Determination and Method Validation ofthe Major Organic Acids in Wine using HPLC and UV Detection by Tunde Meyers Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science in the Chemistry Program YOUNGSTOWN STATE UNIVERSITY May 2003 Determination and Method Validation of the Major Organic Acids in Wine using HPLC and UV Detection Tunde Meyers I hereby release this thesis to the public. I understand that this thesis will be made available from the OhioLINK ETD Center and the Maag Library Circulation Desk for the public access. I also authorize the University or other individuals to make copies of this thesis as needed for scholarly research. Date(j Signature: =i/noliz )/ (2(pz t..s Approval. es H. Mike, Thesis Advisor Dr. Mike Serra, Committee Member Date Peter J. Kasvmsky, Dean of Gradua Date ABSTRACT Determination and Method Validation ofthe Major Organic Acids in Wine using HPLC and UV Detection Tunde Meyers May 2003 Youngstown State University A method for the analysis of the major organic acids in wine was validated. The acids such as: tartaric (TA), malic (MA), lactic (LA), citric (CA), succinic (SA), and acetic acid (AA) were derivatized and detected at 254 nm. Detection at 254 nm was targeted to avoid interference from other wine components at 210 nm. The following validation parameters were studied: linearity, accuracy, precision, usable concentration range, limit of detection (LOD), limit of quantitation (LOQ), robustness, peak identification, and quantitation in wine. iii ACKNOWLEDGMENT I would like to express my sincere gratitude to Dr. James H. Mike for his guidance, support and supervision throughout my research. I would like also to thank Dr. Roland Riesen and my colleague Karen Howard for their help and valuable time spent with this work. In addition, I would like to express my thanks to the following departments that made my research possible: LEERC (Lake Erie Enology Research Center), YSU Chemistry Department, YSU Biology Department, and YSU School of Graduate Studies. IV TABLE OF CONTENTS TITLE PAGE SIGNATURE PAGE ABSTRACT ACKNOWLEDGEMENT TABLE OF CONTENTS LIST OF FIGURES LIST OF TABLES LIST OF EQUATIONS LIST OF ABBRIVIATIONS CHAPTERS I. INTRODUCTION A. CHROMATOGRAPHY B. HPLC C. DERIVATIZATION D. ORGANIC ACIDS IN WINE E. BIOCHEMISTRY OF ORGANIC ACIDS II. HISTORICAL REVIEW III. STATEMENT OF THE PROBLEM IV. MATERIALS AND EQUIPMENT A. REAGENTS Page 11 111 IV V viii x xii xiii 1 1 4 5 5 7 12 16 17 17 V B. EQUIPMENT 18 V. METHOD 19 A. GRAPE MUST 19 B. WINE FERMENTATION 19 C. WINE SAMPLING 20 D. WINE ANALYSIS 20 E. DERIVATIZATION 20 F. HPLCMETHOD 21 G. STANDARDS PREPARATION 22 H. METHOD VALIDATION 23 1. Specificity 23 11. Linearity 24 iii. Accuracy (Recovery) 24 iv. Standard Addition 25 v. Range 26 vi. Precision 26 vii. Limit ofDetection 27 viii. Limit of Quantitation 27 ix. Stability 28 x. Ruggedness 28 Xl. Sensitivity 28 VI. RESULTS AND DISCUSSIONS 29 VIT. CONCLUSION 52 vi Vill. FUTURE WORK XIX. REFERENCES 54 55 vii LIST OF FIGURES Figure Description Page 1. The System Configuration 5 2. Biochemical Pathway of Organic Acids 10 3. HPLC System 18 4. Phenacyl Ester Formation 21 5. Calculation of Peak Symmetry, Theoretical Plate Count, and Resolution 23 6. Overlaid Chromatogram of Wine Samples Spiked With Acids 30 7. Chromatogram of Pinot Oris Wine 31 8. Chromatogram of Chardonnay Wine 32 9. Chromatogram of a Blank 32 10. Lactic Acid Linearity Standards and Standard Addition in Pinot Oris 39 11. Lactic Acid Linearity Standards and Standard Addition in Chardonnay 39 12. Overlaid Chromatograms from Injection Repeatability 41 13. Overlaid Chromatogram for Intra-Assay Repeatability 44 14. Stability ofLactic Acid at Refrigerated Temperature Using Std.#5 45 15. Stability ofLactic Acid at Room Temperature Using Std.#5 46 16. Chromatogram of Robustness Test at 45?C Column Temperature 47 17. Calibration Curve ofTartaric Acid in 15 % Ethanol 49 18. Calibration Curve ofTartaric Acid in 12 % Ethanol in Water containing 250 gIL Sugar 49 viii 19. Spontaneous Fermentation Day 6 20. Spontaneous Fermentation Day 38 51 51 IX LIST OF TABLES Table Description Page 1. Reagents Used in This Research 17 2. Concentrations of Standard 1 to 6, Undiluted 22 3. Standard Addition in Chardonnay 26 4. Standard Addition in Pinot Gris 26 5. LOD and LOQ Using Standards That Were Diluted 1.4 to 10 27 6. LOD and LOQ Using Standards That Were Diluted 1 to 10 27 7. Specificity Test Using Wine Samples 30 8. Solvent Compositions and Gradients Tested to Optimize the HPLC Separation A: H20, B: MeOH / ACN, C: CAN 34 9. Effect ofDerivatizing Agent Concentrations on Peak Area and Shape 33 10. Effect ofDerivatizing Time on Peak Area and Shape 35 11. Effect ofDerivatizing Time at Double the Phenacyl Bromide Concentration on Peak Area and Shape 36 12. Dilution Studied for Determining and Analyzing the Correlation Coefficient 37 13. Dilution Study and Relative Standard Deviations 37 14. Linear Regression Equation for Standard Solutions, Pinot Gris and Chardonnay Standard Additions 38 15. Acid Recoveries in Chardonnay 40 16. Acid Recoveries in Pinot Gris 40 x 17. Range 41 18. Injection Repeatability of Retention Time 42 19. Injection Repeatability of Peak Area 42 20. Intra-Assay Repeatability of Retention Time 43 21. Intra-Assay Repeatability of Peak Area 43 22. LOD of Acids 44 23. LOQ of Acids 45 24. Linear Regression Equations at Varying Ethanol Concentrations 48 25. Linear Regression Equations at Varying Sugar Concentrations 48 26. Acid Concentrations in Pinot Oris During Fermentation 50 Xl LIST OF EQUATIONS Equation Description Page 1. Distribution Constant 1 2. Capacity Factor 2 3. Separation Factor 2 4. Theoretical Plate 2 5. Peak Resolution 3 6. Peak Symmetry 3 7. Column Dispersion Mechanism 3 8. Glucose Conversion to Ethanol 8 9. Acid Concentration Determination 20 10. Percent Recovery 25 xii A AA ACN APT ATP Avg B C CA CoA Cone. FTIR GC-MS GTP gIL H HPLC IR IS k' LIST OF ABBREVIATIONS Eddy diffusion Acidic acid Acetonitrile Absolute Pressure Transducer Adenosine triphosphate Average Longitudinal diffusion Mass transfer Citric acid Coenzyme A Concentration Fourier Transform Infrared Gas Chromatography Mass Spectrometry Guanosine triphosphate Grams per liter Theoretical plate height High Performance Liquid Chromatography Infra Red Internal standard Capacity factor xiii KD Distribution constant L Liter LA Lactic acid LOD Limit of detection LOQ Limit of quantitation M Molarity MA Malic acid MeOH Methanol mg Milligram mg/L Milligrams per liter min Minute mL Milliliter mUmin Milliliters per minute MLF Malolactic fermentation mm Millimeter N Theoretical plate NADH Nicotinamide adenine dinucleotide NB Theoretical plate at the base N1I2 Theoretical plate at the half height nm nanometer PDA Photodiode Array Detector PDAM I-pyrenyldiazomethane ppm Part per million xiv PTFE R %R R2 %RSD RI SA Std SID SI-FfIR TA TCA Temp u Poly-I, 3-dioxole-co-tetraflurorthylene Resolution Percent recovery Correlation Coefficient Percent Relative Standard Deviation Refractive Index Succinic acid Standard Standard deviation Sequential Injection Fourier Transform Infrared Tartaric acid Tricarboxylic acid cycle Temperature Dead volume Retention time Resolution time of component A Resolution time of component B Adjusted retention time of an analyte Linear velocity Ultraviolet Peak width at the baseline A Peak width at the baseline Peak width xv W1l2 Peak width at half height of the peak Micro liter XVI CHAPTER ONE INTRODUCTION A. Chromatography Chromatography is a technique in which components of a mixture are separated based on the rate they travel through a stationary phase with the help of a gaseous or liquid mobile phase. The stationary phase is in equilibrium with the mobile phase as shown below: A mobile phase ?:::> A stat. phase The partitioning between the stationary and mobile phase occurs based on intermolecular forces acting between solutes and the two phases. The equilibrium distribution KD, constant, is defined as: Ko= Cs Cm Equation 1. Distribution Constant Where Cs is the concentration of solute in the stationary phase and Cm is the concentration of the solute in the mobile phase. In reversed-phase chromatography the stationary phase is non-polar and the mobile phase is polar. An example is seen with CI8 or octadecylsilane phases that are composed of I8-carbon alkane chains covalently bonded to silica particles, where the mobile phase is typically composed of mixtures of water with a polar solvent such as an aliphatic alcohol. In these systems the more polar the analytes the longer they will stay in the stationary phase and the larger will be the KD. 1 The following parameters, defined below, characterize liquid chromatography: peak retention, capacity factor, separation factor, number of theoretical plates, resolution, peak symmetry, and column dispersion mechanism. Peak retention time of the analyte, tR, is the time from injection to analyte peak elution. Dead time, to, is the time from injection to detection of an unretained analyte. The adjusted retention time of an analyte, tRO, is tR minus to and is defined in Equation 2. k'= tR-to to Equation 2. Capacity Factor The capacity factor describes the interaction between stationary phase and analyte. The higher the value of k' the more effective the stationary phase is in retaining the analyte. The separation factor a is the ratio of two capacity factors (Equation 3). It is the measure of the spacing between two peaks, or the relative retention. k'a=_2 k'1 Equation 3. Separation Factor Number of theoretical plates, resolution, and peak symmetry define the column efficiency. Theoretical plate number is the ability of the analyte to flow through the column with minimum band broadening. This is usually expressed as the number of theoretical plates, N. Equation 4. Theoretical Plate 2 Where, N is the theoretical plate number, WB the peak width at the base, W1I2 is the peak width at the half height of the peak, and tR is the peak retention time. The theoretical plate number indicates the quality of the packed bed within the column. Peak resolution R is a measure of the separation between two adjacent peaks on a chromatogram. The resolution between two analyte peaks, A and B, is expressed as: R = 2(tRB - tRA ) wA +wB Equation 5. Peak Resolution where, tRB and tRA are the retention times of the components A and B, and WA, WB are the peak widths at the baseline for those components. A resolution factor of 1.5 indicates that the two components are baseline separated. Peaks resolved at a resolution factor of less than 1.0 cannot be quantified reliably. Peak symmetry is a measure ofpeak tailing and peak fronting. Itis measured at 10% of peak height. BPeak symmetry = A Equation 6. Peak Symmetry Where A is the distance from peak front to peak maximum and B from peak max to peak end. A symmetrical peak has a peak symmetry of 1, a fronting peak is <1, and a tailing peak >1. The column dispersion, or degree of band-broadening, is described by the Van Deemter equation as: H=A+ B +Cuu Equation 7. Column Dispersion Mechanism 3 Where, H is the theoretical plate height, A is the eddy diffusion coefficient, B is the longitudinal diffusion coefficient, C is the mass transfer coefficient, and u is the linear velocity of the mobile phase through the column. Band-broadening originates from the following three mechanisms. Eddy diffusion (A) results from the flow path (multiple path effect) of the solute molecules through the column packing material. Longitudinal diffusion (B) is due to simple diffusion, where high concentrations of solutes move spontaneously toward low concentrations. Mass transfer (C) affects band-broadening as solvent molecules cross the boundary between the stationary phase and mobile phase. The solute may be retained while other solvent molecules may travel with the stationary phase, causing unpredictable band-broadening that is dependent upon the speed of the phase transfer. B. HPLC High-perfonnance liquid chromatography, HPLC is a type of chromatography that uses a liquid as the mobile phase and a solid as the stationary phase. Depending upon the nature of the stationary and mobile phases, different types of components can be separated. In this technique, a mixture of analyte solutes dissolved in a solvent matrix is injected onto a packed column of the stationary phase under high pressure. Within the column, the mixture of solutes is separated into its individual components based on the interaction between the stationary phase and the mobile phase. In reversed-phase chromatography the stationary phase is packed with non-polar material and the mobile phase is a polar liquid such as water or methanol. In reversed-phase chromatography non-polar components are retained longer than polar components. An HPLC system configuration is shown in Figure 1. 4 Solvent A Solvent D Gradient Proportioning Valve Solvent B Solvent C Accumulator Piston Chamber Figure 1. The System Configuration C. Derivatization Derivatization of analytes for HPLC analysis is used primarily to introduce a detector-oriented tag. In HPLC, common methods of derivatization are used to improve the detectability of analytes by ultraviolet absorption or fluorescence. Acylation is the conversion of compounds containing active hydrogens into esters, thioesters, and amides through the action of carboxylic acids [1]. D. Organic Acids in Wine The separation and quantitation of organic acids in wine has been studied for many years. They playa major role in the taste and balance of a wine. Some are originally present in the grape while others appear during the alcoholic, or malolactic fermentation [2]. There are six major organic acids present in wine. They are acetic (AA), citric (CA), lactic (LA), malic (MA), tartaric (TA), and succinic (SA) acids. Currently, methods for determining carboxylic acids (organic acids) in wine include enzymatic analysis and HPLC. 5 LA occurs in fermented beverages but is generally not present in grape musts. It is produced from malic acid by bacterial action or by yeast during the alcoholic fermentation process. LA occurs as both L- and D- enantiomers, the origin of which depends upon the fermentation process. Alcoholic fermentation produces 180-400mg/L ofD- (-)-LA. L (+)- LA is produced by lactic acid bacteria during the malolactic fermentation process (MLF), a desired process for certain wines because it decreases the acidity while increasing the biological stability and mouth feel [3]. It is present in the concentration range of (0-5g/L). Malic acid (MA) is found in almost all fruits. In grapes it is an important indicator of the maturation process. In wines, its determination is required to monitor and control malolactic fermentation [4], where MA is converted to LA [2,3]. Citric acid (CA) is present in grape musts and wine in small amounts. It is sometimes added to the wine to avoid precipitation of the iron (Ill) salt. The evaluation of CA in wines is also of great interest for maintaining biological stability when it is present within certain concentration levels [3]. Succinic acid (SA) is produced in small amounts during alcoholic fermentation. The amount produced depends on the condition of fermentation and may range from (0 to 1.5g/L) [3]. Tartaric acid (TA) is one of the main organic acids in grapes and wine. Currently there is no enzymatic method available for its measurement [5,1]. The amount ofTA in grapes depends on the grape variety, the region, and the growing season. The concentration stays relatively constant during the growing season, but decreases during 6 alcoholic fermentation and during cold stabilization. It may undergo degradation by lactic bacteria to lactic and acetic acids [3]. Acetic acid (AA) is another important acid that must be monitored. Nitrogen content and sugar concentration of the must influence AA levels [6]. Oxidation by acetic acid bacteria results in the formation of acetic acid. AA is formed in small amounts during alcoholic fermentation and by lactic acid bacteria during malolactic fermentation. LA bacteria and certain wild yeasts, like Brettanomyces, Hansenula animala, and Kloeckera apiculata, can also produce high levels of acetic acid [6]. Ifacetic acid is present in large amounts, a wine is considered defective [3]. E. Introduction ofBiochemistry ofOrganic Acids Some of the compounds of interest in wine analysis can be found in the various biochemical pathways (e.g., the citric acid cycle) associated with malolactic fermentation and alcoholic fermentation. Alcoholic and malolactic fermentations are anaerobic processes that involve no net oxidation or reduction. The alcoholic fermentation pathway allows the yeast to make small amounts of adenosine triphosphate (ATP) from glycolysis by consuming pyruvate, which allows glycolysis to proceed. The net gain in ATP molecules by the glycolytic process is two moles for each mole of glucose utilized. This accounts for 24,000 calories ofenergy that is transferred to ATP [7]. Also, two molecules of the high-energy electron carrying compound, nicotinamide adenine dinucleotide (NADH) are produced. Glycolysis is the splitting of the six carbon sugars glucose and fructose to two molecules of the three-carbon molecule pyruvate. After glycolysis, pyruvate is converted into 7 ethanol by alcoholic fermentation or under aerobic conditions, it reacts with Coenzyme A (CoA) and enters the tricarboxylic acid cycle (TCA). The malolactic fermentation pathway is an offshoot of the TCA cycle. Malic acid is converted to lactic acid and carbon dioxide. An overview of the process of glycolysis and alcoholic fermentation showing glucose or fructose being converted into ethanol is as follows: C6H120 6 --> 2Cz~O + COz Glucose or Fructose ----+ 2 Ethanol Equation 8. Glucose Conversion to Ethanol In yeast, the pyruvate that enters the alcoholic fermentation pathway is first decarboxylated and the resulting acetaldehyde is converted into ethanol in a step that also oxidizes NADHto NAD+ [8]. This step replenishes NAD+ in the cell so that glycolysis can continue to transform glucose and lor fructose into pyruvate. The breakdown of glycerol from triacylglycerol degradation can be introduced in the glycolytic pathway by the formation of dihydroxyacetone phosphate, which can then enter the glycolytic pathway [7]. Glycerol----+ glycerol-3-phosphate ----+ dihydroxyacetone phosphate ----+ glycolysis. After pyruvate is combined with CoA it may enter the tricarboxylic acid cycle where citrate, malate, succinate, oxaloacetate and other products are formed. The tricarboxylic acid cycle (also called citric acid cycle or the Krebs cycle) starts after pyruvate is converted into acetyl-CoA. Through each tum of the TCA cycle, several molecules of high-energy electron carrying compounds are produced (NADH and FADHz). These high-energy electron-carrying molecules transfer their electrons into the 8 electron transport chain, which generates a proton gradient across the inner mitochondrial membrane of the mitochondria that in tum drives ATP production through oxidative phosphorylation. In the first reaction of the TCA cycle, acetyl-CoA reacts with oxaloacetate acid to produce citrate. The reaction is catalyzed by citrate synthase [7]. Succinic acid is also produced in the citric acid cycle. Succinyl-CoA contains a high-energy bond and uses this to synthesize a GTP rather than ATP. The reaction is catalyzed by succinyl-CoA synthetase and produces succinic acid and a free CoA [8]. Malic acid also is of importance in wine analysis. Its source can also be found in the citric acid cycle. After fumarate is formed in the cycle it accepts a water molecule in a reaction catalyzed by fumarase. The malic acid formed here is then oxidized in a reaction catalyzed by malate dehydrogenase resulting in the production of another NADH [8]. The biochemical pathways utilized by yeast during the process offermentation of glucose to ethanol and C02 are illustrated in Figure 2. 9 Cell Membrane M 2+ K+g , pyruvate kinase (PK) Glucose Fructose Glucose Fructose ATP ADP ~:~:~ (::~ose-6-P1hOl sP:::s~:::ucose isomer~ (PG ATP ADP\ L.. Fructose-6-phosphate (F6P) fructokinase (FK) 1Mg2+ phosphofructokinase (PFK) Fructose-l,6-bisphosphate (FBP) 1l IDdol~ GAP + DHAP'0 triose phosphate isomerase (TIM) NAD+Jlglyceraldehyde-3-phosphate dehydrogenase (GAPDH) NADH + H+ 1,3-Bisphosphoglycerate (l,3-BPG) ADP "'11 pho~fhoglycerate kinase (PGK ATP -.--/ Mg 3-Phosphoglycerate (3PG) phosphoglycerate mustase (PGM) 2-Phosphoglycerate (2PG) H20 Jl enolase Mg'+ Phosphoenolpyruvate (PEP) ADP '\1 ATP~ Pyruvate cont. / 1 cont. cont. 10 Pyruvate L-Lactate 1 Pyruvate lactate I(NADHV CoASH+NAD+ dehydrogenase (LDH~ Nyruvate dehydrogenase NAD+ COz+ NADH Acetyl- CoA Pyruvate pyruvate V" decarboxylase +~ COz Acetaldehyde alcohol I'CADHdehydrogenase + NAD+ Ethanol aconitase + NADH+It NAD ~ Oxaloacetat lr~me citrare malate synatha<;e H20 l' . de.hydrogenase --/ fumarase Fumarate CoASH citrate \( ~H20 dehydration cis-Aconitate succinate dehydrogenase aconitase rehydration Isocitrate NADH+H+Oxalosuccinate CO2 isocitrate dehYd'Og~ a-Ketoglutarate ~ succinyl-CoA synthetase ~e(CoASH J a-ketoglutarate dehydrogenase GDP+Pi Figure 2. Biochemical Pathway of Organic Acids 11 CHAPTER TWO HISTORICAL REVIEW Several analytical methods have been used to separate and quantify the organic acids in wine, but to date complete validation has not been published. This chapter will outline a survey of some of the work done in the analysis and quantitation of organic acids and other components in wine using HPLC. Frayne [9] used a dual cation exchange column on an HPLC with UV and refractive index (RI) detection to analyze the major organic acids, sugars, and alcohols in wine. No sample preparation was needed, and the separation was achieved in 40 minutes. Schneider and coworkers [10] developed a method using a cation exchange resin to analyze citric, tartaric, malic, lactic, and acetic acid in wine. The acids eluted within 12 minutes. Although this was faster than Frayne's separation, succinic acid could not be determined due to coelution with shikimic acid. Tusseau and coworkers [11] used reversed phase chromatography with UV detection to separate organic acids, but only five acids were analyzed. Therefore, acid peaks were resolved but further research was needed to include other acids. Nimura and coworkers [12] prepared 1-pyrenyldiazomethane (PDAM), a new fluorescent labeling agent for carboxylic acids. PDAM readily reacted with carboxylic acids at room temperature without a catalyst to give an intensely fluorescent ester. Although this method was developed for the analysis offatty acids, itmay also be applied to wine acids. 12 Allenmark, [13] developed a method using N-(9-acridinyl)-bromoacetamide derivatives. A phase-transfer-catalyzed esterification was carried out with various carboxylic acids to give highly fluorescent esters. These were separated using reversed phase liquid chromatography in 40 minutes. Calull and coworkers, [14] used ion-exchange chromatography and refractive index detection to analyze sugars, lactic, succinic and acetic acid in wine. Because significant dilution of the samples was necessary the acids could not be quantified. Another ion-exchange method, developed by Lopez [15], was used to determine the major organic acids, sugar, glycerol, and ethanol in wine simultaneously in 45 minutes using UV detection, and refractive index (RI) for malic acid and fructose. Vonach [16] reported a method using a resin-based ion-exchange stationary phase coupled with FfIR detection to determine organic acids. Complete separation was achieved in 20 minutes. The method took advantage of the IR absorption at different wave numbers by different functional groups. The organic acids absorbed at 1260 cm-I , the carbohydrates at 1050 cm-I . Schindler [17] developed a method based upon Sequential Injection Fourier Transform Infrared Spectrometry (SI-FfIR). This method allowed the simultaneous determination of all the major organic components in wine, such as: glucose, fructose, glycerol, citric, tartaric, malic, lactic, acetic acids and ethanol in 3 minutes. Succinic acid was not included in the study. Using this method the results for some organic acids (acetic acid and tartaric acid) were relatively poor, which was explained by the low concentration and the uncharacteristic absorbance of these acids in the investigated spectral range. The relative standard deviation of a single sample was smaller than 8% 13 for all components, except for acetic acid at 30%, and for ethanol, which was less than 2%. The major drawback of this method was that the accuracy for determination of concentration levels was diminished due to the FfIR analysis. The author concluded that FfIR spectroscopy was best suited for identification and screening purposes rather than validation. The calibration procedure was time-consuming because they used seventy two calibration solutions, with more being desirable for accurate analysis. When analyzing synthetic samples, only six components were used for the calibration and nine samples were analyzed. In order to develop a complete method for all compounds of interest, an HPLC method was developed. Another method was developed by Giumani et al. [18] for analyzing wine acids using GC-MS. The formation of phenacetyl esters through derivatization allowed the determination of 20 different acids, many of which had not been recorded previously, in 18 minutes. In liquid chromatography, the only useful method to determine small amounts of carboxylic acids was based on pre-column derivatization with the formation of an ester group, which strongly absorbs UV radiation and / or fluoresces [13]. Borch [19] separated long fatty acids as phenacyl esters by HPLC. The method described allowed a rapid and convenient way to derivatize and subsequently analyze fatty acid mixtures on the microgram scale and gave a high degree of resolution in most cases. This method could be further applied to complex samples matrixes such as wine. Caccamo and coworkers [20] developed a method by derivatizing the major organic acids in wine with phenacyl bromide and crown ether in acetone including lactic, acetic, tartaric, succinic, malic, and citric acids. The major organic acids were 14 determined as phenacyl esters using reversed phase HPLC in sixteen minutes. Recoveries were 95% and higher. The derivatizing agent worked best in a buffered solution, as 0.08 M phosphate buffer. 15 CHAPTER THREE STATEMENT OF THE PROBLEM Wineries need to monitor the concentrations of tartaric, malic, lactic, citric, and acetic acid during the winemaking process to ensure the quality of their wines. Tartaric, citric, and malic acid may all be added legally to wine. The quantitation of citric acid is important as wines destined for Europe must comply with regulations that specify that the citric acid level in wine must be below 1.0 gIL. Malic acid levels must also be monitored closely as many wines undergo a process known as malolactic fermentation, which "softens" the wine as malic acid is converted to lactic acid. Acetic acid is a by-product of the primary and secondary fermentation processes. Acetic acid can also be formed by acidic acid bacteria and other microorganisms. High levels of acetic acid are indicative of high volatile acid, which is considered a wine defect [13]. Many articles have been published on the separation and quantitation of organic acids in wine, but none has been fully validated. The primary purpose of this research was the modification and validation of the method by Caccamo et al. [20]. To accomplish this, it was necessary to optimize the mobile phase and the separation conditions. Methods from the literature were used to derivatize the major carboxylic acids to esters allowing detection at 254 nm. This approach avoids interference from other wine components, which may occur at the commonly used detection wavelength of 210 nm. The method validation includes studying standard curve linearity, accuracy and precision, usable concentration range, analyte detection and quantitation limits and robustness. After validation the method was used to analyze Pinot Gris and Chardonnay wine samples from YSU's Enology laboratory. 16 CHAPTER FOUR MATERIALS AND EQUIPTMENT A. Reagents All reagents used in this research were of the highest grade available and are listed in Table 1 along with relevant purity and source data. Standard Percent Purity Source Lactic acid 85% Fisher Scientific Glacial Acetic acid 100% Fisher Scientific L-(+)-Tartaric acid 99.8% Fisher Scientific DL- Malic acid 99+% Fisher Scientific Succinic acid 99.5+% Aldrich Citric acid 99.5+% Fisher Scientific Methylmalonic acid 99% Aldrich Mobile Phase Reagents Ethyl alcohol 200 Proof Pharmaco Methanol HPLC grade Burdick and Jackson Acetonitrile Optima grade Fisher Scientific Water Purified In-house (Modulab) Derivatizing Reagent Dicyclohexano-18-Crown 6 98% Sigma 2 Bromoacetophenone 98% Aldrich Sodium phosphate anhydrous A.C.S. J.T. Baker Sodium hydroxide 98.4% Fisher Scientific Acetone Pesticide grade Fisher Scientific Hydrochloric acid A.C.S. Fisher Scientific Potassium dihydrogen phosphate 99.14% Fisher Scientific Test Mixture Reagents Uracil Eastman Phenol Mallinckrodt N,N-Diethyl-m-toluamide 98% Aldrich Toluene A.C.S. Fisher Scientific Sensitivity Reagents D(-) Fructose Sigma Dextrose anhydrous Fisher Scientific Table 1. Reagents Used in This Research 17 Source Supplier Aldrich Alltech Burdick and Jackson Eastman Fisher J.T. Baker Mallinckrodt Pharmaco Sigma Aldrich Chemical Company Inc. (Milwaukee, WI, U.S.A.) Alltech Associates, Inc. (Deerfield, IL, U.S.A.) Burdick and Jackson (Muskegon, MI, U.S.A.) Eastman Kodak Co. (Rochester, NY, U.S.A.) Fisher Scientific (Fairlawn, NJ, U.S.A.) J.T. Backer (Phillipsburg, N.J, U.S.A.) Mallinckrodt Chemical Works (St. Louis, MO, U.S.A.) Pharmaco (Brookfield, CT, U.S.A.) Sigma Chemical Co. (St. Louis, MO, U.S.A.) B. Equipment The HPLC system used for this work was a Waters 2695 Separations module (Milford, MA, U.S.A.), equipped with an absolute pressure transducer (APT) and an online degasser. Itincluded an LC-pump, an eluent mixing chamber, an auto sampler, a 100-~L loop injector, a column oven heater, a 996 Photodiode Array Detector (PDA), and Millennium 3.2 software. Solvent Bottle Tray Detector Drip Tray ---11+----+------,/- Syringe Access DoorFront Panel Display _____tt""r-----!".and Keyboard Solvent Delivery Tray Access Door - ~~~?~-Jl_~----+---+- Column Heater ModuleFloppy Disk Drive Sample Compartment -+---l Access Door Figure 3. HPLC System 18 CHAPTER FIVE METHODS A. Grape Must Grape musts were obtained from regional wineries. Twenty-gallons of Chardonnay must was provided by Markko Vineyards, Conneaut, Ohio, from their October 4th, 2001 harvest. Twenty-gallons of Pinot Gris must was provided by the Kingsville Grape Research Branch (OARDC), Kingsville, Ohio, from their October 1st, 2001 harvest. B. Fermentation Musts were divided into 12, 3-gallon glass fermentation vessels, 6 each for Chardonnay and Pinot Gris. Three each for Chardonnay and Pinot Gris were control fermentations (C-1, C-2, C-3), and three each were spontaneous fermentations (S-l, S-2, S-3). Fifty ppm of SOz was added to the control vessels to inhibit the growth of indigenous yeasts; none was added to the musts for the spontaneous fermentations. The musts were allowed to settle by gravitation for 24 hours at 38?C. The clear supernatant was then separated from the sediments and controls were inoculated with commercial freeze-dried yeast (Prise de Mousse, EC 1118) at one gram per gallon. No additional yeast was added to the spontaneous fermentations. The vessels were kept at 22?C during the fermentation. After the fermentations were completed the wine was racked off, the amount offree SOz was adjusted to 40 ppm, cold stabilized at 4 ?c for 3 weeks, and bottled. 19 c. Wine Sampling Each day 10 mL of wine was collected from the center of the vessel into a plastic vial with screw cap and stored frozen at -80?C until the analysis was performed. D. Wine Analysis Wine samples from both the spontaneous and inoculated fermentations were analyzed every fourth day. The wine samples were thawed, thoroughly mixed, and centrifuged for ten minutes to remove yeast and other particulate matter. For the analysis one mLof wine was used and was spiked with an internal standard. The mixture was then diluted 1 to 10 with a mixture of 12% ethanol in water. The acid concentrations in each sample were calculated using Equation 9. (LASample)Area AcidConc. =__ xStd.#5LAconc.g / L (LAStd.#5)Area IS(Std.#5) Equation 9. Acid Concentration Determination E. Derivatization ofOrganic Acids The purpose of derivatization of the major organic acids in the wine samples was to enable their detection using UV spectrophotometry at 254 nm. The acids were derivatized to form phenacyl esters, as shown in Figure 4. 20 Br o Br + ... Br o o o--.lLR Figure 4. Phenacyl Ester Formation The derivatization reaction was carried out in a PTFE (poly-I, 3-dioxole-co- tetrafluoroethylene) lined screw capped glass-tube containing 20 I-lL ofwine sample, 80 I-lL ofphosphate buffer, and 300 I-lL ofderivatizing reagent. The derivatizing reagent was prepared by dissolving phenacyl bromide and dicyclohexane-18-crown-6 in acetone to give concentrations of 30 and 1.5 gIL, respectively [20]. The buffer was prepared by adding 2.7 g ofKHZP04 and 2.84 g NazHP04 in 1 L of deionized water to give a 0.08 M phosphate buffer of pH 6.8. Wine samples were diluted 1 to 10 using a mixture of 12 % ethanol in water. After thorough mixing the mixture was heated for 45 minutes at 95?C. After cooling the mixture to room temperature, a 15.0 ilL aliquot was taken and subjected to chromatographic analysis. F. HPLC Method The chromatographic separation was performed using an 250 mm x 4.6 mm Alltima CIS packed column with a 5-micron particle size. The analytical column was preceded by a 7.5 mm x 4.6 mm CIS guard column. Both the analytical and guard columns were supplied by Alltech Associates (Deerfield, IL, U.S.A.). The following mobile phases were used: solvent A: water, solvent B: acetonitrile-methanol (50:50), and solvent C: acetonitrile. A linear gradient was applied by increasing solvent B (50:50) (methanol: acetonitrile) from 45 to 90% in 25 min., followed by a lO-minute wash with 21 100% acetonitrile. The flow rate was 1.000 ml/min. The oven temperature was 30.0 ?C and UV detection was set at 254 nm. G. Standards Preparation Standards were prepared by dissolving the seven acids of interest in 12 % ethanol in water to give concentrations as shown in Table 2. The standards were then diluted 1 to 10 with 12% ethanol in water. This dilution was applied to all samples to eliminate any potential problems due to the derivatizing agent becoming the limiting reagent in the derivatization reaction. All other standards were weighed individually into a volumetric flask and then diluted 1 to 10 with 12 % ethanol in water. All standards were adjusted to pH 3. For weights less than 80 mg a microbalance was used; otherwise a standard analytical balance was used for all weighings. All glassware was volumetric. Standards LA AA TA MA SA CA IS (2!L) (2!L) (2!L) (2!L) (2!L) (2!L) (2!L) Std#l 0.2528 0.2510 0.6347 1.6166 0.3811 0.1805 1.0004 Std#2 0.4944 0.49664 1.2437 3.268 0.7334 0.2419 1.0004 Std#3 1.0300 1.0057 2.5763 6.488 1.5074 0.49940 1.0004 Std#4 1.5328 1.4112 4.292 9.944 2.2672 0.73792 1.0004 Std#5 1.9856 1.9729 5.428 13.01 3.0509 1.0047 1.0004 Std#6 2.4465 4.2110 5.216 14.43 3.0973 1.1828 1.0004 Table 2. Concentrations of Standards 1 to 6, Undiluted 22 H. Method Validation Method validation is the process of testing an analytical method for specificity, linearity, accuracy, precision, usable concentration range, detection limit, quantitation limit, sensitivity and robustness. Samples used in the specificity study were Standard #5 and Chardonnay diluted 1.4 to 10. Three runs ofeach sample were analyzed in triplicate for retention time, peak resolution, theoretical plate count, peak tailing factor and capacity factor (see Figure 5). To facilitate the identification of peaks, each acid was individually spiked into the sample at a relatively high concentration. Comparison of retention times between spiked and unspiked samples gave verification ofpeak identity. Peak resolution was calculated according to Equation 5 in Chapter 1. 0.399 t 0.24 .1' 0.20 0.054 0.004 3 oE--+--10 ~ 40 --+---;~ o o 3 t ~il. 'iii..c. 0.6 ~ c- '00.5 c o .~ u. 0.0 Figure 5. Calculation of Peak Symmetry, Theoretical Plate Count, and Resolution 23 Theoretical plate count is a measure of column performance. The number of plates is related to the ability of an analyte band to flow through the column with a minimum amount ofband broadening (diffusion). The plate count was calculated according to Equation 4 in Chapter 1. Asymmetry of a peak can occur in two different ways: fronting or tailing. Based upon considerations of the solute distribution coefficient, D, peak-tailing results from Langmuir type absorption isotherms, where the analyte is partitioned more into the stationary phase as the concentration of analyte in the migrating band of molecules increases. Fronting peaks result from anti-Langmuir isotherms, where the analyte partitions more into the mobile phase as concentrations within the band increases. Capacity factor is related to the distribution coefficient, and may be thought of as the absorbing quality of stationary phase relative to the mobile phase. Capacity factor was calculated using Equation 2 in Chapter 1. The evaluation oflinearity was performed using six standard solutions ranging from 50% to 150% of the target analyte concentrations. Each concentration was analyzed a minimum of three times. The linearity range was determined by the correlation coefficient and the "y"-intercept of the linear regression line. A correlation coefficient of > 0.999 was generally considered as evidence of acceptable linearity. Along with these mathematical parameters a visual examination of the calibration curve was required. The accuracy of a method can be measured in several ways. One way is based on recovery as determined by spiking analytes into a blank matrix. Another approach is the standard addition technique, which is used when a blank sample matrix is not appropriate. 24 In this project, accuracy was determined by calculating the recovery of each acid using Equation 10. %R = PA mix xlOO% (PA Std. #5 +PA wine)/2 Equation 10. Percent Recovery The recovery study was set up as shown below: ~ Vial #1. 0.50 ml wine plus 0.50 ml Std. #5; diluted 1:10 with 12% etOH in water ~ Vial #2. 1.00 ml Std. #5 diluted 1:10 with 12% etOH in water ~ Vial #3. 1.00 ml wine sample diluted 1:10 with 12% etOH in water All vials contained equal concentrations of internal standard. Samples were analyzed on three different days in triplicate resulting in 9 injections. The peak areas were averaged. The standard addition was performed with Pinot Oris and Chardonnay wines. To each variety seven different acids were added in increasing concentrations, (see Table 3 and 4). Each individual acid was extrapolated to the negative x-axis to determine the individual acid concentration as an absolute value for each variety of wine. The concentration ofeach acid in wine was determined by calculating the intercept with the x- axis. 25 Std. Add. LA AA TA MA SA CA IS Chardonnay (gIL) (gIL) (gIL) (gIL) (gIL) (gIL) (gIL) Std#l 0.331 0.1889 0.209 0.319 0.203 0.103 1.0004 Std#2 0.605 0.421 0.4038 0.611 0.4132 0.206 1.0004 Std#3 0.841 0.742 0.617 0.901 0.615 0.412 1.0004 Std#4 1.073 0.925 0.808 1.202 0.809 0.608 1.0004 Std#S 1.321 1.16 1.01 1.505 1.04 1.008 1.0004 Table 3. Standard Addition in Chardonnay Std. Add. LA AA TA MA SA CA IS Pinot Gris (gIL) (gIL) (gIL) (gIL) (gIL) (2!L) (2!L) Std#l 0.347 0.200 0.217 0.299 0.210 0.104 1.0004 Std#2 0.589 0.423 0.435 0.605 0.411 0.215 1.0004 Std#3 0.866 0.655 0.6095 0.9114 0.617 0.404 1.0004 Std#4 1.075 0.937 0.8111 1.206 0.824 0.607 1.0004 Std#S 1.324 1.154 1.014 1.517 1.014 1.018 1.0004 Table 4. Standard Addition in Pinot Gris The range of an analytical method is the concentration interval over which accuracy, linearity, and precision are valid. Generally, the range is determined by using the linearity and accuracy data. In this project, the range was determined to be from the limit of quantitation to the highest standard concentration (Standard # 6). Precision is the amount of scatter in the results obtained from multiple analyses of the same sample. One type of precision examined was instrument precision, or injection repeatability. A minimum of 10 injections of one sample was required to test the performance of the chromatographic system. A second type of precision examined was the analyst precision, or the intra-assay repeatability. The analyst repeatedly analyzed 26 independently prepared samples. Ten samples were analyzed and the relative standard deviation calculated. Limit ofDetection (LOD) was the lowest analyte concentration that produced a response detectable at three times the noise level. The limit ofquantitation (LOQ) was the lowest analyte concentration that could be precisely and accurately measured. LOQ is often calculated as the analyte concentration that gives a signal to noise ratio of ten. For the determination ofLOD and LOQ the lowest standard concentration, Standard #1, was diluted to achieve appropriate levels, see Tables 5 and 6. AA and SA were diluted 1 to 10 while all other acids were diluted 1.4 to 10. The signal was calculated as peak areas ofthe analyte minus the peak area blank signal that may have been co-eluting with the peak of interest. The peak areas of the blank signals were averaged from 50 blank samples. The noise was averaged from different dilutions and on different days of analysis. 1.4-10 LA TA MA CA dilution dilution dilution dilution LOn 2-10 0.5-10 1-10 Std#l LOQ Std#l 1-10 3-10 Std#4 Table S. LOD and LOQ Using Standards That Were Diluted 1.4 to 10 1.0-10 AA SA dilution dilution LOn Std#2 3-10 LOQ Std#3 Std#2 Table 6. LOD and LOQ Using Standards That Were Diluted 1 to 10 27 The analysis of the standard stability was performed using freshly prepared Standard #5 and comparing it to old standards, over a period of 43 days, both at room temperature and at refrigerated temperature (4?C). Standards that were derivatized on day one were re-used and compared to standards that were derivatized fresh, daily. The stability was evaluated daily for the first 8 days, and thereafter twice a week until decomposition was observed. Ruggedness tested the use of the analytical method by multiple analysts, multiple instrumentation, on multiple days in one laboratory to determine the method robustness. Different sources ofreagents and multiple lots of columns should be used. The robustness of a method is the ability to remain unaffected by small changes in organic solvent, pH, mobile phase composition, buffer concentration, temperature, and injection volume. These factors can be evaluated one at a time or simultaneously as fractional experiments. In this project column temperatures ranging from 25?C to 45?C and various solvent compositions were evaluated. Sensitivity tests consisted of ethanol and sugar additions to the standards and determining the effect on the slope. The ethanol concentrations studied ranged from 0 to 15 %. The second sensitivity test consisted of adding sugar (1:1) (fructose: glucose) in concentrations ranging from 0 to 250 gIL. All standards were diluted 1.4 to 10 with 12% ethanol water. The slopes of the lines in the ethanol and sugar addition study were analyzed for all six acids. 28 CHAPTER SIX RESULTS AND DISCUSSIONS A. Specificity Specificity is the ability ofan analytical method to accurately measure the analytes in the presence ofall components. Three runs ofeach sample were analyzed in triplicates for retention time, resolution factor, theoretical plate count, tailing factor and capacity factor as shown in Table 7. A resolution factor of 1.15 or better was achieved for all acids meeting the minimum quantitation requirement of>1. For perfect resolution between adjacent Gaussian curves the resolution factor should be grater than or equal to 1.5. The minimum theoretical plate count for an analytical column to be considered good should generally be in the range 1000 to 10,000 and depends upon the analytical requirements ofthe method. Itwas found that for NB was greater than 14691 and N1/2 greater than 12761 for all acids. Asymmetry can occur in two different ways, fronting or tailing. Tailing is the analyte retaining more on the column and fronting is the analyte retaining less on the column. Acids were added individually to wine to determine their retention times (see Figure 6). 29 LA AA TA MA SA CA Rt (min.) 7.727 9.810 12.573 8.742 17.530 21.026 Resolution 1.43 1.15 1.29 2.27 Not needed 1.33 NB (plates/m) 14691 22431 37415 56458 67446 102263 N 1/2 (plates/m) 12761 14768 32167 29282 40510 91062 Symmetry 1.18 1.38 1.20 1.20 1.37 1.31 Capacity factor 4.15 5.54 7.38 8.742 10.68 13.017 Not needed: no other components were in the vicinity ofsuccinic acid Table 7. Specificity Test Using Wine Samples Figure 6. Overlaid Chromatogram ofWine Samples Spiked With Acids 30 In order to be valid a method needs to demonstrate specificity, which means that it will be able to accurately measure the analyte response, and only the analyte response, in the presence ofa sample matrix containing the analyte. Once resolution is acceptable the chromatographic parameters, such as column type, mobile phase, composition, flow rate, and detection mode are considered to be valid. Figures 7 and 8 show the separation ofthe acids in two different wines, Pinot Gris and Chardonnay. The chromatogram ofa blank shown in Figure 9 illustrates a small interference with lactic, acetic, succinic, and metylmalonic acids. 0.50-r-r---------r--.-~-------------. ...- 0.40 0.30 =>? 0.20 0.10 Figure 7. Chromatogram ofPinot Gris Wine 31 N ~<:) .......? 0.25? 0.20 0.15 0.10 0.05 O. 00 +---,---r---r-......--.~_r_~,___,___.___r_..___.___r___r__..___r___::;::..._r=__r___=;::....:::;::.._,__......___r=....:::;=_,___.__I 6.00 Figure 8. Chromatogram ofChardonnay Wine O. 50"TT"'"T---.,...,..------,-~--------r------, 0.40 0.30 8.00 10.00 12.00 14.00 16.00 18.00 20.00 Minutes O. 00 +--T-,.--r-=r---r~...,.._,--r-_,__r__r~...,..._........,..---r-.,...._,___.___r=~_;=._r:_........,.._,...__r___f 6.00 0.10 :=J? 0.20 Figure 9. Chromatogram ofa Blank 32 Several different solvent compositions and gradients, as shown in Table 8, were tested to overcome various interferences. The best composition was found to be solvent A water, solvent B (50:50) methanol: acetonitrile and solvent C acetonitrile, at 1.000 mUmin flow rate. The solvent B was increased from 45% to 90% in 25 minutes followed by a 10-minute was with 100% acetonitrile. This resulted in a 35-minute separation where all analyte peaks were resolved. B. Derivatizing Agent Study The time of derivatization and the concentration of derivatizing agent were studied using the parameters outlined below. Table 9 shows the effect of changing concentrations of phenacyl bromide and dicyclohexano-18-crown-6. Phenacyl Dicyclohexano- bromide (gIL) 18-crown-6 (gIL) A 30.96 3.020 B 34.40 4.500 C 30.60 7.500 D 32.70 15.60 E 32.80 30.60 Table 9. Effect ofDerivatizing Agent Concentrations on Peak Area and Shape 33 Time Gradient start Time Gradient ProceedingTime Gradient ProceedingFlowrateTemp start A:H 2 0(min)A:H 2 0(min)A:H 2 0(m1/min) eC) (min)B: MeOHlACN B:MeOHlCAN B:MeOHlACN C:ACNC:ACN C:ACN AIB/CAIB/CAIB/C 1. 0 80120 10 90/10341.00030.0 2. 070/30 15 10/90351.00030.0 3. 060/402210/904010/901.00030.04.060/402242/584510/901.00030.05.060/403020/8032.510/901.00030.0 6. 060/103510/901.00030.0 7. 060/40 14 49/51 18 47/531.00030.0 8. 070/30 17 10/901.00030.0 9. 070/303410/903810/901.00030.0 10. 060/403010/901.00030.0 11. 055/453010/901.00030.0 12. 095/53065/351.00030.0 13. 0100/03070/301.00030.0 14. 090/103070/301.00030.0 15. 095/5895/53065/351.00030.0 16. 0100/035 75125 1.00030.0 17. 098/23085/151.00030.0 18. 0 75125 3025/751.00030.0 19. 070/303010/901.00030.0 * 20.065/353010/901.00030.021.060/40 25 10/90350/0/1001.00030.022.050/50 25 10/90350/0/1001.00030.023.055/45 25 10/90350/0/1001.00030.0 *Facesealwash,plungersealwaschanged. Table 8.SolventCompositionsandGradientsTestedtoOptimizetheHPLCSeparation A: H 2 0, B: MeOHlACN, C: ACN34 Based on this study it was concluded that the concentrations of these reagents had no effect on the reaction. The concentrations used in the article by Caccamo and coworkers [20] were applied. Various derivatizing times at constant derivatizing agent concentrations were studied as shown in Table 10. It was confirmed that the ideal derivatization time was 45 minutes, as stated in the article. Broad chromatographic peaks appeared when heating the sample for more than 45 minutes, indicating that an undesirable by-product was formed, or that the derivatives were decomposing. Time (min) Derivatizing agent Temp. Peak a,phenacylbromide(gnL) (OC) Observation b, dicyclohexano-18- crown 6 (WL) 30 a,30.32, b,1.508 30.0 Good 45 a,30.32, b,1.508 30.0 Ideal 90 a,30.32, b,1.508 30.0 Wide peaks 120 a,30.32, b,1.508 30.0 Wide peaks 180 a,30.32L, b,1.508 30.0 Wide peaks 240 a,30.32, b,1.508 30.0 Wide peaks Table 10. Effect ofDerivatizing Time on Peak Area and Shape Various derivation times were studied with double the phenacyl bromide concentration while keeping the catalyst (dicyclohexano-18-crown-6) constant. It was concluded that the derivatization time was ideal at 45 minutes as seen in the previous experiment (see Table 11). 35 Time Derivatizing agent Temp. Peak (min) a,phenacyl bromide (gnL) (OC) Observation b, dicyclohexano-18 - crown- 6 (2!L) 45 a,60.1O, b,3.120 30.0 Ideal 90 a,60.1O, b,3.120 30.0 Wide peaks 180 a,60.1O, b,3.120 30.0 Wide peaks 135 a,60.1O, b,3.120 30.0 Wide peaks Table 11. Effect ofDerivatizing Time at Double the Phenacyl Bromide Concentration on Peak Area and Shape C. Linearity and Standard Addition Linearity is the determination of the concentration range where the analyte is potted against concentration and it shows that the calibration curve is linear. Table 14, Figure 10 and Figure 11 show not only the linearity, but also matrix effects by comparing the standard addition curve ofPinot Oris and Chardonnay with the calibration curve. These three regression curves show no significant matrix effect in the samples when diluting the wine 1 to 10. Lactic acid was chosen as an example. In this project linearity was evaluated using six standard solutions varying from LOQ to the highest concentration where the curve was found to be linear. Each concentration was analyzed in triplicates, averaged, and plotted against the signal. Various dilutions were studied to determine which one yielded the best correlation coefficient (see Table 12). 36 Dilutions LA AA TA MA SA CA Corr. Corr. Corr. Corr. Corr. Corr. CoetT. CoetT. Coeff. Coeff. Coeff. CoetT. 1.4 to 10 0.969 0.995 0.990 0.998 0.943 0.999 5.0 to 10 0.997 0.000 0.988 0.996 0.973 0.968 3.0 to 10 0.995 0.965 0.994 0.957 0.983 0.939 1.0 to 10 0.999 1.000 1.000 0.999 0.998 0.998 Table 12. Dilution Studied for Determining and Analyzing the Correlation Coefficient For various dilution injections, the acid peak areas were averaged to determine the derivatizing agent effect and were plotted against the dilutions. The relative standard deviation was high for some acids, indicating the necessity of dilution. (see Table 13). Dilution: LA AA TA MA SA CA 0, 1.33x, 2x, 4x, 7.14x Average 3090363 502211 12565028 4805016 633463 976730 %RSD 9.9 81.2 8.0 5.8 88.9 4.5 Table 13. Dilution Study and Relative Standard Deviations 37 AcidsEquations of linear R~ Equationlinearregression R~ Equationlinear R~ regression,standardsPinotGrisStandardregressionChardonnayAdditionStandardAdditionLA Y=2538679x+9640 0.995 Y=2490900x+135080 0.998 Y=2377004x+83841 0.992 AA Y=5719487x+417947 0.994Y=6089000x+4873400.990Y=6322435x+658662 0.981 TA Y=4902932x-63718 0.992Y=5162300x+3364000 0.979 Y=4313847x+1177499 0.978 MA Y=5035905x+74928 0.993 Y=3214900x+9185000 0.903 Y=4542232x+2190626 0.922 SA Y=5466538x+603885 0.990Y=3450800x+l093400 0.977 Y=5584189x+1131267 0.983 CA Y =4313466x + 7283 0.976Y=3372100x+606050 0.997 Y=478838x+134351 0.996 Table14. LinearRegressionEquationforStandardSolutions,PinotOrisandChardonnayStandardAdditions 38 LA Linearity and Std.Add. of Pinot Gris y = 2490952.0114x + 135088.8433 R2 = 0.9975 800000 600000as ! 400000< 200000 o o 0.1 0.2 Cone. giL ? P.G.Std.Add. ? Standards 0.3 Y= 2538678.94467x + 9639.88470 R2 = 0.99461 Figure 10. Lactic Acid Linearity Standards and Standard Addition in Pinot Gris I '-~'---"---'---'''-------'' y =2.3770E+06x + 8.3842E+04I LA Linearity and Std. Add. R2 =9.9228E-01 of Chardonnay 800000 .. 600000 1100000? 200000 o o 0.1 0.2 0.3 ? Chardonnay Std.Add. ? Standards Cone. gIL y =2.5387E+06x + 9.6399E+03 R2 =9.9461 E-01 ~~---~---- ._-....~~._._~I Figure 11. Lactic Acid Linearity Standards and Standard Addition in Chardonnay 39 D. Accuracy Accuracy is the closeness between the value accepted as true value and the value found. The relative standard deviations in both Pinot Oris and Chardonnay were found to be 95.3-111.4% or better (see Tables 15, and 16). They were calculated from the recovery study according to Equation 1, in Chapter 5. The recoveries were performed with Pinot Oris and Chardonnay. Chardonnay LA AA TA MA SA CA Day 1 116.5 114.3 109.1 90.6 111.5 116.7 Day 2 101.5 101.6 112.5 100.9 105.5 108.7 Day 3 107.0 118.2 96.80 94.30 104.2 93.00 Avg. runs: 108.3% 111.4% 106.1% 95.3% 107.1% 106.1% %RSD 7.0 7.8 7.8 5.5 3.6 11.4 Table 15. Acid Recoveries in Chardonnay Pinot Gris LA AA TA MA SA CA Day 1 104.4 105.6 107.2 107.4 107 109.5 Day 2 92.0 95.4 96.1 96.0 106.1 105.2 Day3 95.3 102.4 103.2 96.0 102.3 89.9 Avg. runs: 97.2% 101.1% 102.2% 99.8% 105.1% 101.5% %RSD 6.6 5.2 5.5 6.6 2.4 10.1 Table 16. Acid Recoveries in Pinot Oris 40 E. Range The range ofan analytical procedure is the interval between the upper and lower concentrations ofanalytes in a sample for which precision, accuracy and linearity are valid. The range was detennined to be from the LOQ to Standard #6 (see Table 17). Range LA AA TA MA SA CA 0.1700 0.1006 0.0427 0.0528 0.0733 0.0864 to to to to to to 2.446 4.211 5.216 14.43 3.097 1.183 Table 17. Range F. Precision Precision is the amount ofscatter between a series ofmeasurements obtained from multiple samplings. For the injection repeatability the relative standard deviation ofthe retention times was < 0.80% and ofthe peak areas was < 5%, (Tables 18, 19 and Figure 12). J ~C'T---r-I------,m--------m-m--------------, J t I \ Figure 12. Overlaid Chromatogram from Injection Repeatability 41 LA AA TA MA SA CA RT RT RT RT RT RT 1. 7.787 9.899 12.741 14.824 17.770 21.315 2. 7.834 9.959 12.792 14.868 17.811 21.325 3. 7.849 9.973 12.797 14.870 17.814 21.331 4. 7.856 9.986 12.839 14.939 17.889 21.414 S. 7.907 10.050 12.889 15.003 17.926 21.420 6. 7.908 10.055 12.929 15.012 17.969 21.512 7. 7.908 10.055 12.932 15.019 17.969 21.512 8. 7.909 10.062 12.939 15.040 17.989 21.520 9. 7.971 10.123 12.949 15.076 17.993 21.530 10. 7.985 10.169 13.009 15.200 18.149 21.734 Av~. 7.894 10.0331 12.882 14.984 17.9279 21.461 STD 0.0610 0.08023 0.08542 0.11288 0.1122 0.1284 %RSD 0.77 0.80 0.66 0.75 0.63 0.60 Table 18. Injection Repeatability of Retention Time Area Area Area Area Area Area LA AA TA MA SA CA 1. 553322 101016 1106823 706529 862530 90687 2. 531297 102138 1145954 705656 885454 87191 3. 555333 101282 1116771 715314 872718 95380 4. 544113 101069 1132580 704854 868731 92664 S. 581858 100013 116083 747974 897762 98030 6. 541488 103416 1115922 700996 872027 99048 7. 539473 103923 1168351 735641 898813 88526 8. 54814 103013 1169406 736264 873933 93999 9. 588907 99784 1172729 684378 850047 87976 10. 594003 98611 1124897 700605 861979 89510 Av~. 557961 101426.5 1136951.6 713821 874399.4 92301 STD 22207.23 1704.69 25.292.6 19873.8 15609.9 4220.3 %RSD 4.0 1.7 2.2 2.8 1.8 4.6 Table 19. Injection Repeatability of Peak Area 42 A second precision test, the intra-assay repeatability, indicated that the relative standard deviations for retention times were < 0,48 %, and for the peak areas < 7.9 %RSD (Tables 20, 21 and Figure 13). LA AA TA MA SA CA IS 1. 6.840 8.889 11.247 13.249 16.235 19.664 17.540 2. 6.851 8.890 11.256 13.265 16.256 19.670 17.552 3. 6.820 8.836 11.186 13.196 16.195 19.625 17.499 4. 6.835 8.836 11.234 13.285 16.312 19.784 17.637 5. 6.888 8.921 11.337 13.347 16.339 19.775 17.650 6. 6.902 8.942 11.361 13.380 16.364 19.810 17.671 7. 6.837 8.867 11.227 13.232 16.255 19.720 17.576 8. 6.840 8.875 11.240 13.239 16.218 19.624 17.516 9. 6.817 8.840 11.209 13.215 16.198 19.614 17.497 10. 6.839 8.879 11.250 13.260 16.245 19.648 17.540 Av~: 6.847 8.8775 11.2547 13.2668 16.2617 19.6934 17.568 Std. Dev. 0.027 0.035 0.054 0.057 0.058172 0.073365 0.063672 %RSD 0.40 0.40 0.48 0.43 0.36 0.37 0.36 Table 20. Intra-Assay Repeatability of Retention Time LA AA TA MA SA CA IS 1. 141691 211828 1164243 2121611 777423 140677 668813 2. 146639 196659 1103760 1865760 723593 124290 586293 3. 163413 211868 1174891 2046940 769230 149572 643218 4. 140533 196601 1085597 1839777 702020 117540 578681 5. 164553 216181 1105794 2039862 787298 144980 642555 6. 133198 223764 1196932 2072466 764942 140599 634651 7. 150032 209568 1140578 1940531 733903 137572 613090 8. 128439 184564 1051680 1756580 749011 131985 544700 9. 141666 194314 1144003 1894986 762329 140164 571509 10. 144347 210241 1115085 1881469 741770 135782 601101 Av~: 145451.1 205558.8 1128256 1945998 751151.9 136316.1 608461.1 Std. Dev. 11563.02 11963.1 44156.07 118429.6 26251.32 9549.515 38849.44 %RSD 7.9 5.8 3.9 6.1 3.5 7.0 6.4 Table 21. Intra-Assay Repeatability of Peak Area 43 (i.:2( Figure 13. Overlaid Chromatogram for Intra-Assay Repeatability G. Limit ofdetection (LOD) and Limit ofQuantitation (LOQ) The limit ofdetection was the lowest amount ofan analyte in a sample which could be detected reliably. The limit ofquantitation is the lowest amount ofan analyte in a sample which can be quantitated reliably. The LODs and LOQs are listed in Tables 22 and 23. LA AA TA MA SA CA SIN 2.9 2.4 4.4 5.0 4.0 3.2 Cone. (gIL) 0.0340 0.0497 0.0213 0.0176 0.0114 0.0216 Table 22. LOD ofAcids 44 LA AA TA MA SA CA SIN 12 10. 8.9 11 11 10. Cone. (gIL) 0.1700 0.1006 0.0427 0.0528 0.0733 0.0864 Table 23. LOQ of Acids H. Stability Stability was the consistency of a response over a specified period of time. The stability of the standards was analyzed for 43 days both at refrigerated and room temperature, as shown in Figures 14 and 15. Lactic Acid Stability at Refrigirated Temp. 1.2 1.1 i'CD 1 ~'0 0.9- o :; 0.8a: ~ .~jt ~ JlP" ~ ~lr~ \ ~~ \ -----"OIIIl ,. 0.7 0.6 o 10 20 Days 30 40 50 Figure 14. Stability ofLactic Acid at Refrigerated Temperature Using Std#5 45 Lactic Acid Stability at Room Temp. ~ ........ ~ JIII~ ~~ I , ~ '\.T .,... ~l~~ 1 """"" J.--- J .l. J. ----- T -y 1.5 1.4 'i' 1.3 ~ 1.2 =c 1.1 .9. 1o ~ 0.9ca a: 0.8 0.7 0.6 o 10 20 Days 30 40 50 Figure 15. Stability ofLactic Acid at Room Temperature Using Std#5 I. Robustness Robustness was the capability of the analysis to remain unaffected by small, but deliberate variations in the methods parameters. Studying the variation of the column temperature it was found that resolution was getting wore at 45 ?C and higher. All acids peaks are shifted to the left and merge into other peaks (see Figure 16). 46 0.25