The Pennsylvania State University. Earth and Mineral Sciences. College of Energy and Mineral Engineering A COMPREHENSIVE EVALUATION AND DEVELOPMENT OF

The Pennsylvania State University Earth and Mineral Sciences College of Energy and Mineral Engineering A COMPREHENSIVE EVALUATION AND DEVELOPMENT OF ALTERNATIVE BIODIESEL ANALYTICAL QUALITY TESTING METHODS A Thesis in Energy and Mineral Engineering by Ryan A. Johnson 2011 Ryan A. Johnson Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science May 2011

The dissertation of Ryan A. Johnson was reviewed and approved* by the following: André Boehman Professor of Fuel Science and Materials Science and Engineering Thesis Advisor Joseph M. Perez Senior Research Scientist of Chemical Engineering Sarma V. Pisupati Associate Professor of Energy and Mineral Engineering Matthew M. Kropf Post Doctorate of Engineering Science and Mechanics Yaw D. Yeboah Professor of Energy and Mineral Engineering Head of the Department of Energy and Mineral Engineering *Signatures are on file in the Graduate School ii

ABSTRACT This thesis had as its objective the evaluation of current commercial techniques used to assess and monitor biodiesel quality in industry. It was also essential to statistically determine the reliability of alternative tests as compared to current ASTM testing methods. Biodiesel quality assurance is a major cost issue for many small scale producers, while being a major concern for engine manufacturers. The critical tests for biodiesel fuel quality, defined by BQ-9000, were deemed the most necessary to develop alternative testing methods which would benefit the biodiesel industry as a whole. The commercial analytical methods evaluated in this study include QTA, i-spec and the phlip test. In addition, methods based on spectrophotometry, dielectric spectroscopy and ultrasonic velocity were developed and explored as potential methods for assessing biodiesel quality. Of all the tests evaluated, most had the potential of acting as a firewall against poor biodiesel quality fuel, but none were found to be capable of predicting whether the fuel would meet ASTM specification consistently. While the QTA FT-IR rapid testing unit can measure most of the critical parameters designated by BQ-9000, it was found that results for key biodiesel quality parameters did not adequately reproduce ASTM results. Yet, the QTA shows promise for potentially carrying out nearly full-range biodiesel analysis in one test. The other commercial apparatus, the i-spec Q-100, showed highly insignificant results overall. While the test claims to have high potential, there were no valid results which indicated so. The spectrophotometer test for total glycerin was found to have mediocre results but has the potential to be a highly inexpensive method to produce reliable results. Dielectric spectroscopy measurements of biodiesel did not establish usable trends, but set the foundation for carrying out experiments in-situ for the monitoring of biodiesel either in the facility or as a standalone method for total glycerin, methanol and free glycerol. The ultrasonic velocity measurements provided potentially accurate data for monitoring the biodiesel reaction, but may be limited by being very feedstock dependent. iii

TABLE OF CONTENTS LIST OF FIGURES... viii LIST OF TABLES... x ACKNOWLEDGEMENTS... xi Chapter 1. Introduction... 1 1.1. Biodiesel as an Alternative Diesel Fuel... 1 1.2. Overview of Biodiesel Production... 2 1.2.1. Base Catalyzed Transesterification... 4 1.2.2. Acid Catalyzed Esterification... 5 1.3. Biodiesel Quality Standardization... 6 1.3.1. ASTM Standardization of Biodiesel... 6 1.3.2. US Biodiesel Quality Programs... 7 Chapter 2. Implications of Biodiesel Properties and Impurities on Engines with a Review of the Standardized Techniques that Measure Them... 9 2.1. Biodiesel Standard Parameters and Fuel Property Measurement Methods... 9 2.1.1. Mono-, di-, and triglycerides (bound glycerin)... 9 2.1.2. Free Glycerol... 10 2.1.4. Methanol Content / Flash Point... 13 2.1.6. Acid Number... 17 2.2. Fuel and Physical Properties... 18 2.2.1. Cold Temperature Properties... 18 2.2.2. Oxidative Stability... 19 2.3. Carryover Elements... 20 iv

2.3.1. Sulfur... 20 Chapter 3. Commercial Alternative Biodiesel Quality Testing Equipment... 22 3.1. Introduction to Alternative Testing Methods... 22 3.2. phlip Test... 22 3.3. Mid Infrared Fourier Transform (FT-IR) QTA System... 24 3.4. i-spec Q-100 Handheld Biodiesel Analyzer... 25 3.5. Methanol Solubility Test (Jan Warnquist s Conversion Test)... 26 3.6. Soap and Catalyst Measurement by Colorimetric Titration... 27 Chapter 4. Analytical Methods Development... 29 4.1. Spectrophotometric Analysis of Biodiesel for Bound Glycerol Determination... 29 4.1.1. General Principles of Spectrophotometry... 30 4.1.2. TSL230R Light to Frequency Converter... 32 4.2. Dielectric Spectroscopy of Biodiesel in MW Regime... 34 4.3. Measurement of the Speed of Ultrasound as a Biodiesel Characterization Technique... 37 4.3.1. Ultrasound Fuel Quality Measurement Background... 37 4.4. Unique In-Column Injection for Total and Free Glycerol Determination by GC... 40 Chapter 5. Results and Discussion of Analytical Fuel Quality Techniques... 44 5.1. Test Samples... 44 5.1.1. Commercial Biodiesel Samples... 44 5.1.2. Small Scale Batch Biodiesel Samples from Various Feedstocks... 44 5.2. Past Studies of Correlating Two Instruments... 45 5.2.1. 2004 NREL Survey Two Rancimat Instruments... 45 5.3. Methods for Comparing Two Instruments that Measure the Same Parameter... 46 5.4. Alternative Testing Technique Analysis... 48 5.4.1. phlip Test... 48 5.4.1.1. Upper Phase (Glycerin Detection)... 49 5.4.1.2. Lower Phase (Acid Value Detection)... 49 5.4.2. Near Infrared QTA System... 51 v

5.4.2.1. Total Glycerin... 51 5.4.2.2. Methanol Content... 52 5.4.2.3. Acid Number... 54 5.4.2.4. Free Glycerol... 55 5.4.3. I-Spec Q100... 57 5.4.3.1. Total Glycerin... 57 5.4.3.2. Methanol Content... 59 5.4.3.3. Acid Number... 60 5.4.4. Methanol Solubility Test... 61 5.4.5. Spectrophotometric Analysis of Biodiesel for Bound Glycerol Determination... 63 5.4.6. Dielectric Spectroscopy of Biodiesel for Total Glycerin... 70 5.4.7. Ultrasonic Velocity Measurements in Biodiesel for Bound Glycerol Determination... 72 5.4.8. Unique In-Column Injection Method for Total and Free Glycerol Determination by GC 74 5.4.8.1. Total Glycerin... 74 5.4.8.2. Free Glycerol... 76 Chapter 6. Discussion... 78 Chapter 7. Conclusions and Future Work... 83 7.1. Qualitative Testing Method... 83 7.1.1. phlip... 83 7.2. Quantitative Testing Methods... 83 7.2.1. QTA... 83 7.2.1.1. Total Glycerin... 84 7.2.1.2. Methanol... 84 7.2.1.3. Acid Number... 84 7.2.1.4. Free Glycerol... 85 7.2.2. I-Spec... 85 7.2.2.1. Total Glycerin... 85 7.2.2.2. Methanol... 85 vi

7.2.2.3. Acid Number... 86 7.2.3. Spectrophotometer... 86 7.2.4. Dielectric Spectroscopy... 86 7.2.5. Ultrasound... 86 7.3. Future Work... 87 Appendix A. Sample Sets for Biodiesel Quality Testing... 89 Appendix B. Raw Data of Analytical Instruments... 94 Appendix C. Calculations for Statistical Representation of Results... 96 References... 100 vii

LIST OF FIGURES Figure 1: Biodiesel Production in the United States (7) Figure 2: Conventional Biodiesel Production Process Figure 3: Transesterification Reaction (10) Figure 4: Representative Chromatogram for Determining Free and Bound Glycerol (36) Figure 5: Diagram of Pensky-Martens Closed Cup Flash Point Tester (37) Figure 6: Left (54): phlip standard test vial Pass. Center two samples (54): Two samples that fail. Right: Highly Pure FAME (distilled) Pass. Figure 7: phlip Linear Color Shift with ph in Acidic Range (54) Figure 8: I-Spec Q100 Handheld Unit (60) Figure 9: 27/3 Bound Glycerol Test Result (Fail Sample) Figure 10: Biodiesel Spectrophotometer Apparatus Figure 11: TSL230R Block Diagram (68) Figure 12: TSL230R Spectral Responsitivty at Various Wavelengths (68) Figure 13: TSL230R Output Frequency (khz) as a Function of Irradiance (uw/cm2) (68) Figure 14: Dielectric Response Mechanisms (75) Figure 15: Dielectric Storage Permittivity and Loss Permittivity (78) Figure 16: Dielectric Loss of Canola B100 ( ) vs. Frequency (Hz) Figure 17: Loss Tangent of Glycerol ( ) vs. Frequency (Hz) Figure 18: Dielectric Loss of Methanol ( ) vs. Frequency (Hz) Figure 19: Frequency Sweep of Vegetable Oil from 1.2-1.5 MHz Figure 20: Ultrasonic Velocity Measurement Apparatus (80) Figure 21: Free Glycerol Concentration vs. FID Response Figure 22: Monoolein Concentration vs. FID Response Figure 23: Diolein Concentration vs. FID Response Figure 24: Triolein Concentration vs. FID Response Figure 25: Typical Chromatogram of the Modified GC method for Sample A (sec. 5.1.1) viii

Figure 26: Bound Glycerol Conversion Curve for Various Feedstocks Figure 27: Rancimat Value (Bosch) vs. Rancimat Value (SwRI) (23) Figure 28: QTA vs. ASTM D 6942 for Total Glycerin Measurement Figure 29: QTA vs. EN 14110 for Methanol Content Measurement Figure 30: QTA vs. ASTM D 664 for Acid Number Measurement Figure 31: QTA vs. ASTM D 664 for Free Glycerol Measurement Figure 32: I-Spec vs. ASTM D 6942 for Total Glycerin Measurement Figure 33: I-Spec vs. EN 14110 for Methanol Content Measurement Figure 34: I-Spec vs. ASTM D 974 for Acid Number Measurement Figure 35: Absorbance vs. Total Glycerin at 10:1 Ratio Figure 36: Absorbance vs. Total Glycerin at less than 0.5 wt. %, 10:1 ratio Figure 37: Spectrophotometer (10:1 Ratio) vs. ASTM D 6942 for Total Glycerin Figure 38: Absorbance (9:1 Ratio) vs. Total Glycerin Figure 39: Spectrophotometer (9:1 Ratio) vs. ASTM D 6942 for Total Glycerin Figure 40: Absorbance (8:1 Ratio) vs. ASTM D 6942 for Total Glycerin Figure 41: Spectrophotometer (8:1 Ratio) vs. ASTM D 6942 for Total Glycerin Figure 42: Dielectric Loss at 7.47 GHz vs. Bound Glycerol Weight Percentage Figure 43: Maximum Dielectric Loss vs. Bound Glycerol Weight Percentage Figure 44: Speed of sound measurements of Canola Biodiesel vs. Temperature (80) Figure 45: Speed of Sound Measurements at 25 o C vs. Bound Glycerol Mass Percentage Figure 46: Modified GC Method Total Glycerin vs. Commercial Testing Lab Total Glycerin Figure 47: Modified GC Method Free Glycerol vs. Commercial Testing Lab Free Glycerol ix

LIST OF TABLES Table 1: Common Fatty Acid Chains Table 2: Acid Number of Various Vegetable Oil Feedstocks Table 3: ASTM Biodiesel Quality Limits and Testing Procedures Table 4: Repeatability and Reproducibility for Total and Free glycerol values Table 5: Repeatability and Reproducibility of Methanol Mass Percentage Values Table 6: Pioneering Experiment for Optical Density Tester of Used Cooking Oil Table 7: phlip Glycerin Analysis Table 8: phlip Acid Number Analysis Table 9: 27/3 Methanol Solubility Test of WVO Samples Table 10: 81/9 Methanol Solubility Test of WVO Samples Table 11: Statistical Evaluation of QTA Acid Number Correlation Table 12: Statistical Evaluation of QTA Free Glycerol Correlation Table 13: Statistical Evaluation of I-Spec Total Glycerin Correlation Table 14: Statistical Evaluation of I-Spec Methanol Content Correlation Table 15: Statistical Evaluation of I-Spec Acid Number Correlation Table 16: 27/3 Methanol Solubility Test of WVO Samples Table 17: 81/9 Methanol Solubility Test of WVO Samples Table 18: Statistical Evaluation of Spectrophotometer (10:1 Ratio) Total Glycerin Correlation Table 19: Statistical Evaluation of Spectrophotometer (9:1 Ratio) Total Glycerin Correlation Table 20: Statistical Evaluation of Spectrophotometer (8:1 Ratio) Total Glycerin Correlation Table 21: Statistical Comparison of Alternative Testing Techniques for Total Glycerin and Glycerol Free Table 22: Statistical Comparison of Alternative Testing Techniques for Methanol Content and Acid Number x

ACKNOWLEDGEMENTS I would like to express my gratitude to my project advisor, Dr. Joseph Perez, for his continued support and guidance through my undergraduate and graduate career here at Penn State. I want to thank Dr. Matthew Kropf for his utmost support for my endeavors, where his sincere hands-on assistance and encompassing knowledge allowed me to complete the work that I could not have accomplished on my own. Furthermore, Dr. Kropf brought me into his own projects of which I have learned so much from. I would also like to thank the United Soybean Board for funding for my initial semesters of graduate schooling for working on the Alternative Testing Methods for Biodiesel Project. Howell Rigley with Knightsbridge Biofuels was so kind as to devote his time and resources for providing the QTA Infrared data in this thesis. Furthermore I would like to give my greatest support and appreciation of the Chemical Engineering Biodiesel Research Group, administered by Dr. Joseph Perez and Dr. Wallis Lloyd, for providing me with the capabilities to carry out biodiesel reactions in a controlled and safe manner and with extensive analytical equipment for analyzing vegetable oil and biodiesel fuels throughout this project and beyond. xi

Million Gallons of B100 Produced Chapter 1. Introduction 1.1. Biodiesel as an Alternative Diesel Fuel Biodiesel is a clean burning, alternative diesel fuel produced domestically from various oil seed crops or rendered animal fats (1-3). The ideal characteristics of biodiesel fuel for our current transportation infrastructure make it a promising future alternative energy source. The most common method for producing biodiesel is through a process known as transesterification, which has been carried out for decades (4-5). By definition, biodiesel is described in ASTM D 6751-10 as the mono alkyl esters of long chain fatty acids derived from renewable lipid feedstocks, such as vegetable oils and animal fats, for use in compression ignition (diesel) engines (6). Recent incentives for producing domestic fuels in the United States with a low carbon footprint from renewable sources have allowed for the quick inception of biodiesel production, as depicted in Figure 1, with many states mandating 2-5% blend of biodiesel into petroleum diesel (7). As of December 2009, there are over 122 active biodiesel facilities capable of producing 2 billion gallons per year. Yet, in 2009, U.S. biodiesel facilities were only running at 25% capacity, with 506 million gallons produced (8). Some of the reasons for decreased biodiesel production can be attributed to the low cost of diesel relative to feedstock cost, the loss of subsidies and tax credits and the stricter limits on biodiesel exports into the lucrative European market. 800 700 600 500 400 300 200 100 0 670 477 506 246 92 31 2004 2005 2006 2007 2008 2009 Year Figure 1: Biodiesel Production in the United States (7)

1.2. Overview of Biodiesel Production Conventionally, biodiesel is produced through a transesterification reaction of a triglyceride (animal fat or vegetable oil) with a short chain alcohol (typically methanol) in the presence of a base catalyst (usually sodium or potassium methoxide). While there are many different configurations of biodiesel processing equipment and methods, the method utilized at The Pennsylvania State University (PSU) is depicted in Figure 2. Since PSU produces biodiesel from used cooking oil (yellow grease), a two-step alkali process is carried out. Figure 2: Conventional Biodiesel Production Process Plant oils, animal fats and used cooking oils (UCO) are the main feedstocks used to produce biodiesel. Through precedent classifications, fats are defined as a solid at 20 o C, whereas oil is liquid at 20 o C, even though they are both chemically defined as triglycerides. Animals synthesize fats for their own energy storage whereas plants synthesize oils for energy requirements of the next generation plant. Oils tend to be concentrated in seeds and nuts. The molecular structure of oils and fats are triglycerides, also known as 2

acylglycerols. They are esters of three long chain fatty acids connected to a three carbon molecule known as glycerol. Fatty acids are long chain aliphatic carboxylic acids, ranging from 12-20 carbon atoms in length. Biodiesel will contain a distribution of fatty acid types, known as a fatty acid profile, which is inherent of the feedstock used. The chemical and physical properties of the oil and the biodiesel obtained from them vary in relation to the fatty acid type. Fatty acids can vary in molecular weight and the amount of double bonds along the aliphatic chain, which is highly dependent upon the vegetable or animal source. Fatty acids are often abbreviated to define these two characteristics. By convention, fatty acid chains are more easily denoted as CX:Y, where X is the amount of carbon atoms and Y is the quantity of double bonds. If zero double bonds are present, such as in palmitic acid shown in Table 1, the fatty acid chain is fully saturated. If double bonds are present then the fatty acid is considered unsaturated, such as in oleic acid shown in Table 1. The fatty acid length and degree of unsaturation also reveals some inherent qualities of the fatty acids. If a fatty acid molecule contains more than two double bonds, it will have better cold-flow properties but is susceptible to oxidative stability. Subsequently, fatty acid molecules with zero or one double bond will have poor cold flow properties yet are more stable toward oxidation. Table 1: Common Fatty Acid Chains (9) Fatty Acid Acronym (Cx:y) Formula Mol. Weight (g/mol) Melting Point ( o C) Palmitic acid C16:0 C16H32O2 256.428 63-64 Stearic acid C18:0 C28H36O2 284.481 70 Oleic acid C18:1 C18H34O2 282.465 16 Lioleic acid C18:2 C18H32O2 280.450-5 Linolenic Acid C18:3 C18H30O2 278.434-11 The transesterification reaction occurs step wise, with one fatty acid ester chain being removed first (forming one mono alkyl ester and a diglyceride), the second fatty acid ester removed next (forming two mono alkyl esters and a monoglyceride), and lastly, reaction of the third fatty acid ester as shown in Figure 3. 3

Figure 3: Transesterification Reaction (10) The resulting products are three fatty acid methyl esters (FAME) and glycerin. Glycerin is removed and further purified into a valuable co-product (11). The desired results of the biodiesel process are to break down the high molecular weight acyglycerols into individual fatty acids and remove the glycerol chain. This reduces the viscosity of the fuel to the ASTM specified viscosity range for modern diesel engines. Other benefits of using FAME include reduced tailpipe combustion emissions (12). Many different types of alcohols can be used to carry out the transesterification reaction, but methanol is the most often utilized. Ethanol and iso-propanol have also been used to produce biodiesel with different pros and cons, such as being able to be produce ethanol from renewable sources yet a major downside is ethanol s azeotrope with water causing process difficulties. This thesis will focus only on biodiesel production using methanol. Methanol has a relatively low boiling point compared to the other biodiesel processing components and should be removed FAME and glycerol streams. It can be completely removed by distillation. Methanol s low flashpoint also categorizes it as a Class I-b liquid, being highly flammable. 1.2.1. Base Catalyzed Transesterification In order for the transesterification reaction to be carried out to completion in a timely manner, a base catalyst is required. The most common bases are sodium and potassium hydroxide (NaOH and KOH). These 4

are then converted to the desired catalyst by adding them to methanol to create a solution of sodium or potassium methoxide (NaOCH 3 or KOCH 3). NaOH + CH 3OH NaOCH 3 + H 2O [1] The above reaction is actually not desired since the water content of the catalyst can cause complications during the biodiesel process, such as excess soap formation. The usage of commercially prepared sodium methoxide solution is ideal because there is no preparation required and no water present. As determined in literature, 0.5% wt. sodium hydroxide (0.664% wt. sodium methoxide) to oil is used for catalysis while an additional amount of base catalyst is required to react with any free fatty acids (FFA) into soap. This excess amount of catalyst is determined by titrating the oil, and if not taken into account a certain amount of catalyst will be consumed according to the following reaction. FFA + CH 3O-Na Soap + MeOH [2] The sodium methoxide catalyst is necessary for two reasons. The presence of the base catalyst increases mass diffusion of the reactants for enhanced reaction rates. The increased mass diffusion is caused by an increase in alcohol solubility into the less polar oil phase, since alcohol is hardly soluble in oil under normal conditions (13). The methoxide anion of the catalyst is also responsible for cleaving the original ester bond linkage located on the acylglycerols. Since the transesterification reaction is an equilibrium reaction, thus it can proceed in reverse. The reverse reaction of FAME into acylglycerols is inhibited by adding excess methanol (or ethanol) to force the reaction to the products. Once the methanol and base methoxide are added to the vegetable oil at elevated temperature, the stepwise reaction in Figure 3 occurs. 1.2.2. Acid Catalyzed Esterification According to many sources (14-16), the oil or fat used in alkaline-catalyzed transesterification reactions should contain no more than 1% FFA. If the FFA content exceeds this threshold, saponification occurs which hinders separation of the ester from the glyceride, facilitates emulsification, consumes the alkali 5

catalyst, and reduces the yield and formation rate of FAME. In a typical reaction of oil with low (<1%) FFA content, about.5% by mass NaOH is used as catalyst and to neutralize whatever FFAs may be present. As shown below in Table 2, many feedstock oils exceed this threshold and other pre-treatment steps must be carried out Table 2: Typical Acid Number Range of Various Vegetable Oil Feedstocks (17) Oil Canola Rapeseed Soy Jatropha Cuphea FFA (% mass in oil).4-1.2.5-1.2.5-1.6 3-14.09-5 Many sources available deal with this issue by applying the acid pre-treatment esterification method which converts FFA into FAME. The reaction is catalyzed by concentrated sulfuric acid (14-15), as shown in reaction 3, R1-COOH + R2OH H 2O + R1-COO-CH 2-R2 [3] The goal of this step is to reduce the FFA content in the oil to <1% prior to the alkaline-catalyzed reaction, while inhibiting the breakdown of triglycerides. 1.3. Biodiesel Quality Standardization 1.3.1. ASTM Standardization of Biodiesel Before biodiesel can be sold as a fuel or blending stock, it must first meet a defined standard to ensure the fuel does not damage engine components. Therefore, the quality control of biodiesel is a necessity for the successful commercialization of this fuel and its blends (18).As described previously, the transesterification reaction that produces unpurified FAME will also contain glycerol, alcohol, catalyst, tri, di- and monoglycerides as well as free fatty acids (19). The American Society for Testing and Materials (ASTM) Biodiesel Task Force was formed in 1994 to agree upon the fuel quality requirements of biodiesel (20). Since then, over 10 iterations of the standard, Table 3, have occurred through collaboration between biodiesel producers, consumers, researchers and engine manufacturers (6, 21). Due to the misrepresentation of biodiesel which has sometimes been referred to as pure vegetable oil, a mixture of vegetable oils, esters of natural oils and mixtures of esters and petrodiesel, the ASTM Biodiesel Task Force decided that a written description of biodiesel was essential. The Task Force adopted the definition of biodiesel as stated in section 6

1.1. It was also decided to develop a standalone specification for pure biodiesel and for blends of biodiesel into petrodiesel, D 6751-10 (Table 3) and D 7467, respectively. Table 3: ASTM D 6751-10 Biodiesel Quality Limits and Testing Procedures (21) 1.3.2. US Biodiesel Quality Programs The National Biodiesel Board (NBB), a collaboration of biodiesel groups in industry, has been very influential in developing biodiesel quality programs (22). NBB has supported the necessity of conforming to ASTM specifications by encouraging its use and adoption. The NBB has taken biodiesel fuel quality one step further, however, by sponsoring a voluntary fuel supplier certification program called BQ-9000. This accreditation program provides producers with a good housekeeping seal of approval, to leverage their sales and by increasing the confidence of engine companies and consumers that BQ-9000 certified marketers will meet ASTM specifications NREL in collaboration with the National Biodiesel Board has also conducted quality surveys to assess the quality of fuel being distributed through the United States. The results of three surveys in recent years summarized key issues of biodiesel quality control. The main result of the biodiesel quality survey in 2004 (23) was that out of 27 B100 samples, 85% met all of the required ASTM D 6751-03a parameters. The 4 of samples that failed were either out of spec. for acid number, total glycerin or phosphorus (one sample exceeding total glycerin limit by 5 fold). 7

Furthermore, it was found that 26 out of 27 samples would fail the EN oxidative stability limit by Rancimat, requiring an induction period of 6 hours. The main result of the B20 blend quality survey was that 36% of the samples contained biodiesel contents between 7-98% (outside the range of 18-22%). It was determined that the cause was that conventional splash blending methods were not providing homogenous mixtures. All B20 blend samples passed quality testing requirements in accordance to D 975. A subsequent survey was carried out in 2006 which discovered high failure rates against ASTM D 6751, with 59% of the samples not meeting specification (24). The majority of failing samples exceeded the allowable total glycerin (33%) or did not meet the minimum flash point specification (30%). The samples obtained were derived from soy, canola, palm, rapeseed or animal fat. The 2006 failure rate was alarming which showed serious quality control issues. The conclusion of the study influenced the National Biodiesel Board to stress fuel quality and subsequently released several informational documents to augment fuel quality programs (25). NREL surveyed B20 blends once again in 2008 (26) and were tested with the new B6-B20 ASTM specification, D 7467. In this study, it was found that 40% of 33 samples were not between 18-22% biodiesel, compared with 36% in 2004. Most samples that were below 18% contained B2, B5 and B10 making it necessary to enforce pump labeling. No B100 samples were analyzed in 2008. 1.3.2.1. BQ-9000 The BQ-9000 National Biodiesel Accreditation Program is a cooperative and voluntary program which subjects the facility to test each production lot of B100 with full specification testing under ASTM 6751-08b until there is sufficient confidence that the production process consistently produces fuel that is up to standard. The quality system addresses many aspects of biodiesel production such as storage, sampling, testing, blending, shipping, distribution and fuel management practices (27). Once a producer becomes certified under BQ-9000, each lot of fuel produced is subjected to critical specification testing. Every six months the facility is subjected to full range ASTM 6751 testing of their fuel, while once a month a sodium/potassium and calcium/magnesium test shall be run (27). 8

The critical tests were deemed most important for biodiesel reliability. Furthermore, many of the other testing parameters in D 6751 are not as likely to go out of specification once the process method is fully developed. The critical tests are as follows (27): Alcohol control, water and sediment, cloud point, acid number, free glycerin, total glycerin, sulfur, oxidative stability, visual appearance and cold soak filterability test. Chapter 2. Implications of Biodiesel Properties and Impurities on Engines with a Review of the Standardized Techniques that Measure Them 2.1. Biodiesel Standard Parameters and Fuel Property Measurement Methods The development of internal combustion engines over the past century has resulted from the complimentary refinement of the engine design and fuel properties. As such, engines have been developed to utilize the properties of the fuels that were available. Replacement of existing fuels with new fuel formulations requires understanding the critical fuel properties. To ensure that the new fuels can be used effectively requires consistent fuel quality. The critical tests for biodiesel fuel quality defined by BQ-9000 were deemed the most necessary by our group for which to develop and test alternative testing methods. Being able to use alternative tests for the parameters that need to be tested the most often would alleviate financial burdens on biodiesel producers, reduce turnaround times for sample analysis all while ensuring biodiesel quality. Discussed in this section will be some key fuel properties as well as the ASTM and EN methods to measure these properties, as required in D 6751, Table 3. 2.1.1. Mono-, di-, and triglycerides (bound glycerin) As seen previously in Figure 3, the biodiesel transesterification process is a three step equilibrium reaction. The equilibrium constants for each reaction are pushed towards the product by optimizing the reaction conditions with excess methanol and the correct ratio of alkaline catalyst to triglycerides. Since the chemical reaction is reversible, there will almost always be left over un-reacted acylglycerols in the final product which are in the form of mono-, di- and triglycerides. The amounts will depend on process conditions. There are no commercial separation techniques for removing un-reacted acylglycerols, with one 9

exception that lower concentrations can be achieved if the final ester product is distilled (28). The ASTM test method uses gas chromatography (GC) to analyze the three types of un-reacted esters and combines them into a term known as bound glycerol. Bound glycerol accounts for only the glycerol backbone to be counted toward the impurity concentration. The following calculations show how each mono-, di- and triglycerides is converted to bound glycerol for quantification (29), Bound glycerin = Gl M, Gl D, Gl T) [4] Gl M = 0.2591 monoglyceride, mass%) [5] Gl D= 0.1488 diglyceride, mass%) [6] Gl T = 0.1044 triglyceride, mass%) [7] Where Gl M, Gl D and GL T are the mass percentage concentrations of monoglycerides, diglycerides and triglycerides, respectively. It is significant to realize that the fatty acid molecule attached to the glycerol backbone is the majority of the molecular weight of the un-reacted ester, which is taken into account in equations 5-7. While ASTM does not set explicit limits for individual partial glycerides, the EN standard does. Recently it has been found that large proportions of monoglycerides have been the cause of cold weather issues which gave rise to the importance of implementing the cold soak filtration test into the ASTM standard (6). 2.1.2. Free Glycerol As shown in Figure 3, glycerol is a major product of transesterification being approximately 10% by weight of the biodiesel product. Thus separating glycerol sufficiently is a major concern for fuel quality (30). Free glycerol in significant concentrations will separate out of the biodiesel either in storage or in the fuel tank. Due to glycerol s hydroscopic properties, it will attract other polar compounds such as water, monoglycerides and soap. The increased concentration of these compounds will augment damage to the injection system (31). Concentrated free glycerol may also clog up the fuel filter and can result in increased aldehyde emissions (32). 10

2.1.3. Total Glycerin For clarity, the ASTM requirement for bound glycerol is combined with the amount of free glycerol into a term known as total glycerin. This places all glycerol backbone moieties into one category. Biodiesel fuel that is out of specification for total glycerin can lead to engine coking which will cause the formation of deposits on injection nozzles, pistons and valves (33). While determining bound glycerol in biodiesel using GC has been previously proven (34), the GC method was augmented to include the determination of both free and bound glycerol to suffice for the ASTM standard (35). 2.1.3.1. ASTM D 6584: Standard Test Method for Determination of Free and Total Glycerin in B-100 Biodiesel Methyl Esters By Gas Chromatography The ASTM D 6584 standard quantitatively determines the amount of free glycerin in the range of 0.005 to 0.05 mass% and total glycerin in the range of 0.05 to 0.5 mass% by GC. Detection limits are 0.001% mass% for free glycerin and 0.02% mass% for mono-, di- and triglycerides. The ASTM GC procedure is as follows (29): Column: Non-Polar, high-temperature capillary column coated with 95% dimethyl 5% diphenylpolysiloxane stationary phase, 10-15 m length with 0.32 mm inner diameter and 0.1 mm film thickness. A guard column is recommended for robustness of the column due to potential sample contaminants and high oven temperatures. Injection: 1-2μL, Cool on-column injection. Detection: Flame Ionization, 380 o C Carrier gas: Helium or Hydrogren, 4mL/min. Oven Temp.: 50 o C (hold 1 min) to 180 o C @ 15 o C/min (hold 7 min) to 230 o C @ 30 o C/min to 380 o C @ 30 o C/min (hold 5 min). Sample Preparation: Since glycerol and bound glycerol are essentially non-volatile compounds, they need to be treated with a silyating agent to enable them to be vaporized during the separation. The free hydroxyl 11

groups of the sample are silyated with N-methyl-N-trimethylsilytrifluoracetamide (MSTFA) by shaking for three minutes and let stand for 20 minutes. The sample mixture is then diluted with heptane to quench the silyation reaction. Standardization: Internal standards are utilized to account for any potential interference during the injection or column degradation. 1,2,4 butanetriol is the standard for free glycerol and 1,2,3 tricaproylglycerol (tricaprin) is the standard for mono-, di-, and triglyceride. Calibration: The detector response is calibrated with known concentrations of glycerol, mono-, di- and triglycerides alongside the internal standards, butanetriol and tricaprin. Five different concentrations are run to develop a linear calibration curve for each component as well as to designate retention times of each compound. Furthermore, a mixture of monopalmitin, monostearin and monoolein need to be run to detect separate retention times to distinguish between saturated and unsaturated monoglycerides. Below is a chromatogram of a biodiesel sample injection with labeled component peaks. Calculation: In utilizing the peak areas of each compound and internal standards, as labeled in Figure 4, the mass percentage of each component is determined by the previously determined calibration curves. Figure 4: Representative Chromatogram for Determining Free and Bound Glycerol (36) 12

2.1.4. Methanol Content / Flash Point The determination of the amount of methanol either by GC or flash point ensures that the majority of methanol used in production is removed from the fuel. Methanol content is of concern due to both fire safeties during transport and storage as well as the corrosive nature of methanol. Furthermore, methanol makes biodiesel a toxic substance. Biodiesel does have a beneficial characteristic in that its flashpoint is over twice that of its petroleum diesel counterpart, with values between 130 o C and 200 o C whereas petroleum diesel is approximately 64 o C (33). Yet, the high flashpoint of biodiesel will decrease rapidly with increasing amounts of residual methanol. Since methanol content and flashpoint are correlated, the biodiesel ASTM requirements allow for either the determination of flashpoint of biodiesel or the mass percentage of methanol. The ASTM spec. for flashpoint, 130 o C, limits the amount of methanol to approximately 0.1-0.2% by mass in the fuel. The removal of residual methanol can be accomplished by distillation or repeated water wash steps. 2.1.4.1. ASTM D 93: Standard Test Methods for Flash Point by Pensky- Martens Closed Cup Tester A key property in determining the flammability of a fuel, and in this case methanol content, is to determine the fuel s flash point. The flashpoint is the lowest temperature to which an ignition source applied above the liquid surface layer will cause the fuel vapors to ignite. The ASTM method D 93 restricts the flash point of biodiesel to 130 o C, which will ensure that the methanol content is below 0.2% by mass. The Pensky-Martens is the most widely used flashpoint apparatus and can be run manually or automatically. The fuel sample is heated at a regulated rate with stirring and a flame is passed over the fuel sample in certain intervals. When the fuel sample reaches the flashpoint, the fuel vapors will ignite due to the presence of the flame and air. The ignition is easily detectable by human sight or by a pressure sensor. Figure 5 depicts the flash point apparatus where a sample cup is placed in a heating block with an agitator. A temperature ramp occurs where an external flame is applied at specific intervals with an adequate air-to-fuel 13

ratio to detect flammable vapors. The Pensky-Martens Closed Cup Flash Point Tester is depicted below in Figure 5. Figure 5: Diagram of Pensky-Martens Closed Cup Flash Point Tester (37) 2.1.4.2. EN 14110- Residual Methanol in B100 Biodiesel by Headspace-Gas Chromatography An alternative method to measuring flashpoint is to determine the mass percentage of methanol present in biodiesel by GC (38). The biodiesel ASTM 6751 requirement of methanol is 0.2% by mass carried out as per EN 14110. The method can determine methanol concentrations from 0.01 to 0.5% mass. If an automatic headspace injector is available, internal standardization is not required. Manual headspace injection utilizes a small constant amount of 2-propanol as an internal standard to account for variances in sample heating, syringe handling and injection. Column: Non-polar, capillary column, 100% dimethyl polysiloxane. 14

Injection: 250 μl manual headspace injection into split injector (flow rate: 50mL/min). Detection: Flame Ionization, 240 o C Carrier gas: Helium. Oven Temp.: 50 o C, isothermal Sample Preparation: 5mL of biodiesel and 5uL of 2-propanol are added into a hermetically sealed headspace vial and heated at 80 o C for 45 minutes. This will allow the vapor phase to reach equilibrium with the sample mixture. Standardization: An internal standard is utilized to account for any potential interference during the manual injection procedure. 5μL of 2-propanol is added to each sample. Calibration: Three calibration standards are made, all starting with the same prepared biodiesel solution with no methanol content. The beginning biodiesel blendstock is prepared by either vacuum distillation or multiple distilled water washing steps with subsequent drying. Known amounts of high purity methanol are mixed into the blendstock. The blendstock with methanol is then diluted twice for a total of three samples along with 5μL of 2-propanol as an internal standard (internal standard for manual injection only) to produce a 3 point linear calibration curve. From the linear FID response of methanol, a calibration factor, F, is determined for use in the following calculation section. Calculation: In utilizing the peak areas of both methanol and the internal standard 2-propanol, the mass percentage of methanol is calculated as shown in equation 8, where, [8] F is the calibration factor obtained from the linear response curve S m is the peak area of methanol 15

C i is 2-propanol content added to the sample, expressed in % mass S i is the peak area of 2-propanol 2.1.5. Water and Sediment Water can be present in the final biodiesel product either due to water being present in the feedstock or due to water washing steps. Furthermore, biodiesel is hygroscopic so it can absorb up to 1500 ppm before it becomes saturated, thus it needs to be reduced well below the limit by drying (28). Water content above 1500 ppm will eventually separate during storage, which is known as free water, and will promote microbial growth causing eventual sludge and slime formation. The sludge that is formed is then known to block up fuel filters (28). Furthermore, free water is also associated with hydrolysis reactions, converting biodiesel into free fatty acids, causing an increased acidity of the fuel which can both block up fuel filters or cause corrosion. Water can also be a leading cause of corrosion of chromium and zinc parts located in the engine (31). High water content may also cause poor combustion, plugging and smoking. The respective maximum concentration of water for fossil diesel fuel is less than half of the values required for biodiesel water content, but it is easily met due to the non-polar nature of the fuel, thus the water will sink to the bottom of the fuel tank. 2.1.5.1. ASTM D 1796 Standard Test Method for Water and Sediment in Fuel Oils by Centrifuge Method Water and sediment is a test that determines the volume of free water and sediment in middle distillate fuels having viscosities at 40 o C in the range of 1.0-4.1 mm 2 /s and densities in the range of 700 to 900 kg/m 3. The test ensures that there will be no free water present to settle out during storage and also acts as a firewall for the cleanliness of a fuel by measuring sediment. The described biodiesel quality standard limits water and sediment to 0.05% by volume (39). The ASTM test method centrifuges 100mL of biodiesel in a conical centrifuge tube between 500 and 800 relative centrifugal force (rcf) for ten minutes. Water and sediment are visible below the biodiesel layer and are measured quantitatively. 16

2.1.5.2. EN ISO 12937 While the ASTM D 6751 standard does not utilize EN ISO 12937, this test is carried out at PSU as well as many other biodiesel production companies. The analytical procedure for the determination of water in both biodiesel and fossil diesel fuel involves titration, by means of a Karl Fischer Coulometric Titration apparatus. The basic principle of this procedure is a reaction between I 2 and SO 2, which only occurs in the presence of water. Since the water and sediment content is limited to 0.05% by volume, this can be adjusted for the coloumetric titration to be approximately 500 ppm by weight (40). 2.1.6. Acid Number The acid number of biodiesel fuel is a measurement of the free fatty acids (FFA) or mineral acids present in the fuel. It is expressed in mg KOH required to neutralize 1g of sample. The acidity of the fuel can exceed the limit due to a variety of factors during the production process. Any FFA that is present in the starting feedstock is converted to soap during the transesterification process. The resulting soap can either be washed out of the fuel phase or in some cases is reverted back to FFA using ion exchange resins. Also, the acid treatment of soaps will form FFA. Furthermore, FFA can indicate that the fuel has oxidized past its stability point or be due to hydrolytic cleavage of ester bonds. FFA content higher than the prescribed limit is known to clog fuel filters, lead to engine deposits in fuel injector, catalyze polymerization in hot recycling fuel loops and lead to corrosion (41). acids. 2.1.6.1. ASTM D 974 Standard Test Method for Acid and Base Number by Color-Indication Titration This method does not distinguish between acidity caused by mild carboxylic and strong mineral Titration Reagent: 0.1 M alcoholic KOH standardized with ph electrode against oxalic acid to detect molarity up to 0.0005. Color indicator: 10 g/l p-naphtholbenzein Indicator (in titration solution) used for color indication from orange to green even in highly opaque samples. 17

Titration Solvent: 100:99:1 toluene, isopropanol, water Procedure o 100mL of titration solvent is blanked to determine acidity of solvent. o Approx. 20g of fuel is weighed to the nearest 0.001mg o Fuel sample is added to 100mL of titration solvent o 0.5mL of color indicator added to titration mixture. o Stirred and titrated from red to dark green/brown using micro-burette with 0.05mL markings. o Calculation of acid number is as follows [9] where, A= ml KOH solution required for titration B= KOH soln. required for blank M = molarity of KOH (0.1 M) W = weight of sample titrated, g. 2.2. Fuel and Physical Properties 2.2.1. Cold Temperature Properties Biodiesel can be the source of major fuel reliability issues at low temperature due to the nature of the fatty acid molecules present in certain feedstocks. Solidification points of FAME depend on chain length and degree of unsaturation, with long-chain saturated FAME having the least favorable cold-temperature properties. Thus, biodiesel fuels derived from feedstocks rich in these compounds, such as tallow and palm methyl esters, may even be problematic at room temperature, 18

The partial solidification due to crystal formation or gelling of the fuel at low temperatures can lead to blockage of the fuel filter, which can cause issues during both engine start-up and operation. The cloud point is an important parameter to monitor for biodiesel since biodiesel fuels will cause operational issues at higher temperatures than petroleum diesel fuel. 2.2.1.1. ASTM D 2500 Cloud Point of Petroleum Products The cloud and pour point of biodiesel fuel and biodiesel blends are of particular importance during cold weather usage, especially for emergency vehicle reliability. As described in ASTM D 2500, the cloud point is the temperature at which a cloud is formed due to the crystallization of fatty acid chains that will appear in the liquid upon cooling (44). The cloud point is measured by cooling the sample at a specific rate and visually inspecting for a haze to begin forming. The pour point stands for the lowest temperature to which the sample may be cooled while still retaining its fluidity. 2.2.2. Oxidative Stability Due to the chemical composition of FAME, biodiesel fuel has inherent instability in the presence of oxygen. Oxidation of FAME is augmented with increased amounts of unsaturated fatty acids, as the methylene groups adjacent to double bonds are particularly susceptible to radical attack, which is the first step of fuel oxidation processes (45). The formed hydroperoxides from radical attack may polymerize with other free radicals to form insoluble sediments and gums, which are associated with fuel filter plugging and deposits within the injection system and the combustion chamber (46). The products of fuel oxidation are accompanied by an increase in viscosity. Furthermore, the oxidation of hydroperoxides may also result in the formation of aldehydes, ketones and short-chain carboxylic acids, which are linked to increased corrosion of the injection system caused by the low ph (45). Apart from the fatty acid composition of the feedstock, the content of natural antioxidants, such as carotenes and tocopherols, has been identified as beneficial components for oxidative stability. In general, antioxidant concentrations are high in non-distilled fuels prepared from fresh vegetable oils, whereas hardly any antioxidants are contained in distilled samples or in samples prepared from used frying oil. The addition 19

of synthetic antioxidants has been identified as a viable means of improving oxidative stability. A few that have been realized for increased oxidative stability are tert-butyl hydrochinone (TBHQ), pyrogallol BHT and propylgallate (47). Since Rancimat induction period have been found to decrease after prolonged periods of storage, antioxidants are added in comparatively high concentrations to ensure that fuels will still meet the specifications when ready for consumption (47). 2.2.2.1. ASTM and EN Biodiesel Stability Methods 2.2.2.1.1. EN 14112 - Rancimat The standard analytical method for the determination of biodiesel oxidation stability is a method derived from food chemistry, which known as the Rancimat (48). In the Rancimat procedure a fuel sample is placed in the presence of elevated temperatures (110 C) and air to accelerate the oxidation process. The effluent gases are sparged into a cell of distilled water, of which the conductivity is constantly recorded. When the sample reached a critical point in oxidation, a sharp increase of conductivity can be observed. The period of time up to this point is called induction period (IP) and is expressed in hours. Systematic tests showed that Rancimat induction period is well correlated to other biodiesel quality parameters, such as peroxide value, anisidine value, kinematic viscosity, ester content, acid value, and polymer content (49). 2.3. Carryover Elements 2.3.1. Sulfur Fuels with high sulfur content have been associated with negative impacts on human health and on the environment, which is the reason for the current tightening of national limits as per the Clean Air Act. Vehicles operated on high-sulfur fuels produce more sulfur dioxide and particulate matter (50). Furthermore, fuels with high sulfur levels may increase engine wear and reduce the efficiency and the life span of oxidation catalytic converters and/or denitrification after-treatment systems. Biodiesel fuels are inherently sulfur-free, as only trace amounts of sulfur can be detected from minor components within the feedstock, such as glucosinolates and contamination of protein material (51). Secondly, it is also possible that when an acid esterified fuel is produced, sulfuric acid can carry over into the 20

final fuel. Due to FAME s inherent low sulfur quality, biodiesel exhibits tremendous advantages over petrodiesel in terms of sulfur dioxide emissions of which may lead to particle-bound mutagenicity (51). Ultra low-sulfur diesel has been mandated in recent years, but it was found that the resulting fuel lacks in lubricity, which can lead to injection pump failure. (52) This phenomenon is due to the removal of nitrogen and oxygen compounds, normally responsible for lubrication, during the desulfurization process. The addition of small proportions of biodiesel has been found to alleviate the lack in lubricity, due to FAME s lubricating qualities. 21

Chapter 3. Commercial Alternative Biodiesel Quality Testing Equipment 3.1. Introduction to Alternative Testing Methods The United Soybean Board s Biodiesel Technical Workshop identified the immediate need to evaluate and develop analytical methods that could potentially be quicker and less expensive for determining biodiesel quality. It was designated to produce data on current commercial or unidentified analytical techniques that could be incorporated into D 6751. Furthermore, field tests that provide qualitative results, which by nature could never be adopted into D 6751, were also targeted. Field tests are becoming a necessity since they could provide useful and immediate feedback to distributors and users. The benefits of doing so could boost consumer confidence by ensuring biodiesel reliability. Our project group also developed and evaluated potential testing techniques that were aimed to analyze critical parameters of biodiesel quality. In doing so we identified techniques that could either be used in-situ for current biodiesel production equipment to monitor biodiesel quality in real time, or to be a standalone test to potentially replace ASTM techniques. 3.2. phlip Test The phlip test was developed to provide a qualitative means for detecting off-spec fuel due to a variety of trace contaminants in a quick and inexpensive manner. The phlip test is provided by CytoCulture International Inc (53). While this test can be only used as a firewall since it cannot provide quantitative results, thus would not be considered relevant to ASTM methods, it can serve as a tool for ensuring low quality biodiesel does not get used commercially on-road. Figure 6 depicts the reference vial on far left which indicates how a high quality B100 sample should appear after shaking and let stand for 10 minutes. The top phase of the vial, being the organic phase, will contain the biodiesel sample. The phase interface in the middle should be clean and have a mirror finish. Lastly, the bottom phase, or the polar phase, contains an indicator solution that can change color if the biodiesel contains an acidic or basic contaminant. The two samples in the center can easily be determined as 22

a sample that failed due to a variety of contaminants. The sample on the right is a BQ9000 commercial sample of utmost high quality and passed the phlip test similarly to the reference vial. Figure 6. Left (54): phlip standard test vial Pass. Center two samples (54): Two samples that fail. Right: Highly Pure FAME (distilled) Pass. If the upper phase which contains the biodiesel becomes foggy or opaque, it indicates that the biodiesel contains contaminants that can readily absorb moisture. These contaminants include bound glycerides, free glycerol and oxidized esters. Secondly, the middle phase separation can contain emulsions or debris that indicates monoglycerides or free glycerol contaminates the biodiesel. The main contaminants that can be observed in the lower polar phase are catalyst contaminants, FFA or oxidized fuel, which is shown by the color indicator solution. The alkaline catalyst will turn the cherry red bottom phase to purple. The acidic FFA or oxidized fuel will cause the bottom phase to turn orange to yellow. The color shift due to ph change is indicated in Figure 7. While it can be beneficial to estimate the acidity or alkalinity of the fuel, this test can be interpreted differently by different users. The linear change in color due to ph makes it very hard to determine if a sample is below or above the ASTM limits for acid, as it will be somewhere between orange and yellow. Alternately, if the bottom phase becomes cloudy or opaque, this will indicate residual soaps contaminate the biodiesel. 23

A 580nm COLOR 3 2.5 2 1.5 1 0.5 0 3.0 4.0 5.0 6.0 7.0 8.0 ph Figure 7: phlip Linear Color Shift with ph in Acidic Range (54) 3.3. Mid Infrared Fourier Transform (FT-IR) QTA System The Cognis QTA (Quality Trait Analysis) System is service providing a rapid, on-site analysis and management system (55). The system communicates over the internet with the Cognis QTA systems central processor providing rapid analyses to the user. While FT-IR equipment is widely available, the Cognis QTA system is unique since it provides sample analysis for the operator. The QTA system for biodiesel analysis begins by digitizing the infrared spectra of a biodiesel sample using mid-infrared technology. The algorithm for determining quality of B100 from various feedstocks is formed though the database containing hundreds of samples. The QTA On-Demand measures many of the necessary parameters required to assess the quality of a B100 batch including free and total glycerin, acid number, cloud point, moisture, mono-,di-, and triglycerides, oxidative stability, sulfur and methanol content. The cost of the system was prohibitive for this study but samples were analyzed through one of the collaborating laboratories. Mid-infrared (4000 cm -1 625 cm -1 ) spectroscopy analyzed the vibrational states of compounds through the covalent bonds between atoms. Previous groups have identified FT-IR as an informative tool to 24

monitor the transesterification reaction (56). Different bonds at different energy states will absorb IR energy at specific frequencies, which is detected and analyzed by Fast Fourier Transform to produce a spectrum of absorption. Table 4 depicts various covalent bonds contained in biodiesel fuel which may give rise to information on concentrations of impurities. Table 4: Various Functional Groups That May Give Rise to Biodiesel Quality Parameters (57-59) Compound Biodiesel/ Acylglycerol Biodiesel Conversion Glycerol /MeOH MeOH FFA FFA Unsaturation (Cloud Point) Functional Group RCOOR, Aliphatic Ester Hydroxid e Alcohol RCOOH, carboxylic acid RCOOH, carboxylic acid HRC=CRH, Alkene Bond C=O C-O R-O-H C-O C=O O-H C=C-H Wavenumber 1735 cm -1 1300-1060 3600-1050 cm -1 3200 cm -1 cm -1 1710 cm -1 3000 cm -1 3300-3000 cm -1 3.4. i-spec Q-100 Handheld Biodiesel Analyzer The i-spec TM is a hand held instrument, Figure 8, of Paradigm Sensors (60) that utilizes an electrical measurement technique known as Impedance Spectroscopy (IS). A small amplitude of AC voltage of varying frequency is applied to the sample under analysis and the measured response of the individual frequencies are incorporated into proprietary algorithms which allow analysis of blend composition (2-99%), total glycerin (in B6-B99 & B100), methanol content (B100), and acid no (B100). The algorithms are the result of correlations derived between the electrical characteristics of a broad range of biodiesel samples and their corresponding physio-chemical attributes as determined from appropriate reference standard analyses. The cost of the unit was prohibited for this study. However, collaboration of Paradigm resulted in use of a unit. Sample cartridges were purchased for the study. The I-SPEC operates using a single use test cartridge that is inserted into the hand-held unit. The sample to be tested is injected into a cartridge after a 25

field calibration is performed on the empty cartridge. Sample size required is approximately 0.5 ml. Test results are displayed on an LCD screen and can be printed using a built-in IR link. Figure 8: I-Spec Q100 Handheld Unit (60) 3.5. Methanol Solubility Test (Jan Warnquist s Conversion Test) The Jan Warnquist s conversion test is carried out by many small scale producers to act as a quick pass/fail test (61). It is highly discussed and revered on many biodiesel community forums and home-brew process guides. The premise is that bound glycerol, mainly triglycerides, are not soluble in methanol at room temperature at a vol:vol ratio of 27 parts methanol and 3 parts biodiesel. The test is carried out at 20-25 o C (room temperature) in a tall glass jar where biodiesel is added to methanol and then shaken vigorously. If any bound glycerol settles out on the bottom, Figure 9, which is known as a precipitation, then the fuel will fail ASTM specification. If there is no precipitation, then there is a high chance the fuel will pass the 0.24 % wt. ASTM limit. It is known to work on washed and unwashed biodiesel. Another aspect to the test is that some B100 mixtures will make the MeOH in the 27/3 test turn cloudy, or opaque. While the small-scale community says this is due to a multitude of factors that provides no qualitative information, the experiments run here where bound glycerol was the only impurity showed the methanol clarity was dependent on conversion. 26

Figure 9: 27/3 Bound Glycerol Test Result (Fail Sample) Factors that may influence the test making it irreproducible: Fatty acid profile o Unsaturated fatty acids may be harder to solubilize in methanol. Bound glycerol distribution o Triglycerides will much more easily fall out of solution than monoglycerides, thus a fuel that has a large amount of monoglycerides may pass the 27/3 test yet still fail ASTM bound glycerol limits. 3.6. Soap and Catalyst Measurement by Colorimetric Titration During process optimization and product quality assessment, it can be useful to know the amount of soap formed, where the catalyst resides, and how effective the washing process is in removing these two compounds. A simple titration procedure can be used to measure the amount of soap and catalyst. The titration procedure consists of titrating biodiesel, wash water or glycerol with a 0.01 N solution of HCl to the phenolphthalein end point. This gives an estimate of the amount of catalyst. Then, a few drops of bromophenol blue indicator are added and the titration continued to the color change for that indicator. This gives an indication of the amount of soap. In the first titration, the HCl neutralizes the alkali catalyst, so when the phenolphthalein indicates that the solution has become neutral, then all of the catalyst has been measured. Then, if the titration is continued, the HCl, as a strong acid, begins to split the soap molecules to free fatty acids and salt. When the ph reaches about 4.5, where the bromophenol blue changes color, then this 27

indicates that the HCl has split all of the soap. It is now lowering the ph, so it has protons to donate since the soap has all been split. The procedure utilized at Penn State is a modified version of AOCS method Cc 17-79, soap in Oil (62). 28

Chapter 4. Analytical Methods Development 4.1. Spectrophotometric Analysis of Biodiesel for Bound Glycerol Determination Initial studies were carried out at PSU to optimize the methanol solubility test to no avail, but a modified version is summarized in Table 6. The optimization of the solubility test found that when the volumetric proportion of biodiesel to methanol at 1 to 9 was chilled to 10 o C, the mixture becomes turbid even if there are small quantities of bound glycerol, but will remain clear if the sample below the ASTM limit of bound glycerol. This phenomenon is hypothesized to take place due to the bound glycerol becoming emulsified in methanol at decreased temperatures in certain concentrations. An emulsion is defined as a system of two immiscible liquids, one being dispersed in the other in the form of small droplets (63). The droplet distribution causing the turbidity depends on several conditions such as temperature, degree of agitation and the length of time the precipitates are allowed to stand (64) The proportion of biodiesel to methanol and the temperature were optimized to be able to detect bound glycerol in and near the ASTM standard for conversion by a function of cloudiness, as shown in Table 5. Carrying out the turbidity experiment at 10 o C and at a methanol to FAME ratio of 9:1, it is very promising to be able to tell the difference between fuel that passes ASTM standard and fuel that does not. It was of the essence to be able to convert these values into a quantitative measurement. To do so, use of a light sensor was considered. The light sensor will function by measuring the degree of absorbance of light as a function of cloudiness though a disposable spectrophotometric cell. 29

Table 5: Pioneering Experiment for Optical Density Tester of Used Cooking Oil Conditions: 10 o C, 9:1 methanol to B100, shaken vigorously Bound Glycerol, % wt. 10 min 20min 0.745 Highly cloudy, opaque. No dropout Highly cloudy, opaque. No dropout 0.48 Cloudy, Can see through. No dropout Cloudy, Can see through. No dropout 0.359 Slightly cloudy. No dropout Slightly cloudy. No dropout 0.265 Slightly cloudy. No dropout Faintly cloudy 0.227 Clear Clear 0.165 Clear Clear Other spectrophotometric techniques were also established for the determination of biodiesel content in blends (65). The authors concluded that UV spectroscopy was the most reliable wavelength band for inspection of biodiesel blend, which provided simple, fast and reliable results (65, 66). 4.1.1. General Principles of Spectrophotometry Many molecules absorb specific wavelengths of radiant energy when monochromatic light passes through a solution containing such molecules (solutes). In the case of detecting turbidity caused by the emulsification of bound glycerol in this specific application, the spectrophotometer will actually be calculating optical density, which is due emulsions of oil and methanol scattering the photons in the light beam rather than absorbing them. The degree of absorption or scattering is directly proportional to the logarithm of the concentration of solute as well as the length of light path as described by the Beer-Lambert Law (69), equation 12. Even though there are spectrophotometers commercially available, it was of the essence to develop an inexpensive modified version geared toward biodiesel analysis at decreased temperatures. As can be seen from Figure 10, the incident light from the monochromatic laser diode is initially diffused through HDPE #2 plastic. The diffuser will provide a wide uniform light beam in order to remove the potential interference of refractive index of various samples. The light is transmitted through a standard disposable polystyrene 1 cm spectrophotometric vial containing the sample. Any light that is not absorbed is 30

measured in a light-to-frequency converter and then translated using an inexpensive microcontroller onto a serial display. Figure 10: Biodiesel Spectrophotometer Apparatus The measurement of absorption is made by comparing the intensities of incident ( ) and transmitted ( ) light passing through pure methanol and test solutions, respectively. The term transmittance ( ), is the ratio of the radiant power transmitted by a sample to the radiant power incident on the sample as shown in equation 10. (10) equation 11. The Logarithm of the reciprocal of the transmittance is termed absorbance ( ) as shown in (11) Fundamentally, there are two laws of colorimetry, Lambert s Law and Beer s Law. Lambert s Law states that the amount of light absorbed is directly proportional to the logarithm of the length of the light path. Beer s Law states that the amount of light absorbed is directly proportional to the logarithm of the concentration of solute (67). Thus, the combination of the two laws gives 31

(12) where, is the extinction coefficient of the solute, is the concentration of the solute and is the length of the light path. There exists a linear relationship between absorbance and concentration of solute when the path length through the cell is constant. A plot of absorbance ( ) vs. concentration of solute ( will yield a straight line passing through the origin indicating conformity to the Beer-Lambert Law. 4.1.2. TSL230R Light to Frequency Converter A block diagram of the light to frequency converter is shown in Figure 12. The light sensor will convert irradiance into frequency. The square wave produced by the sensor has a 50% duty cycle, allowing for inexpensive equipment, such as an Arduino microcontroller, to register the high pulses. The output can be scaled using S2 and S3 pins shown in Figure 11. The full range of frequency output is from 1 Hz to 1 MHz depending on irradiance. Figure 11: TSL230R Block Diagram (68) Figure 12 depicts the spectral response of sensor. A simple and inexpensive red laser diode was chosen due to the high response of the light sensor at frequencies near red laser wavelength (650-670nm). Furthermore, the 600 nm wavelength is typically used for applications in biological samples such as measuring the optical density to determine cell concentration (69). 32

Figure 12: TSL230R Spectral Responsitivty at Various Wavelengths (68) Figure 13 depicts the sensitivity control of the light sensor. The sensitivity control will determine how many receptors are active on the chip at once. The sensitivity can be controlled by S0 and S1 pins as shown in Figure 11. When set at a high sensitivity, it will be able to detect smaller amounts of light but will lose the ability to register large amounts of light, and vice-versa for low sensitivity. The full range of irradiance that can be measured is from 0.001 μw/cm 2 to 0.1 W/cm 2. Figure 13: TSL230R Output Frequency (khz) as a Function of Irradiance (μw/cm 2 ) (68) The microcontroller was chosen for its capabilities of programming the light sensor as well as controlling the laser diode. The code used for the spectrophotometer system, for use in open source Arduino coding system, allowed for optimal light measurements to be obtained using a red laser diode. 33

4.2. Dielectric Spectroscopy of Biodiesel in MW Regime A dielectric is a material that has the ability to store energy when an electric field is applied, also known as permittivity (2). Dielectric relaxation spectroscopy (DRS) probes the molecular dipole moment of materials over a wide frequency range (1 mhz 30 GHz) (70). Information on chemical structure, molecular chain length and distribution and detection of contaminates in real time can be obtained (71-73). The relative permittivity is expressed as a complex function in Equation 8, (13) where, ε is the real part of the permittivity, which is a measure of how much energy from the external electric field is stored in the material per cycle. ε is the imaginary part of the permittivity, called the loss factor, and is a measure of how dissipative a material is when in the presence of an external electric field per cycle. A useful quantification tool that can also be used is known as the loss tangent, which is the phase angle between the excitation voltage and the current through the cell, is given by the quotient between the imaginary and real parts of shown in equation 14 (74), (14) There are four different types of dielectric mechanisms over a broad range of frequencies. In respect to measuring the dielectric loss factor of biodiesel, the dipolar mechanism was studied in the frequency range of 200 MHz-20 GHz. During the applied electric field, a polar molecule will exhibit the dielectric material to align and misalign to the electric field causing the permittivity to sharply decrease and the loss factor to peak at certain frequencies, as shown in Figure 14. 34

Figure 14: Dielectric Response Mechanisms (75) During isothermal conditions, the dielectric relaxation is identified by a peak of the loss permittivity, e, over a range of frequencies, Figure 15. The peak will correspond to a step decrease in the storage permittivity, e. Figure 15: Dielectric Storage Permittivity and Loss Permittivity (78) Initial studies on biodiesel and materials that impact biodiesel quality were investigated at PSU along the whole frequency span of 200 MHz 20 GHz, shown in Figures 16-18. A broadband vector network analyzer was utilized for the frequency generation and a coaxial probe is placed into a small beaker containing the FAME. Initially the coaxial probe is blanked with a short and with air. It has been found in previous studies that dielectric measurements can be used to detect levels of biodiesel blends (79-80), similar to onvehicle analysis of detecting alcohol levels in gasoline which also utilize dielectric spectroscopy (81). 35