. Ultimately, though, the same efficiency is realised although at the expense of time in the case of simple SDME without the air bubble. This behaviour is demonstrated better in Figure 82 below.

The efficiency realised with BID-SDME is clearly lost as extraction increases. This could be attributed to the growth of the air-bubble leading to some solvent loss during the retraction of the droplet back into the syringe before injection. This is the case at 30 minutes extraction time, where the extraction without the air-bubble appears to be more efficient than that with an air-bubble. However, this is brought about by losses in the BID-SDME set- up rather than the efficiency of the SDME analogue

148 50%+NaCl 50%BID+NaCl 75%Basic+NaCl 75%-Bid-NaCl 80

40 Response (x1000) 20

0 5 10 15 20 25 30 35 Extraction time (min)

Figure 82 An expanded picture showing the extraction-time profile with 10% w/v sodium chloride

Interestingly, the BID-SDME profiles show a drastic drop-off following the maximum extraction, possibly due to excessive evaporation of the solvent into the already pre-formed air-bubble. Due to increased buoyancy, there is a high tendency of the air-bubble to float to the top of the solution as the droplet is being sucked back into the syringe which compounds the problem and reduces the reliability of the method at that extreme.

4.3.8 The effect of temperature on air bubble and solvent mixture on extraction efficiency The extraction temperature is an important variable in micro extraction as solubility is a function of temperature, as is mass transfer. Too high a temperature could lead to high solvent loss with a concomitant setback in the efficiency. However, lower temperatures could also lead to reduced mass transfer due to poorer kinetic energy. The effect of temperature on the extraction efficiency was assessed for several air-bubble sizes, the results of which are summarised in Figure 83 below.

149

Figure 83 the effect of temperature and air bubbles on extraction efficiency

From Figure 83, it can be observed that temperature decreases efficiency as expected, given that solubility of the solvent increases with temperature, as does its volatility. The results for the effect of air bubbles is consistent with those obtained earlier in that 0.5 µL air bubbles still afford better efficiency. It was hoped that losses in the efficiency of extraction that are attributable to the growth of the air bubble due to evaporation of the droplet solvent, would be reduced by the drop in temperature. The reduced temperature led to some technical difficulties in visualising the droplet due to water condensing on the walls of the beaker which led to inaccuracies that counteracted the desired effects.

4.3.9 Characterisation of the urine for particulate/solutes matter Urine is a solution/suspension of many salts and other compounds being excreted out of the animal body, so it is important to determine the total amount of dissolved/suspended salts in the urine suspension. The amount of solid matter dissolved or suspended in the urine sample was determined by air-drying the urine sample and weighing the residue.

6 x 1 mL samples were poured into the pre-weighed vials and air-dried to a constant mass. The total mass was calculated as a difference between the empty vial and the vial with residue after evaporation. The results are shown in Table 27 below. It can be seen that the average level of dissolved matter in the urine is about 2.5% w/v.

150 Table 25 The amount of particulate matter in the urine following air drying Sample Initial mass (g) Final mass (g) Difference (g) % mass 1 9.9823 10.0065 0.0242 2.42 2 9.8461 9.8704 0.0243 2.43 3 9.9892 10.0135 0.0243 2.43 4 9.9984 10.0234 0.025 2.50 5 9.8978 9.9238 0.026 2.60 6 9.8787 9.9029 0.0242 2.42

4.3.10 Effects of NaCl at ng/mL level using the matrix-matched samples It has been noted that added NaCl led to extraction inefficiency when applying BID-SDME to the extraction of hormones-containing solutions. Given the drop in efficiency with addition of sodium chloride, it was deemed worth exploring the effect of NaCl at lower concentration levels of the hormones given that urine contains significant levels of salts. To achieve this, it was decided to use the actual urine samples instead of using water so that the matrix is matched appropriately.

The reasoning behind this move is that it was possible that the added NaCl was causing a salting out effect with the very non-polar compounds, the effect of which would be reduced at lower analyte concentrations. It was thought that the salting out may lead to deposition of the organic non-polar analyte onto the surface of the glass vial, rendering it inaccessible for extraction.

Urine solutions were diluted 1/10 times and spiked to 50 ng/mL with a 1 µg/mL of the hormones mixture (DES and HEX) as opposed to the single component standard used in the prior work using flame ionisation detector. One solution was used as a blank to determine the presence of the hormones in the urine. The extraction was carried out using the 75% chloroform-toluene mixture with 50 ng/mL 4,4'-Dihydroxybiphenyl - CAS 92-88-6 (DHB) as an internal standard. The chromatogram of the extracted blank urine (see Figure 84 below) showed that the urine was free of the hormones. This was expected since the urine sample was from a “hormone-free” cow.

151 (x100,000) 135.00 (1.00) 268.00 (1.00) 2.5 2.0 1.5 1.0 0.5 0.0 10.5 10.6 10.7 10.8 10.9 11.0 11.1 11.2 11.3 11.4 11.5 11.6 Figure 84 A SIM chromatogram of the urine extracted without spiking (The highlighted area represents the area where the hormones were expected to appear).

There is a small peak towards the tail end of the retention time of interest (11.25 minutes). However, it is important to note the scale in this chromatogram has been expanded to enhance visualisation of the area. This peak was too small to be integrated. Most importantly it is not related to HEX since the “qualifying ion”, m/z 107, was not detected.

The effect of NaCl on extraction efficiency for both hormones is shown in Figure 85 below.

Figure 85 Effect of NaCl content on extraction efficiency at 50 ng/mL hormones spiking level

Interestingly, these results contradict those obtained using 5 µg/mL concentration. The decrease observed with the higher concentration could be due to extreme salting out of the hormones by adsorbing onto the glass walls. The 5% NaCl plus the suspended solutes approximates 7.5% NaCl which is close to 10% NaCl that was found to be ideal for BID-SDME

152 of triazines in our earlier work. Therefore it was shown the improvement offered by the presence of NaCl in solution could still be captured at lower concentrations of the non-polar analyte.

4.3.11 The effect of pH on extraction efficiency Hexestrol and diethylstilbestrol are phenolic compounds with a potentially acidic OH group, and as such their extraction is suspected to be possibly pH dependent. This is due to the fact that the OH group could easily ionise to O- and H+ in basic medium resulting in enhanced water solubility and reduced extraction. The pH of the original urine solution was 8.6, possibly due to urea.

To evaluate the pH dependence of the extraction efficiency, pH of the original urine solution (diluted 10 times) was measured and thereafter three other solutions were prepared with pH varying from 3.5 to 10.2 using either ammonium hydroxide or formic acid solutions; these solutions were spiked to provide 50 ng/mL concentrations of the hormones.

Figure 86 The effect of pH on extraction of the hormones

The 5% in Figure 86 legend indicates the presence of 5% NaCl (w/v) in the solution. As can be seen from Figure 86, extraction decreases with the increasing pH, which is in agreement with the expectation. This is because under basic conditions the phenolic OH group may be ionised so the hormones become more soluble in water than under acidic conditions.

153 4.3.12 Determination of enrichment factor at optimum conditions The optimum conditions for extraction of these hormones have been determined as follows: BID-SDME (1 μL droplet with 0.5 μL air-bubble), 3:1 chloroform:toluene mixture, 20 minutes static extraction, pH 3.5, 5% NaCl, room temperature (20 °C) for a 1 mL solution. The enrichment factor is determined as a ratio of the concentration (response) after extraction to the concentration of the initial solution, expressed mathematically: 0 E = Cx/Co

where Cx represent the concentration post extraction and Co denotes the initial concentration in the aqueous solution.

Since the aqueous solution cannot be directly injected into the GC, the corresponding solution was prepared in methanol and used as a reference solution. The other solutions were prepared as follows: 1. Methanol (reference solution) 2. 5% w/v NaCl aqueous solution 3. 5% w/v NaCl 1/10 acidified to pH 3.5 urine solution 4. blank 1/10 urine sample

Solution 1 was injected as a reference solution against which to compare the enrichment factor. Solutions 2 and 3 will provide the enrichment factor for both normal SDME and BID- SDME, while solution 4 provides a blank for the urine solutions.

It is evident from the results (see Table 26 below) that BID-SDME improves the classical SDME by a factor of 1.6 in aqueous solutions while in acidified urine the average improvement is 1.5. The enrichment factors in urine supersede those in water. This is possibly due to other dissolved salts in the urine that may stabilise the BID-SDME arrangement, leading to the improvements noted.

154 Table 26 Enrichment factors of hormones from various conditions using 50 ng/mL solution Enrichment factor HEX cis-DES trans-DES reference 100 100 100 aqueous 1281 888 2041 urine 3612 3827 9043 BID-SDME (water) 1935 1546 3570 BID-SDME (urine) 6212 6022 13816

BEF* (aqueous)# 1.5 1.7 1.7 BEF (urine)$ 1.7 1.6 1.5 * BEF: Bubble enrichment factor # ratio of the extraction of the BID-SDME to normal SDME using aqueous solutions $ ratio of the extraction using BID-SDME to normal SDME using urine samples

One of the most important observations regarding this analysis is the narrow dynamic range observed with the analytes. The chromatogram of 50 ng/mL directly injected into the GC-MS gives very small peaks that are almost masked by the background as can be seen in the following chromatogram (Figure 87).

(x100,000) 2.00 135.00 (1.00) 268.00 (2.61) 1.75 HEX 1.50

1.25

1.00 DES 0.75

0.50 0.25

10.9 11.0 11.1 11.2 11.3 11.4 11.5

Figure 87 An expanded chromatogram of a 50 ng/mL solution of hormones in methanol

When the corresponding solution was extracted using the best conditions, it resulted in the chromatogram as shown in Figure 88. The latter chromatogram is 10 × the scale of the other, to place the matter into perspective.

155 (x10,000) 135.00 (1.00) 8.0 268.00 (2.50) 186.00 (1.00) HEX 7.0 6.0 DES 5.0 4.0 3.0 2.0 1.0

10.7 10.8 10.9 11.0 11.1 11.2 11.3 11.4 11.5

Figure 88 An expanded chromatogram of 50 ng/mL extracted using the optimized conditions for BID-SDME

In an effort to improve the S/N ratio for the 50 ng/mL in order to be able to quantify the other isomer of DES, it was decided to use 100 ng/mL solution. However, the extraction from the corresponding 100 ng/mL urine resulted in saturation of the detector when hexestrol (m/z 135, black trace in Figure 88) eluted. This technical issue led to the use of this 50 ng/mL solution as described above.

4.3.13 Determination of linearity, limits of detection and quantification Linearity was determined in the range of 0.5 ng/mL to 10 ng/mL. This was informed by the fact that the CRM concentration was about 10 ng/mL. The calibration curve is obtained as follows in Figure 89.

HEX cis-DES trans-DES net DES 1500

1250

1000

750

500 relative response relative 250

0 0 2 4 6 8 10 concentration (ng/mL)

Figure 89 Concentration versus relative response for the hormones

156 The following table illustrates the analytical data as obtained from the calibration curve above. The net in Figure 89 and Table 27 below refers to the sum of the two DES isomers (cis and trans) as indicated by the CRM.

Table 27 Table showing the analytical data obtained from the calibration curves

HEX cis-DES trans-DES net DES R2 0.9993 0.9991 0.9994 0.9994 Intercept 101.77 10.838 13.51 24.34

SDintercept 9.15 0.86 1.36 2.03 Slope 118.93 9.65 19.02 28.68

SDslope 1.79 0.17 0.26 0.40

LODintercept (ng/mL) 0.35 0.40 0.32 0.32

LODstd err (ng/mL) 0.23 0.28 0.21 0.21 LOQ (ng/mL) 0.77 0.93 0.70 0.71

Seemingly, the LOD values obtained show that the responses obtained with a 0.05 ng/mL standard were below the detection limits. This could point towards the controversy surrounding the detection limits determination where other researchers use the concentration that gives S/N ratio of 3 and use that as a detection limit, despite the subjectivity of this method. However, one of the important aspects that is consistent is the fact that the LOD obtained using the standard error of the intercept are different from those obtained with the standard error of the calibration.

The chromatogram obtained from the extraction of a 0.05 ng/mL solution is shown in Figure 90 below. As can be seen, the peak intensities at retention times 11.0 minutes and 11.025 minutes are very low.

157 (x1,000,000) 1.75 135.00 (1.00) 268.00 (1.00) 186.00 (1.00) 1.50 1.25 1.00 0.75 0.50 0.25 0.00 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0

Figure 90 A SIM chromatogram of a 0.05 ng/mL mixture

A better presentation of the expanded chromatogram is shown in Figure 92 below.

(x10,000) 135.00 (1.00) 8.0 268.00 (2.50) 186.00 (1.00) HEX 7.0 6.0 DES 5.0 4.0 3.0 2.0 1.0

10.7 10.8 10.9 11.0 11.1 11.2 11.3 11.4 11.5

Figure 91 An expanded SIM chromatogram for the 0.05 ng/mL mixture following extraction

From the expanded chromatogram (Figure 91), it is evident from visual inspection that the 0.05 ng/mL solution still gives a response that is almost at the detection limit (S/N = 3) from with respect to diethylstilbestrol (DES), while the response is higher than the detection limit with respect to hexestrol (HEX). This further demonstrates the importance of stating which approach has been used to determine the detection limits other than just stating the values.

Other researchers have worked in the area of hormones detecting and quantification. Some pertinent features of those studies are summarised in Table 28.

158 Table 28 A table showing some hormones and the matrix from which they were determined from Hormone Matrix Extraction LOD Recovery (%) Accuracya Reference technique (%)

HEX, DNS Bovine SPE 0.19 ng/gb Not 98-102 Malone et al.41 muscle determined

DES, DNS, milk MIP-SPE 2.5, 3.3, 3.3 83.7 – 90.6 Not Liu et al.42 HEX µg/L determined DES, HEX, urine SPE 0.5 ng/mL 45, 47, 54 Not Cassademont DNS determined et al.43 a Percentage value relative to the CRM value b decision limit (CCα) determined 2.33 × standard deviation of the intercept

The obtained LOD values of the present method are generally better than those cited from literature and presented in the Table 28. For example, Cassademont et al.43 obtained LOD of 0.5 µg/L for the same hormones from urine using SPE, which is expected to be more sensitive due to the fact that it is exhaustive rather than equilibrium-based. Liu et al.42 achieved between 2.5 and 3.3 µg/L LOD values for the same hormones including dienestrol using highly selective molecularly imprinted polymers from milk. Most importantly these obtained LOD values for the BID-SDME method are better than the EU required performance limit of 1 µg/kg (µg/L for liquids) reported by Malone et al.41

4.3.14 Evaluation of reproducibility Evaluation of reproducibility was achieved by using the two sets of 2 x 1 ng/mL standards sampled daily for two days followed by the other set for the remaining two days making the analysis span four consecutive days as depicted in the Table 29 and Figure 92. The results demonstrate good reproducibility of the extraction with an average %RSD value of 7 for n = 24. This value of n = 24 is achieved since each of the solutions entered in the first column was extracted a minimum of 3 times.

159 Table 29 Reproducibility of the extractions using 1 ng/mL hormones solutions: individual and relative extractions

Individual extractions Extractions relative to solution 1 day 1 HEX cis-DES trans-DES HEX cis-DES trans-DES Day 1 S1 196.96 18.81 30.27 100 100 100 Day 1 S2 225.91 21.02 35.36 114.7 111.7 116.8 Day 2 S1 199.82 20.80 30.71 101.5 110.6 101.4 Day 2 S2 213.34 21.52 33.59 108.3 114.4 111.0 Day 3 S3 198.53 18.70 33.26 100.8 99.4 109.9 Day 3 S4 202.20 19.93 32.41 102.7 105.9 107.1 Day 4 S3 196.13 18.76 28.94 99.6 99.7 95.6 Day 4 S4 197.40 17.90 28.23 100.2 95.1 93.3 Average 203.79 19.68 31.60 103.47 104.61 104.38 SD 10.51 1.32 2.46 5.33 7.03 8.14 %RSD 5.2 6.7 7.8 5.2 6.7 7.8

The information in Table 29 above can be presented graphically as in Figure 92 below.

Figure 92 Reproducibility of the extractions using four solutions extracted over four days

The same results could also be depicted as normalised ratios of the first solution to demonstrate clearly the reproducibility of diethylstilbestrol isomers at the same scale as the hexestrol.

160

Figure 93 Relative reproducibility of four solutions extracted over four days

4.3.15 Evaluation of the accuracy of the method The method accuracy was determined using the 1 ng/mL standard and compared against the CRM at 1.28 ng/mL following dilution of the original CRM which was at 12.8 ng/mL. The CRM contained both cis- and trans- diethylstilbestrol isomers with no description of the ratio of the two isomers, hence it was prudent to combine the two isomers and to determine the concentration using the combined responses as it has already been shown in the preceding sections. Table 30 illustrates the relative responses obtained when analysing the CRM and the corresponding concentrations for seven replicate extractions.

Table 30 Relative responses for extractions of the CRM and the corresponding calculated concentrations

Trial CRM extraction Concentration (ng/mL)*

CRM 1 62.17 1.32 CRM 2 54.06 1.04

CRM 3 62.45 1.33 CRM 4 60.43 1.26 CRM 5 56.33 1.12

CRM 6 58.98 1.21 CRM 7 64.51 1.40 CRM 59.85 1.24 SD 3.67 0.13 %RSD 6.14 10.34 *Concentrations are calculated from the calibration curve equation: y = 28.478x + 25.732

161 The concentration of the CRM obtained by the calculation of individual relative responses using the calibration curve data (calibration equation) was obtained as 1.24 ng/mL (see Table 30) with the %RSD value of 10%. The confidence interval for the calculated concentration of the CRM (calculated as CI = ± t6,005 × s/√n yields the value of 0.099 ng/mL; where t = critical statistical t-value for 95% confidence level for 6 degrees of freedom, s = standard deviation, n = number of replicates. Thus the CRM concentration is calculated as 1.24 ± 0.10 ng/mL. This value is within the limits of the CRM (1.024 – 1.536 ng/mL) as depicted by the confidence interval of 0.24 ng/mL for the 1.28 ng/mL CRM.

Table 31 The comparison of the calculated concentrations to the CRM

Concentration of CRM = 1.28 ± 0.24 ng/mL Concentration of closest standard = 1 ng/mL Calculated concentration from calibration: 1.24 Range of CRM 1.024 1.28 1.536

The student t-test was performed to validate the accuracy of the obtained value to the CRM value using the following formula:

t = (x - µ)√n/s

where represents average concentration, µ denotes the CRM value, n denotes the number of replicates, and s denotes standard deviation.

The value obtained using the values: (1.24), µ (1.28), n (6) and s (0.13) was 0.7536 compared to the critical t-value of 2.45 for n = 7. This illustrates the sufficiently good accuracy of the method.

The reliability of the extraction assessed through the standard deviation of the CRM at the concentration of 1.28 ng/mL analysed seven times (n = 7). The method shows good reliability with the %RSD of the relative responses of about 6% and the calculated concentration yielding the %RSD of about 10% for a combined diethylstilbestrol, which falls within the acceptable limits of trace analysis. Most importantly, both lower and higher

162 confidence limits of the calculated concentration fall comfortably within the limits of the CRM.

4.4 Application of the method to other biological matrices Following successful optimisation with regards to the urine sample, the method was extended to assess its applicability on milk samples. The pH of different samples was measured and the results are presented in Table 32. The pH of the urine is also included for ease of reference.

Table 32 Different samples and their pH values sample pH value Acidified pH value water 7.7 3.4 urine 8.6 3.5 milk 6.5 3.6

5 mL portions of each sample were taken and used to measure the pH of the original solution. Thereafter the samples were each diluted ten times to reduce the colouration that would affect the visualisation of the droplet. 2 mL of these portions were taken for further analysis, while the remaining 3 mL portions were acidified to pH around 3.5 using formic acid. 2 mL aliquots from the acidified solutions as well as the original solutions were spiked with 50 µL of 1 µg/mL hormones standard stock solution to make samples with 25 ng/mL concentration of the hormones. These solutions were allowed to stand for 3 hours to achieve equilibration. Thereafter, 1 mL aliquots were transferred to the correspondingly labelled vials each containing 0.05 g NaCl.

Following NaCl addition to this solution, a jelly-like yellowish brown precipitate formed. A white jelly-like suspension also formed after addition of 0.05 g of NaCl to the acidified milk. These two solutions were filtered through a 0.45 µm nylon filter to offer clear colourless solutions, before being extracted using the optimized BID-SDME extraction method. The extraction results are presented in Table 33.

163 Table 33 Comparison of extraction efficiencies of the hormones from water, urine and milk samples Sample Relative intensity after extraction Hexestrol cis-diethylstilbestrol trans-diethylstilbestrol Dienestrol Water 929.03 (11.1)a 41.00 (3.2) 84.38 (5.2) 8.10 (12.9) Urine 994.02 (15.8) 71.74 (7.9) 133.36 (6.4) 13.02 (19.5) Acid Milk 473.91 (31.9) 42.27 (39.1) 68.05 (39.8) 3.07 (93.8) Relative extraction (%) compared to extraction efficiency in water Water (Ref) 100 100 100 100 Urine 157 175 158 161 Acid Milk 51 103 81 38 a Values in parentheses represent the % relative standard deviations

The observed low recoveries with regards to the milk sample could be due to the curdled precipitate that was filtered out of the solution. It is likely that this proteinacious material would adsorb and absorb some analytes thereby removing them from the filtrate. The visibility of the droplet was also very poor, which might be responsible for low the precision in the case of milk sample.

In a study by Dagnac et al.44 the recovery of 77.5% for atrazine-D5 was reported for extraction of pesticides from raw milk. In this study the dispersive solid-phase micro- extraction method was used. However, the matrix effect was reduced through extraction with the organic solvents, methanol and acetonitrile before the final extraction. At this stage of its development, it appears as if the SPME method would be preferred over the BID- SDME method currently under development. However, it is quite likely that future work in this area will be rewarded with successful outcomes.

4.5 General discussions and conclusions The BID-SDME method has been developed and validated for analysis of stilbene hormones from bovine urine. This method seems to be sufficiently accurate, reproducible and reliable with good linearity. This is the first report of the BID-SDME application on analysis of hormones in biological systems. The other important aspect is the use of mixed solvents in BID-SDME. The important aspect in using mixed solvents, much as in the case of individual solvents, that the density of the solvent mixture is critical since the use of solvents less dense than water has been shown to be incompatible with the BID-SDME extraction due to

164 buoyancy that results in wicking of the droplet up the syringe. The bubble in the BID-SDME has proven useful for increasing the surface area and providing a thin film for mass transfer.

BID-SDME has been proven to be effective for extraction of stilbene hormones from urine samples without any negative matrix effect. pH of the matrix is quite important as the hormones are relatively acidic (through the phenolic OH groups) hence ionise easily in slightly basic media. This results in reduced solubility in the extracting solvent and concomitant reduced extractability.

Future work in this area could include the application of the method to the quantification of hormones in other animal products such as meat, blood and milk. It is anticipated that some methods of sample pre-concentration will be necessary before the BID-SDME method can be applied with confidence. Nevertheless, the method shows promise and has already proven successful to other areas on different matrices.

165 4.6 References

1. Dzięcioł, M.; Staoczyk, E.; Noszczyk-Nowak,A.; Niżaoski, W.; Ochota, M.; Kozdrowski, R. Research in Veterinary Science, 2012, Article in press; D.O.I.: 10.1016/j.rvsc.2012.02.009 2. Ramasamy, R.; Ricci, J.A.; Palermo, G.D.; Gosden, L.V.; Rosewaks, Z.; Schlegel, P.N. The Journal of Urology, 2009, 182, 1108-1113 3. Engel, J.B.; Schally, A.V.; Dietl, J.; Rieger, L.; Hönig, A. Molecular Pharmaceutics, 2007, 4, 652-658 4. ul Haq, I.; Cáceres. C.; Hendrichs, J.; Teal, P.; Wornoayporn, V.; Stauffer, C.; Robinson, A.S. Journal of Insect Physiology, 2010, 56, 1503-1509 5. Regal, P.; Anizan, S.; Antignac, J.-P.; Bizec, B.L.; Cepeda, A.; Fente, C. Analytica Chimica Acta, 2011, 700, 16-25 6. Shang, Y. Cell Research, 2007, 17, 277-279 7. Madhunapantula,S.R.; Mosca, P.; Robertson, G.P. Cancer Biology and Therapy, 2010, 10, 765-766 8. Noppe, H.; Le Bizec, B.; Verheyden, K.; De Brabander H.F. Analytica Chimica Acta, 2008, 611, 1-16 9. Segal, D. South African Medical Journal, 2009, 99, 187-195 10. Sauerwein, H.; Breier, B.H.; Gallaher B.W.;, Götz, C.; Küfner, G.; Montag, T.; Vickers, M.; Schallenberger, E. Domestic Animal Endocrinology, 2000, 18, 145–158 11. Dodge, T.L.; Jaccard, J.J. Journal of Adolescent Health, 2006, 39367-39373 12. Hildebrandt, T.; Alfano, L.; Langenbucher, J.W. Journal of Psychiatric Research, 2010, 44, 841-846 13. Hansen, M.; Jacobsen, N.W.; Nielsen, F.K.; Björklund, E.; Styrishave, B.; Halling-Sørensen, B. Analytical and Bioanalytical Chemistry, 2011, 400, 3409–3417 14. Raul, J.-S.; Cirimele, V.; Ludes, B.; Kintz, P. Clinical Biochemistry, 2004, 37, 1105–1111. 15. Bévalot, F.; Gaillard, Y.; Lhermitte, M. A.; Pépin, G. Journal of Chromatography B, 2000, 40, 227–236 16. Brunet, B.R.; Barnes, A.J.; Scheidweiler, K.B.; Mura, P.; Huestis, M.A. Analytical and Bioanalytical Chemistry, 2008, 392, 115-127 17. Jurge, M. H.; Hagey, L. R.; Jurke, S.; Czeakala, N. M. Primates, 2000, 41, 311-319 18. Song, G.; Qin, T.; Liu, H.; Xu, G.-B.; Pan, Y.-Y.; Xiong, F.-X.; Gu, K.-S.; Sun, G.-P.; Chen, Z.- D. Lung Cancer, 2010, 67, 227-231 19. Deng, C.; Zhang, X.; Li, N. Journal of Chromatography B, 2004, 808, 269-277 20. Riazanskaia, S.; Blackburn, G.; Harker, M.; Taylor, D.; Thomas, C.L.P. Analyst, 2008, 113, 1020–1027 21. Kaklamanos, G.; Theodoridis, G.; Dabalis, T. Journal of Chromatography B, 2009, 877, 2330–2336 22. Aman, C.S.; Pastor, A.; Cighetti, G.; de la Guardia, M. Analytical and Bioanalytical Chemistry, 2006 , 386 1869–1879

166

23. Plenis, A.; Konieczna, L.; Olędzka, I.; Kowalski, P.; Bączek, T.Molecular BioSystems, 2011, 7, 1487–1500 24. Germano, J.M.; Molinia, F.C.; Bishop, P.J.; Cree, A. Theriogenology, 2009,72, 663–671 25. Akre, C.; Fedeniuk, R.; MacNeil, J. D. Analyst, 2004, 129, 145-149 26. Görög, S. Journal of Pharmaceutical and Biomedical Analysis, 2011,55, 728–743 27. Izumi, Y.; Okazawa, A.; Bamba, T.; Kobayashi, A.; Fukusaki, E. Analytica Chimica Acta, 2009, 648, 215–225 28. Xu, C.-L.; Peng, C.-F.; Liu, L.; Wang, L.; Jin, Z.Y.; Chu, X.-G. Journal of Pharmaceutical and Biomedical Analysis, 2006, 41, 1029–1036 29. Petrovic, M.; Tavazzi, S.; Barcelo, D. Journal of Chromatography A, 2002,971, 37–45 30. Zhao, C.; Ji, Y.; Shao, Y.; Jiang, X.; Zhang, H. Journal of Chromatography A, 2009, 1216, 7546–7552 31. Kataoka, H. Analytical and Bioanalytical Chemistry, 2010, 396, 339–364 32. Vidal, L.; Chisvert, A.; Canals, A.; Salvador, A.; Journal of Chromatography A, 1174 (2007) 95–103 33. Xie, S.; Xiang, B.; Zhang, M.; Deng, H. Microchimica Acta, 2010, 168, 253-258 34. Williams, D.B.G.; George, M.J.; Meyer, R.; Marjanovic, L. Analytical Chemistry, 2011, 83, 6713-6716 35. Medir, M.; Mackay, D. The Canadian Journal of Chemical Engineering, 1975, 53, 274–277 36. Samaratunga, S.S.; Nishimoto, J.; Tabata, M. Analytical Sciences : The International Journal of the Japan Society for Analytical Chemistry, 2005, 21, 1073-1078 37. Zhang, M.; Huang, J.; Wei, C.; Yu, B.; Yang, X.; Chen, X. Talanta, 2008, 74, 599–604 38. Qian, L.-L.; He, Y.-Z. Journal of Chromatography A, 2006, 1134, 32–37 39. Bagheri, H.; Khalilian, F. Analytica Chimica Acta, 2005, 537, 81-87 40. Lin, C.-Y.; Huang, S.-D. Journal of Chromatography A 2008, 1193, 79-84 41. Malone, E.M.; Elliot, C.T.; Kennedy, D.G.; Reagan, L. Analytica Chimica Acta, 2011, 637, 112-120 42. Liu, M.; Li, M.; Qiu, B.; Chen, X.; Chen, G. Analytica Chimica Acta, 2010, 663, 33-38 43. Cassademont, G.; Pérez, B.; García Regueiro, J.A. Journal of Chromatography B, Biomedical Applications, 1996,686, 189-198 44. Dagnac, T.; Garcia-Chao, M.; Pulleiro, P.; Garcia-Jares, C.; Llompart, M. Journal of Chromatography A, 2009, 1216, 3702-3709

167 Chapter 5: Development of a dispersed solvent-assisted headspace

5.1 Introduction

Headspace analysis of any solution is an ideal way of pre-concentration of analytes before analysis as it avoids the extractant from coming into contact with the sample matrix, which is often very complex and can generate negative matrix effect.1,2 As it depends heavily on the volatility of the analytes, mass transfer becomes a bottleneck of the method.3 Mass transfer in the headspace, however, is arguably fast due to higher diffusion coefficients in the gas phase than in condensed phases (about 104 greater),4 but moving the analytes into the headspace is still a challenge. Headspace analysis has been documented for various compounds most of which are volatile and “semi-volatiles” on both liquid and solid-phase techniques.5

Mass transfer and other kinetic and thermodynamic parameters governing headspace sampling have been reviewed in detail by Kalua et al.6 They demonstrated the importance of principles of chemical potentials and of Henry’s law of vapour pressures for mass transfer and dilution phenomena in the headspace and in the bulk of the aqueous solution as well as the SPME extraction. In simple terms, partitioning of the analytes between the bulk of the solution and the headspace depends on a partition coefficient K (expressed as a ratio of concentration in the gas phase to that in the condensed phase) and phase ratio ß (expressed as a volume ratio of headspace to sample solution).7, This clearly shows the relationship between the K and ß and the importance of volume of both sample solution and the headspace for pre-concentration and sampling in general.

Solid-phase micro-extraction (SPME) is a well-developed and widely accepted pre- concentration and clean-up technique for trace analysis, which has recently been reviewed by Quintana et al.8 SPME requires sorbents that ensure good molecular interactions with the analytes and these interactions depend on the number, size and nature of the active sites in the polymer network, these active sits therefore act as a limiting factor with regards to efficiency but may offer some degree of selectivity. Several stationary phases are reported for various samples and matrices.9

168 Solvent micro-extraction has made a recent entry into the micro-extraction techniques and has received a great deal of attention leading to a number of publications due to its simplicity, affordability and efficiency.6 The efficiency of SME depends on analyte solubility in the organic solvent. As such solvent changes are the most highly investigated parameter to optimise the extraction. However, the use of this technique for headspace sampling still is a challenge due to the volatility of most of the common organic solvents available in most labs.

Rezaee et al.10 have developed a new solvent micro-extraction technique called “dispersive liquid-liquid micro-extraction” wherein the organic solvent is dispersed into the aqueous sample with the aid of a second organic solvent that is miscible with the aqueous sample as well as the organic solvent being used as an extractant. Due to the dispersed organic solvent, there is an increase surface area that improves the mass transfer of the analyte into the extractant. The importance of surface area to volume of extractant in pre-concentration techniques has also been demonstrated in earlier sections of the present study when using drop-based solvent micro-extraction.11

Herein, the potential of using dispersive liquid-liquid micro-extraction as an aid for mass transfer of analytes to the headspace is explored in an effort to improve what is termed the headspace capacity.12 Evaporation of the dispersed organic solvent droplets into the headspace at elevated temperatures could afford a means of transport for improved mass transfer of the “pre-extracted” analytes into the head-space resulting in what would be termed assisted volatilisation. Once the compounds have been volatilised this way they can be sampled directly if their concentration levels are sufficiently high or they can be pre- concentrated further using SPME or SME techniques. The model analytes used in this preliminary experiment are chlorophenols.

5.2 Experimental 5.2.1 Chemicals and standards used The phenols (phenol, 4-chlorophenol, 3,5-dichlorophenol, 3-methyl-5-chlorophenol and 2,4,6- trichlorophenol) were obtained as standard compounds from Sigma-Aldrich (Seelze, Germany). All solvents used in this study were of analytical grade or higher: methanol, chloroform, water were obtained from Riedel-de Haën (Seelze, Germany); hexane,

169 diethylether, ethylacetate, NaCl, sodium chloride AR grade were obtained from Sigma-Aldrich (Seelze, Germany).

Preparation of a standard solution of chlorophenols: 10 mg of phenol (Phe), 4-chlorophenol (CP), 3,5-dichlorophenol (DCP), 3-methyl-5-chlorophenol (MCP) and 2,4,6-trichlorophenol (TCP) were weighed individually and dissolved simultaneously in 10 mL of HPLC grade methanol and stored in a fridge. Dilutions of this stock solution were carried out to prepare the working solutions prior to analyses. The dilution was made by adding appropriate amount of the methanolic phenols solution to the HPLC grade water.

5.2.2 Apparatus A multi-vial compartmental water bath with magnetic stirrer facility (Pierce, Rockford Illinois USA) was used for heating; 100 µm PDMS fibres (Supelco, Bellefonte, PA, USA) used for sampling.

5.2.3 Chromatography and mass spectrometry instrumentation

Analyses were carried out using a Shimadzu QP2010 gas chromatograph mass spectrometer (Kyoto, Japan) fitted with a Zebron 35ms column with 30 m × 0.25 mm × 0.25 μm dimensions. The carrier gases used were nitrogen and helium (99.999%, Afrox, South Africa), respectively, passing through the GC columns at a constant column flow rate of 1 mL/min.

The GC program was as follows: Injector temperature: 250 °C; transfer line 250 °C; Column program: 70 °C for 4 min, ramped at 10 °C/min to 120 °C, then ramped at 30 °C/min to 160 °C, then ramped at 40 °C/min to 260 °C and held for 3.17 min (total run time = 16 minutes).

MS settings were as follows: EI 70 eV with 1.5 kV detector voltage; Ion source temperature: 200 °C; Interface temperature: 240 °C; Acquisition window: 6 – 18 min; Scan mode: 50 – 350 at scan rate of 416 and dwell time of 0.50 sec. For SIM analyses the molecular ions were selected as they produced sufficiently intense peaks than the fragmental peaks as can be observed in the mass spectra below.

170 5.2.4 Extraction procedure

The following parameters were evaluated using 5 ng/mL aqueous solutions of the 5- component phenols mixture which was thermally equilibrated using the thermo bath: ideal solvent, volumes of solvent (the ideal volume of spiked organic solvent), ionic strength (effect of added NaCl), repeatability of extractions and eventually extraction efficiency for the combined properties was calculated.

Detailed procedures for each part of the experiment will be outlined briefly before the results of such parts.

5.3 Results and discussions 5.3.1 Development of chromatography and identification of compounds

A reference solution of chlorophenols (5 µg/mL ) was prepared using HPLC grade methanol and injected into the GC-MS to develop a GC program that sufficiently separates the compounds within relatively short time (11 minutes) without compromising resolution and chromatographic parameters. Typical tailing behaviour expected with phenolic compounds is evident in the resulting chromatogram shown in Figure 95 below.

(x10,000,000) TIC 1.25 MCP DCP TCP 1.00 CP 0.75 Phe 0.50

0.25

0.00

5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0

Figure 94 A TIC chromatogram of the mixture of 5 chlorophenols

The compounds were identified from their fragmentation patterns using the NIST 2005 Library incorporated in the Shimadzu QP2010 GC-MS. The m/z values were selected for SIM analyses based on the individual base peaks from the various phenols noting the mass spectra, which are reproduced below: 94, 128, 162, 107 and 196 respectively.

171

phenol (Phe) 4-chlorophenol (CP)

2,6-diclorophenol (DCP) 2-methyl,4-chlorophenol (MCP)

2,4,6-trichlorophenol (TCP)

Having identified the base peak m/z values from individual mass spectra, the MS was set on selected ion monitoring (SIM) mode and the following chromatogram was observed:

(x1,000,000) 3.0 162 (DCP) 142 (MCP) 2.5 94 (Phe) 128 (CP) 196 (TCP) 2.0 1.5 1.0 0.5 0.0

5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0

Figure 95 The SIM chromatogram showing the 5 components and their corresponding m/z values

5.3.2 Demonstration of the proof of concept A standard headspace protocol was established as follows: a sample was prepared in aqueous solution, spiked with the organic solvent, then shaken vigorously before being heated to the required temperature. Sampling was performed once the thermal equilibrium had been attained. All of the extractions were static.

172 To demonstrate the concept, one aqueous solution containing 50 ng/mL chlorophenols concentration, was prepared in water by pipetting 10 µL of a 5 µg/mL stock and diluted to 1 mL using HPLC grade water (about 900 µL). This solution was spiked with 90 µL of methanol resulting in 10% MeOH. The other solution was prepared using methanol instead of water resulting in 100% methanol. These solutions were subjected to headspace sampling for 20 min. Thereafter similarly prepared solutions were heated to 45 °C and subjected to headspace sampling as well. The results are shown in Figure 97 below.

Figure 96 The effect of temperature on headspace extraction of aqueous and methanol solutions

As can be seen from Figure 96, increasing temperature results in a slight increase in the extraction efficiency from the aqueous solutions. Interestingly, the increase in temperature did not seem to improve the efficiency from the solution prepared in neat methanol.

Two more aqueous solutions were prepared and spiked with 100 µL of acetone and diethylether, respectively, resulting in about 10% organic solvent besides the methanol in which the stock solution was prepared. These solutions were shaken vigorously before being heated to 45 °C using a thermobath and thereafter subjected to headspace extraction as before. The results of these two sets of solutions were compared and are presented in Figure 98 below.

As can be observed from Figure 97, there was a remarkably high extraction following diethyl ether spiking of the solution and increasing the temperature to 45 °C (highlighted bars). The signal integration was only possible for phenol (Phe) and monochlorophenol (CP) –

173 highlighted bars; whilst the rest of the compounds saturated the detector as such they could not be integrated! That means the pre-concentrated level was too high for the MS detector settings. This observed increase in extraction was attributed to the organic solvent dispersed in the aqueous solution, and this dramatic result serves as a demonstration of the potential of this method to significantly improve upon current headspace practices.

MeOH MeOH 10% Aqueous (45 °C) 10%Acet( 45°C) 10%Met (45 °C) 10%Ether(45°C) 25000 Detector 20000 saturated 15000

10000 peak areas areas peak 5000

0 Phe CP DCP MCP TCP compound

Figure 97 Headspace extraction of aqueous and MeOH solutions after diethylether spiking

The 100% MeOH solution was prepared to assess if the compounds could volatilise with the vapour of MeOH as it has a higher vapour pressure (lower boiling point) than water. As the compounds are dissolved in MeOH, they should expectedly vaporise with some MeOH molecules as the temperature increases. The fact that no significant increase was observed with increase in temperature could be due to the fact that methanol dissolves in water so the compounds are evenly dispersed in the aqueous solution. This solution provides good basis from which to build the arguments in the proceeding sections.

5.3.3 Choice of solvent To evaluate the principle using the water immiscible solvents, several 50 ng/mL aqueous solutions were prepared and spiked with 100 µL of water immiscible organic solvents, specifically diethylether and dichloromethane (DCM). These solutions were heated to 50 °C

174 using a thermobath and subsequently subjected to 20 minutes static headspace extraction as outlined earlier.

Figure 98 Effect of different solvents on organic solvent-assisted headspace extraction

The higher efficiencies observed with diethylether and dichloromethane (DCM) signal the importance of the dispersed water-immiscible solvent in this scenario. The slight drop noted in the case of DCM could be attributed to the fact that DCM is denser than water and settles at the bottom of the vial as soon as the vial is allowed to settle after the shaking. However this does not seem to be too dramatic as this solvent nevertheless leads to substantial enhancement to the headspace concentrations of the phenols. It must be borne in mind that the aqueous solution still contains some residual MeOH used in spiking the solutions with the standard solution, but this apparently had no dramatic effect on the outcome of the analyses. These results quite clearly demonstrate the proof of concept for the principle under investigation.

Despite the high extraction efficiencies observed, there was a nasty technical catch in that there was notable fibre swelling leading in some cases in the loss of the fibre while trying to retreat it back into the barrel. Various efforts were made to reduce this swelling effect but were left fruitless.

175 5.3.4 Optimisation of the added solvent volume The volume of the diethyl ether was varied from 10 to 100 µL and the solutions were headspace extracted. The error bars in Figure 100 below represents the standard deviations for n = 3 extractions from the same solution.

Figure 99 Effect of the organic solvent volume on efficiency

From Figure 100 above it can be seen that there is an increase in efficiency with increasing organic solvent amount through a maximum at 50 µL followed by a drop in efficiency. The increases notable relate to pre-extraction of the phenols into the organic solvent droplet and the possible saturation of the solvent (equilibrium extraction) by the phenol analytes. A larger volume of organic solvent would typically be able to extract a larger amount of solutes. Exactly why the enhancement is noted to taper off at volumes above 50 µL is not known. It is possible, though, that the headspace becomes saturated with the solvent vapour and that some solvent is retained in the solution. This would have the effect of retaining the solutes in the aqueous solution (in the liquid state of the organic solvent) and reduce the amount of the solutes in the headspace.

There are two important postulations that could possibly explain the drop in efficiency in excess of the 50 µL of the dispersed organic solvent:  incomplete vaporisation of the solvent thereby making the compounds already pre-concentrated to remain in the bulk solution

176 If this is the case, it means that the solutes will remain in the non-evaporated organic solvent that is still dispersed in the aqueous solution.

 reduced adsorption efficiency of the SPME fibre in the presence of an organic vapour-saturated headspace Since the efficiency of adsorption into the fibre relies on evenly advantages gained upon transfer of the sample to the fibre, the net gain of which changes with variation of the constitution of the headspace, it is possible that this is affected by an organic solvent vapour-saturated headspace. In such an instance, the headspace gases may act as in improved “solvent” for these vapourised solutes, thereby limiting their ability to adsorb to the fibre.

An interesting observation was made when using smaller vial. Whereas it was hoped that the smaller vial would result in increased extraction efficiency, the results were contrary to this expectation. This observation could support the gas phase equilibrium/dissolution idea described above.

To evaluate if the extraction would be exhaustive after several extractions, intra-vial extractions were performed and the results are as follows. The solution was spiked once with the organic solvent (50 µL) and sampled n = 8 without further spiking. The results are demonstrated in Figure 100 below.

These results demonstrate that the extraction efficiency is sufficiently repeatable, suggesting a non-exhaustive extraction. If the extraction was exhaustive, the results would not be repetitive. This directly implies that the fibre could not extract all of the analytes in the headspace, which remain available for further extraction/sampling with the fibre.

177

Figure 100 Repeatability of the extraction following addition of 100 µL of diethylether

5.3.5 The effect of ionic strength of the aqueous solution The increase in extraction efficiency with an increase of the solution’s ionic strength is due to the well-understood phenomenon of salting-out. However, there are contradicting observations regarding the effect of addition of salt to the extracted aqueous mixture for various micro-extraction techniques, including instances in this study. To investigate this phenomenon, different NaCl solutions were prepared and subjected to the head-space extraction.

The results demonstrate an increase in extraction efficiency from 0% to 5% NaCl (m/v) with sufficient repeatability (n = 3 with RSD averaging about 10%). From 10% NaCl, the results show erratic behaviour, including high RSD values (RSD > 20%). The extraction seems to be exhaustive in the first instance (the bars labelled with asterisks in Figure 102 legend) while for the subsequent extractions, there seems to be much lower efficiency. The reason for this phenomenon is presently unknown.

178

Figure 101 Effect of ionic strength on extraction

5.3.6 Extraction as function of time

SPME extractions are characterised by longer extraction times (longer than 30 minutes) due to lower mass transfer compared to solvent-based counterparts like liquid-phase micro- extraction where the extraction times are within 30 minutes.13,14 To assess the response of the current method on varied extraction time, an extraction-time profile for was set-up and the results are presented in Figure 102 below.

Figure 102 The extraction-time profile for solvent assisted headspace sampling

179 From Figure 103, it is evident that this method does reduce the sampling time since the maximum extraction efficiency is reached within about 15 - 20 minutes for CP and MCP. This improves on the examples reported in the literature where optimum results are secured after extraction times at or longer than 30 minutes.15,16 This behaviour is consistent with the results obtained by Zhao et al.17 using an ionic liquid-based SPME where the equilibrium was reached within 20 minutes due to higher mass transfer rate. The same authors also reported a drop in extraction efficiency of some compounds with increasing extraction time due to competition. A similar phenomenon was noted in the present study. Figure 102 shows that the extraction efficiency for phenol and chlorophenol declined with increasing extraction time, more dramatically so for phenol. It may be that this observation may also relate to competitive binding. In general, it can be concluded that indeed this technique does offer the benefit of extraction time reduction.

5.3.7 Exploration of smaller volume of samples To assess the effect of larger headspace volume to sample volume, the sample volume was reduced to 0.5 mL in the same 1.8 mL vial, implying almost double the headspace volume compared to when 1 mL of the aqueous sample was used.

Figure 103 The effect of organic solvent volume on 0.5 mL solution

Figure 103 shows that the highest overall efficiency is obtained when the volume of the organic solvent is 20 µL, which is almost the same ratio as the 50 µL for 1 mL (4 and 5% respectively). The same trend is obtained with the smaller volume.

180 5.3.8 Evaluation of different fibre phases In an effort to reduce fibre swelling, two different fibre coatings were evaluated (polyacrylate - PA and polydimethylsiloxane – PDMS) while still using diethylether as a dispersant. The extraction time reduced to 15 minutes. Accordingly, samples were made up containing 50 ng/mL of each phenol. These solutions were shaken and heated the same way as before, and subsequently extracted using the PA fibre. The results shown in Figure 104 demonstrate that the extraction efficiency is not solely a function of solvent modification of the fibre since PDMS gives much higher efficiency that PA fibre. If the role of the solvent outweighs that of the fibre itself, then, one would expect results that were more closely matched.

Figure 104 Extraction as a function of a fibre phase

As can be seen from Figure 104, the extraction efficiency for the PDMS fibre is almost 10 times that for phenol compared to the PA fibre and the ratio decreases to about twice for trichlorophenol. That this is the case argued for fibre involvement in the extent and efficiency of uptake of the phenolic analytes. Indeed, this is also attested to by the competitive effect shown by the various phenols for the PDMS fibre, as demonstrated in Section 5.3.6.

5.3.9 Comparison of different solvent for use as dispersants Solvents of different polarities were tested in an effort to discover a milder solvent that would be compatible with the SPME fibre to reduce the swelling if not completely eliminate it. However, amongst the tested solvents (hexane, cyclohexane, ethylacetate, butylacetate, acetonitrile) no complete elimination of the fibre swelling was realised. The extraction

181 efficiency was not evaluated during this exercise since relatively old fibres were used to prevent sacrificing these relatively expensive items. The swelling of the fibres was followed by viewing under low power microscope. It was quite easy to see the swelling effect on the fibres using this method.

Future work could carefully address this aspect in the hope of establishing suitable fibre/solvent pairing that retain the headspace enhancement effects summarised here.

Important questions still to be answered The fact that diethyl ether improves the extraction so much could be due to the following properties that are presumed to be key in the principle: a high vapour pressure (boiling point 35 °C) makes it easily vaporised, lower density than water (0.71 g/mL) making it easy to disperse in the aqueous solution with a propensity to float on the aqueous phase thereby carrying any dissolved analytes to the surface with it, immiscibility with water and of course its polarity index for the polar chlorophenols.

Another phenomenon that could also lead to improved efficiency is modification of the surface of the fibre which would also offer improved mass transfer through liquid medium in the pores of the polymer network of the fibre. This is further supported by the fact that there was significant fibre swelling especially with increasing salt composition.

5.4 General discussions and conclusions The most important aspect of this work is the effect of solvent on the fibre. It is not yet clear what the effect of the solvent is, in terms of whether it modifies the fibre surface for higher mass transfer or it actually penetrates the fibre in which case it does not only offer mass transfer benefit but also provides extra volume for improved extraction capacity of the fibre, or in fact if it merely transfers the analytes into the headspace as was hypothesised in the conception of the idea. Future experiments will be necessary to explore the merits of this idea, and to overcome the technical challenges experienced.

182 5.5 References

1. Pawliszyn, J.; Pawliszyn, B.; Pawliszyn, M. The Chemical Educator, 1997, 2, 1-7 2. Hakkarainen, M. Journal of Biochemical and Biophysical Methods, 2007, 70, 229–233 3. Saison, D.; De Schutte, D.P.; Delvaux, F.; Delvaux, F.R. Journal of Chromatography A, 2009, 1216, 5060-5068 4. Penapereira, F.; Lavilla, I.; Bendicho, C. Spectrochimica Acta Part B: Atomic Spectroscopy, 2009, 64, 1-15. 5. Lambropoulou, D.A.; Konstantinou, I.K.; Albanis, T.A. Journal of Chromatography A, 2007, 1152, 70–96 6. Kalua, C. M.; Boss, P. K. Journal of Chromatography. A, 2008, 1192, 25-35. 7. Snow, N.H.; Bullock, G.P. Journal of Chromatography A, 2010, 1217, 2726-2735 8. Quintana, M. C.; Ramos, L. Trends in Analytical Chemistry, 2008, 27, 418-436. 9. Spietelun, A.; Pilarczyk, M.; Kloskowski, A.; Namieśnik, J. Chemical Society Reviews, 2010, 39, 4529-4537 10. Rezaee, M.; Assadi, Y.; Hosseini, M.-R. M., Aghaee, E.; Ahmadi, F., Berijani S. Journal of Chromatography A, 2006, 1116, 1-9 11. Williams, D.B.G.; George, M.J.; Meyer, R.; Marjanovic, L. Analytical Chemistry, 2011, 83, 6713-6716 12. Jeannot, M. A; Przyjazny, A.; Kokosa, J. M. Journal of Chromatography A, 2010, 1217, 2326-2336 13. Huang, S.-P.; Huang, S.-D. Journal of Chromatography, A. 2007, 1176, 19-25. 14. Lambropoulou, D. A; Albanis, T. A. Journal of Biochemical and Biophysical Methods, 2007, 70, 195-228 15. Simões, N. G.; Cardoso, V. V.; Ferreira, E.; Benoliel, M. J.; Almeida, C. M. M. Chemosphere, 2007, 68, 501-10 16. Campillo, N.; Peñalver, R.; Hernández-Córdoba, M. Journal of Chromatography, A. 2006, 1125, 31-7 17. Zhao, F.; Lu, S.; Du, W.; Zeng, B. MicroChimica Acta, 2008, 165, 29-33

183 Chapter 6: General Conclusions

This chapter draws the work together and discusses the lessons learned in the study and the response towards the overall objectives. From the initial stages of Chapter 2, the BID-SDME has been fully validated using the US EPA triazine mixture TP 619 as model pesticides. Validated parameters include linearity, limits of detection, repeatability, reproducibility, robustness and accuracy. The method has demonstrated a high degree of accuracy and a wide dynamic range. The accuracy of the method was demonstrated using different concentrations of atraton CRM and it passed in all respects. This was significant since the BID- SDME phenomenon has not been reported before so it was indeed paramount that full validation be carried out.

Some basic principles governing the success of BID-SDME as a technique have been demonstrated in Chapters 3 and 4 where in the importance of solvent density has been demonstrated. For the BID-SDME to be successful there has to be a balance between buoyancy and gravity pulling the solvent down. As opposed to simple SDME where the solubility of the analytes in a solvent is the main consideration for choice of solvent, in BID- SDME, that choice is also dependent on the density of the solvent. The applicability of mixing organic solvents to meet the above requirements has also been demonstrated (Chapter 4).

Another important fundamental principle that this study has shed light on is the importance of high surface area to volume. BID-SDME has consistently offered an average bubble enrichment factor of 1.5 compared to the simple SDME without the air-bubble. This is in part due to the increased surface area and the formation of thin film of the organic solvent around the air bubble.

The applicability of the method outside of laboratory set-up has been successfully demonstrated using atrazine as the analyte representing the triazine mixture used in method development. This was further extended to metolachlor, a different class of herbicides. The method showed applicability for high concentration levels (immediately after spraying) as well as trace levels (at harvesting period) for the two herbicides (atrazine and metolachlor). Furthermore, it was demonstrated that this method can equally be applied in monitoring of

184 pollution from unknown source as demonstrated by analysis of field soils and water with an unknown application history to the author. The importance of hot water extraction over simple room temperature extraction has been demonstrated and been shown to be better than room temperature extraction using sonication, a technique which is widely applied in the extraction of pesticides from soil matrices.

The analyses of soils have also brought light to an omission in many papers regarding spiking of the surrogate in the assessment of the recovery of analytes, especially from soil matrices. This study has demonstrated the importance of carrying out trace-less spiking. This was achieved through the use of an aqueous surrogate that reduces the use of organic solvents since these have been demonstrated to affect the recovery of the analytes thus leading to an overstatement of the recovery values due to the non-traceless spiking. This is a vital contribution since the use of organic solvents in spiking is widely reported even in official methods, not only in academic literature.

As a clean-up technique BID-SDME can be equally applied to different classes of organic compounds as evidenced in analysis of organochlorine pesticides from water and the analysis of synthetic hormones from bovine urine. A drawback of the method is the limited visibility of the droplet in highly particulate and coloured matrices as demonstrated in analysis of milk and undiluted urine samples. However, besides the visibility limitation the method shows promise.

In all cases of different chemical species and matrices, the BID-SDME extraction method has been successfully validated and some few applications reported. The limits of detection obtained for the analysed compounds match those obtained with other methods except for the organochlorines pesticides. The particular failing here is attributed mainly to the use of a generic positive electron impact mass spectrometric (EI-MS) detector as opposed to the ECD, negative electron impact and/or negative chemical ionisation mass spectrometry (NCI-MS) that are more suitable to the compounds. As such, improvements will readily come when the method is coupled to these detection modes. However, it is clear that the results are still of high quality judging by the reproducibility and accuracy validation results.

185 The other important issue that this study has revealed is the need for the analytical chemistry community to set up guidelines for the determination of limits of detection and quantification. Currently, this is left to the analyst to determine a preferred approach, which gives a measure of subjectivity in the data. This is more so in chromatography where blanks cannot be analysed to obtain a background signal. These guidelines would remove the subjectivity of the results obtained since a uniform approach will be used.

With regards to the headspace, the concept of using an organic solvent for facilitating mass transfer has shown promise. This concept needs to be developed further for a better understanding of the principles that lead to the observed improvements. Further development can also be pursued to develop a deeper understanding of technical limitations and possible outcomes thereto.

In conclusion, a simple, affordable, greener, yet effective pre-concentration method has been developed and validated for the analyses of environmental and biological matrices. The already cheaper and sustainable method employing only a few microliters of organic solvents has been improved through the simple use of air-bubble. The method has demonstrated potential for applications in a number of matrices.

The SDME section has thus far contributed one article published in Analytical Chemistry (Williams, D. B. G. George, M. J. Meyer, R.; Marjanovic, L. Analytical Chemistry. 2011, 83, 6713-6.); three more manuscripts are in preparation for publication. This work has also been disseminated at various international conferences as follows: 1. South African Chemical Institute (SACI) Conference, Jan 2011, Johannesburg, South Africa 2. Analytika Conference, Dec 2010, Stellenbosch, South Africa 3. Southern and Eastern African Network of Analytical Chemists (SEANAC) International Conference, July 2009, Izulwini, Swaziland 4. Chromatography Society of South Africa – Society of Mass Spectrometry (ChromSAAMS) International Conference, October 2008, Warm Baths, South Africa 5. Eastern and Southern Africa Analytical and Environmental Chemistry (ESSAEC) Conference, Dec 2007, Victoria Falls, Zimbabwe (initial stages of the present study was presented)

186