Análisis de Biodiesel Mono Di y Triglicéridos

A Final Report to the National BiomEL BoA’Dl

Sample Analysis From Biodiesel Test

Final report submitted by the University of Missouri-Columbia, Agricultural Engineering Department

Designed to analyze samples taken during a NBB 1000 hour durability test

PRINCIPAL INVESTIGATOR

Leon G. Schumacher, Ph.D. Nancy Elser, Research Assistant

235 Agricultural Engineering Building Department of Agricultural Engineering

University of Missouri-Columbia Columbia, Missouri 65211

Phone: 573-882-2126 FAX: 573-884-5650

June 12,1997

Sample Analysis From Biodiesel Test

ExecutiveSummary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Review of Literature ......................................................... 7 Summary of Analytical Me thodr .......................................... 7 SummaT of Fueling Issues and Problems .................................. 10

Experiment& Procedures .................................................... 13 Ana&ses Performed at the Universiq of Missouri ............................ 13

Filter Samples .................................................. I3 X-Ray Microanalysis ...................................... I3 Scanning Electron Microscopy (SW) Analysis .................. I4 SoxhIet Extraction ........................................ I4

FueISampies ................................................ ..I 5 Inductively Coupled Plasma (ICP) Analysis ..................... I5 Filtration of Fuel Samples .................................. I6 Analysis of Filter Paper Samples ............................. I7 Nitrogen Analysis ......................................... 18 OrganicAnalysis ......................................... I9

Analyses Performed at SLS ............................................. i9 Organic Analysis ............................................... i9 Metals Analysis ................................................ 20 Karl Fischer Moisture ........................................... 20 Peroxide Value ................................................. 20 Total Acid Number .............................................. 21

Analyses Performed at ORTECH, Inchcape Testing Services, and Iowa State Univ. . . 21 FTIR Analysis ................................................. 22 Oxidative Stability .............................................. 24 Tests Performed at Iowa State University ............................. 25

Discussio>l of Results of Experimenlal Procedures . . . . . . . _ . . . . . . . . . _ _ _ _ _ . . . _ . . _ . . _ 2 7

Summary of Key Facts Noted and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Implications of Results . . . . . _ . . . . . . . . , _ . . , . . . . . . . _ . . . . . _ . . . . . . . . _ . . . . . _ . . . . . . 33

Recommendations _.._.._...._......._,_._._...._......._._...._...__....... 36

Refererlces................................................................ 38

Appendix I

Review of Literature ........................................................ 47 Review of Analytical Methods ........................................... 4 7

I. Analysis of Fatty Acid Alcohol Esters and Other Fatty Acid Derivatives ... 51 Gas Chromatography (CC) .................................. 51

Infrared Spectroscopy. ..................................... 52 OtherMethods ......................................... ..5 3

2. Analysis of Free and Bonded Glycerol, SteroIs and Methanol ........... 54 Mono-, Di- and Triglycerides ................................ 55

Gas Chromatography (CC) ............................ 56 Infrared Spectroscopy (?R) ............................ 5 7 High Performance Size EMusion Chromatography (HPSEC) . . 57 Other Method ..................................... 58

Free Glycerol ............................................. 58 Sterois ................................................. 59 Methanol ............................................... 60 OtherMethods ......................................... ..6 0

3. Analysis of Free Fatty Acids .................................... 61 Gas Chromatography ...................................... 61 Derivatization Methods. .................................... 62 Infrared Spectroscopy ...................................... 62 Other Method ........................................... 63

4. Analysis of Mixtures of Vegetable Oils with Hydrocarbon Oils .......... 64 5. PoIymerization Phenomena and Analysis of Polymerization Products Form

by Thermal or A&oxidation of Vegetable Oils ................... 66 Review of Literature Focusing on Fueling Issues and Problems .................. 7/

Appendix 2

Organic Analysis Performed at the University of Missouri . . . . . . . . . . . _ . . . _ _ . . _ . . 8 I

Diagram I

System and Sample Points _ . _ _ . _ . . . . . _ _ . . . _ . . . . . . . . . . . . . . _ _ . . . . _ _ _ . _ . . _ _ 8.5

Appendix 3

Results of Sample Analvsis Performed at University of Missouri Table I : List of Samples Analyzed at the Universiv of Missouri . . . _ . _ . . . . . 86 Table 2: Metals Analysis by ICP . . . . . . . . . . . . . . . . . . . . . . . . _ _ . . . _ . . . . . 89 Table 3: Nitrogen Analysis . . . . . . . . . _ . . . . _ . . . . . . _ . . . . . . . _ _ . . . _ . . . . . 91

Appendix 4

. . X-Ray Microanalyses of Filter Samples . . . . . . . . . . . . . . . . . . . . . . . . .

Appendix 5

Scanning Electron Microscopy of Filter Samples . . . _ , _ _ . . . . . _ . . _ . . . . . . _ _ . . . . . . . . . . . . . . . . . . . .

........ 92

....... 100 . .

Appendix 6

X-Ray Microanalysis of Filtered Fuel Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

Appendix 7

Results of Sample Analysis Performed at SLS Table I: List of Samples Analyzed at SLS . . . . . . . . , . . . . . . . . . . . . . . . . . . 156 Table 2: Organic Analysis . . . . . . . . . . . . . . ~ . _ . . . . . . . . . . . . . . . _ . . . . . . 157 Table 3: Metals Anaiysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 Table 4: Acid Number, Peroxide Number, and Moisture AnaIyses . . . . , . . . . 159

Appendix 8

FTIR Analysis Performed at ORTECH Corporation . . . . . . . . . . . . . . . . . . . . . . _ . . 160

Appendix 9

Oxidative Stabilig Performed at Inchcape Testing Services and Iowa State Univ. - . . I67

Sample Analysis From Biodiesel Test

Executive Summary

The overall purpose of this project was to analyze samples taken during a NBB 1000 hour

durability test. The researchers analyzed samples so as to provide the information necessary to

duplicate and subsequently eliminate the formation of fuel system residues that were noted during the

project. Specifically, the objectives of this project were to: 1) determine fuel properties/contaminants

so as to make it possible to duplicate the fuel conditions that caused fuel filter plugging and residue

formation; 2) determine chain length of the residue samples in order to determine whether the

deposits were of biodiesel or of petroleum origin; and 3) determine the elemental composition and

compounds present in the fuel and residue samples. Based on these analyses, the researchers

attempted to determine the fuel additives and contaminants that were most likely present.

During this project the researchers 1) reviewed literature concerning analyses of various

components and products formed during the transesterification of vegetable oils to fatty acid methyl

esters and research on the fueling of vehicles with biodiesel or biodiesel blends; 2) performed

laboratory analyses on samples taken at Ortech corporation after fueling a Cummins N14 diesel

engine with biodiesel and biodiesel blends; 3) attempted to identify the compounds present in the

residue formed during fueling; and 4) attempted to determine the origin of the residue formed during

fueling and replicate the conditions which caused residue formation.

In order to-achieve these objectives, the researchers performed a series of analyses at the

University of Missouri. Other analyses were contracted out to System Lab Services, Inc. (SLS) of

Kansas City, KS. Fuel properties (acid number, peroxide number, moisture content and metals

content) were performed on seven fuel samples at SLS. Included in these samples were two diesel

fuel samples from the 7000 gallon tank (see Diagram 1). One sample was taken without flushing.

Three 200,/o biodiesel/ 80% diesel fuel blend samples (denoted “B20”), from the top and bottom of

one barrel and from the 1000 gallon blend storage tanlq and four 100% biodiesel samples (denoted

“B 100”) were also induded. Organic analysis was performed on the four B 100 samples. Metals

content of both the unfiltered and filtered fuel samples was determined by Inductively Coupled Plasma

(ICP) at the University of Missouri Experiment Station Chem Labs (ESCL) . A comparison of these

values was made in order to determine if the metals were contained in the fuel or in the residue.

Organic analysis, which revealed the origin of the residue, was performed at both the University of

Missouri and at SLS. However, due to instrumentation problems at the University of Missouri, the

amount of free and bonded glycerol (mono-, di- and triglycerides) in the fuel samples could not be

determined. Vaiues for these components were obtained at SLS. Selected samples were also

analyzed for nitrogen content at the University of Missouri ESCL.

Analyses performed before the researchers received the samples included FTIR analysis

undertaken at ORTECH Corporation; oxidative stability (ASTM D 2274) at Inchcape Testing

Services and at SLS; test for particulate contamination (ASTM D 5452) at Inchcape; and simple

analytical tests at Iowa State University and at BRABMIJ, Inc. of Pittsburg, MO.

Results of the anaiyses revealed the following problems:

1. FTIR analyses indicated that the residue found in the float bowl, test ceil filter wall, and in

the methyl soyate fuel were the same. These analyses also indicated that the methyl ester

(carbonyl functional group) present in the fire1 was absent in the residue, indicating

carboxylate salt formation, Other bands supported a possible formation of oxidative polymers

2

in the residue.

2.

3.

4.

5.

The B 100 samples analyzed at SLS ail contained elevated levels (approx. 1000 ppm) of water,

as indicated by the ASTM D 4928 (Karl Fischer Moisture) values. In addition to water, all

four BlOO samples had high acid values according to ASTM D 664, with values ranging corn

0.89 to 0.96 mg KOWg.

Peroxide values (PV) measured by ASTM D 3703 indicated that all seven samples analyzed

at SLS were highly oxidized, with PVs ranging from 52 to 224. The PV of 224 that was

determined for the diesel fuel,sample from the 7000 gallon tank (taken without flushing)

resulted fi-om poor sampling techniques. This sample was therefore not representative of the

diesel fuel from the 7000 gallon tank.

Free glycerol values for the biodiesel samples analyzed at SLS were all under the National

Biodiesel Board specification value of 0.02%. In addition, bonded glycerol levels were not

higher than expected, leading the researchers to beiieve that free and bonded glycerol did not

contribute to the observed residue formation.

Inductively Coupled Plasma analysis performed at the University of Missouri indicated that

the fuel samples were contaminated with the metals iron, copper, zinc and aluminum.

Filtration of the fuel removed most of these metals, whose presence was confirmed on the

hlters by qualitative X-Ray Microanalysis. The most significant contamination was found in

the B20 from the 1000 gallon storage tank, which contained 26 ppm copper and 4 1 ppm zinc.

XRh4 of the filter also revealed high levels of copper and zinc. The high levels of zinc and

copper noted in this sample remains unexplained. The discrepancy between values obtained

at SLS by spectrochemical analysis and ICP at the University of Missouri can be explained

3

6.

by the differences in the techniques used.

Nitrogen analysis of the residue taken from the 1000 gallon B20 storage tank indicated the

formation of an ammonium salt.

The researchers were informed that a corrosion inhibitor and conductivity improver were

present as fuel additives in the diesel fuel used for the blended fuel samples. However, the researchers

could not determine specifically which compounds had been added, only the general types of

compounds in each class of additives. As such, we can only speculate if such additives contributed

to the residue formation observed.

Based on the information revealed in the analyses and on the information obtained from a

thorough literature search, the researchers feel the that a combination of oxidative processes of the

biodiesel and ammonium carboxylate sait formation contributed to the formation of the residue found

in the samples.

Sample Analysis From Biodiesel Test

Introduction

The National Biodiesel Board, during the fall of 1995, contracted the ORTECH Corporation

in Toronto, Canada to prep a N14 Cummins Engine for a 1000 hour durability test. The fuel that was

selected was a 20 %/ 80% blend of biodiesel and #2 petroleum low sulfur diesel fuel (320). A

durability cycle that was prepared by Cummins Engine Company was used to cycle the engine.

The engine performed according to specifications until it reached the 200 hour mark. Shortly

after 200 hours, however the engine experienced a drop in power. The trouble shooting that followed

revealed that the fuel filter on the engine was restricted. This phenomenon did m occur again on

the engine, as the subcontractor (ORTECH) replaced the filter on the test cell wall with a finer filter.

The engine operated in a normal manner until the 650th hour mark. At this point the engine

again experienced a drop in power. Troubleshooting that followed suggested that the PT pump on

the N14 was not providing the fuel flow required of the engine. The PT pump was replaced and

power output returned to normal.

The PT pump was disassembled by Cummins technicians. A tan-brown residue coated many

of the parts. The residue had accumulated until it restricted the ability of the pump to fuel the engine.

Representatives from NBB and ORTECH systematically gathered fuel and filter samples at

this point-in-time for the study. Samples were taken at five (5) different locations in the fuel

handling/delivery system. Samples were taken from storage tanks, mixing or blending tanks, day

tanks, fuel filters, and residue found in / or on some of the these tanks.

The NE3B subsequently contracted the University of Missouri - Columbia Chemistry

5

Department to conduct an analysis of these fuel and fiiter samples.

6

Review of Literature

While reviewing the literature concerning the production and use of biodiesel, the researchers

found information pertaining to two general areas. These included 1) analytical methods and 2)

fueling issues and problems encountered during the use of the biodiesel. In order to determine which

analytical techniques were most applicable to achieve the objectives of the study, a thorough review

- of the available literature was completed. A complete review of related literature can be found in

Appendix 1:

Summary of Analytical LMethods

A variety of analytical methods exist which monitor components present and/or which can

form during the production of and fueling with “biodiese!.” During the production process, vegetable

oils are reacted with an alcohol and catalyst, usually potassium hydroxide. Therefore, analytical

procedures are used to monitor the presence of un- or partially reacted starting materials (mono-, di-

and triglycerides or bonded glycerol, alcohol and catalyst), as well as the glycerol and the fatty acid

alcohol esters which are formed by the reaction. Any free fatty acids present in the soybean oiI

sample can react with the catalyst to form soaps (100). Undesirable components in the fatty acid

alcohol ester fuel include these un- or partially reacted starting materials, glycerol, and other

“contaminants” such as unsaponifiable matter (for example, sterois), free fatty acids, soaps, catalyst,

and water (97). In addition, any products which have formed during fkeling (such as polymers and

other products of oxidation) must also be monitored to determine fuel quality. Mixtures of the

biodiesel with diesel hel pose special analytical problems due to interference of the biodiesel

7

components from the diesel fuel hydrocarbons (62,63).

In addition to the known ASTM methods used to determine moisture content, peroxide value,

acid number, and other standard combustion characteristics, the researchers felt it necessary to

thoroughly investigate the availability of methods which directly analyze the triglyceride nature of the

biodiesel fuel. These incIude methods to analyze the following compounds:

1. Fatty acid alcohol esters and other fatty acid derivatives;

2. Bonded glycerol (mono-, di- and triglycerides);

3. Free fatty acids;

4. Mixtures of vegetable oils and/or FAMES with hydrocarbon (such as diesel) oils; and

5. Polymerization products formed by thermal or autoxidation of vegetable oils.

The most common methods involve chromatographic separations (1,4-17,21-23,25-29,32-33,36-

38,44-52,56-57,59-60). The type of separation used depends upon the type of compounds monitored.

For compounds or derivatives which are volatile, gas chromatography (GC) is the method of

choice (1,4-17,26-29,36-38,44-49,52,60). The two most common detection systems are flame

ionization (FID) and mass spectrometry (MS). Analyses for fatty acid methyl esters (FAMES) and

free fatty acids (FFAs) utilize gas chromatography. For compounds of higher molecular weight or

lower volatility such as the bonded glycerol components, a derivatization is necessary if GC analysis

is to be undertaken. In order to analyze bonded glycerol components by GC, silylation is performed.

This reaction adds a trimethylsilyl group to all hydroxyl (-OH) groups present in the fuel sample.

Two silylating agents are commonly used: N-methyl-N-trimethylsilyltrifluoroacetamide (MSTFA)

has been used by Plank and Lorbeer (27,28); Freedman and Pryde (26) and Plank and Lorbeer (1)

use N,O-bis(trimethyisilyl)trifluoroacetamide (BSTFA); and Bondioli et al. (29) use BSTFA with 1%

8

trimethylchlorosilane (TMCS). The reaction of BSTFA with a monogiyceride is shown:

Me

psi-Me

CF3C he 0

Me ‘\N CHTOH

+ A.&-O-tH -. ‘Si

M& 'Me

I

CHPH

0 II

CHZO-SicMe),

Ak-C-0-CH I CHZO-Si' Me) 3

Silyhred Monoglyceride

This addition increases the volatility of the compounds which contain the hydroxyl group. Since the

fatty acid methyl esters and triglycerides contain no free hydroxyl groups, they are not affected. Only

mono- and digiycerides and free fatty acids are silylated. This allows for GC analysis with adequate

separation of the components.

Compounds of higher molecular weight and volatility can also be analyzed by high pressure

liquid chromatography (HPLC) and high pressure size exclusion chromatography (HPSEC)

(25,32,38,50-51,56-57). HPSEC separates the components of a mixture based on size and is

therefore useful to examine oxidation products such as polymers (2). The use of thin-layer

chromatography (TLC) and supercritical fluid chromatography (SFC) has also been documented (21-

23,33,44,46,59-60). Other methods such as infrared (IR) spectroscopy, nuclear magnetic resonance

(NMR) and differential thermal analysis (DTA) have also been utilized in the analysis of compounds

which could be present in biodiesel(3,l l-12,1S-19,24,30-31,34,41-43,53-55,60).

One reference which has proved invaluable is the Hmdbook of Analytical Methods for Fatty

9

Acid Methyl Esters Used as Diesel Fuel Substifutes (1). It contains three chapters describing

analytical methods used for the quality control of vegetable oil alcohol ester fuels, including methods

to test for combustion parameters, degree of transesterification, bonded and free glycerol, iodine

number, water content, and fatty acid distribution and degree of unsaturation.

For the biodiesel, diesel fuel and blended fuel samples, the GC methods described by C. Plank

were determined to be the most applicable to achieve the objectives listed (1,27,28,38).

An article published by C. Evans et al. in 1970 provided the basis for metals analysis of the

fuel samples. The char-ashing of giy.ceride oils as a means of preparing a sample for atomic

absorption analysis was described (94). Samples of the biodiesel and biodieseVdiese1 fuel blends were

carbonized slowly in special crucibles and then ashed in a heated muffle furnace. Inductively coupled

plasma analysis (ICP) was used instead of atomic absorption at the University of Missouri. Another

article by Black (102) compared three atomic absorption techniques for determining metals in

soybean oil. This method proved less applicable due to the limitations using the direct aspiration

method, in which it is difficult to prepare reference solutions in a matrix similar to that of the fuel

samples.

Summary of Literature Focusing on Fueling Issues and Problems

During fueling tests performed with vegetable oils as diesel fuel substitutes (77-84), some

problems were encountered, mainly during tests of longer (> 10 h) duration (79,86). Injector coking

was the most common problem described (78,79). The injector nozzle deposits and carbon residues

observed differed in appearance from the dry carbon residue formed during diesei fuel tests. The

residue formed during fLeling with vegetable oil fuels was described as hard and shiny with a sticky

10

oily appearance (78)

Due to such problems encountered while fueling with vegetable oil fkels, the use of

transestetied fiels increased. Transesterikation of the oil provided a reduction in viscosity, thereby

improving the physical properties of fuel and improving performance (91,100). Although gum

formation problems were noted while fueling with transesterified vegetable oil tieIs, which lead to

fuel filter plugsng (86,91,97), no analysis of the filters was reported.

Storage and material compatibility issues while heling with biodiesel were addressed by a

number of studies (83,87,89,90,92,97,99,107). Du Plessis et al. investigated the influence of air,

temperature, light, antioxidants and contact with metal on the acid, peroxide and anisidine values,

ultraviolet (UV) absorption, viscosity, and induction periods of sunflower oil ethyl and methyl esters.

From the results, he deduced practical guidelines for the storage of fatty acid ester fiels (92).

Thompson et al. of the University of Idaho conducted a two-year storage study on both methyl and

ethyl esters of rapeseed oil (107). They discovered that, on average, properties of the esters such as

acid value, peroxide value and viscosity tended to increase over the two year time period. Of special

interest were the acid and peroxide values, because “both of these values [were] related to

autoxidation” of the fuel and “the acid values naturally increase with an increase in peroxides.”

Thompson et al. also found that the acid values for both the RME and REE fuel were not significantly

affected by the type of container used.

Other studies included data obtained Corn a French trial in which transit buses were fUeled

with a 30% rapeseed methyl ester (RIVE) / 70% diesel fuel blend (87). It was reported that lead-

coated fuel storage tanks were damaged, and that the detergent effect of the blend released deposits

that had accumulated on storage tank walls and pipes. As a result, filter replacements were necessary.

11

A review of the results of material compatibility studies indicated that nitrile rubber and

polyurethane foams undergo deterioration upon contact with biodiesel and biodiesel blends (90).

Korbitz suggested that components made from these materials be replaced with fluorine rubber (89).

Bessee et al. indicated also that nitrile rubber, nylon 6/6 and high-density polypropylene were

incompatibie with biodiesel and biodiesel blends, but concluded that Teflon and Viton materials were

the least affected of the elastomers they tested (96). Reed et al. observed no degradation of

aluminum, brass, steel or phosphatized tiei tanks in their material compatibility study (90). Bessee

et al. however noted heavy gum formation in samples of biodiesel or biodiesel blends with copper-

containing materials during their metals compatibility study (96). They also observed a sharp increase

in acid number over a six month period for these fuels when a steel or aluminum coupon was added.

‘Brown gum” and “light brown film” formation were noted for biodiesel blends with bronze and steel

(96). Although Masjuki et al. found increased levels of aluminum, chromium, lead and copper in

lubricating oil after fueling with blends of palm oil methyl ester and diesel fuel, values of wear metals

were considered to be within the “normal” range (83).

In a report to the National Biodiesel Board from April 3, 1997, System Lab Services noted

that failures occurred while operating Cummins PT pumps on off-spec, high acid value, blended fuel.

According to Van Gerpen, all internal surfaces of the PT pump fueled with the out-of-spec blend had

very noticeable coatings of a greenish-gold varnish (104).

12

Experimental Procedures

Diagram 1 indicates the test cycle of the Cummins engine used for the ORTECH Corporation

1000 hour durability study. A complete list of samples and the point where each sample was taken

is given in Diagram 1 and Appendix 3, Table 1. Samples were taken at three locations: prior to

blending; tier blending but before running through the system; and after blending and running

through the system (just before fueling in test cell). Samples with the designation “Inchcape” in

Appendix 3, Table 1 were previously analyzed for oxidative stability at Inchcape Testing Services in

Toronto, Canada.

Analyses Performed at the University of Missouri

Appendix 3, Table 1 provides a complete list of the samples analyzed at the University of

Missouri.

Filter Samples

X-Ray Microanalysis

X-Ray microanalysis (XRM) was performed with an Amray 1600 scanning electron

micrograph with a wavelength resolution of 2 nm and equipped with energy and wavelength

dispersive detectors. This procedure was used to qualitatively investigate the metals present in the

clogged fuel filter samples. Small sections of each filter were cut away and placed in evaporating

dishes for dry& The dishes were placed in an oven held at 105°C for 2-5 days until sufficiently dry.

A piece of an unused filter was dried as well for a blank sample. Samples were then transferred into

13

capped vials and sent for analysis. Analyses were performed by Lou Ross, Senior Electron

Microscope Specialist, of the University of Missouri Department of Geology. Results are shown in

Appendix 4. AU filter samples contain elevated levels of zinc, with smaller amounts of copper present

in about half of the samples.

Scanning Electron Microscopy (SEM) Analysis

The samples prepped as described above were also subjected to SEM analysis on the Amray

1600 SEM. This allowed for investigation into the image of the surface of the clogged filter samples.

Once again, analyses were performed by Lou Ross of the Geology Department at the University of

Missouri. Results are shown in Appendix 5. A comparison of the clean unused filter image with the

images obtained from the clogged filter samples shows the presence of a globular-like substance.

Soxhlet Extraction

A SoxhJet apparatus consisting of a Soxhlet extraction tube, Allihn reflux condenser, 125 mL

round-bottomed flask, and heating mantel (Fisher, St. Louis, MO) was used to extract the oily dark-

brown residue from the fuel filter samples. Extraction was performed with 100 mL of different

solvents (Fisher, St. Louis, MO): hexane (analytical grade), toluene (analytical grade), methylene

chloride (analytical grade) and methanol (Optima), Approximately 5 g of the fuel filter were weighed

into a Whatman 25 mm x SO mm cellulose thimble (Fisher, St. Louis, MO) and placed in the Soxhlet

extraction apparatus. The contents of the thimble were allowed to extract for either 2-5 hours or

overnight until the extraction solution ran clear into the round-bottom flask. Extraction of the filter

samples was attempted using a variety of extractants. Methanol and toluene proved successful in

14

earlier preliminary attempts with a similarly-dogged filter received from the University of Missouri.

However, these solvents resulted in a cloudy solution; a clear solution is necessary for organic

analysis. The cloudy solution obtained after extraction with toluene was subjected to solvent

evaporation using a Buchl Rotovapor R apparatus. Methylene chloride was added to the oily residue

left on the side of the flask. (In an earlier experiment on a similarly residue-clogged MU fuel fiiter,

it was discovered that methylene chioride dissolved the polyester adhesives corn the fuel filter,

making it useless as a primary extractant.) The oily residue did not, however, go into solution, even

after the solvent was gently heated. The non-polar solvent hexane was then probed. This solvent

extracted even less material off of the filter, as evident by the light color of the obtained solution, as

well as by the appearance of the filter after extraction. This would indicate that the residue was more

polar in nature. Due to the difficulty in obtaining a clear solution, further attempts at extraction were

abandoned.

Fuel Samples

Inductively Coupled Plasma (ICP) Analysis

Fuel samples were prepared for ICP analysis by carbonization and ashing, based upon a

method described by C. Evans et al. (94). Sampies were agitated on an Eberbach shaker until used.

Approximately 5 or 10 mL of the sample containing a representative amount of residue were weighed

into 30 or 50 mL Vycor crucibles and carbonized slowly on a hot plate until only a black carbon

residue (for the diesel fuel samples) or a sticky brown-black resin (for biodiesel and biodiesel blends)

remained. Some of the samples were spiked with either a 2 ppm aqueous iron solution or a solution

containing 2 ppm iron, aluminum, zinc and copper, The carbonization required approximately 3-5

15

days. The crucibles were allowed to cool slightly, and were then placed in a muffle fitmace and

heated to 500 + 15 “C overnight until only ash remained. Approximately 5 mL concentrated HCl

(Tracemetal grade, Fisher, St. Louis, MO) was added to each crucible with an Eppendorf pipette.

The crucibles were transferred to a hot plate and warmed gently for approximately % hour. Acid

solutions were quantitatively transferred to 10 mL volumetric flasks, and the crucibles were rinsed

twice with deionized water to ensure recovery of any remaining solution. The 5 mL flasks were filled

to volume tith deionized water. Samples were sent to the Experiment Station Chemistry Labs

(ESCL) at the University of Missouri for-ICP analysis of iron, aluminum, zinc, and copper. Analyses

were performed by Dr. Thomas Mawhinney, Director of the ESCL, on a Fisons ARL Model 3410

ICP unit and Mini-torch with a wavelength resolution of CO.02 nm. Results are tabulated in Appendix

3, Table 2.

It is interesting to note that all fuel samples contain uniform levels of aluminum, ranging from

approximately 1.5 to 4.5 ppm. The B20 sample 95El lB0027-9, taken from the 1000 gallon storage

tank, has elevated levels of copper (26 ppm), zinc (41 ppm) and iron (4 ppm), when compared to the

other samples. This sample was clear but did have reddish-brown particles adhering to the bottom

of the sample container. An Interchem B 100 sample (sample number 95El lB0077-18), taken from

bottom of barrel 95El lB0077-5 for comparison to the B 100 used to formulate the blends fueling the

test cell, contained a very extensive amount of reddish-brown matter. This sample showed the

highest level of metals contamination: 180 ppm iron

Filtration of Fuel Samples

In order to determine if the metals found in the fuel samples were contained in the residue,

16

filtration of the samples was undertaken All fuel samples were filtered through Whatman 2V

prepleated filter paper (Fisher, St. Louis, MO), which retained all particles with a diameter greater

than Spm. The atered fIrei samples were prepared for ICP metals analysis as described above. The

samples were once again analyzed by the ESCL at the University of Missouri. Results are given in

Appendix 3, Table 2, along with a comparison to the values obtained from the unf?ltered samples.

These results clearly indicate the almost all of the metals contained in the unfiltered fuel samples were

removed by filtration.

Analysis of Filter Paper Samples

The filter paper containing the retained particulates of the fuel samples was analyzed by Lou

Ross of the Geology Department of the University of Missouri using the X-ray microanalysis (XRM)

preparation and procedure described above. Small pieces of the fiiters were dried, then put into

capped vials and sent in for analysis. Each sample was scanned twice, using two different crystals.

Thallium acid phthalate (TAP) was used for energies ranging from 1.42 to 1.52 keV, to detect

aluminum; lithium fluoride (LiF) was used from 5.32 to9.5 keV, to detect other metals such as iron,

copper, zinc, nickel and chromium. Portions of each filter were analyzed, regardless if visible residue

was present on the filter. Diagrams are shown in Appendix 6.

There is a strong correlation between the existence of the metals in the unfiltered samples and

the presence of the same metals on the paper used to filter these fuel samples. For example, XRM

of the filter used with the B20 sample from the 1000 gallon storage tank indicated extensive copper

and zinc. Copper and zinc were also present in the unfiltered sample. It is interesting to note that

some samples contained very large amounts of aluminum, although the values for aluminum in the

17

unfiltered samples were essentially the same for all of the fuel samples. XRM of one BlOO fuel

sample filter (95EllB0077-14, from the top of barrel 95El lB0077-13) indicated extensive

contamination by aluminum., yet the sample was light and clear with no visible residue. This

correlates well with Bessee et ai.‘s findings that indicate high total acid numbers when aluminum was

exposed to biodiesel and biodiesel blends for a lengthy period of time, yet no discoloration was

observed (96). In general, the paper used in the filtration of most of the diesel fuel samples indicate

little metal contamination except for the sample taken without flushing (95E 1 lB0079- 1).

Nitrogen Analysis

Dr. Jim Waters, Senior Research Chemist at the ESCL of the University of Missouri,

performed nitrogen analysis on a Lice FP428 Nitrogen Analyzer. Selected samples were analyzed.

These included 95El lB0077-3 (engine filter from O-250 hrs); 95El lB0085-1 (residue kom 1000

gallon B20 storage tank); a diesel fuel sample (95El lB0077-20a) horn the 7000 gallon tank; a B20

sample (95E 1 lB0027-6a) fkom the top of barrel 27-6; and a B 100 sample, from the bottom of barrel

77-13 (95El lB0077-15). The B 100 sample had visible pieces of reddish-brown residue floating in

it. In addition, a clean piece of engine fuel filter and pure methyl soyate fuel from the University of

Missouri were analyzed for comparison.

Test results are given in Appendix 3, Table 3. High levels of nitrogen could indicate the

formation of an ammonium salt. The residue from the 1000 gallon B20 storage tank showed

unusually high levels of nitrogen when compared to the other samples. Jon Van Gerpen of Iowa

State University indicated that akenyl succinic amines (among other compounds) are commonly used

as corrosion inhibitors in diesel fkei, indicating a possible source of amines (105). These amines could

18

have combined with tiee fatty acids in the biodiesel to form ammonium salts.

Different levels of nitrogen were noted when analyzing the deposits on the filter sample as

compared to the residue taken from the 1000 gallon B20 storage tank. The engine fuel filter from

O-250 hrs did not show elevated levels of nitrogen when compared to the clean unused fuel filter.

This could indicate that the elevated temperature of the fuel may have inhibited the formation of the

ammonium salt. However, this hypothesis remains untested and should be interpreted with caution.

Organic Analysis

Organic analysis performed at the University of Missouri was inconclusive and selected

samples were therefore sent to System Labs, Inc. of Kansas City, Kansas, for analysis of bonded

glycerol components. For a complete description of the procedure used for organic analysis

performed at the University of Missouri, see Appendix 2.

Analyses Performed at System Lab Services, Inc.

Appendix 7, Table 1 includes the samples selected to be analyzed by System Labs. In addition

to organic analysis, peroxide value (ASTM D 3703) acid number (ASTM D 664) Karl Fischer

Moisture (ASTM D 4928) and total heavy metals (spectrochemical analysis) were performed.

Results are tabulated in Appendix 7 for all analyses performed at System Labs.

Organic Analysis

Four B 100 samples were analyzed at SLS for bonded glycerol and total glycerol, following

a procedure outlined by C. Plank (28). N-Methyl-N-trimethylsilyltrifluoroacetamide (MSTFA) was

19

used as a silylating agent. The internal standards tricaprin and 1,2,4-butanetriol were both 99% pure.

Standard solutions were prepared as described, and allowed to react at room temperature. Analysis

followed on a gas chromatograph equipped with flame ionization detection. Results are tabulated

in Appendix 7, Table 2. It is interesting to note that organic analysis revealed normal values for

bonded glycerol components, and the content of free glycerol was below 0.02%, which meets the

specification set by the National Biodiesel Board.

Metals Analysis

At SLS, a Baird Spectra Emission spectrometer was used for the heavy metals analysis, with

the following detection limits: 1 ppm for iron, chromium, lead, copper, zinc, vanadium, tin,

aluminum, nickel, silver, molybdenum, silicon, boron, sodium, phosphorous and titanium; 10 ppm for

calcium. barium, and magnesium; and 20 ppm for potassium. Samples were directly aspirated into

the arc, so no sample preparation was necessary. Results are given in Appendix 7, Table 3. In

general, values for metals are low compared to the analyses performed at the University of Missouri.

Karl Fischer Moisture

ASTM method D 4928 was used to determine the amount of water in the samples. Appendix

7, Table 4 lists the values obtained. The B 100 samples all contained high (approx. 1000 ppm) levels

of moisture.

Peroxide Value

ASTM method D 3703 was used to determine peroxide values for the samples. As indicated

20

by Appendix 7, Table 4, all fuels exhibited an advanced state of oxidation. The high value of 224

Meq /kg for the diesel fuel sample taken without flushing is unusual and probably resulted from the

sampling technique that was used. The ORTECH technician drew a sample which was not

representative of the diesel fuel in the 7000 gallon tank.

Total Acid Number

SLS followed the procedure outlined in ASTM method D 664 to determine the acid number

of the samples. Results are given in Appendix 7, Table 4. The values ranged from 0.01-0.03 mg

KOWg for the pure diesel fuel samples and from 0.10 to 0.27 for the B20 samples; much higher

values (0.89 to 0.96) were obtained for the BlOO samples. It was interesting to note that these values

were lower than the values reported by ORECH and Twin Rivers. Their analysis of the acid number

indicated acid values of 1.2 mg KOH/g.

Analyses Performed at ORTECH Corporation, Inchape Testing Services, and

Iowa State University

Before the researchers received the samples containing the residue under investigation, some

analyses were performed elsewhere. This included Fourier transform infrared spectroscopy (FTIR)

performed at ORTECH; oxidative stability performed by System Lab Services, Inc. (SLS) and

Inchcape Testing Services; and solubility studies and thin-layer chromatography (TLC) at Iowa State

University.

21

FTIR Analysis

FIB2 analyses were performed by ORTECH Corporation on October 23 and 24, 1995. The

spectra obtained are shown in Appendix 8. Two samples of B 100 (Interchem SME, 95E 1 lB0070- 10

and Twin Rivers SME, 95El lB0028-13a), two diesel fuel samples (Durability diesel, 95El lB0072-3

and Centrification diesel, 95El lB0073-1) and three residue samples were scanned from the

wavenumbers 4000 to 500 cm-’ and from 2000 to 400 cm“. The residue samples were from the test

cell fuel filter #l wall (95El lB0046-6) from the test cell float bowl (95El lB0056-lb) and one

residue sample was taken from BlOO fuel (95El lB0057-5a).

Dr. I-Iillar Auksi of the ORTECH corporation presented results of these analyses on October

26, 1995. He confirmed that all three residues were the same, and that they therefore originated in

the biodiesei and not the diesel fuel. Dr. Auksi suggested that the residue profiles were consistent

with a carboxylic acid salt derived from the biodiesel. Upon inspection of the spectra, some bands

do support this theory, although other information gleaned from the analyses indicated that this does

not appear to be the case. The intense C=O stretching band of the methyl ester, which usually

appears at 1750 + 50 cm-’ (106) has essentially disappeared in the residue spectra compared to the

B 100 samples, which is consistent with carboxylate COO- ion formation. In soaps, the usual C=O

stretching of the acid or the ester is replaced by two bands which arise from symmetrical and

antisymmetrical stretching vibrations of the COO- ion (1 1). Two weak bands at 1742 and 1584 cm-’

in the residue from the cell filter strengthen this argument. However, notable was the appearance of

a broad band in the O-H stretching region (approximately at 3400 cm-‘) in the residue samples. This

weakens the hypothesis of a metal type salt forming, because this band should be absent in such a

case. In addition, metal analyses perCorrned at both SLS and the University of Missouri revealed only

22

catalytic amount of metals, with no metal value high enough to have formed a salt with the

carboxylate ion.

. .

The researchers therefore hypothesize that one of two phenomena could be indicated by the

FTIR analyses: 1) the presence of an -0-H stretching band at 3450 cm-’ is indicative of autoxidation

of the firei, which has lead to polymer (i.e. residue) formation. Hydroperoxides formed during such

oxidative process would have given rise to characteristic -0-O-H stretching vibrations at

approximately 3 570 em-’ (IO). When Maier and.Tappel used IR to investigate products formed up05

the oxidation of unsaturated fatty acids (1 l), they found that bands from 3300 to 3570 cm“ indicated

the presence of hydroperoxyl and hydroxyl groups. In other studies utilizing IR spectroscopy in the

analysis of fatty acid derivatives, the ultraviolet-Light catalyzed oxidation of methyl linoleate gave rise

to monomers, dimers and trimers (10). The dimers exhibited a low degree of unsaturation but

contained an appreciable amount of -0-H groups. We feel that this is especially applicable to our

study, because B 100 contains approximately 50% methyl linoleate. In addition, some of the bands

arising from the unsaturation in the B 100, seen at 3009 cm -’ (-HC=CH- stretching) and

approximately 700 cm-’ (out of plane bending) in the Interchem SME sample, decreased in intensity

in the residue spectra, therefore indicating a decrease in unsaturation. This is consistent with

oxidative polymerization.

However, during a study of the IR spectra of highly unsaturated drying oils during

autoxidation, the ester carboxyl band at 1740 cm-’ was seen to increase and widen, “indicating

formation of compounds containirq other C=O groups not completely resolved in their spectra from

the ester carboxyl groups.” (10) In the FTIR analyses performed at ORTECH, the C=O band had all

but disappeared, indicating more likely carboxylate ion formation

23

This led to the possibility of 2) ammonium salt formation. Ammonium salts of carboxylate

acids exhibit spectra similar to those found in the ORTECH anaiyses. The region formerly identified

as containing the -0-H stretching band could in actuality be the -N-H stretch of the aminium group.

Typical for amine or ammonium salts is a very broad aminium band in the range between 3600 and

2000 cm-‘, which can be seen in all the residue samples (106). Since water was present in high

quantities (approximately 1000 ppm) and metals were found in catalytic amounts, we hypothesized

that hydrolysis of the fatty acid methyl ester occurred. Amines are common diesel fuel stability

enhancers and corrosion inhibitors; an ammonium salt could have formed between the carboxyiate

ion from the free fatty acid and the amine in the diesel fuel additive. This compound would give rise

to the observed spectra: weak bands of the COO- ion at approximately 1740 and 1590 cm-’ , and the

broad band at 3400 cm-’ , attributable to the -N-H stretch. It is interesting to note that ammonium

salts also contribute acidity to a sample, and the high acid value obtained by SLS also supports this

hypothesis.

Oxidative Stability

Inchcape Testing Services performed oxidative stability tests on the fuels in this study, using

ASTM D 2274. For a complete list of the fuels analyzed and the results, see Appendix 9. ASTM D

5452, “Standard Test for Particulate Contamination in Aviation Fuels by Laboratory Filtration,” was

also performed on the fuels. Results from Inchcape indicated that the B 100 fuel had good thermal

stability, and the B20 blend had intermediate thermal stability. Only one diesel fuel sample,

95El lB0013-10, had a low thermal stability, with 92.22 mg/lOO mL insolubles forming. Upon

receiving this result, the same fuel was analyzed at SLS, where the same test yielded the result of

24

0.69 mg/lOO mL insolubles (Appendix 9). It was therefore decided that the previous result Tom

Inchcape was an anoma.Iy, “caused by biodiesei contamination of the sampIe. This was probabIy

caused by biodiesel leaking into the sample Tom the hose connected to the mixing tank.” (103).

Tests Performed at Iowa State UniversiQ

Dr. Inmok Lee of Iowa State University performed some simple analytical tests on three

samples: 1) 9.5El lB0060-1, residue from mixing tank; 2) 95El lB0059-1, B20 from float bowl; and

3) 95El lB0057-1, BlOO from bottom dump. Dr. Lee performed inorganic anaiysis on sample 1, the

liquid portion of sample 2, and the residue from sample 2, which was recovered by decanting off the

liquid portion. He also included Interchem methyl soyate (B 100) in these analyses. Samples were

heated in a muflle fin-nace at 550°C for 3 hours, to bum off the organic materials. He believed the

recovered inorganic material to be metals, because the material showed magnetic properties.

Spec&aIly, he found the following weight percentages to be composed of inorganic material:

1) Sample 1: 9.82%

4 Residue in Sample 2: 15.76%

3) Liquid in Sample 2: 0%

4) Interchem methyl soyate: 0%

After centrihgation of sample 3, thin layer chromatography (TLC) was performed on the

liquid and the residue, and the free glycerol content was also determined for both. The liquid phase

had a free glycerol content of 0.01 $6, which is below the specification of 0.02% set forth by the

National Biodiesel Board. TLC of this liquid phase, however, showed the existence of a compound

other than the methyl esters (denoted compound “X”). No significant amounts of partial glycerides

25

were detected, and by using fke t&y acids as a standard, it was determined that free fatty acids were

also absent in the liquid portion. Compound “X” was located between the free fatty acids spot and

triglycerides standard spot, and Dr. Lee was unable to identify this compound.

The residue portion of sample 3 contained 6.7% free glycerol. The greatest portion of the

residue was found to be the fatty acid methyl esters of the fuel. Dr. Lee also performed simple

analytical tests on the residue. He found that the residue did not melt on a hot plate, which suggested

that the residue was not composed of partial glycerides, phospholipids or free fatty acids. The residue

did not dissolve completely in water, alcohol, chloroform, or hexane. However, the residue could

be separated into two portions, which differed in density. The heavy portion sank in chloroform,

which indicated that its density was greater than that of chloroform (d=1.492). The lighter portion

exhibited a density between 1.266 and 1.492, because it floated in chloroform but sank to the bottom

in a 2: 1 mixture of chloroform and methanol. Dr. Lee believed that the heavier portion consisted of

metals, but was unsure of the composition of the lighter portion.

26

Discussion of Results of the Experimental Procedures

Upon review of the results of analyses performed at the University of Missouri, System Lab

Services, Inc., Inchcape Testing Services, ORTECH Corporation and Iowa State University, certain

patterns were observed. In the samples for this study, elevated values of the metals iron, aluminum,

zinc and copper were found in some of the fuel samples. Filtration of the fuel removed most of these

metals. This is evident upon comparing the quantitative values obtained by inductively coupled

plasma (ICP) analysis before and after filtration Qualitative SEM and X-ray microanalyses revealed

the content and, to a certain degree, the,extent of metals contamination in the titers used to filter the

fuel samples.

There is a strong correlation between the values obtained by ICP and the qualitative metals

found on the filters. Similarly, emission spectroscopy performed by SLS indicated contamination by

the metals aluminum, iron, copper, chromium, tin, lead, zinc and molybdenum. The SLS values are,

however, substantially lower than those found by ICP analysis performed at the University of

Missouri. The discrepancies in these values were most probably a result of sample preparation

differences. At MU, representative samples with visible residue were carbonized and ashed. At SLS,

samples were aspirated directly into the emission spectrophotometer. It is doubtful that the red-

brown residue particles were aspirated properly into the instrument, which would have possibly led

to the plugging of the aspirator line. This would result in metal values that were much lower than

those obtained at MU.

Similarly, SEM and X-Ray Microanalysis of the cell wall and engine fuel filters used in the

test cell at ORTECH indicated elevated levels of zinc and copper, indicating contamination of the

biodiesel and biodiesel blends used to fuel the engine. The source of these metals was unknown, but

27

the high levels of zinc and copper noted in the fuel sample taken from the 1000 gallon tank clearly

suggests that contamination either occurred during blending or during storage in the tank.

AU B 100 fuel samples analyzed at SLS exhibited an advanced state of oxidation, high acid

number values, and high moisture content. This is consistent with results obtained as biodiesel ages

over longer storage periods (97,100). These characteristics all promote ammonium salt formation

and polymerization of the biodiesel, and as a result, residue formation in the blended B20 fuel was

inevitable.

28

Summary of Key Facts Noted and Conclusions

1.

2.

3.

4.

5.

6.

The amount of total and free glycerol did not appear to impact the formation of gum-like

materials. The biodiesei met the NBB specification concerning free and total glycerol.

The amount of mono-, di- and triglycerides did not appear to impact the level of gum

formation noted with the biodiesel. SLS did not find that this variable was out of

specification for the biodiesel.

As noted in the review of literature, the peroxide value (PV) of biodiesel in storage should

first increase and then decrease over time. PV is therefore only an incomplete indication of

the oxidative state of the sample.

It appears that the acid value of the fuels as tested by ORTECH and Twin Rivers was either

in error, or the value declined over time. Both ORTECH and Twin Rivers reported acid

values of 1.2 mg KOWg* for the fuel prior to testing, yet SLS determined that the acid value

one year later was between 0.89 and 0.96 mg KOWg.

*Although this value is above the current NBB specification, the fuel met the specification

for the test (NE3B specification: maximum acid number of 1 mg KOWg.)

Stratification of biodiesel occurs when stored for extended periods of time. This was evident

by the fact that the acid number for the biodiesel consistently differed between top and bottom

samples.

Proper sampling procedures, as with any testing procedure, must be used when preparing fuel

samples for analysis. The acid value and the oxidative stability of samples drawn without

using appropriate sampiing procedures were invalid.

29

7.

8.

9.

10.

Catalytic amounts of metals, which can lead to an elevated acid number, were acquired during

transport of the fuel. This was evident in that the samples were stored in plastic vessels prior

to use, yet the fire1 contained in these vessels had catalytic amounts of metals. For example,

all the samples had catalytic amounts of aluminum present. This can only be explained ifthe

biodiesel came into contact with aluminum during transit since none of the biodiesei was

stored in aluminum vessels or pumped through aluminum tubing.

Catalytic amounts of metals were acquired during storage of biodiesel blends. A significant

amount of the catalytic metals present in the blended fuel resulted from storage in the 1000

gallon red blend storage tank. This sample had significantly higher values for copper, zinc,

and iron than the other samples analyzed.

Steel and aluminum, as well as copper and zinc, act as catalysts to further the biodiesel

oxidation process and elevate acid number. This was found repeatedly in the review of

literature.

After careful review of the literature related to the polymerization of unsaturated vegetable

oils, it seems very likely that the conditions associated with the use of transesterified vegetable

oils as fuel are conducive to polymerization. FTIR analysis performed at ORTECH

Corporation supported the hypothesis that the residue formed during fueling with biodiesel

and biodiesel blends resulted from oxidative polymerization of the fuel. The intensity of the

bands in the Interchem and Twin Rivers SME (BlOO) samples resulting from unsaturation

(at 3009 cm-’ for -HC=CH- stretching and at 700 cm-’ for out-of-plane bending) decreased

significantly in the residue samples. This is indicative of a decrease in unsaturation, which is

consistent with oxidative polymerization.

30

11. Filtration of biodiesel and biodiesel blends can effectively remove the metals that are present

in catalytic amounts. MU filtered the samples and verified that the metals were either totally

removed or significantly reduced.

12.

13.

14.

X-Ray Microanalysis should be considered to analyze the metals content of any residue which

has formed from the biodiesel, as it worked effectively to qualitatively analyze biodiesel

residue on fuel filters and filter paper. A very strong correlation was noted between the

qualitative assessment of the residue and the quantitative analysis of the liquid sample.

A strong possibility exists that the catalytic amounts of metals present in the biodiesel

samples, combined with the 1000 ppm of water in the biodiesel, caused the hydrolysis of the

fatty acid methyl esters. Free fatty acids would have then been formed. These free fatty acids

could then have combined with metal ions, forming metal carboxylate salts, or with amine

compounds, forming ammonium salts. This would explain the presence of the weak

carboxylate (COO-) bands at approximately 1740 and 1580 cm-’ in the FTIR spectra of the

residue samples. tine compounds could originate from the diesel fuel, as these types of

compounds are often used as corrosion inhibitors and oxidative stability enhancers.

In the residue samples from the float bowl, cell wail filter and methyl soyate (B 100) fuel that

were analyzed by FTIR at ORTECH Corporation, ammonium salts were more likely present

than metal carboxylate salts as previously reported by ORTECH. The broad band at

approximately 3350 to 3400 cm-’ is indicative of either an -0-H or -N-H group, which is not

present in metal carboxylate salts.

15. Ammonium salts were formed while preparing the fuel for use in the engine. The residue

Corn the red 1000 gallon blend storage tank that was analyzed at MU had nitrogen levels that

would indicate the formation of such salts.

32

Implications of Results

The researchers believe that the preceding sections indicate that the residue in question arose

due to 1) polymerization of the biodiesel and 2) formation of carboxylate salts, most likely of

ammonium.

It is known 1) that vegetable oil methyl esters underwent polymerization reactions at elevated

temperatures, 2) that metal catalysts (Al, Fe) promoted polymerization and 3) that metals were found

in the sampies analyzed for this study. Therefore, the researchers believed that these factors

facilitated polymer formation during fueling with biodiesel and biodiesel blends. In addition, earlier

studies indicated that oxidative polymerization resulted in an increase in the polar fraction of the oil.

This could provide an explanation for the observations made during the attempted Soxhlet extraction

ofthe engine and cell wall filter samples collected at ORTECH Corporation. Extraction was much

more complete and resulted in a much cleaner filter after extraction with the more polar solvents

toluene and methanol, whereas the non-polar solvent hexane removed very little of the residue from

the filter, even after extracting overnight. This would indicate that the residue was indeed more polar

in nature, and could therefore have resulted from oxidative polymerization of the unsaturated

FAMES.

The study by Adams et al. very clearly showed the role that catalytic amounts of metals played

in the oxidative polymerization reaction. Using mixtures of soybean oil, motor oil, and water, no

hardening of the heated, a$tated mixtures occurred until catalytic amounts of metal were added. We

therefore hypothesized that a similar phenomenon would be observed when mixtures of a

hydrocarbon oil (i.e. diesel fuel), unsaturated soybean oil derivatives (FAMES), metals, and moisture

(evident by the Karl Fischer Moisture ASTM D 4925) are subjected to high temperatures and

33

agitation, which occurs during fUe!ing.

Organic analysis of the tie! samples did not reveal higher than expected levels of bonded

glycerol (mono-, di- and triglycerides) or free giycero!. The weight % of free glycerol in the selected

samples analyzed by SLS was less than 0.02% in each sample, which meets the stringent specification

for biodiesel set by the National Biodiesel Board. This led the investigators to believe that the residue

and gum formation phenomena was not directly a result of increased glycerol. This is consistent with

Peterson et a!.‘~ findings (86), where they reported an ? value of less than 0.01 between the tota!

glycerol content and injector coking. This study concluded that the injector coking was more likely

related to the molecular weight and viscosity of the tie!, which would be consistent with fuel

polymerization. Increase in molecular weight and viscosity are indications of polymerization (65).

FTIR analyses performed at ORTECH Corporation supported the hypothesis that

polymerization occurred. The presence of a broad band at approximately 3400 cm-’ in the residue

samples is usually consistent with the stretch of the hydroxy! (-0-H) group, indicative of autoxidation

of the tie!. Hydroperoxides formed during oxidation give rise to such characteristic bands. Studies

have shown that the presence of bands in the region 3300 to 3570 cm-’ indicated hydroxy! and

hydroperoxy! groups. These bands appeared after the oxidation of unsaturated fatty acids.

ORTECH suggested that the residue under investigation was composed of a carboxylate salt,

due to the appearance of two weak bands in place of the usual intense C=O of the carbonyl methyl

ester at 1750 + 50 cm-’ . However, it does not appear that the salt in question is a metal carboxylate -

salt, due to the presence of the -0-H group stretch. This band would be absent in a metal carboxylate

salt. However, the aminium group in an ammonium salt exhibits a very broad band in the range

between 3600 and 2000 cm-‘, leading the researchers to believe that an ammonium salt could have

34

formed. The high water content of the biodiesel samples, along with catalytic amounts of metals,

could have hydrolyzed the fatty acid methyl esters. The free fatty acids which resulted could have

reacted with amine compounds present as additives in diesel fuel to form such an ammonium salt.

This also provided an explanation as to why more gum residue was encountered when fueling with

blends of biodiesel and petroleum diesel fuel.

35

Recommendations

1) The investigators believe that acid number should be strictly controlled from the beginning

of production. High acid number levels of biodiesel lead to corrosion of storage containers and

engine components with biodiesel.

2) The amount of metals in the biodiesel fuel that can act as catalysts for a variety of reactions,

including oxidative polymerization, should be minimized.

3) A study should be designed to investigate metal deactivators for biodiesel. The only other

alternative to this would be to transport biodiesel in non-metal or stainless steel containers.

4) Jfmetal deactivators prove to be ineffective, filtration of the biodiesel prior to use might be

a viable alternative to reduce the amount of catalytic metals acquired through transit of the fuel to the

end user.

5) A more thorough study of biodiesel polymerization tendencies should be undertaken. This

investigation shouid include an analysis of commercially available anti-polymerization and

antioxidative agents.

6) Biodiesel should be agitated prior to use after extended storage periods.

7) Based on the review of literature and the fact that catalytic amounts of metals were present

in our samples, aged biodiesel should be not used for fuel purposes.

8) Traditional methods used to analyze petroleum diesel fuels for metals content, such as Atomic

Absorption and Inductively-Coupled Plasma spectroscopy, must be modified to analyze biodiesel

samples with visible residue particles. Directly aspirating the sample into the instrument is

insufficient, because the metals contained in the residue will either not make it into the instrument,

or the residue will plug the aspirator line.

36

9) The oxidative stability enhancers and corrosion inhibitors commonly added to diesel fuel must

be chosen with care before blending with biodiesel. Further, we hypothesize that these additives in

the presence of catalytic amounts of metal and water promote high acid levels that lead to gum

formation. Any compatibility study, either planned or underway, should include catalytic amounts

of metals in the tiei samples. Without a catalyst, some of the possible carboxylate salts will less likely

form, thereby rendering results obtained from such compatibility studies less useful.

10) Storage tanks for biodiesel and biodiesel blends should be chosen very carefully, particularly

if the fuei has aged. Catalytic levels of aluminum, iron, copper and zinc were noted in the analysis

in spite of the fact that the fuels were stored in high density polypropylene containers.

37

References

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

Bailer, J. et al. “Handbook of Analytical Methods for Fatty Acid Methyl Esters Used as Diesel Fuel Substitutes,” Research Institute for Chemistry and Technology of Petroleum Products, University of Technology, Vienna Austria, Fichte Publishing, 1994.

Skoog, D. and Lear-y, J. “Principles of Instrumental Analysis,” Saunders College Publishing, Fourth Edition, 1992.

Panford, J. et al. “Determination of Oil Content of Seeds by NIR: Influence of Fatty Acid Composition on Wavelength Selection,” JAOCS (1990), 67(8), 473-482

Bohov, P. et al. “Analysis of Fatty Acid Methyl Esters on an SP 2340 Glass Capillary Column” Journal of Chromatography, 286 (1984) 247-252

Herman, I. et al. “Routine Gas Chromatographic/‘Mass Spectrometric Analysis of Fatty Acid Methyl Esters Using an Ion Trap Detector,” Biomed. Environ. Mass Spectrom. (1989), 18(1 l), 1016-22

Murata, T. “Analysis of Fatty Acid Methyl Esters by a Gas-Liquid Chromatography- Chemical Ionization Mass Spectrometry Computer System,” J. Lipid Res. (1978) 19(2), 166-71

Vine, J. “Analysis of Fatty Acid Methyl Esters by High-Resolution Gas Chromatography-Chemical Ionization Mass Spectrometry,” Journal of Chromatography, 196 (1980) 415-424

Scribe, P. et al. “Identification of the Position of the Stereochemistry of the Double Bond in Monounsaturated Fatty Acid Methyl Esters by Gas Chromatography/Mass Spectrometry of Dimethyl Disulfide Derivatives,” Anal. Chem. 1988, 60, 928-93 1

Kolb, B. et al. “Radio Gas-Chromatographic Analysis of Fatty Acid Methyl Esters and Steroids,” Fresenius’ Z. Anal. Chem. (1968) 243, 129-42

Hofmann, M. “Gas-Chromatographic Analysis of Fatty Acid Methyl Esters Having 12-18 Carbon Atoms with a Thermal Conductivity Cell,” Z. Med. Labortech. (1969) 10(2), 75- 83

Doumenq, P. et al. “GC/FT-lR Analysis of Fatty Acid Methyl Esters,” Applied Spectroscopy (1990),44(S), 1355-1359

3s

(12)

(13)

(14)

(16)

(17)

(18)

(19)

(20)

(21)

(22)

(23)

Lanser, A.C. et al. “Comparison of FTlR and Capillary Gas Chromatographic Methods for Quantitation of frans Unsaturation in Fatty Acid Methyl Esters,” JAOCS (1988), 65(g), 1483-1487

Craske, J. et ai. Letter to the Editor, JAOCS (1988), 65(7), 1190-I 19 1

Bannon, C. et al. “Analysis of Fatty Acid Methyl Esters with High Accuracy and Reliability. VI*. Rapid Analysis by Split Injection Capillary Gas-Liquid Chromatography,” Journal of Chromatography, 407 (1987) 23 l-241

Albertyn D. et al. “Analysis of Fatty Acid Methyl Esters with High Accuracy and Reliability. I. Optimization of Flame-Ionization Detectors with Respect to Linearity,” Journal of Chromatography, 247 (1982) 47-6 1

Bannon, C. et al. “Analysis of Fatty Acid Methyl Esters with High Accuracy and Reliability. V. Validation of Theoretical Relative Response Factors of Unsaturated Esters in the Flame Ionization Detector,” JAOCS (1986) 63(l), 105-l 10

Fabien, R. et al. “Quantitative Analysis of Synthetic Mixtures of Triacylglycerols with Fatty Acids from Caprylic to Stearic,” JAOCS (1993) 70(6), 55 l-554

O’Connor, R. “Application of Infrared Spectrophotometry to Fatty Acid Derivatives,” JAOCS (1956), 33(l), l-15

O’Connor, R. “Recent Progress in the Applications of lR Absorption Spectroscopy to Lipid Chemistry,” JAOCS (196 l), 38( 1 l), 648-659

Zirrolli, J. et al. “Low-Energy Tandem Mass Spectrometry of the Molecular Ion Derived from Fatty Acid Methyl Esters: a Novel Method for Analysis of Branched-Chain Fatty Acids,” J. Am. Sot. Mass Spectrom. (1993) 4(3), 223-9

Cocks, S. et al. “Analysis for Fatty Acid Methyl Esters by Using Supercritical Fluid Chromatography with Mass Evaporative Light Scattering Detection,” Anal. Proc. (Lo&x$ (1991), 28(l), 11-12

Nomura, A. et al. “Supercritical Fluid Chromatographic Determination of Fatty Acids and Their Esters on an ODS-Silica Gel Column,” Anal. Chem. 1989, 6 1, 2076-2078

Svetashev, V.I. et al. “Analysis of Labeled Fatty Acid Methyl Esters by Argentation and Reversed-Phase Two-Dimensional Thin-Layer Chromatography,” Journal of Chromatography, 330 (1985) 396-399

39

(24) Rey, F. et al. “Characterization of Fatty Acid Methyl Esters by Thermal Analysis,” J. Therm. Anal. (1993), 40(2), 463-73

(25) Fillieres, R. et al. “Ethanolysis of Rapeseed Oil: Quantitation of Ethyl Esters, Mono-, Di- and Triglycerides and Glycerol by High-Performance Size-Exclusion Chromatography,” JAOCS (1995), 72(4), 427-432

(26) Freedman, B. et al. “Quantitation in the Analysis of Transesterified Soybean Oil by Capillary Gas Chromatography,” JAOCS 1986, 63 (lo), 1370-1374

(27) Plank, C. et al. “Simultaneous Determination of Glycerol, and Mono-, Di- and Triglycerides in Vegetable Oil Methyl Esters by Capillary Gas Chromatography,” Journal of Chromatography A, 697 (1995) 46 l-468

(28) Plank, C. “Gas Chromatographic Determination of Glycerol, Mono-, Di- and Triglycerides in Vegetable OiI Methyl Esters,” Method published by the Institute of Organic Chemistry of the University of Vienna.

(29) Mariani, C. et al. “Vegetable Oil Derivatives as Diesel Fuel Substitutes. Analytical Aspects. Note 1: Determination of Methyl Esters, Mono-, Di- and Triglycerides,” Riv. Ital. Sostanze Grasse (199 l), 68( lo), 549-55 1

(30) O’Connor, R. “Near-Infrared Absorption Spectroscopy--A New Tool for Lipid Analysis,” JAOCS (1961) 38(1 l), 641-648

(3 1) Chapman, D. “Infrared Spectroscopic Characterization of Glycerides,” JAOCS (1960), 37(2), 73-77

(32) Christopoulou, C. et al. “High Performance Size Exclusion Chromatography of Fatty Acids, Mono-, Di- and Triglyceride Mixtures,” JAOCS (1986), 63(5), 679-684

(33) Freedman, B. et al. “Thin Layer Chromatography/Flame Ionization Analysis of Transesterified Vegetable Oils,” JAOCS (1984), 61(7), 1215-1220

(34) Yamazaki, M. et al. “High Pressure Differential Thermal Analysis (HE’DTA) of Fatty Acid Methyl Esters and Triglycerides,” JAOCS (1980) February, 59-60

(35) Bondioli, P. et al. “Vegetable Oil Derivatives as Diesel Fuel Substitutes. Analytical Aspects. Note 2: Determination of Free Glycerol,” Riv. Ital. Sostanze Grasse (1992), 69(l), 7-9

40

(36) Mittelbach, M. “Diesel Fuel Derived from Vegetable Oils, V [l]: Gas Chromatographic Determination of Free Glycerol in Transesterified Vegetable Oils,” Chromatographia (1993), 37(1 l/12), 623-626

(37) Hayes, L. et al. “Cholesterol Interference in Analysis of Fatty Acid Methyl Esters,” Lipids (197 1), 6(l), 65-6

(38) Plank, C. et al. “On-Line Liquid Chromatography-Gas Chromatography for the Analysis of Free and Esterified Steroids in Vegetable Oil Methyl Esters Used as Diesel Fuel Substitutes,” Journal of Chromatography A, 683 (1994) 95-104

(39) Bondioli, P. et al. “Vegetable Oil Derivatives as Diesel Fuel Substitutes. Analytical Aspe.cts. Note 1: Determination of Methanol,” Riv. Ital. Sostanze Grasse (1992), .69(g), 467-469

(40) Basu, H. et al. “Preparation of Glycol Derivatives of Partially Hydrogenated Soybean Oil Fatty Acids and Their Potential as Lubricants,” JAOCS (1994), 71(1 I), 1227-1230

(4 1) Wollenberg, K. “Quantitative High Resolution 13C Nuclear Magnetic Resonance of the Olefinic and Carbonyl Carbons of Edible Vegetable Oils,” JAOCS (1990), 67(8), 487- 494

(42) Konishi, K. et al. “Quantitative Analysis of 1,2-Propanediol Monoesters of Long Chain Fatty Acids by Fluorine Magnetic Resonance,” Anal. Chem. (1968), 40(12), 1881-3 494

(43) Gelbard, G. et al. “‘H Nuclear Magnetic Resonance Determination of the Yield of the Transesterification of Rapeseed Oil with Methanol,” JAOCS (1995), 72( IO), 123 9- 124 1

(44) Ulberth, F. et al. “Simplified Method for the Determination of tram Monoenes in Edible Fats by TLC-GLC,” JAOCS (1992), 69(8), 829-83 1

(45) Williams, M. et al. “Rapid Determination of Free Fatty Acids in Vegetable Oils by Gas Liquid Chromatography,” JAOCS (1983), 60(8), 1507-l 509

(46) Oette, K. et al. “Mikromethode zur Schnellen Umestenmg von Lipoiden auf Duennschichtplatten mit Natriummethylat fuer die Gas-Chromatographische Analyse der Fettsaeuremethylester,” J. Chromatog. 32 (1968) 439-450

(47) Ballesteros, E. et al. “Determination of Free Fatty Acids in Dairy Products by Direct Coupling of a Continuous Preconcentration Ion-Exchange-Derivatization Module to a Gas Chromatograph,” Anal. Chem. 1994, 66, 628-634

41

(48) May, W. et al. “An Automated Gas-Liquid Chromatographic Method of Measuring Free Fatty Acids in Canola,” JAOCS (1993), 70(3), 229-233

(49) Jie, M. S. F. Lie Ken et al. “CC-MS of Saturated and Unsaturated Even and Odd- Number Long Chain Fatty Acids by Analysis of the Picolinyl Esters. Application to a Chinese Medicinal Seed (Cuscuta Semen) Oil,” J. Int. Fed. Clin. Chem. (1991), 3(3), 122, 124, 126, 128

(50) Miwa, H. et al. “High-Performance Liquid Chromatographic Analysis of Serum Long- Chain Fatty Acids by Direct Derivatization Method,” Journal of Chromatography, 416 (1987) 237-245

(51) Fatica, M. “HPLC Determination of Long Chain Saturated Fatty Acids in Tall Oil Rosin,” JAOCS (1989), 66(4), 572-575

(52) Vreeken, R. et al. “The Derivatization of Fatty Acids by (Chloro)Alkyl Chloroformates in Non-Aqueous and Aqueous Media for GC Analysis,” Journal of High Resolution Chromatography (1992) 15( 12), 785-790

(53) Sato, T. et al. “Nondestructive Determination of Fatty Acid Composition of Husked Sunflower (Helianthus annua L.) Seeds by Near-Infrared Spectroscopy,” JAOCS (1995) 72(10), 1177-1183

(54) Sato, T. et al. “Near Infrared Spectral Patterns of Fatty Acid Analysis from Fats and Oils,” JAOCS (1991), 68(1 l), 827-833

(55) Lanser, A. et al. “FTlR Estimation of Free Fatty Acid Content in Crude Oils Extracted from Damaged Soybeans,” JAOCS (1991), 68(6), 448-449

(56) Teng, J. “Reversed-Phase High Performance Liquid Chromatography of Underivatized Fatty Acids by Fatty Acid Analysis Column,” Journal of Liquid Chromatography, 15(g), 1473-1485 (1992)

(57) Suzuki, T. et al. “Determination of Free Fatty Acids in Marine Phytoplankton Causing Red Tides by Fluorometric High-Performance Liquid Chromatography,” JAOCS (1995), 72(10), 1211

(58) Caldwell, K. et al. “Nondiscriminatory Fast Atom Bombardment Matrix for Fatty Acid Mixture Analysis,” Anal. Chem. 1989, 6 1, 494-496.

(59) Ohshima, T. et al. “Cod Lipids, Solvent Systems and the Effect of Fatty Acid Chain Length and Unsaturation on Lipid Class Analysis by Iatroscan TLC-FID,” JAOCS (1987) 64(2), 2 19-223

42

(60) Favier, J. et al. “Detection of Total Trans Fatty Acids Content in Margarine: An _ Intercomparison Study of GLC, GLC+TLC: FT-KR, and Optothermal Window (Open Photoacoustic Cell),” Anal. Chem. 1996, 68, 729-733

(61) Canham, J. et al. “Automated Free Fatty Acid Determination Using Flow Injection Analysis Solvent Extractions,” JAOCS (1987), 64(7), 1004-1007

(62) Bondioli, P. et al. “Vegetable Oil Derivatives as Diesel Fuel Substitutes. Analytical Aspects. Note 4: Determination of Biodiesel and Diesel Fuel in Mixture,” Riv. Ital. Sostanze Grasse (1994), 7 l(6), 287-289

(63) Jha, J. “IR & UV Spectroscopic Studies of Mixtures of Mustard Oil & Some Other Triglyceride Oils with Hydrocarbon Oils,” J. Nep. Chem. Sot., 1988, 8, 16-30

(64) Prankl, H. et al. “Influence of the Iodine Number of Biodiesei to the Engine Performance,” Proceedings of the Third Liquid Fuel Conference, 15-17 September 1996, Nashville, TN

(65) Bailey’s Industrial Oil and Fat Products, Daniel Swem, Editor. Third Edition, John Wiley and Sons, 1964.

(66) Wexler, H. “Polymerization of Drying Oils,” Chemical Reviews (1964), 64(6), 59 l-6 11

(67) Korus, R. et al. “Polymerization of SafYlower and Rapeseed Oils,” IAOCS (1984), 61(3), 537-540

(68) Liu, H. et al. “Oxidative Stability of Soybean Oils with Altered Fatty Acid Compositions,” JAOCS (1992), 69(6), 528-532

(69) Liu, H. et al. “High-Temperature Stability of Soybean Oils with Altered Fatty Acid Compositions,” JAOCS (1992), 69(6), 533-537

(70) Sinchez-Muniz, F. et al. “Sunflower Oil Used for Frying: Combination of Column, Gas and High-Performance Size-Exclusion Chromatography for Its Evaluation,” JAOCS (1993), 70(3), 235-240

(71) Sebedio, J. et al. “Heat Treatment of Vegetable Oils I. Isolation of the Cyclic Fatty Acid Monomers from Heated Sunflower and Linseed Oils,” JAOCS (1987), 64(7), 1026-I 032

(72) Mkquez-Ruiz, G. et al. “Quantitation and Distribution of Altered Fatty Acids in Frying Fats,” JAOCS (1995), 72(10, 1171-l 176

43

(73

(74)

(75)

(76)

(77)

(78)

(79)

(80)

(81)

(82)

(83)

(54)

Chtist0p0ulou, C. et al. “Chromatographic Studies on Fatty Acid Dimers: Gas-Liquid Chromatography, High Performance Liquid Chromatography and Thin-Layer Chromatography,” JAOCS (1989) 66(9), 1353-1359

Christopoulou, C. et al. “Dimer Acids: Synthesis and Mass Spectrometry of the Tetrahydroxy, Dihydroxy, and Diketo Dimers of Methyl Stearate,” JAOCS (1989), 66(9), 1344-1352

Christopoulou, C. et al. “High Performance Size Exclusion Chromatography of Monomer, Dimer and Trimer Mixtures,” JAOCS (1989), 66(9), 1338-1343

ASTM Method D 2274, “Standard Test Method for Oxidation Stability of Distillate Fuel Oil (Accelerated Method),” 1995 Annual Book of ASTM Standards, Vol. 05.01, 776-780

Engler, C. et al. “Effects of Processing and Chemical Characteristics of-Plant Oils on Performance of an Indirect-Injection Diesel Engine,” JAOCS (1983), 60(8), 1592-l 595

Ziejewski, M. et al. “Laboratory Endurance Test of a Sunflower Oil Blend in a Diesel Engine,” JAOCS (1983) 60(8), 1567-l 573

Pryde, E. “Vegetable Oils as Diese! Fuels: Overview,” Paper presented at the 73rd AOCS Annual Meeting, May 2-6, 1982

Rakopoulos, C. “Olive Oil as a Fuel Supplement in DI and IDI Diesel Engines,” Energy (1992), 17(8), 787-790

Rewolinski, C. “Suntlower Oil Diesel Fuel: Engine Wear Implications,” JAOCS (1985), 62(1 l), 1598-1599

Humke, A. et al. “Performance and Emissions Characteristics of a Naturally Aspirated Diesel Engine with Vegetable Oil Fuels. Part 2,” Sot. Automot. Eng., [Spec. Publ.] SP (198 l), Sp-495 (Diesel Combust. Emiss., Pt. 3), 25-35

Masjuki, H. et al. “A Rapid Test to Measure Performance, Emission and Wear of a Diesel Engine Fueled with Palm Oil Diesel,” JAOCS (1993) 70( lo), 102 l-1025

Bettis, B. et al. “Fuel Characteristics of Vegetable Oil from Oilseed Crops in the Pacific Northwest,” Agron. J. (1982), 74(2), 335-9

Adams, C. et al. “Investigation of Soybean Oil as a Diesel Fuel Extender: Endurance Tests,” JAOCS (19X3), 60(8), 1574-I 579

44

(86) Peterson, C. et al. “Processing, Characterization and Performance of Eight Fuels from Lipids,” ASAE Paper No. 94653 1

(87) van Walwijk M. “Biodiesel Particulate Emissions and Fuel Storage,” IEA/AFIS Publication, October 1995

(88) Horstmann, B. et al. “Rape Methyl Ester (RME) and Soy Methyl Ester (SME) Test Experience.”

(S9) Korbitz, W. “The Technical, Energy and Environmental Properties of BioDiesel.”

(90) Reed, T. et al. “Development and Commercialization of Oxygenated Diesel Fuels from Waste Vegetable Oils,” Biomass and Bioenergy (1992), 3(2), 11 l-l 15

(91) Clark, S. et al. “Methyl and Ethyl Soybean Esters as Renewable Fuels for Diesel Engines,” JAOCS (1984), 61(10), 1632-1638

(92) Du Plessis, L. et al. “Stability Studies on Methyl and Ethyl Fatty Acid Esters of Sunflowerseed Oil,” JAOCS (1985) 62(4), 748-752

(93) Lee, I. “Use of Branched-Chain Esters to Reduce the Crystallization Temperature of Biodiesel,” JAOCS (1995) 72(10), 1155-1160

(94) Evans, C. et al. “Char-Ashing of Glyceride Oils Preliminary to the Atomic Absorption Determination of Their Copper and Iron Contents,” JAOCS (1971), 48, 840

(95) ASTM Method D 664, “Standard Test Method for Acid Number of Petroleum Products by Potentiometric Titration,” 1995 Annual Book of ASTM Standards, Vol.05.01, 237-243

(96) Bessee, G. B. et al. “Compatibility of Elastomers and Metals in Biodiesel Fuel Blends,” SAE Technical Paper Series 971690, 1997.

(97) Van Gerpen, J. et al. “Determining the Influence of Contaminants on Biodiesel Properties,” SAE Technical Paper Series 971685, 1997.

(98) Steiner, I. et al. “Chemische Untersuchungen an Speiseolen under besonderer Beriicksichtigung polyensaurereicher idle,” Fat Sci. Technol., 95. Jahrgang, Nr. 12, 1993

(99) “Determination of Additive Compatibility and Efficacy Project,” Final Report to the National Biodiesel Board, 1997.

(100) Erickson, D.R., Editor, “Practical Handbook of Soybean Processing and Utilization,” AOCS Press, Champaign, IL, 1995.

45

(101) Tian, L.L. et al. “Antipolymerization Activity of Oat Extract in Soybean and Cottonseed Oils Under Frying Conditions,” JAOCS (1994), 71(10), 1087-1094

(102) Black, L. T. “Comparison of Three Atomic Absorption Techniques for Determining Metals in Soybean Oil,” JAOCS 52, 88 (1975).

(103) Schumacher, L. and Van Gerpen, J. N14 Trip Report, PT Pump Analysis and Fuel Sampling, 1 l/3/95.

(104) “Diesel Fuel Pump Evaluation and Analysis,” Final Report to the National Biodiesel Board, 413197.

(105) Personal Communication with Jon Van Gerpen of Iowa State University, 5/23/97.

(106) Roeges, N. “A Guide to the Complete Interpretation of Infrared Spectra of Organic Structures,” John Wiley and Sons Publishing, West Sussex England, 1994.

(107) Thompson, J.C. et al. “Two Year Storage Study with Methyl and Ethyl Esters of Rapeseed,” Proceedings of the Third Liquid Fuel Conference, 15-17 September 1996, Nashville, Tennessee, pp 104- 114

46

Appendix 1

Review of Literature

Review of Analytical Methods

“Biodiesel,” fatty acid methyl esters used as diesel fuel substitutes, is obtained by the alkali-

catalyzed transestetication reaction of vegetable oils with alcohols:

0 II

%I--k-0-CH 12 ?

0 CH-O-c- R? I

R3 -;-0-CH2

Vcgcublc Oil

+ %-OH

Alcohol

KOH

0

R1 -;-O-R4

0 RZ-&O-Rsi

0

R3 -;-O-R.,

l

FH20H

CHOH

CH20H

Glyccml

Fmy Acid Alcohol Esters

Usually, 0.1 to 0.5 % (based on oil) potassium or sodium hydroxide is dissolved in 99% methanol,

which is then added to the oil, either at room temperature or heated (typically SO’C). Two times the

stoichiometric amount of methanoi is required for complete transesterification. The mixture is then

stirred and allowed to stand. Two phases form almost immediately; the glycerol formed settles to the

bottom. The ester layer is usually vacuum-distilled to remove the unreacted alcohol and washed with

water to remove the catalyst and any remaining glycerol and soaps which have formed (97,100).

Fatty acid alcohol esters are essentially the only desirable components of a biodiesel fuel.

Typical “contaminants” are water, free glycerin, bonded glycerin, free fatty acids, catalyst,

unsaponifiable matter, soaps, and the products of oxidation (97). Water content can be determined

using ASTM D 4928 (Karl Fischer Moisture) (97). Water is considered a major source of

contamination, because it can cause corrosion of engine fuel system components through increasing

acidity, and it also contributes to microbial growth (97). Van Gerpen et al. recently determined that

47

methyl soyate can contain up to 40 times more dissolved water than diesel &el, and that if methyl

soyate fuel comes into contact with water, it will absorb up to 1500 ppm (97).

Acid number (ASTM D 664) measures all components contributing to acid number, which

includes free fatty acids as wetI as other compounds such as inorganic acids, lactones, esters, phenolic

compounds, resins, salts of heavy metals, and addition agents such as inhibitors and detergents (95).

Bonded or bound glycerol (or glycerin) includes the products of incomplete transestetication:

mono-, di- and triglycerides. Soaps are the alkali salts that form when the catalyst reacts with any

tiee fatty acids present (100). Soaps may also form as a result of moisture absorption during storage,

which reacts fbrther with the biodiesel (100). Unsaponifiable materials include tocopherois, sterols,

phytosterols and hydrocarbons (97,100). Jn addition, very small quantities of pigments and minerals

may be present (97). The initials products of oxidation are hydroperoxides, which can then induce

polymerization, forming insoluble gums and sediments (97). The Peroxide Value (ASTM D 3703)

denotes the miiliequivalent of peroxide-bound oxygen in a 1000 g sample, and gives an indication of

the extent of dynamic oxidative damage (98,100). As oxidation proceeds, the unstable peroxides

tirther decompose into aldehydes and short-chain acids (97,98,100). The decomposition and

secondary products of oxidation can be determined using thiobarbituric acid. This test is based on

a condensation reaction between the aldehyde product and thiobarbituric acid, which forms a red

compound with an absorption maximum at 532 nm (98).

In order to ensure quality standards that meet the specifications required of automotive diesel

fuel substitutes, a variety of analytical methods exist which monitor the compounds yielded during

the production process. In Austria, where rapeseed oil methyl ester (RME) fuel has been produced

on an industrial level for several years, standard specifications denoted in &or-m C 1190 were

48

established by the Austrian Institute of Standardization in order to define quality criteria for the

production of vegetable oil methyl ester &els (1). Work completed by several researchers in Austria

are discussed in detail in the Handbook of Analytical Methods for Fatty Acid Methyl Esters Used

as Diesel Fuel Substitutes. Included in Chapter I are detailed listings of the available testing methods

for such standard combustion characteristics as density, viscosity, flashpoint, cold filter plugging

point, sulfated ash and carbon residue. Chapter II deals with quality parameters from the production

of biodiesd, including degree transesterification and important methods used to determine the

concentration of undesirable components such as mono-, di- and triglycerides and free glycerol.

Additional testing methods for RME and fatty acid methyl esters (FAMES) are provided in Chapter

III. These testing methods included: iodine number, indicating the degree and amount of

unsaturation present in the sample; water content; fatty acid distribution and degree and amount of

unsaturation; content of unsaponifiable matter; and the sterols content of the sample. Common ASTM

methods are included. The Appendix of this reference contains detailed tables of the known chemical

and physical properties of RME fuel.

In addition to this handbook and the known ASTM methods, many articles were found

that describe analytical methods for quality control and analysis of the components of biodiesel.

Some deal specifically with the use of such compounds as fuel, while others analyze FAMES from

other sources. These can be divided into five sections:

1. Analysis of fatty acid alcohol esters and other fatty acid derivatives;

2. Analysis of bonded glycerol (mono-, di- and triglycerides);

3. Analysis of free fatty acids;

4. Analysis of mixtures of vegetable oils and/or FAMES with hydrocarbon (such as diesel) oils;

49

5.

and

Polymerization phenomena and analysis of polymerization products formed by thermal or

autoxidation of vegetable oils.

Since analysis of a residue such as that which formed in the samples collected at ORTECH

Corporation was previously undescribed and unanalyzed, and not a question of quality control, the

researchers felt it both necessary and important to explore analytical methods involved with all five

compound classes listed above. In addition, each class of compounds were analyzed using a variety

of different analytical tools. Chromatographic separations are the most common. These type of

separations involve dissolution of the sample in the “mobile phase,” which sweeps the sample over

a “stationary phase.” The interactions between the compounds in the mixture and the stationary phase

are responsible for the separation of the compounds. The sample must be dissolved or miscible with

the mobile phase; this is a limitation to these methods.

One of the most common instruments used for the analysis of lower molecular weight

compounds is the gas chromatograph (GC), which separates compounds in a mixture based upon

diiering volatilities. This method is also known as gas-liquid chromatography (GLC), because the

separation is based upon the partition of the compounds of interest between a gaseous mobile phase

and a liquid phase immobilized on the surface of an inert solid (2). Methods of detection include

flame ionization (FID) and mass spectrometry (MS).

For compounds of higher molecular weight, methods such as high performance liquid

chromatography (HPLC) and high performance size exclusion liquid chromatography (HPSEC or

SEC) provide good separation and therefore quantitation of complex mixtures. These two methods

50

are similar in that the mobile phase in which the sample is dissolved is a liquid. HPLC methods are

the most widely used of ail analytical separation techniques, due to their sensitivity, ready adaptability

to accurate quantitative determinations, and suitability for separating nonvolatile species or thermally

unstable ones (2). The liquid chromatography methods differ in the type of separation achieved.

Most HPLC separations described in fatty acid chemistry involve bonded phase separation, mainly

in the analyses of non-ionic polar compounds of low to moderate molecular weight (~3000) (2).

HPSEC, also called gel permeation chromatography, is particularly applicable to species of high

molecular weight, making it especially useful in polymer characterization (2).

Other useful methods of analyses include thin-layer chromatography (TLC), supercritical fluid

extraction (SE) or chromatography (SFC), and thermal methods including therrnogravimetry (TG)

and differential thermal analysis (DTA). Near infrared spectroscopy or Fourier transform

spectroscopy (NIR or FTR), another common analytical tool, is based on the absorption of light

energy at a given frequency by molecules or functional groups on a molecule having a permanent

dipole which vibrates at the same frequency (3).

I. Analysis of Fatty Acid Alcohol Esters and Other Fatty Acid Derivatives

Gas Chromatography (GC)

Gas chromatography (GC) with either mass spectrometric (MS) or flame ionization detection

(FID) was the most common analytical method used in the analysis of fatty acid esters. GC with

mass spectrometric detection (GC-MS) was useful for identification of separated compounds. A

mass spectrum is obtained by converting a compound into ionic fragments and separating these based

upon their mass-to-charge ratio. Compounds yield very specific fragmentation patterns, allowing for

51

identification (2). Fragmentation was achieved by using electron impact (5) or chemical ionization

(6,7). Murata et al. used chemical ionization in the analysis of polyunsaturated fatty acid methyl

esters such as C225 and C22:6, whose molecular weights were difficult to determine by electron

impact MS (6). CC-MS analyses were also performed on dimethyl disulfide derivatives of

monounsaturated fatty acid methyl esters to identify the position and stereochemistry of the double

bond (8). Other GC analyses included the use of 14C- and 3H-radio labeled fatty acid methyl esters

of palmitic, .oleic and linoleic acids (9), GC with thermal conductivity detection for FAMES with 12-

18 carbon atoms (lo), and CC-FTIR analysis to determine the structure of fatty acid mixtures found

in a marine environment (11). One study also compared the quantitation of b-ans unsaturation of fatty

acid methyl esters determined by using an FTIR instrument to those obtained by the AOCS Official

Method (peak height) and by capillary gas chromatography (12). Craske and Bannon et al. of

Unilever Australia, Ltd. also published a series of articles pertaining to theoretical as well as practical

considerations in the GC analysis of fatty acid methyl esters (13-17).

Infrared Spectroscopy

Work published by R O’Connor in 1956 presented a sweeping review of 116 research articles

(1 S). Included are tables of absorption bands employed in the applications of IR spectroscopy to fatty

acid chemistry. O’Connor, Field and Singleton published quantitative spectra of the fatty acids and

their methyl esters in chloroform, reporting absorptivities at significant wavelengths, which permitted

relative quantitative comparisons. According to the researchers, a band at 9pm could be used to

distinguish ethyl esters fi-om methyl esters. Different groups of researchers also reported on spectra

of saturated and unsaturated fatty acids and esters. Another review published by O’Connor in 1961

52

highlighted advances made in the time between these articles (19). Ramanathan et al., as reported

in O’Connor’s review, obtained IR absorption curves of fatty acid methyl esters before and after

thermal oxidation. Little current data was found that used JR techniques alone for the analysis of

fatty acid methyl esters, however, hyphenated methods such as GC-FTR were more commonly used

Other Methods

Zirrolli et al. discovered a novel method in the analysis of branched-chain fatty acids using

low-energy tandem mass spectrometry of the molecular ion derived from fatty acid methyl esters by

electron impact ionization. This method can be used to determine methyl or alkyd branching positions

in a saturated fatty acid methyl ester (20).

Supercritical fluid chromatography (SFC) was used by two groups. Cocks et al. achieved

separation of both unsaturated and saturated fatty acid methyl esters, including a near baseline

separation of the Cl 8: 1 isomers, using mass evaporative light scattering detection (2 l), while Nomura

et al. used SFC with CO? as a mobile phase to extract lipids from fungus and to separate their

esterified products (22). SFC with FID detection proved to be advantageous over GC due to the

nonvolatility of longer-chain acid derivatives and the thermal instability of unsaturated products. This

method also circumvented the poor detection limits of HPLC analysis due to the inability of common

HPLC detection systems such as UV absorption and fluorescence to detect saturated compounds

(22).

L’C-labeled fatty acid methyl esters were recovered from complex samples by reversed-phase

two-dimensional thin-layer chromatography after introduction of 14C-acetate into the sample (23).

53

Rey et al. characterized the thermal stability of selected straight-chain (C6-C 14) esters of fatty acids

by thermogravimetric-differential thermogravimetric analysis (TG-DTG) and differential thermal

analysis (DTA). He related the results of the analyses with ignition points, molecular weights and

boiling points of the compounds (24). High performance size exclusion chromatography (HPSEC)

was used to monitor the ethanolysis of rapeseed oil and quantity the ethyl esters formed (25).

2. Analysis of Free and Bonded Glycerol, Sterok and Methanol

Mono-, Di- and Triplvcerides ,

According to Mittelbach, “a main criterion for the quality of a FAME fuel...is the content of

mono-, di- and triglycerides.” Long term engine tests showed that “a higher content of glycerides,

e.g. a transesterification degree below 95%, may cause the formation of deposits at the injection

nozzle, at the piston and at the valves.” (1) Van Gerpen et al. stated that “high levels [of bonded

glycerol] can cause crystallization and increased viscosity.” (97) High-temperature gas

chromatography (GC) with either mass spectrometric (MS) or flame ionization detection (FID) was

the most common analytical method used for the quantitation of bonded glycerol (mono-, di- and

triglycerides) contained in fatty acid alcohol esters (predominantly fatty acid methyl esters, or

FAMES) fuel samples. Thin-layer chromatography (TLC) and high performance liquid

chromatographic CHpLC) methods were also described. Extensive work was conducted by C. Plank

and E. Lot-beer of the University of Vienna. Two other groups working in this area included a group

of Italian researchers who published similar studies in the Italian journal La Rivista Italiana Deiie

Sostanze Grasse, and the researchers Freeman and Pryde of the USDA in Peoria, IL.

54

Gas Chromatography (GC)

In the GC analysis of bonded glycerol components, four methods were found in the literature.

AU four were based upon the silylation (attaching trimethylsilyl groups) of the free hydroxyl groups

of mono- and diglycerides. This increased their volatility and allowed the researchers to perform GC

analysis. Derivatization also prevented rearrangement reactions and prevents chromatogram peak

skewing (26). Triglycerides and the fatty acid methyl ester components remained unchanged.

“Trimethyl~ylation... followed by GC analysis using a short thin film capillary column [allowed] for

the determination of all the analytes, varying considerably in polarity and volatility, in a single GC

run.” (27) These four methods varied only in the use of solvents, internal standards, and silylating

agents.

Two basic silylating agents were used: N-methyl-N-trimethylsilyltrifluoroacetamide (MSTFA)

was used by Plank and Lorbeer (27,28); Freedman and Pryde (26) and Plank and Lorbeer (1) used

N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA); and Bondioli et al. (29) used BSTFA with 1%

trimethylchlorosilane (TMCS). According to Freedman and Pryde, “one outstanding advantage of

BSTFA compared to some silylating reagents [was] that both the reagent and reaction by-products

[were] highly volatile and [did] not interfere with the analysis.” (26) Different compounds were also

used as internal standards: tridecanoin (TD, also called tricaprin) (1, 26); Cl7 methyl ester, Cl 7

monoglyceride and C3S diglyceride (29); and TD and 1,2,4-butanetriol(27, 28). Plank and Lorbeer

used the addition of the second internal standard 1,2,4-butanetriol to serve “as a very sensitive

indicator of incomplete derivatization. In case of insufficient silylation (not all of the three hydroxyl

groups are siiylated), the peak of 1,2,4-butanetriol [appeared] split and drastically reduced in height.”

(27) All methods required a high-temperature GC column (stable to 350 or 370”(Z), and achieved

good separation of the analytes. Identification of the peaks was made by the use of appropriate

reference compounds. For rapeseed methyl ester fuels, monoolein, diolein and trioIein were used,

because oleic acid is the most common fatty acid in rapeseed oil (1, 27, 28). For studies with

soybean oil methyl ester fuels, monolinolein, dilinolein and trilinolein were used (26). Quantitation

was performed by determining the area of component (AJ / area of internal standard (A,J ratios

(denoted AJAJ. From plots of A&A, vs. weight of component (W,) / weight of internal standard

(WJ (denoted W/w,), intercepts and slopes were determined and used to quantify the samples (1,

26).

Infrared Spectroscopy (IR)

In the late 1950s and early 196Os, researchers in England published definitive reviews of IR

research in lipid chemistry. Inciuded were data and spectra of mono-, di- and triglycerides. Chapman

published work where he characterized glycerides by IR, and stated that “in particular, polymorphic

form, chain length, type of unsaturation and configuration [were] all revealed by the spectra.” (3 1)

Studies and a review by O’Connor mentioned earlier also contain sections describing the

characterization of glycerides (l&30). O’Connor, DuPre and Feuge compared the IR spectra of

mono-, di- and triglycerides, and noted that the O-H stretching region could be used to confirm the

presence or absence of mono- or diglycerides in a sample of triglycerides, because triglycerides did

EX contain any free hydroxyl groups and therefore did not exhibit this O-H stretching band. The

maximum absorption band in the C-O stretching region for mono- and diglycerides also occurred at

different wavenumbers; triglycerides exhibited no bands at these wavenumbers (IS).

56

High Performance Size Exclusion Chromatography (HPSEC)

HPSEC was used to evaluate different variables affecting transesterification and to quantitate

bonded glycerol components by two different researchers (25,32). It had an advantage over GC

analysis by saving the time-consuming step of silylation. It also eliminateed the multitude of near-

overlapping peaks in the chromatograms obtained by GC analysis. HPSEC is “based on the selective

retention of moIecules according to their size when they enter the pores of the polymer matrix” of

the columns (25). Refractometer detection was used. It was noted that in the case of

transesterification, the difference in mqlecular weight (and hence size) of the glyceride components

was about 250 atomic mass units (amu). This resulted in each class of components eluting at the

same time, giving only one peak.

In order to separate the entire range of sizes of the compounds, columns are often connected

in series. Christopoulou et al. used two columns packed with 5pm styrenejdivinylbenzene copolymer,

with an upper and lower exclusion limit of 5000 amu and 100 amu, respectively (32). Fillikres et al.

also connected two styrenddivinylbenzene copolymer columns with a molecular weight range of 50-

1000 amu and 500-10,000 amu (25). In both cases, quantitation was performed using correction

factors calculated by using glycerol (25) or monolaurin (32) as an internal standard.

Other Methods

Freedman and Pryde developed a quantitative method for analyzing glyceride mixtures from

the transesterification of vegetable oils by thin-layer chromatography with flame ionization detection

(TLC-FID) (33). Th ey reported that this method was plagued by lower accuracy and precision

compared to GC and HF’LC techniques, particularly in the case where the analytes of interest are

present in quantities of l-2%. This made it less usetil to quantitate glyceride components left in the

sample after the transesterification process. However, they did report that this method could be used

to satisfactorily monitor the transesterZcation reaction process (33).

In addition, separation and quantitation of the triglycerides tricaprylin (C10) to tristearin

(C 18) was achieved by non-aqueous reversed-phase KPLC with refractive index detection (17).

Yamazaki et al. published a study using high pressure differential thermal analysis (HPDTA) of fatty

acid methyl esters and triglycerides (34). HPDTA could be used to measure oil stability and effects

of sample containers and air pressure 0~ heating and oxidation of the methyl esters and triglycerides

of palmitic, oleic and linoleic acids.

Free Glvcerol

The presence of free glycerol (glycerin) in fatty acid methyl ester fuel may “cause problems

in the fuel system, e.g. deposition of glycerol in the fuel filter, or can lead to injector fouling or to the

formation of higher emissions of aldehydes.” (1). For this reason, it is of vital importance to monitor

the amounts of glycerol, the triglyceride backbone remaining after deavage of the fatty acids, in the

fUe1. The content of free glycerol in biodiesel is limited to 0.02% (1). According to the Handbook

of Analytical Methods used as Diesel Fuel Substitutes, the most sensitive method (with a detection

threshold of 0.0001%) for monitoring free glycerol was a GC method involving derivatization.

Additionally, HPSEC (using the methods described above (25)) and enzymatic methods exist (1).

Bailer and de Hueber described an enzymatic method for the quantitation of glycerol using

commercially available enzymatic test kits (1).

Free glycerol can be determined using siiylation as described above (1,27,28). Bondioli et al.

58

developed a CC method for the determination of free glycerol in biodiesel fuel samples, without

derivatization. They added glycerol in alcoholic solution (to make the glycerol soluble) to biodiesei

fuel samples in order to ensure recovery and used 1,4-butanediol as an internal standard.

Reproducibility was obtained only when the concentration of glycerol was higher than 0.02% in the

fuel (35). Mittelbach described a GC method in which glycerol in a sample is siiylated with an excess

ofBSTFA (36). FID or MS detection was used, and this method had a minimal detectable limit of

0.00001% using MS or 0.0001% using FID.

S terols

According to Hayes et al., cholesterol and related compounds can interfere in the analysis of

fatty acid methyl esters when the methyl esters are prepared by transesterification (37). In order to

obtain qualitative and quantitative information about this group of compounds, Plank and Lorbeer

developed an on-line liquid chromatographic-gas chromatographic method (TLC-G-C) to detect free

and esterified sterols in vegetable oil methyl esters used as diesel fuel substitutes (38). Sterols are

“the main constituents of the unsaponifiable matter of most vegetable oils” and “remain in the fatty

acid methyl ester phase” where they “are recovered both as free sterols and as esterified sterols in the

transesterification product.” It is thought that the presence of these minor components may influence

combustion characteristics of the fuel, as well as storage properties and other physical and chemical

properties. Plank and Lorbeer’s method involved silylation with MSTFA and use of stigmasterol,

cholesteryl oleate, cholesteryl palmitate and cholesteryl stearate as reference substances, as well as

betulinol as an internal standard. LC was used to remove the large amounts of fatty acid methyl

esters present prior to GC analysis.

59

Methanol

The transesterification process for methyl esters is carried out with an excess of methanol

(CH,OH), usually ranging from 1.5 to 2 times the stoichiometric amount (39). A high content (0.15 -

0.20%) of methanol left in the fuel increases the flash point of the fuel. Therefore, Bondioli et al.

developed a method to detect methanol in biodiesel fire1 samples. Distilled water and a citric acid

solution were added to the weighed fuel sample. The solution was then distilled almost until

completion, at which point the internal standard (absolute ethanol, C,H,OH) was added. The

distillate was analyzed by GC analysis.

Other Methods

Basu et al. prepared glycol diesters and mixtures of mono- and diesters from methyl esters

of partially hydrogenated soybean oil to test their potential as lubricants. Testing was done with TLC

and viscosity determinations (40). Three groups of researchers used nuclear magnetic resonance

(NMR) to quantitate different components in vegetable oils. Wollenberg used “C NMR to quantitate

the amounts of different triglycerides based upon the acyi distribution and acyl positional distribution

(I,3-acyl and 2-acyl) in vegetable oils by monitoring the olefinic and carbonyl carbons (41). 1,2-

Propanediol monoesters of long-chain fatty acids were determined using ‘T NMR after

trifluoroacetylation (42). And finally, ‘H NMR was used to monitor the rate of transesterification and

quantify methyl esters without knowing the exact amount of the intermediary mono- and diglycerides

(43).

60

3. Analysis of Free Fatty Acids

Free fatty acids initially present in a soybean oil sample are usually removed during refinement

of the oil prior to transesterification or in the washing step of production after transesterification

(100). Any free fatty acids remaining or formed in the biodiesel fuel can be detected using ASTM

methods D 974, D 3339, and D 664 as a component contributing to acid number, however, other

compounds such as inorganic acids, lactones, esters, phenolic compounds, resins, salts of heavy

metals, and additives such as inhibitors and detergents also contribute acidity to a sample (95). Little

has been reported about other analytical methods to determine the free fatty acid (F’FA) content of

biodiesel, presumably because no practical analytical method existed for the analysis of FFA in a

mixture of fatty acid methyl esters. In the presence of moisture, fatty acid alcohol esters can be

hydrolyzed to alcohol and the free fatty acid. As such, the researchers deemed it necessary to

investigate possible methods to quantify the FFA content. Unfortunately, many studies quantitating

FFA from other sources (19, 44-47) utilized the transesterification reaction to convert the FFA to

FAMES, which were then analyzed as described above. However, several publications were secured

and reviewed, in hopes of finding a procedure which tests for isolated FFAs.

Gas Chromatography

Bohov et al. used a glass capillary column coated with a cyanopropylsiloxane stationary phase

and FlD detection for the separation of geometric isomers of unsaturated fatty acids (4). May et al.

described a GC method for analyzing the FFA extracted from spring canola seed grown in Canada,

that produced an oil with FFA levels that exceed the 0.3% nor~~Ily found in canola grown in western

Canada (48). The FFA content from several cultivars ranged from 0.14% up to a high of 12%,

61

although most were below 4%. GC-FID analysis was performed on the extracted samples using an

internal standard of heptadecanoic acid. Peaks were identified by internal standards of palmitic,

stearic, oleic, linoleic and linolenic acids. The results were linearly related to those obtained with the

traditional method of Soxhlet extraction and titration.

Derivatization Methods

Four studies were found which characterize and quantify FFA content by derivatization using

picolinyl esters of saturated and unsaturated acids and CC-MS (49); Z-nitrophenylhydrazine

hydrochloride derivatives and HPLC (49); p-bromophenacyl bromide derivatives and HPLC (50); and

(chloro)alkyl chloroformates and CC-electron capture detection (ECD) (52). All of these methods

were very sensitive for FFA and had very low detection limits (100 fin01 to 200 fin01 or 400 fmol to

1 pmol derivative per injection (50), 0.05% (5 1) and 0.05 ppm (52)). However, it is doubtful if three

of the methods could be used in samples containing fatty acid methyl esters, because both fatty acids

and fatty acid methyl esters would form the same derivatives. However, Fatica mentioned that the

determination of FFA in tall oil resin could be achieved even in the presence of their esters by

omitting a step in the derivatization with p-bromophenacyl bromide (5 1).

Infrared Spectroscopy

Three groups of researchers used IR to characterize the fatty acid content of seed oils non-

destructively (3,53,54). However, the researchers question whether the methods utilized here could

yield information about FFA content in samples containing fatty acid methyl esters, although the

fUnctional hydroxyl group of the FFA is not present in the FAMES. Lanser et al. utilized computer-

62

assisted Fourier transform lR (FTR) to determine FFA content via the triglyceride ester and FFA

carbonyi bands (55). A comparison was then made to the values obtained by the standard methods.

They found that the FTIR method “had the largest variance f?om the standards when low (0.7%) FFA

content was determined,” because identifying “a small shoulder on the ester absorbance and trying

to obtain an accurate area calculation [was] extremely difiicult.”

Other Methods

Other methods existing for FFAzquantification include HPLC without derivatization (56,57),

mass spectrometry (58) and TLC (59, 60). One interesting method used copper soap calorimetry to

determine the FFA content of vegetable oil samples (61). The sample was mixed with an organic

solvent and then with an aqueous copper reagent (tetrapyridine copper (II)), forming a copper soap.

The copper soap was mainly non-polar in character, due to the long chain hydrocarbon section of the

fatty acid. Mixing the solution caused the copper soap to enter the organic phase. The aqueous and

organic phases were then separated and the absorbance of the organic phase was measured at 716

nm. Absorbance at this wavelength is due almost entirely to the copper soap which formed. The

authors of this study make no mention of the usefulness of this method for the analysis of free fatty

acids in a fatty acid methyl ester mixture. However, due to the nature of the copper soap formation,

the ester should remain unreacted (i.e. form no copper soap). Therefore, the absorbance of the

orvganic phase would come only from the free fatty acids in the mixture with no interference of the

fatty acid methyl ester fuel. To prevent conversion of the ester to the free fatty acid, the pH of the

mixture should be kept as close to neutral as possible. The authors of this study used a buffered

solution of pH 6.1.

63

4. Analysis of Miires of Vegetable Oils with Hydrocarbon Oils

A group of Italian researchers led by P. Bondioli developed two different methods for

determining the amount of biodiesel in biodieseudiesel fuel blends. This had important implications

not only in a fiscal sense, but also by considering that the best performance of a diesel engine can be

achieved using a mixture of biodiesel in diesel fuel in a ratio ranging between 5 to 30% (62). One

method used utilized the relationship between the saponification value (SV) (method NGD C33-

1976) of the sample and the percent of biodiesel in blends. The saponification value indicates the

amount of hydrochloric acid in mL that reacted per g oil, which is titrated volumetrically using

bromphenol blue as an indicator (98). A good agreement was discovered between % diesel fuel and

SV. However, it must be pointed out that the researchers had determined the SV for the pure

biodiesel used in the blends; this value represented 0% diesel or conversely 100% biodiesel. If a

sample of the pure biodiesel used in the blend is not available, then no SV can be determined for

100% biodiesel. The second method attempted to determine SV via fatty acid composition of the

biodiesel. Gas chromatography analysis was problematic due to the presence of the diesel fuel

components whose volatility was similar to one of the biodiesel components. This problem was

solved by performing a pre-separation of the hydrocarbon fraction on a SEP-PAK silica column (62).

J.S. Jha of the Royal Nepal Academy of Science and Technology published work on infrared

(JR) and ultraviolet (UV) spectroscopic studies of mixtures of vegetable oils with hydrocarbon oils

(63). He reported that not much work has been done for detecting hydrocarbon (HC) oils in

triglycerides (TG) by instrumental techniques. This study was of interest due to the similarities of a

blend of biodiesel (vegetable oil) with diesel fire! (hydrocarbon oil). Previously, work had been

64

undertaken to detect adulteration of edible oils and fats. In order to further investigate the suitability

of this method, mixtures of triglyceride oil (mustard, groundnut, soybean coconut, linseed and

margosa oils; ghee, cheuri and hydrogenated vegetable fats) and diesel oil were prepared in blends

from 0.5% TG : 99.5% diesel up to 99.5%TG : 0.5% diesel v/v. All TG oils had similar IR spectra,

showing prominent absorption bands at 715, 1115, 1170, 1375, 1460, 1745, 2855 and 2930 cm-‘.

The TG oils of mustard, groundnut, soybean, coconut and linseed had an additional band at 3010

cm-‘, which was due to the higher percentage of unsaturated acids. The spectra of four HC oils

(kerosene, petrol, diesel and mobil oils) showed two strong bands at 1375 and 1460 cm-‘, due to

symmetrical and asymmetrical C-H bending of the hydrocarbon oil molecules. A weak band was

present at 1605 cm-’ in the diesel fuei spectrum. Upon examination of the IR spectra of the mixtures

of TG and HC oils, it became apparent that the presence of small amounts of HC oils in TG oils could

be detected with IR spectroscopy because all the bands present in a HC oil spectrum were also

present in a TG oil spectrum. HC oils could only be detected quatitatively in TG oils if their

percentage was higher than 15% in the mixture. Conversely, the C=O absorption band at 1745 cm-’

of TG oils was an indication of TG oils, even if as little as 0.5% TG oils were present in the mixture

with HC oils.

Jha also conducted studies comparing the ultraviolet (UV) absorption spectra of the same

mixtures (63). He suggested that “ifthe absorbance value of a 0.1% solution of mustard oil sample.. .

in carbon tetrachioride [was] more than 0.1 at 257 MI,” which corresponds to the )c- of the mustard

oil, then “it may be due to the presence of some impurity” such as a hydrocarbon oil in it, as the

absorption of the TG solution alone was only 0.09. Jha concluded that “results obtained from IR and

UV spectroscopic studies of TG oil samples may substantiate some useful information to the other

65

infom-ntion already obtained” by traditional methods and “would be helptil” in determining the purity

of a TG sample.

5. Polymerization Phenomena and Analysis of Polymeriiation Products Formed by Thermal or

Autoxidhtion of Vegetable Oils

It is well known that a high degree of unsaturation in the fatty acid methyl esters used as diesel

fuel substitutes can lead to polymerization (64). In addition, much research has been conducted into

the polymerization phenomena of vegetable fats and oils, especially in the paint and varnish industries.

A very important reference which discusses all aspects of lipid chemistry in industrial applications is

Bailey’s h.hstria.i Oil and Fat Froa?.~ts, edited by D. Swem (65). Another informative review was

published by Wexler in 1964. He investigated in depth the physical-chemical theory behind vegetable

oil polymerization (66). Additionally many publications were secured which investigated the

polymerization phenomena of different vegetable oils under heating and frying conditions (67-72).

The manufacture of paints, varnishes and other protective coatings comprises one of the three

major fields of fat and oii utilization (65). Drying oils, coatings which “dry” or polymerize to film

after application, belong in this class. The film-forming abilities are closely associated with their

degree of unsaturation, since it is through the unsaturated centers of double bonds that polymerization

takes place. Wexler stated that “it is well known that the rates of oxidative polymerizations vary with

the number and degree of conjugation of the unsaturated linkages.” (66) Swem made the distinction

between heat or thermal polymerization and oxidative polymerization. Thermal polymerization “is

considered to produce direct carbon-to-carbon linkages between the unsaturated acid chains, whereas

polymerization accompanied by oxidation presumably occurs to a considerable extent through the

66

establishment of carbon-oxygen-carbon linkages, as well as carbon-carbon.” (65) Oxygen interacts

with the double bonds, forming hydroperoxides. Conjugation of double bonds occurs to stabilize

hydroperoxides. As the hydroperoxides decompose, a free-radical chain reaction and subsequent

chain-growth polymerization give rise to cross-linked products with a high molecular weight (67).

A decrease in the relative percentages of the unsaturated fatty acid content and an increase in the

relative percentages of the saturated fatty acid content can be observed (68). Korus et al. reported

that the rates of thermal polymerization showed a stronger dependence on the degree of unsaturation

than oxidative polymerization, and polymerization occurred in a step-growth process by a Diels-

Alder reaction or free-radical mechanism (67).

The Peroxide Value (PV) denotes the milliequivalent of peroxide-bound oxygen in a 1000 g

sample, and gives an indication of the extent of dynamic oxidative damage (98,100). As oxidation

proceeds, the unstable peroxides further decompose into aldehydes and short-chain acids (97,98,100).

Van Gerpen et al. pointed out that the hydroperoxides formed during oxidation of biodiesel were

unstable and attack elastomers; the acids produced upon decomposition caused corrosion of the

engine fuel system (97). It is important to note that the peroxide value of a sample initially increases

as oxidation proceeds, then decreases as the peroxides decompose. The PV therefore provides only

incomplete data on the oxidative state of a sample (98). One way to enhance the information

contained in the PV is to determine if the peroxides initially measured have decomposed and formed

other secondary products such as aldehydes. These secondary products of oxidation can be

determined with thiobarbituric acid. This test is based on a condensation reaction between the

aldehyde product and thiobarbituric acid, which forms a red compound with an absorption maximum

at 532 MI (98). The ASTM D 3703 method is used to determine PV.

67

To determine modifications in vegetable oils due to repeated heating,, such as in frying, both

the non-polar triglycerides, comprising both the unaltered part of the oil and the products arising from

thermal oxidation in the absence of oxygen, and the polar fraction, products of oxidative alteration,

were analyzed (70,72). Total alteration is defmed as the sum of the polar fiaction plus the

unrecoverable fraction. Heat treatment of fats such as sunflower oil led to an increase in compounds

which were polar in nature (such as polar triglycerides). These compounds were therefore part of

the polar fraction (70).

The rates of polymerization canbe monitored via viscosity. Bulk viscosities of vegetable oils

were used as a measure of the degree of both thermal and oxidative polymerization by Korus et al.

in their study of the polymerization of safflower and rapeseed oils. They also determined the

relationship between viscosity and molecular weight by size-exclusion chromatography (SEC) for

thermal polymerizations. Eight separate peaks were detected for the triglyceride monomer through

the octomer. Oligomer retention times were used to establish calibration curves for molecular

weights between 2,500 to 11,000 amu (67).

Christopoulou et al. published a series of papers on the analysis of fatty acid monomer, dimer

and trimer mixtures by SEC, TLC and HPLC (73-75). “SEC can be used as a measure and indication

of the extent of heating and polymerization of heated fats and oils.” (75)

Wexler reported extensively about the effects of different compounds which act as catalysts

in poIymerization reactions (66). Among these are metallic compounds, free radicals, organic

compounds containing at least three aromatic rings and lipoxidase, an enzyme found naturally in

soybeans. Lipoxidase is an “especially powerful and specific oxidation [catalyst] for unsaturated

fatty acids containing methylene interrupted multiple bond systems...in the cis configuration.“(66)

68

Organic compounds actually act as “isomerization catalysts, which assist in moving adjacent double

bonds to conjugated positions, to permit linkage of fatty acid chains by a Diels-Alder reaction” (65).

It was reported that “thermal polymerization is favored by the presence of conjugated double bonds

and, consequently, in thermal polymerization, the rate-determining step appears to be the

isomerization of nonconjugated to conjugated forms.” (67) Many metal compounds exhibit catalytic

effects. Most active are cobalt, lead and manganese, as well as cerium, copper, chromium, iron, tin

vanadium and zirconium (66). In addition, the ASTM society stated in method D 2274, which tests

the oxidative stability of fuel oil, that “oxidation [was] a major chemical process causing adherent and

filterable insolubles to form,” and that the presence of metals “such as copper or chromium that

catalyze oxidation reactions will cause greater quantities of insolubles [e.g. polymerization products]

to form.” (76)

Although most of the literature focused on polymerization reactions of the pure vegetable oil,

it stands to reason that the same type of oxidative polymerization reactions will occur with vegetable

oil methyl esters. It was stated that the degree of unsaturation and double bond position are decisive,

and that the double bonds are the site where the polymerization reaction occurs. These sites remain

unchanged after the transesterification process. In addition, Swem addressed the polymerization of

fatty materials other than glycerides. Ln the manufacture of polybasic acids, the reactions were

preferably carried out on free fatty acids, soaps or monoesters, for a variety of reasons. He stated

that “the presence of water also [favored] dimerization at the expense of the formation of higher

polymers. ..” In the polymerization of methyl esters of mixed soybean oil fatty acids, the ratio of

dibasic acids to tribasic acids was 2.6 to 1. “The production of polymerized monoesters, by thermal

reaction” at 300” C, was described, and a process exists to polymerize sodium salts of fatty acids at

69

300” C (65).

Tocopherols, which are minor components of soybean oil, are anti-oxidants (100). Van

Gerpen et al. pointed out that the natural anti-oxidants present in soybean oil are part of the

unsaponifiable matter and were removed upon distillation of the oil prior to transesterification (97).

They performed studies on the oxidation of non-distilled versus distilled soybean methyl esters to

determine the effect of distillation on peroxide value. They found that the distilled methyl esters

oxidized much more quickly, requiring six days to reach a PV of 96 after an initial PV of 0. The

freshly-prepared, non-distilled methyl esters, with an initial PV of 6, required 24 days to reach a PV

of 80.

In a study by Tian et al., the anti-polymerization activity of a methanolic Noble oat (Avena

sativa L.) extract was tested under frying conditions (101). It was reported that a specific sterol

found in oats, A5-avenasterol, was effective in retarding soybean oil deterioration at 180°C.

Reportedly, a tertiary free radical in an ethylidene side group reacts “with free radicals from the

heated oils to produce a stable, isomerized allylic free radical, which interrupts the oxidation.”

HPSEC was used to determine the amount of polymeric fractions formed during heating of soybean

and cottonseed oils. The decrease of high molecular weight fractions observed after addition of the

oat extract, as. well as other commercially available anti-oxidants, was measured.

70

Review of Literature Focusing on Fueling Issues and Problems

Much has been written about the use of vegetable oils as diesel fuel substitutes (77-84).

Although in many cases short term tests (<lo h) were successful, many problems surfaced during

tests of longer duration (79, 86). The origins of these problems could be traced to the oil’s high

viscosity, low volatility, and reactivity due to unsaturated double bonds (79). These problems were

more pronounced in direct-injection engines compared to precombustion (indirect injection) engines

(79). The most common observation was that of injector coking, which eventually led to improper

fuel atomization and decrease in combustion efficiency (78,79). This in turn led to other problems

such as piston ring sticking, crankcase oil dilution and thickening, and gelling of the lubricant oil (77-

80,86). Humke et al. wrote that “injection nozzle deposits with vegetable oils and vegetable oil

blends with #2 diesel fire1 decrease[d] engine performance. ..I’ After the carbon deposits were scraped

off, original performance values were obtained (82). Bettis et al. conducted durability tests with

x&lower oil (high in unsaturated linoleic acid), and reported that after 830 hours, extensive carbon

deposits formed in the combustion chamber and heavy gum deposits were found on the injector and

compression rings (84).

Adams et al. reported that the majority oftests where the problems described above occurred

were conducted using a blend of 50% p&t oil and 50% diesel fuel or 100% plant oil (85). In a 25%

sunflower oil / 75% diesel fuel blend test, the lacquer and carbon residue found on the third land was

“significantly heavier” than the residue found on the land after the diesel fuel test (78). The authors

described this residue as hard and shiny; it did not “flake off as was found with the dry carbon buildup

formed during the run on diesel fuel.” (78) In the same study with the same 25/75 blend, the residue

fi-om the turbocharger had a sticky oily appearance (78).

71

Increased engine wear (as evident by wear metals in oil) when fueling with vegetable oil heIs

was noted. In a study by Adams et al. in 1983, wear metals from engine components were detected

in the lubricating oil, and increased values for manganese, cobalt, nickei and lead were observed.

Manganese, which is a component of steel, is known for its catalytic activity in oil polymerization

(85). Adams et al. then attempted to replicate the phenomenon which caused the described

thickening and gelling of the lubricant oil while fueling a diesel engine with a 50150 blend of soybean

oil and diesel fuel (85). They mixed varying amounts of motor oil, soybean oil, and water. The

mixtures were agitated by aeration for 10 days and heated to 85-95°C. No significant hardening (i.e.

polymerization) took place until catalytic amounts of cobalt and manganese compounds were added.

Ail solutions containing more than 10% soybean oil became dark in color and highly viscous 94 hours

after addition of the metal. At that point, stirring and aeration were discontinued. After 24 hours the

solutions had changed into semi-solid gums. Similar tests that were conducted by Rewolinski, where

oxygen was percolated through tests cells of contaminated oil in the presence of a copper catalyst,

resulted in an increase in viscosity of the oil (an indication of polymerization) (8 1).

As a result of the problems encountered with the use of vegetable oils, alternatives to neat

vegetable oil were probed. Transesterification of the oil provided a reduction in viscosity, thereby

improving the physical properties of fuel and improving performance (9 1,100). Clark et al. stated that

methyl and ethyl esters exhibited properties similar to that of diesel fuel, except for the problem of

gum formation, “which manifested itself in problems with the plugging of fuel filters.” The authors

also noted that the gums formed from the ethyl and methyl ester fbels varied slightly in color and

texture, and that the “methyl ester [experienced] greater carbon and varnish deposits on the pistons.”

The ASTM D 2274 16-hr test for oxidative stability was used to investigate the formation of gum

72

(i.e. insolubles) which formed upon oxidation of the fuel. “This test indicate[d] that the soyates

[methyl and ethyl esters] are much more susceptible to gum formation problems than diesel fuel...”

An increase in the total gum content typically accompanies biodiesel fuel oxidation, forming either

soluble or insoluble gums (99). Fuel filter plugging results from insoluble gums, while soluble gums

can cause the formation of deposits on injector tips (99). In a study by Vinyard et al. from 1982,

degummed sunflower oil ethyl ester produced unacceptable injector coking after 50 hours on part

load operation, even when the fuel was diluted with diesel (86). Another cause of fuel titer plugging

was noted when saturated fatty acid esters underwent crystallization at low temperatures. These

crystals plugged fuel filters and fuel lines (93).

Du Plessis et al. conducted storage stability studies on sunflower oil ethyl and methyl esters

(92). The influence of air, temperature, light, an antioxidant and contact with steel on acid value,

peroxide value (PV), ultraviolet (JJV) absorption, anisidine value, viscosity, and induction period

were evaluated. In order to obtain an accurate representation of the results, statistical analyses were

performed for all of the parameters of the study. They concluded that thermal stability differences

existed between the methyl and ethyl esters. An increase in acid number was observed for both esters

over a 9Oday storage period. At 20°C only a slight increase was observed; this increase was more

evident at elevated temperatures (30 and 50°C). The acid number increased sharply for both esters

upon exposure to air, and light treatment resulted in a similar increase for the ethyl ester. Upon

exposure to air, the following observations in PV were made for both esters: a moderate increase

at 2O”C, a sharp increase at 3O”C, and a very sharp increase at 50°C. Once again, exposure to light

increased the PV of the ethyl ester only. Steel increased the anisidine value of all samples at all

temperatures only slightly; methyl esters were affected significantly at 50°C. UV absorption was

73

increased upon exposure to air due to the formation of conjugated dienes. According to the authors

ofthis study, viscosity measurements reflected the oxidative condition of the esters. Viscosity values

rose for both esters at all temperatures (20, 30 and 50°C). Viscosity increased more significantly for

the ethyl esters upon exposure to air. The time needed for oxidative reactions to begin (induction

period) was reduced upon exposure to air. During the induction period, the fuel slowly oxidizes,

followed by rapid acceleration of oxidation (99). However, no significant decrease in induction

period was observed under air-tight storage conditions, with or without steel strips. In each case, the

effect of the anti-oxidant tertiary butylhydroquinone (TBHQ) was investigated. Addition of 0.04%

TBHQ prevented oxidation of samples stored under moderate conditions.

Du Plessis et al. concluded their study with practical guidelines for the storage of fatty acid

ester fuels, based upon the results obtained. These included storage in airtight containers at a storage

temperature of <3O”C. According to the authors, rust-free mild steel containers would be suitable,

and the addition of TBHQ increased the oxidative stability of the samples. A three year study of

storage stability of vegetable oil fuels was conducted by Klopfenstein et al. in the 1980s (107). Their

results indicated that underground storage in plastic-lined containers, in addition to the use of 0.5%

butylated hydroxy toluene (BHT) as an anti-oxidant, was the best option for vegetable oil fiels.

Some storage problems of the biodiesel fuels have been addressed. Van Walwijk published

answers to questions addressed by the US Department of Energy concerning the storage of biodiesel

(87). Although no problems were reported in Austria for the storage of rapeseed methyl ester @ME)

fuel in double walled metal tanks for two years, either above or below ground, lead coated fuel

storage tanks were damaged by a 30% RME / 70% diesel fuel blend in transit buses in France. It was

also reported that the detergent effect of this 30% RME / 70% diesel fuel blend released deposits that

74

had accumulated on storage tank walls and pipes, which made filter replacements necessary. While

fueling a diesel engine with RME, Horstmann and Stumpf observed the development of sedimentation

in the area of the plunger of an inline pump during extended shut down periods. This was attributed

to fuel leaking into the pump lubricant (88).

Thompson et al. of the University of Idaho conducted a two-year storage study on both

methyl and ethyl esters of rapeseed oil (107). Triplicate samples of rapeseed methyl ester @ME) and

rapeseed ethyl ester (REE) were stored in both glass and lined steel containers at room temperature

(inside) and at the local ambient outdoors temperatures (outside) for a total of twenty-four months.

At the beginning of the study and every three months thereafter, the samples were analyzed for typical

biodiesel properties such as peroxide value, acid number, density, viscosity, and heat of combustion.

They discovered that, on average, these properties of the esters tended to increase over time. One

exception was noted: the heat of combustion decreased over time.

Of special interest are the acid number values and the peroxide values. Thompson pointed

out that “both of these values [were] related to autoxidation” of the fuel and that “the acid values

naturally increase with an increase in peroxides because the esters first oxidize to form peroxides

which then undergo complex reactions including a split into more reactive aldehydes.” These

aldehydes are then oxidized into carboxyiic acids.

Thompson et al. also found that the acid values for neither the RME or RJZE fuel were

significantly affected by the type of container used. In general, the acid values remained fairly

constant for the first six months and then increased significantly. At the end of the study, the “acid

value was 10.3 times higher for RME and 9.2 times higher for REE compared to the beginning

values.” In addition, the acid value of the outside samples lagged behind the inside samples. They

75

also noted that the acid values for the RME increased at a faster rate than the REE values tier six

months, as was also the case with the peroxide values of the RME.

Material compatibility issues of biodiesei and biodiesel blends compared to diesel were also

addressed by a number of authors (37, 88-90). Reed et al. tested ten common polymers, and found

that eight were unaffected or acceptable (90). Excess deterioration was observed with nitrile rubber

and polyurethane foams. Korbitz suggested that parts composed of nitrile rubber be replaced with

fluorine rubber materials, which “show sufficient resistance.” (89)

Reed et al. observed no degradation of aluminum, brass, steel or phosphatized fuel tanks in

their material compatibility study (90). However, when Masjuki et al. examined the lubricating oil

from blends of palm oil methyl ester and diesel fuel for metallic wear particles, they found increased

values for aluminum, chromium, lead and copper. However, in general, “wear metal levels for the

blended fuels and pure POD [palm oil diesel] fuel were considered to be normal throughout...” (83)

Bessee and Fey of the Southwest Research Institute recently published a study assessing the

compatibility of biodiesel fuel blends with elastomers and metals (96). They placed tensile bars of

different elastomers (Teflon, nylon 6/6, nitrile rubber, Viton A401C, Viton GFLT, fluorosilicon,

polyurethane, and polypropylene) into sealed glass containers containing ten different fuels, including

100% diesel, 100% biodiesel, and 20% biodiesel / 80% diesel and 30% biodiesel / 70% diesel blends

of the two. After storage at 5 1.7”C for periods of 0, 22, 70, and 694 hours, each bar was examined

for tensile strength, elongation, hardness, and swell. Their results indicated that the Teflon and Viton

materials seemed to be the least affected by contact with the biodiesei. They indicated that nitrile

rubber, nylon 6/6 and high-density polypropylene were affected by biodiesel and biodiesel blends.

In the same study, metal compatibility with biodiesel and biodiesel blends was also probed.

76

After a storage period for a total of six months at 5 1.6”C, changes in metal coupons placed in the ten

different fuels were evaluated. This included a monthly visual inspection, as well as total acid number

(TAN) determination at four months for any sample which had discolored. At the conclusion of the

storage test, TANS were determined for all samples. They noted heavy gum formation which adhered

to the coupon or settled to the bottom in samples ofbiodiesei blends and copper-containing materials,

as well as high TANS for samples of biodiesel blends with steel and aluminum. It appeared that

aluminum and biodiesel blends did m cause a gum formation, although “brown gum” and “light

brown film” formation were noted for biodiesel blends with bronze and steel (96).

In a report to the National Biodiesel Board published in early 1997, the compatibility of

common fuel additives with biodiesel and biodiesel blends was analyzed (99). In this study, common

diesel fuel additives were evaluated for their efficacy in biodiesel and biodiesel blends, and the

compatibility of multiple additives were also probed. Cetane number, conductivity, corrosiveness,

oxidative stability and the applicability of common testing methods were evaluated.

Of most interest to our study were the results obtained from the evaluation of the corrosivity

and oxidative stability of the biodiesel blends with the common diesel fuel additives. To study

corrosion, the National Association of Corrosion Engineers (NACE) International Standard TM0 17-

93 test and the ASTM D 130, Standard Test Method for Detection of Copper Corrosion from

Petroleum Products, were used. In the NACE test, a 300 mL fuel sample is exposed to a polished

steel spindle for 30 minutes at 100°F with continuous stirring. After the initial 30 minute period, 50

mL water is introduced into the fuel. The spindle is then further exposed to the fuel/water mixture

for an additional 3 hours, again at 100°F. A NACE rating scale is used to evaluate the amount of

corrosion (rust formation) on the spindle caused by the fuel/water mixture. The water added is

77

considered to be the causative agent, along with the compounds that it extracts from the fuel.

Additive efficiency can be compared to a blank sample containing no additives. The results of this

study indicated a general increase in corrosivity as the percentage of biodiesel in the blends was

increased. However, little change in the results were noted as the level of corrosion inhibitor

changed. The ASTM D 130 results indicated that none of the blends tarnished the copper strip.

The results of stability testing in the study indicated that a common oxidative stability test for

diesel fuei, the Du Pont F21 pad rating test, was inappropriate for biodiesel and biodiesel blends. This

test is based on the characteristic darkening of a diesel fuel sample which has been heated at

atmospheric pressure to 300 “F for 90 minutes. The sample is filtered, and the discoloration of the

fuel, as evident by the pad color, is compared to a series of standard filter pads. However, the

biodiesel and biodiesel blends did not exhibit the same discoloration as the diesel fuel samples. A

moditied ASTM D 525, Standard Test Method for the Oxidation Stability of Gasoline, was found to

be appropriate for the testing of biodiesel and biodiesel blends.

Stability additives used in diesel fuel often contain metal deactivators (to prevent metal-

catalyzed oxidative reactions) and dispersants (which prevent sediments from forming particles large

enough to settle and cause plugging of fuel system lines and filters). However, the compatibility

study for the National Biodiesel Board did not include either one of these products in the stability

segment of the testing. The researchers in the study felt that these additives were unnecessary,

because 1) the base fuels used contained no trace metals and also did not come into contact with

metals during the course of the study; and 2) dispersants do not interfere with the chemical

mechanisms of fuel instability. However, the presence of metals in the samples analyzed in our study

reduces the usefulness of these results for our objectives.

78

In a study utilizing high pressure differential thermal analysis (HPDTA) of fatty acid methyl

esters and triglycerides to measure oil stability, the effects of using different metal containers were

investigated. Thermal analysis provides information about such reactions as oxidation and reduction,

and has potential use in identifying polymers (94). The transition point of the exothermic curve

(TIE) was affected by sample containers made of copper, iron and platinum, indicating an interaction

between the fatty acid methyl esters and these metals.

Peterson et al. investigated the relationship between injector coking and glycerol content,

viscosity, and molecular weight (86). In a study of ten different fuels, data for viscosity, percent oil

esterified, total glycerol, and heat of combustion were compared after rating the injector coking

tendencies. Linear regression was performed to compare each parameter. The authors hypothesized

that the content of total glycerol would be responsible for the increase in injector coking; however,

“the r-squared value was less than 0.0 1 between these parameters.” More probably, injector coking

was related to molecular weight and viscosity, with r-squared values of 0.61 and 0.68, respectively.

Summary of Fueling Issues and Problems

In summary, metals such as steel, lead, copper, and aluminum, in small amounts, have acted

as a catalyst to induce gum formation with biodiesel and biodiesel blends. The catalytic metals

significantly increased the rate at which the TAN increases. The potential for gum formation

increased as the total acid number (TAN) of the fuel increased.

The procedures used to transesterify biodiesel also impacted the TAN of the biodiesel and

biodiesel blends. The TAN of biodiesel that was distilled during transesterification and aged was

much more likely to be higher than non-distilled biodiesel. Distillation removed the natural anti-

79

oxidants present in soybean oil, which normaily inhibit the change in TAN.

Some commercial anti-oxidants (such as TBHQ) h ave been used successMy to reduce the

rate at which the TAN of biodiesel increases. Based on this review, additional testing should be

conducted to re-evaluate both natural and commercial anti-oxidants which inhibit the oxidation of the

fuel will reduce the likelihood of gum formation when fueling with biodiesel and biodiesel blends.

80

Appendix 2

Organic Analysis at the University of Missouri

Organic Analysis at the University of Missouri

At the University of Missouri, the procedure described by Plank et al. (1) was used.

Chemicals and Reagents

The reference standards monolinolein (ML), dilinolein (DL), and trilinolein (TL) were

obtained from Sigma (St. Louis, MO), and were declared by the manufacturer to be 99% pure. The

internal standard tridecanoin (TD) was procured from Fluka Feinchemikalien (Buchs, Switzerland),

and was declared >99% pure. The silylating agent N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA)

(declared 99+% pure, derivatization grade) was purchased from Aldrich Chemicals (Milwaukee, WI)

in sealed ampules and was transferred under nitrogen flow into sealed vials for later use. Pyridine and

toluene were of analytical grade and were procured from Sigma (St. Louis, MO) and Fisher (St.

Louis, MO). Derivatization and dilution were performed in 9 mL reaction vials with teflon-lined

screw caps which fit into a Thermolyne Type 17600 Dri-Bath heating block.

Preparation of Standard Solutions

ML, DL and TL were received in 100 mg ampules. The content of each ampule was

transferred quantitatively with a Pasteur pipette and rinsed with pyridine into a 25 mL volumetric

flask, and filled to volume with pyridine. This yieided a stock solution with a concentration of 4

mg/mL 500 mg of the TD internal standard were weighed into a 50 mL volumetric flask using an

analytical balance and fiiled to volume with pyridine, yielding a stock solution concentration of 10

mg/mL All stock solutions were refrigerated at 4°C. These stock solutions were used to prepare

standard solutions containing 0.05 mg/mL, 0.10 mg/mL, 0.15 mg/mL and 0.20 mg/rnL ML, DL, and

81

TL. The concentration of the TD internal standard was 0.25 mg/mL in all standard solutions. The

standard solutions were prepared as follows: Using a calibrated Rainin motorized microliter pipette

(edp Plus), appropriate volumes (100, 200, 300 and 400 pL) of each standard were transferred to a

9 mL reaction vial. To silylate the free hydroxyl groups on the mono- and diglyceride, pyridine and

BSTFA were added in the following volumes: for the 0.05 mg/mL standard, 400 PL pyridine and 200

PL BSTFA; for the 0.10 mg/mL standard, 500 yL pyridine and 300 PL BSTFA; 600 PL pyridine and

400 PL BSTFA were used for the 0.15 mg/mL standard; and finally, for the 0.200 mg/mL standard,

700 PL pyridiie and 500 PL BSTFA were added. The entire reaction mixture and the BSTFA were

flushed with nitrogen during the procedure, to prevent hydrolysis of the BSTFA by air. The vials

were capped and mixed on a Thermolyne Maxi Mix II vortex, and pIaced in the heating block

maintained at 70 *‘3 C for 20 minutes. The samples were then allowed to reach room temperature

before dilution with toluene to a final volume of 8 mL.

Sample Preparation

Approximately 0.10 to 0.12 g of each filtered fuel sample were weighed into a reaction tube.

200 PL TD, 400 PL pyridine and 200 PL BSTFA were added under nitrogen flow, and mixed and

heated as described above for the standard solutions. Toluene was added to make up a final volume

OfS mL.

Instrumentation and Operating Confiitions

Analyses were performed on a Shimadzu 17A Gas Chromatograph equipped with an AOC-17

Autosampler, QP-5000 (quadrupole) Mass Spectrometer with Electron Impact ionization source set

82

to 1.40 keV and Class 5000 Software system. The mass spectrometer was set to detect fragments

in the range of 30 to 350 amu. The gas chromatograph was fitted with a J&W fUsed silica DB-5ht

high temperature column of the dimensions 30 m x 0.25 mm i.d. coated with 0.1 pm film of 5%

phenyl- and 95% methyl silicone. The temperature program was varied until an optimal method was

obtained. The detector interface was set at 345°C and the injector temperature was maintained at

390°C. The beginning oven temperature was 115”C, followed by an increase to 300°C at a rate of

30”C/minut~. The temperature was then increased at a rate of 10”Uminute to 39O”C, and was heid

there for ten minutes, followed a brief cooling down period. Total run time was 27 minutes.

Results and Discussion

For the quantitative determination of the mono-, di-, and triglycerides present in the B 100 and

B20 fuel samples, a calibration with the reference standards ML, DL and TL was attempted. First,

each component was injected separately to determine retention time and to allow for identification.

An injection of the TD followed. A serious problem was then observed: the triglyceride could not

be detected, even after repeated injections. After consultation with technical support at Shimadzu,

it was assumed that the 350 “C temperature limit of the GC-MS interface was causing the problem.

It was then thought that the high molecular weight of the triglyceride (> 900 amu) caused the

problem: ionization did not lead to Fragments in the preset range of 30 to 350 amu. This theory,

however, was abandoned based upon the fact that SOFIA fragments of low molecular weight should

have formed and would be detectable by the system. An attempt was made to set up a set of

calibration injections based only upon the mono- and diglyceride standards. Standard solutions were

to be injected three times each in order to determine the ratio of the area of each component (AJ /

83

area of the internal standard (AJ (denoted AJAJ. From plots of Aj& vs. weight of component

(WJ I weight of internal standard (W,) (denoted WJW,), intercepts and slopes were determined and

used to quantify the samples. However, the high temperature requirements of the analysis caused

unexpected difficulties with the entire instrumentation, and fkther analysis had to be abandoned.

84

Diagram 1

System and Sample Points

Indoors

Converted to copper 7 from this point on.

Valve - Filter -

Copper pipe 7

\ \

Float tank 4

Heat exchanger

-- Galvanized pipe

Outdoors

- I?-m,r.t,r,d (Loed h)

77-f 77.2

\b”+-BD 20 Mixing Plastic Tank -Plastic

(1 ~~~~a’lOn ) Underground

- 820. 27.61 (InchcapsJ

\ 02sT.9 Supply line

I- Nl4 Cummins engine

- 77.3 (0.253 hJ 77-J &‘503% h) 77-7 (%X-653 h)

Filters L Return line

Appendix 3

Sample Analyses Performed at University of Missouri

Table 1

List of Samples Analyzed at the University of Missouri

Sample Number:

95Ell BOO1 3-6a

95E11 B0027-6a

95EllBOO27-7

95EilBOO27-8

95EllBOO27-9

95EllBOO28-16a

95E11 B0028-26

95EllBOO28-27

95Ell B0046-4a

95E11 B0046-5a

95Ell B0072-2a

95EllBOO77-1

Type:

Diesel

820

820

820

820

BIOO

BIOO

BIOO

Filter

Filter

Diesel

Twin Rivers BlOO

Taken From:

Diesel from sample 95Ell BOO1 3-6 (Inchcape)

820 taken from sample 95EilBOO27-6 (Inchcape)

820 from top of barrel 95EilBOO77-17

820 from bottom of barrel 95EllBOO77-17

820 from red 1000 gallon storage tank

Homogenized SME from barrel 95EllBOO28-16 (Inchcape)

Homogenized SME from barrel 95Ell B0077-4

Homogenized SME from barrel 95EllBOO77-4

Cell wall filter 400 to 500 hrs

340-650 hrs off outside

Diesel from 72-2 (Inchcape)

Bottom of TRT 95Ell B0077-4 (Non-homogenized)

Physical Description:

Sample clear

Sample clear

Clear; small amount of reddish particles adhere to bottom of sample container

Clear with reddish brown particles

Clear with few reddish-brown particles

Sample clear

Sample clear with light brown suspension upon swirling

Sample clear; upon shaking, swirly reddish- brown particulate matter seen

Filter clogged with brown residue

Brown-residue clogged filter

Sample clear

Sample light in color with considerable amount of reddish- brown particulate matter

Sample light in color with bits of reddish brown suspended particles which settle to bottom; goes into suspension upon shaking

95EllBOO77-2 Twin Rivers Top of TRT 95EllBOO77-4 8100 (Non-homogenized)

Filter Engine filter 0 to 250 hrs Filter clogged with dark brown (almost black) residue .

95EI 180077-3

lnlerchem BlOO

Top of lnterchem SME Sample clear with reddish residue which adheres to bottom of container

95EllBOO77-6

Filter Cell wall filter 500-650 hrs Filter clogged with brown residue 95EI 180077-7

95EllBOO77-7

95EllBOO77-8

95EllBOO77-9

95EllBOO77-12

Filter Engine filter 500-650 hrs Filter clogged with brown residue

Filter Engine filter 250-500 hrs Filter clogged with dark brown residue

Filter clogged with brown residue Filter

Twin Rivers 6100

Cell wail filter O-400 hrs

Top of TRT 95EllBOO77-11 Sample clear and light in color with large reddish-brown particles

Sample light and clear 95EllBOO77-14 Twin Rivers Top of TRT 95EllB0077-13 BlOO

Twin Rivers BIOO

Bottom of TRT 95EllBOO77-13 Sample slightly cloudy with particulate matter

95EllBOO77-15

Twin Rivers BIOO

Bottom of TRT 95EllBOO77-11 Large amount of reddish-brown particulate matter

95EllBOO77-16

Very large amount of reddish-brown suspended material

95EllB0077-18 lnterchem 8100

Bottom of lnterchem 95EllBOO77-5

Diesel Diesel from 95Ell BOO1 3-8 Sample clear 95EllBOO77-19

95EI 1 B0077-20a

95Ell B0077-20b

95EllBOO77-2Oc

Diesel Diesel from 7000 gallon tank Sample clear

Sample clear Diesel Diesel from 7000 gallon tank

Diesel Diesel from 7000 gallon tank Sample clear

95E1180077-20d Diesel

95E11 BOO794 Diesel

95EllB00851 Residue

Diesel from 7000 gallon tank

Diesel from 7000 gallon tank with contaminant; no flush

Gummy residue from bottom of 1000 gallon 820 storage tank

Sample clear

Sample clear

Dark brown sticky residue

Table 2

Metals Analyses of Filtered Fuel Samples by ICP and Comparison to Unfiltered Samples in mglkg Fuel

Sample ID

95EllB0013-6a'*

95EllB0027-6a**

95EllB0027-7**

95EllB0027-8

95EllB0027-9

95EllB0028-16a**

95EllB0028-26

95EllB0028-27

95EllB0072-2a**

95EllB0077-l*'

95EllB0077-2"

95EliB0077-6*

sample

Type

Diesel

820

820

820

820

SME

SME

SME

Diesel

SME

SME

SME (Inter.)

Sampled From

Barrel 13-6

(Filtered) Unfiltered1 IFiltered)

0.251 1.963 co.01 co.04 2.122 co.01

[Unfiltered)

mm Iron

[Filtered) ‘Unfiltered)

co.01 0.227 1.285 co.01 0.105 0.904

‘Filtered)

0.045 0.014

Unfiltered) -

0.028 0.029

Barrel27-6 0.292 2.059 0.021 co.01 0.324 1.694 0.061 0.012 co.04 1.749 co.01 0.022 0.011 0.839 0.009 0.013

Topof 0.178 1.914 0.022 0.103 0.219 3.736 0.042 0.120 Barrel 77-17 0.130 2.003 0.010 co.01 0.080 1.1701 0.032 0.038

Bottom of co.04 3.764 co.01 <O.Ol co.008 0.185 <0.005 0.056 barrel 77-17 0.189 4.657 0.016 co.01 0.041 0.312 0.036 0.076

1000 gallon co.04 3.634 0.711 29.388 0.146 4.725 0.048 47.250 red tank 0.415 3.728 1.543 21.823 0.191 3.581 0.053 35.124

Barrel28-16 <0.04 1.888 co.01 0.123 0.042 1.582 co.005 0.045 0.306 2.060 0.019 0.250 0.528 6.956 0.044 0.062

Topof co.04 4.039 co.01 <O.Ol ~0.008 0.412 co.005 0.191 Barrel77-4 <0.04 3.792 co.01 co.01 0.008 0.357 co.005 0.117

Bottom of co.04 3.638 0.011 co.01 0.469 0.469 0.007 0.174 Barrel77-4 co.04 4.132 0.019 0.039 0.438 0.507 0.008 0.119

Barrel72-2 0.315 2.008 0.021 0.023 0.312 0.919 0.051 0.107 0.306 2.004 0.013 0.016 0.403 0.602 0.061 0.188

Bottom of Barrel co.04 1.644 co.01 0.390 0.099 0.875 0.007 0.023 77-4 non-homo. 0.312 1.664 0.025 0.676 0.585 2.990 0.022 0.073

Top of Barrel 77-4 non-homo.

co.04 co.04

0.000 0.000

1.609 1.675

<OS01 co.01

0.000 0.000

0.098 0.037 1.180 0.008 0.145 0.091 0.087 1.089 0.007 0.028

Topof Barrel 77-5

1.414 1.363

0.080 0.000 2.757 0.000 0.127 0.087 0.000 2.768 0.000 0.095

wm Muminum

mm Aluminum

wm Copper

wm Copper

wm Iron

wm Zinc

pm Zinc

Sample ID

95EllBOO77-12*'

95Ell BOO77-14**

95EllBOO77-15*,*'

95Ell B0077-16*

95EllB0077-18

95EllBOO77-19*

95Ell B0077-20a

95Ell B0077-20b

95EllBOO77-2Oc

95Ell B0077-20d*

95El-l B0079-1

Sample

Type

SME

SME

SME

SME

)

SME (Her.

Diesel

Diesel

Diesel

Diesel

Diesel

Diesel

i

Sampled From

Top of Barrel 77-l 1

Top of Barrel 77-l 3

Boltom of Barrel 77-13

Bottom of Barrel 77-l 1

Bottom of Barrel 77-5

Barrel 13-8

7000 gallon tank

7000 gallon tank

7000 gallon tank

7000 gallon tank

7000 gallon tank no flush

mm wm Aluminum Aluminum (Filtered) (Unfiltered)

co.04 1.396 co.04 1.541

wm Copper (Filtered)

co.01 co.01

mm wm f-v-n Copper Iron Iron

(Unfiltered) (Filtered) (Unfiltered)

0.075 0.054 'I .381 0.166 0.127 1.859

wm mm Zinc Zinc

(Filtered)- prlfgurl~g~ *

co.005 0.032 0.007 0.410

co.04 1.662 <O.Ol 0.101 0.168 1.389 0.007 0.185 co.04 1.830 <O.Ol 0.453 0.069 2.758 0.014 0.346

0.000 1.790 0.000 0.213 0.000 1.296 0.000 0.032 0.000 1.871 0.000 0.594 0.000 1.280 0.000 0.146

0.000 1.697 0.000 0.728 0.000 2.345 0.000 0.113 0.000 1.608 0.000 0.880 0.000 2.693 0.000 0.230

0.200 3.648 3.654

0.021 0.272 0.303

0.705 197.734 162.949

0.034 0.833 0.502

0.000 1.547 0.000 0.011 0.000 0.327 0.000 0.104 0.000 1.561 0.000 0.025 0.000 0.462 0.000 0.125

0.225 3.601 0.015 0.028 0.062 0.359 0.036 0.076 0.210 3.515 0.022 co.01 0.051 0.189 0.021 0.090

0.181 3.744 0.011 co.01 0.026 0.065 0.035 0.065 0.202 3.592 0.015 <O.Ol 0.059 0.267 0.060 0.184

0.199 3.728 0.059 <O.Ol 0.773 0.135 0.071 0.063 0.189 4.225 0.022 co.01 0.035 0.437 0.056 0.844

0.000 0.000

co.04 co.04

1.575 0.000 0.024 0.000 0.134 0.000 0.069 1.552 0.000 co.01 0.000 1.181 0.000 0.155

4.061 6.025 0.059 0.023 0.146 0.023 0.259 3.802 <O.Ol co.01 0.066 0.148 0.032 0.187

Detection limits of method are as follows: Al, 0.04 ppm; Cu, 0.01 ppm; Fe, 0.008 ppm; Zn, 0.005 ppm.

*Filtered samples were spiked with 2 ppm Al, Cu, Fe and Zn to ensure recovery; 2 ppm has been deducted from values, **Duplicates of unfiltered samples were spiked with 2 ppm Fe; 2 ppm has been deducted from Fe values.

If no data is given for duplicate, then only one sample analyzed. (Inter.) denotes lnterchem SME which was sampled for comparison

Table 3

Nitrogen Analysis [in w/w% nitrogen]

Sample Name

Engine Filter Blank

Engine Filter O-250 hrs 95EllBOO77-3

820 Tank Residue 95Ell BOO851

B20 from bottom of barrel 27-6 95Ell B0027-6a

BIOO from bottom of barrel 77-l 3 95EllBOO77-15

Methyl Soyate Blank

Diesel Fuel from 7000 gallon tank 95Ell B0077-20a

Nitrogen Content [w/w%]

0.276

0.272

2.114

0.083

0.061

0.043

0.044

Appendix 4

X-Ray Microanalyses of Filter Samples

Lif crystal range 5.5-8.8 KeV for

Cr, Mn, Fe, Co, Ni, Cu and Zn

33

25

17

9

I I 5.756 6.276 6.795 7.315 7.835

Diagram 2

X-Ray Microanalysis

Control (Clean Unused Fuel Filter)

47 -

36 -

Lif crystal range 5.5-8.8 KeV for

Cr, Mn, Fe, Co, Ni, Cu and Zn Zn Ka

, I

5.743 6.250 6.757 7.264 7.770 8.277

Diagram 3

X-Ray Microanalysis

Outside Filter 340-650 hrs

1338

1026

715

403

Lif crystal range 5.5-8.8 KeV for

Cu Ka

I I

5.771 6.305 6.840 7.375 7.909

Cr, Mn, Fe, Co, Ni, Cu and Zn Zn Ka

Diagram 4

X-Ray Microanalysis

Cell Wall Filter O-400 hrs

Lif crystal range 5.5-8.8 KeV for

Cr, Mn, Fe, Co, Ni, Cu and Zn

Cu Ka

Zn Ka

- I

5.762 6.288 6.84 4 7.33,4 7.8S5

Diagram 5

X-Ray Microanalysis

Cell Wall Filter 400-500 hrs

1228-

942 -

656 -

370 -

Lif crystal range 558.8 KeV for

Cr, Mn, Fe, Co, Ni, Cu and Zn Zn Ka

I-

Fe Ka Cu Ka

L

I 5.787 6.337 6.887 7.437 7.988

Diagram 6

X-Ray Microanalysis

Cell Wall Filter 500-650 hrs

Lif crystal range 5.5-8.8 KeV for

Cr, Mn, Fe, Co, Ni, Cu and Zn

1050-

803 -

555 -

308 -

Zn Ka

Cu Ka

I----- - h

5.785 6.334 6.883 7.432 7.981

Diagram 7

X-Ray Microanalysis

Engine Filter O-250 hrs

1474

1132

790

448

Lif crystal range 5.5-8.8 KeV for

Cr, Mn, Fe, Co, Ni, Cu and Zn Zn Ka

5.754 6.271 6.788 7.306 7.823

Diagram 8

X-Ray Microanalysis

Engine Filter 250-500 hrs

P-

i-

15;

11E

80

44

Lif crystal range 5.5-8.8 KeV for

Cr, Mn, Fe, Co, Ni, Cu and Zn Zn Ka

I

1 I 5.771

I 6.305 6.840 7.375 7.909

Diagram 9

X-Ray Microanalysis

Engine Filter 500-650 hrs

Appendix 5

Scanning Electron Microscopy of Filter Samples

Diagram 10

Scanning Electron Microscopy

Control (Clean Unused Filter)

Diagram 1 1

Scanning Electron Microscopy

Outside Filter 34060 h-s

Diagram 12

Seaming Electron Microscopy

Cell Wall Filter O-400 h-s

Scanning Electron Microscopy

Cell Wall Filter 400-500 hrs

Diagram 14

Scanning Electron Microscopy

Cell Wall Filter 500-650 hrs

Diacram 15

Scanning Electron Microscopy

Ellsine Filter O-250 h-s

Diagram 16

Scanning Electron Microscopy

Engine Filter 250400 h-s

Diagram 17

Scanning E Iectron Microscopy

E@e Filter 500-650 h-s

Appendix 6

X-Ray Microanalysis of Filtered Fuel Samples

Control

500

400

300

200

100

0

Clean Unused Whatman 2V Filter

1.42 1.44 1.46 1.48

Energy [key

1.5 1.5

Diagram 18

X-Ray Microanalysis (TAP)

Control Clean Unused Whatman 2V Filter

400

c/l 300 -E

z 0 200

100

5.32 6.32 7.32 8.32 9.32

Energy [kev]

Diagram 19

X-Ray Microanalysis (LiF)

95Ell BOO1 3-6a

100

Diesel from 13-6 (Inchcape)

0 1.42 1.46 1.48

Energy [key 1.5 1.5

Diagram 20

X-Ray Microanalysis (TAP)

95Ell BOO1 3-6a

500

400

300

200

100

0

Diesel from 13-6 (Inchcape)

5.32 6.32 7.32 8.32 9.32

Energy [key

Diagram 2 1

X-Ray Microanalysis (LiF)

95Ell B0027-6a .- 500

400

300

200

100

0

B20 from 27-6 (Inchcape)

1.42

Energy [keV]

1

1.5

Diagram 22

X-Ray Microanalysis (TAP)

95El I B0027-6a

500

400

c/l 300 E 3 0

0 200

100

5.32

FEE I------ l---L-- I---+--

820 from 27-6 (Inchcape)

6.32 7.32 8.32 9.32

Energy [keV]

Diagram 23

X-Ray Microanalysis (LiF)

100

0

95Ell B0027-7 Top of B20 Barrel 7747

i I

4

T I - i

:

~~~~~~

<::. .:>:: :.: .:* .,, I: ‘:.:.:::.::::::::-:::::::“::2::::~,::~~::::.::~:

I

1.42 1.44 1.46 1.48

Energy [key

i I . .._..... . .._.... :-..:.:.: .i.._ @ I! !

qqq ’ /

I

1.5 1.5

Diagram 24

X-Ray Microanalysis (TAP)

95Ell BOO27-7

500

400

300

200

100

0

Top of 820 ‘Barrel 77-17

5.32 5.82 6.32 6.82 7.32 7.82 8.32 8.82 9.32

Energy [key

Diagram 25

X-Ray Microanalysis (LiF)

95Ell B0027-8

500

400

300

200

100

0 1.42

Bottom of 820 Barrel 77-17

( I

I. 1

: I I I

i I I I 1

1.44 1.46 I .48

Energy [keV]

1.5 1.52

Diagram 26

X-Ray Microanalysis (TAP)

95Ell B0027-8 Bottom of 820 Barrel 77-17

5.32 6.32 7.32 8.32 9.32

Energy [key

Diagram 27

X-Ray Microanalysis (LiF)

95Ell B0027-9 820 from 1000 Gallon Red Tank

500

400

300

200

100

0 1 1.42 1.46 1.48

Energy [key

1.5 I.52

Diagram 28

X-Ray Microanalysis (TAP)

95El I BOO2799 820 from 1000 Gallon Red Tank

0 5.32 5.82 6.32 6.82 7.32 7.82 8.32 8.82 9.32

Energy [key

Diagram 29

X-Ray Microanalysis (LiF)

95Ell BOO284 6a I31 00 from 28-16 (Inchcape)

500

400

300

200

100

0 1.42 1.44 1.46 1.48

Energy [kev]

1.5 1.52

Diagram 30

X-Ray Microanalysis (TAP)

500

400 !- I

s 300 F

I I

100 t-

0 I-&-

5.32

95Ell BOO284 6a I31 00 from 28-l 6 (Inchcape)

I I I

6.32 7.32 8.32 9.32

Energy [key

Diagram 3 1

X-Ray Microanalysis (LiF)

500

400

300

200

100

0 1.42

95El I B0028-26 Top of Homogenized SME Barrel 77-4

1.46 I .48

Energy [key

1.5 1 .:

Diagram 32

X-Ray Microanalysis (TAP)

95Ell B0028-26 Top of Homogenized SME Barrel 77-4

5.32 6.32 7.32 8.32 9.32

Energy [keV]

Diagram 33

X-Ray Microanalysis (LiF)

95El I BOO28-27 500

400

300

200

100

0

Bottom of Homogenized SME Barrel 77-4

1.42 1.44 1.46 1.48

Energy [keV]

Diagram 34

X-Ray Microanalysis (TAP)

500

400

cn 300 -z z 0 200

100

0

95EllB0028-27 Bottom of Homogenized SME Barrel 77-4

-i---l- 5.32 6.32 7.32 8.32 9.32

--i--

T I 1

I 1

i

--l--l-

Energy [key

Diagram 35

X-Ray Microanalysis (LiF)

0

95El I B0072-2a Diesel from 72-2 (Inchcape)

1.42 1.44 1.46 1.48

Energy [keVj 1.5 I.52

Diagram 36

X-Ray Microanalysis (TAP)

95El I B0072-2a Diesel from 72-2 (Inchcape)

500

400

cn 300

G (‘) 200

100

0 5.32 6.32 7.32 8.32 9.32

Energy [keV]

Diagram 37

X-Ray Microanalysis (LiF)

500

400

300

200

100

0

95Ell BOO774 Non-Homogenized BlOO from Bottom 77-4

1.42 1.48 1.5

Energy [kev]

Diagram 38

X-Ray Microanalysis (TAP)

95El I BOO774 Non-Homogenized BIOO from Bottom 77-4

500

400

300

200

100

0 5.32 6.32 7.32 8.32 9.32

Energy [kev]

Diagram 39

X-Ray Microanalysis (LiF)

95EllBOO77-2

14000 I Non-Homogenized BIOO from Top 77-4

12000 I\ I

10000 \ \

8000 I

6000 ' ;

\

4000 I \ I I

I 2000 I !

I /

0 m-Hew-m-> 1.42 1.44 1.46 1.48 1.5 r

Energy [kev]

Diagram 40

X-Ray Microanalysis (TAP)

95El I B0077-2 Non-Homogenized 6100 from Top 77-4

400 r 300 I- 200 I--- 100

I----

Oh 5.32

I I I I I n . ndU

6.32 7.32 8.32 9.32

Energy [key

Diagram 4 1

X-Ray Microanalysis (LiF)

95Ell B0077-6 lnterchem 8100 from Top of Drum

500

400

300

200

100

0 1.42 1.44 1.46 1.48 1.5

Energy [keV]

Diagram 42

X-Ray Microanalysis (TAP)

95El I B0077-6

500 - lnterchem BIOO from Top of Drum

400

0 ------ 5.32 6.32 7.32 8.32 9.32

Energy [key

Diagram 43

X-Ray Microanalysis (LiF)

95EllB0077-12 BIOO from Top of Drum 2 (77-l 1)

1.42 1.44 1.46 1.48

Energy [keV] 1.5 1.’

Diagram 44

X-Ray Microanalysis (TAP)

cn 300 E

0” 200

95Ell BOO774 2 BIOO from Top Drum 2 (77-l I)

5.32 5.82 6.32 6.82 7.32 7.82 8.32 8.82 9.32

Energy [keVj

Diagram 45

X-Ray Microanalysis (LiF)

6000

95El I B0077-14 6100 from Top of Drum 3 (77-l 3)

5000

4000 cn E 2 3000

c) 2000

1000

0 1.42 1.44 1.46 1.48 1.5 1 :

Energy [kev]

Diagram 46

X-Ray Microanalysis (TAP)

500

400

m 300 -2

z (2 200

100

0

95Ell B0077-14 Bl 00 from Top of Drum 3 (77-l 3)

5.32 6.32 7.32 8.32 9.32

Energy [key

Diagram 47

X-Ray Microanalysis (LiF)

6000 ‘.

95El I BOO774 5 95El I BOO774 5 BIOO from Bottom of Drum 3 (77-l 3) BIOO from Bottom of Drum 3 (77-l 3)

1.46 1.46 1.48 1.48

Energy [key Energy [key

1.5 1.5 I.52 I.52

Diagram 48

X-Ray Microanalysis (TAP)

95EllBOO77-I 5 8100 from Bottom of Drum 3 (77-13)

0 5.32 5.82 6.32 6.82 7.32 7.82 8.32 8.82

Energy [key

Diagram 49

X-Ray Microanalysis (LiF)

100

0

95EllB0077-16 BIOO from Bottom of Drum 2 (77-l I)

1.42 1.44 1.46 1.48

Energy [key 1.5 I .52

Diagram 50

X-Ray Microanalysis (TAP)

95El I BOO774 6

‘_ .

500

400

300

200

100

0

6100 from Bottom of Drum 2 (77-l 1)

5.32 6.32 7.32 8.32 9.32

Energy [kev]

Diagram 5 1

X-Ray Microanalysis (LiF)

.

95Ell BOO774 8 Bottom SME Barrel 77-5 (Interchem)

1.42

Energy [key

Diagram 52

X-Ray Microanalysis (TAP)

,-, .‘; .

;.’ -?, ^ ::,

:

:

95Ell BOO774 8 Bottom SME Barrel 77-5 (Interchem)

5.32 5.82 6.32 6.82 7.32 7.82 8.32 8.82 9.32

Energy [keV]

Diagram 53

X-Ray Microanalysis (LiF)

500

400

300

200

100

0

95Ell BOO774 9 Diesel from 13-8

/ ,

1

/

- ‘; 1.42 1.44 1.46

I T

t I I 1.48

Energy [keV] 1.5 1.52

Diagram 54

X-Ray Microanalysis (TAP)

500

400

300

200

100

0 5

95El I BOO774 9 Diesel from 13-8

32 6.32 7.32 8.32 9.32

Energy [keV]

Diagram 5.5

X-Ray Microanalysis (LiF)

95Ell BO Diesel from 7000 Gallo

379-l 1 Tank No Flush

T-

-t

1.42 1.44 1.46 7

1.48

tnergy [key

f /

-

1

:.:.y.:...: .:.:.:.:.:. gg F . . . . . . . . . . . . . . . . . . . -. . ..A.. . . . . w 1.5 IF .L

Diagram 56

X-Ray Microanalysis (TAP)

95EllB0079-1 Diesel from 7000 Gallon Tank No Flush

800

200

0 i

5.32 6.32 7.32 8.32 9.32

Energy [key

Diagram 57

X-Ray Microanalysis (LiF)

420

cn 320 Y

r 3 s 220

120

I I 1

95Ell B0077-20a Diesel from 7000 gallon tank

I I ! i

-l---l- I I

i I I I

I

1.46

I

- /

I !

1.48 r tnergy [key

Diagram 5s

X-Ray Microanalysis (TAP)

95Ell B0077-20a Diesel from 7000 gallon tank

500

400

c/l 300 r

2 0 200

100

0 5 u-32 6.32 7.32 8.32 9.32

Energy [keV]

Diagram 59

X-Ray Microanalysis (LiF)

500

400

300

200

100

0

95Ell B0077-20b Diesel from 7000 gallon tank

-

1.42

Energy [keV]

Diagram 60

X-Ray Microanalysis (TAP)

95Ell B0077-20b Diesel from 7000 gallon tank

5.32 6.32 7.32 8.32 9.32

Energy [keV]

Diagram 6 1

X-Ray Microanalysis (LiF)

500

400

300

200

100

0 I .42

95EllBOO77-2Oc Diesel from 7000 gallon tank

1.46 I .48

Energy [kev]

1.5 I.52

Diagram 62

X-Ray Microanalysis (TAP)

500

400

300

200

100

0 5.32 6.32

95Ell BOO77920c Diesel from 7000 gallon tank

1

7.32 8.32 9.32

Energy [key

Diagram 63

X-Ray Microanalysis (LiF)

500

400

300

200

100

0 1.42

95Ell B0077-20d Diesel from 7000 Gallon Tank

---I---

-----z +

1.44 1.46 1.48

Energy [keV]

I I

1.5 1.52

Diagram 64

X-Ray Microanalysis (TAP)

500 ~ r. 400 I---

0- 5.32

95Ell B0077-20d Diesel from 7000 Gallon Tank

6.32 7.32

C

Energy [kev]

Diagram 6.5

X-Ray Microanalysis (LiF)

Appendix 7

Sample Analyses Performed at System Lab Services, Inc.

Table 1

List of Samples Analyzed at System Lab Services, Inc.

Sample Number:

95EllBOO27-7

95EllBOO27-8

95EllBOO27-9

95Ell B0077-1

95EliBOO77-2

95EllBOO77-12

95Ell B0077-16

95Ell B0077-20b

95EllBOO79-1

Description:

B20

B20

820

TRT SME

TRT SME

TRT SME

TRT SME

Diesel

Diesel

Taken From:

B20 from top of barrel 95EllB0077-17

820 from bottom of barrel 95EllBOO77-17

B20 from red 1000 gallon fuel tank

Bottom of TRT 95Eil B0077-4 (Non-homogenized)

Top of TRT 95EllBOO77-4 (Non-homogenized)

TopofTRT95EllB0077-11

Bottom of TRT 95EilBOO77-11

Diesel from 7000 gallon tank

Diesel from 7000 gallon tank with contaminant; no flush

Physical Description:

Clear; small amount of reddish particles adhere to bottom of sample container

Clear with reddish brown particles

Clear with few reddish-brown particles

Sample light in color with considerable amount of reddish-brown material

Sample light in color with bits of reddish brown suspended particles which settle to bottom; goes into suspension upon shakin

Sample clear and light in color with large reddish-brown particles

Large amount of reddish-brown particulate matter

Sample clear

Sample clear

Table 2

Organic Analyses of Fuel Samples by SLS in wt. % (C. Plank method)

Top 77-4

20522 95EllBOO77-12 SME Twin River 0.144 Top 77-11

II I 20523 95EllBOO77-16 1 SME 1 ,X&W;, 11 0.141 0.027

Diglycerides Triglycerides Free Glycerin Total Glycerin wt. % wt. o/d wt. % wt. %

0.028

0.032

0.032

0.006

0.010

0.025

0.005

0.015 0.055

0.008 0.049

0.007

0.012

0.052

0.053

Detection Limits: glycerin, 10 ppm (0.001 wt. %); glycerides, 200 ppm (0.02 wt. %).

-.. ^ I able 3

Metals Analyses of Fuel Samples by SLS in ppm [mg/kg Fuel]

SLS No. -

20515

20516

20517

20518

20519

20520

20521

20522

20523

Sample ID

5E11 B0077-201

95EilBOO27-7

95EllBOO27-8

95EllBOO27-9

95EllBOO79-1

95EllBOO77-1

95EllBOO77-2

35EllBOO77-12

35EllBOO77-18

Sample

Type

Diesel

820

820

820

Diesel

SME

SME

SME

SME

Sampled From

7000 gallon tank

Top of barre 77-17

Bottom of barrel 77-l i

1000 gallon red tank

7000 gallon tank

Bottom of barrel 77-4

Top of barrel 77-4

Top of barre 77-l 1

Bottom of barrel 77-l 1

Al

=

1

2

1

2

1

4

1

1

2

211

=

0

0

0

5

0

0

0

0

0

Fe

=

1

1

1

1

1

1

1

1

1

Zn

E

0

Cr

-

1

Pb

=

1

0

1

0

0

0

1

0

0

Sn

E

0

Ni

=

0

0

0

0

0

0

0

0

0

%I

=

0

0

0

0

0

0

0

.

0

0

Mn

-

0

0

0

0

0

0

0

0

0

Si

E

1

B

=

0

0

1

0

0

1

1

1

1

Na Mg

-

1

1

1

0

1

1

1

1

1

-

0

0

0

0

0

0

0

0

0

Ca

=

1

Ba P

=

1

0

2

0

0

0

0

0

0

=

7

0

11

2

1

2

8

8

0

40

=

1

0

1

0

1

0

0

0

0

Ti

=

0

0

0

0

0

0

0

0

0

v

=

0

0

0

0

0

0

0

0

0

K

=

12

11

0

12

15

0

0

0

0

Detection limits: iron, chromium, lead, copper, zinc, vanadium, tin, aluminum, nickel, silver, molybdenum, silicon, boron, sodium, phosphorous, titanium: 1 ppm calcium, barium, magnesium: 10 ppm; potassium, 20 ppm.

Table 4

Acid Number, Peroxide Value and Karl Fischer Moisture of Fuel Samples by SLS

SLS Number

20515

20516

2051

20519

20520

2052

20522

20523

Sample ID

95Ell B0077-20b

95E 1180027-7

95E 1180027-8

95E 1180027-9

95E 1180079-l

95E 1180077-l

95E 11 B0077-2

95EllBOO77-12

95El 180077-16

Sample Sampled

Type From

Diesel

820

820

820

Diesel

SME

SME

SME

SME

7000 gallon tank

Top of barrel 77-17

Bottom of barrel 77-l 7

1000 gallon red tank

7000 gallon tank

Twin River Bottom 77-4

Twin River Top 77-4

Twin River Top 77-11

Twin River Bottom 77-l 1

Acid Number Peroxide Value Karl Fischer Moisture ASTM D 664 ASTM D 3703 ASTM D 4928

mg KOH/g MEG/kg wm

0.01

0.27

0.27

0.10

0.03

0.96

0.91

0.89

0.94

52

111

92

54

224

100

162

154

198

For detection limit information, see the Annual Book of ASTM Standards, Volumes 05.01 and 05.02.

231

233

196

283

129

994

951

1000

1011

Sensitivities for these ASTM methods are intended for conventional petroleum products only; detection limits in conjunction with testing of biodiesel and biodiesel blends have yet to be determined

10

SO

70

$0 l------

I / JSOO

I 301

-.. I 16UO

, 1000

, SOD

90

Diagram 67

FTIR Analysis of Durability Diesel

Wavanumbar [cm-l)

2 .u A 0

I i I I , / I I 18oD 1600 1400 1200 loon 600 600

Wavenumber [cm-11

Diagram 68

FTIR Analysis of Twin Rivers SME

91

7c

LnY 1 LN

a&o , 3500 , 4 2100 2000

, 1500 law 500

Wavenumber (cm-l)

----I

-- ii00

- I

i iao I I

1600 --

1200 1000 sio 600 w.wenumtler (cm-r,

Diagram 69

FTIR Analysis of Interchem SME

56 ,

2000 I 1800 1600 I I 7400 1200 lob0 a00 6bO

Wavenumbnr (cm-l)

Diagram 70

FTIR Analysis of Cell Wall Filter

? \1

1

14bl I I I

, 1800 1600 1 zoo 1000 400 600 \‘dovenumber.&m-11

Diagram 71

3

FTIR Analysis of Float Bowl Residue

IWO I

JO00 I I

2500 2000 Wannumber (cm-l)

I I 1600 1000 40

-.-_-__ --- ---

I / I I / I

0 1800 1600 1400 1200 1000 ai0 600 Wdvanumk (cm-i)

Diagram 72

FTIR Analysis of SME Residue

Appendix 9

Oxidative Stability Tests Performed at Inchcape and System Lab Services

Oxidation Stabi!itv Sample Summarv

SamQle Numbef Stability Particuhte

Oar9 Sampled Sample Type Sample Procedure Result ReSUlt mg!lCOml mgf100mi

A 95Ell&0013-6 Aug 11 Diesel - sample taken allw tx~st pump tram 7000 g.Mo+-~ norage lank 52.3 0 - fuec liltar was removed to eh’minata possb(a SME contamination - q4em was fktshed with - 1 g&m ol djesel - amber glass mmpie bcde was rinwd wiih diesel - one Iitre sample taken .labelL& ard logged

e 95E11ec#27-6

I A-38 e1encl - sampka taken altBr Most pump from mixing fank 24.9 8.6

. fuel fl!ln. kSr, 1.~ ‘k !v.% SC?!!= -- . -

I-

- pump. litter and hose flushed pmr lo taking s8mple -amber glass nm@e borne was rinsed with blend - one utro scunpls taksn .laberled end logged

C 95El180028-16 sepc 13

I

SME - sample fdu31-1 dlrealy irom drum 1.6 57.: - amber gia6.c bomple bo0Is was finsed with SME - one Ike samole taken Jabe4ed and -sod

0 ¶5El16ml3-3 se01 27

- 5 gailon sample taken. labsled and kqped . 40 gallon sam taken

E 95EllBcQ13-10

I

j o=2 1 Oies* 1: f~~s~~ sample taken after boost prmp from 7OC0 gallon slorage rank 9222 3.7 fuel filler kepr In UIM (or Ihis sample 86 ~6 I&W and h~lty

F 9% 1180072-Z a320 Diesel

- no flushing as fully tI&md Sept 27 - am&~ gLau sample boUbe was tinsod with diesel - one litc9 sarngle LlLren Jabbed and w

- sample taken from 5 galrOn saw of SW! 27 - steel sample can was rinsed tim diezel . one fiwe sample taken .labs4ed and Icqgod

I

0.2

-

G 95Errecx373.2

Submitted by:

Date Reported: Sampled By:

System Lab Sekvices (I &d&m of WXanu P+e Line iZ&my

Brian Manicom Ortech Corporation 2395 Speakman Drive Sheridan Science Tech Pk. Mississauqa, Ontario Canada L5K 7 83

1 O/l 6/l 995 Submitter

Lab Test Number: 05083 Sample Identification: Ortech Sample # 95E’l160073-11 Sample Date: 7 opl/1995

Test Requested Result D 2274 0.69 mg/lOOml

Senior Research Chemist

(913) 621-3603 Fax (913) 621-0609 lO%ASunohincRoad, KansasCity,E(S 66[15