Journal of Chromatographic Science, Vol.

45, November/December 2007

Determination of Total Biodiesel Fatty Acid Methyl,

Ethyl Esters, and Hydrocarbon Types in Diesel Fuels
by Supercritical Fluid Chromatography–Flame
Ionization Detection
John W. Diehl and Frank P. DiSanzo*
Analytical Sciences Laboratory, ExxonMobil Research and Engineering, Paulsboro Technical Center, Paulsboro, NJ 08066

Abstract analytical methodologies for such an analysis. Methods based on

Fourier-transform infrared spectroscopy (FTIR) may yield
A method for determining total biodiesel methyl and ethyl ester approximate values because it is difficult to obtain an accurate
content in diesel fuels by supercritical fluid chromatography–flame response factor (RF) for all of the potential esters that may be
ionization detection (SFC–FID) is developed. A silica column produced in biodisels. Chromatographic separation of the
typically used for determining aromatics in conventional diesel biodiesel esters from the diesel hydrocarbons and detection with
fuels by ASTM D5186 is back-flushed after separation of the conventional universal mass detectors, such as the flame ioniza-
hydrocarbons to allow elution of the various esters as a single tion detector (FID) is more desirable. Currently, D5186 (1) is a
“total biodiesel” distinct peak. The modification concurrently
popular technique, used in regulation by California’s Air
allows the determination of total aromatic hydrocarbons and their
distribution as mono- and polynuclear compounds, as described in
Resource Board for the determination of total aromatics and aro-
the current version of D5186. The instrument response is linear matic ring distributions in conventional diesel fuels. D5186 uses
from 1.0% to 50% biodiesel esters with a signal-to-noise ratio of supercritical carbon dioxide, a normal phase silica column, and
25 at the 0.1% level. The short-term relative standard is 0.8%. FID, but has not been fully evaluated for the analysis of the
Normalized percent quantitation using a hydrocarbon response hydrocarbon types in the presence of co-blended biodiesel esters.
factor of 1.00 and an ester response factor of 1.19 provide an The use of supercritical fluid chromatography in the analysis of
average percentage error of 1.8% when measuring actual hydrocarbons is well-documented in the literature (2–4).
biodiesel/hydrocarbon fuel blends. The ester response factor is the Presented here is the investigation into determining total fatty
average of the response factors of 10 pure ester compounds. These acid methyl and ethyl esters biodiesel in diesel fuels by super-
responses are calculated from respective solutions of each ester critical fluid chromatography–flame ionization detection
and the four compounds, hexadecane, toluene, tetralin, and
naphthalene, as used for the D5186 response factor mixture.

Introduction

The proposed use of biodiesel esters derived from a variety of

biological sources, such as canola, corn, fish, and other oils, as
diesel fuel blending components has led to the need within the
petroleum industry to determine these compounds in the pres-
ence of the diesel fuel hydrocarbon matrix. Fatty acid methyl and
ethyl esters ranging in carbon number from C10 through C26
would be produced by trans-esterification of the mono-, di-, and
tri- glycosides in the oils with methanol and/or ethanol. The
product of this reaction would then be blended with conven-
tional hydrocarbon-based diesel fuels.
Figure 1. Valve diagram of a dual D5186/D6550 SFC system used for modi-
Currently, there are a limited number of easy and accurate
fied D5186 for simultaneous determination of hydrocarbon types and
biodiesel esters (valve 2 not used and set in “off” position).
* Author to whom correspondence should be addressed.

690 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher’s permission.
Journal of Chromatographic Science, Vol. 45, November/December 2007

(SFC–FID), with the incorporation of “backflush” for analysis of specifications. (v) Column Pressure: 2176 psi (15.0 Mpa). (vi)
biodiesel esters as well as hydrocarbon types. Column flow: 2 mL/min. (vii) Run time: 45 min. (viii) Column
oven temperature: 40°C isothermal. (ix) Detector: FID at 350°C.
Air flow = 421 mL/min, H2 flow = 66 mL/min, expanded CO2 flow
= 32 mL/min. Detector gas flows were set to provide analysis of
Experimental

An Analytical Controls (Rotterdam, The Netherlands) combi- Table III. Accuracy Tests of Pure Compounds in a Diesel
nation D5186 (1) and D6550 (5) SFC system (Figure 1) was con- Fuel with the Average Response Factor*
figured as follows. (i) Injection volume: 0.5 µL with a Berger
Instruments (Thar, Pittsburgh, PA) Auto-sampler (all samples Compound Determined Actual % Relative Error†
were injected neat). (ii) Supercritical fluid: carbon dioxide. In-
Methyldecanoate 9.2 ± 0.1 10.2 9.8%
house commercial source, filtered over silica gel scrubber ahead Methylundecanoate 9.2 ± 0.1 10.0 8.0%
of pumping system, blank analysis yielded zero interference. (iii)
Pressure control: downstream. (iv) Columns: 25 cm × 4.6 mm
Methyldodecanoate 9.8 ± 0.1 10.5 6.7%

i.d. × 5 µm silica (Analytical Controls). Columns meet D5186

Methyltridecanoate 0.3 ± 0.1 10.9 5.5%
Ethylmyristate 10.3 ± 0.1 10.5 2.9%
Methylpentadencanoate 10.3 ± 0.1 10.0 3.0%
Methylpalmitate 9.8 ± 0.1 9.8 0.0%
Table I. Response Factors Calculated for Each Ester Methyloleate 10.0 ± 0.1‡ 10.0 0.0%
Ethyloleate 10.3 ± 0.1 10.2 1.0%
Compound Response Factor Methylarachidate 10.0 ± 0.1 9.8 2.0%

* Results are in mass %.

Methyldecanoate 1.30 † % Relative Error = [Absolute Value (Determined – Actual)/Actual] × 100. Average
Methylundecanoate 1.30 Relative Error = 3.9%.
‡ ± two times the standard deviation.
Methyldodecanoate 1.18
Methyltridecanoate 1.25
Ethylmyristate 1.17
Methylpentadecanoate 1.13
Methylpalmitate 1.13
Methyloleate 1.14
Ethyloleate 1.12
Methylarachidate 1.13

Mean 1.19
Norm.

RSD (n – 1) 6.0%

Table II. Precision Data of a Methyloleate Spiked Diesel

Fuel with the Average Ester Response Factor*

Injection Monoaromatics PNAs Total Aromatics Methyloleate

Time (min)
1 25.5 4.2 29.7 10.1
2 25.6 4.2 29.6 10.0 Figure 2. SFC chromatogram of 10% esters in diesel fuel.
3 25.6 4.2 29.8 10.1
4 25.6 4.2 29.8 10.0
5 25.5 4.1 29.6 10.0
6 25.4 4.3 29.7 10.0
7 25.4 4.3 29.7 10.0
8 25.5 4.2 29.6 10.0
9 25.5 4.2 29.7 10.0
10 25.5 4.1 29.6 10.0

% RSD† 0.3 1.6 0.3 0.4

Mean 25.5 4.2 29.7 10.0
Actual N/A‡ N/A N/A 10.0
% Recovery N/A N/A N/A 100.0

* Results are in mass %.

† N – 1, n = 10 over two days.
‡ N/A = Not applicable. Figure 3. External standard calibration plot (area vs. g/mL) for methyloleate.

691
 Journal of Chromatographic Science, Vol. 45, November/December 2007

hydrocarbons as described in ASTM method D5186 (1). (x) (Figure 1) was programmed to perform this function. The silver
Backflush valve time (valve 1 off): 30 min. Valve 2 remained off loaded column used in the D6550 configuration was placed in
throughout the separation, keeping the silver column of D6550 by-pass mode and not used; that is, only the six port valve was
isolated from the carrier stream. A valve diagram is shown in used to backflush the esters.
Figure 1. (xi) Linearity test: 0.10, 0.30, 0.50, 1.00, 3.0, and 5.00 g Table I lists the RFs calculated for each of the 10 esters used in
of methyloleate were weighed to the nearest 0.0001 g into this work. Ethyl esters were included because trans-esterifica-
respective 10.0-mL volumetric flasks, which were diluted to the tion with ethanol as well as methanol has been proposed for the
mark with n-heptane and mixed 3 min on a vortex mixer. This production of biodiesels. The data in Table I shows a generally
process produced solutions ranging from 0.01 g/mL to 0.50 decreasing RF with an increasing ester carbon number.
g/mL. The six solutions were injected, and a curve of area vs. con- However, for esters in the range of C15–C20, the RFs are relatively
centration was generated. (xii) Calibration: pure compounds constant and well within the precision of the determination. The
were purchased from Aldrich Chemical Co. (Milwaukee, WI). Ten average ester response factor was 1.19 with a relative standard
mixtures similar to the performance mix required in ASTM deviation (RSD) of 6.0%. For further evaluation, the quantitative
D5186-3 except with an ester added were prepared as follows: analysis was simplified as in D5186 by using normalized percent,
70.0 mass% hexadecane; 20.0 mass% toluene; 3.0 mass% with all non-, mono-, and poly-nuclear aromatics response fac-
tetralin; 2.0 mass% naphthalene; and 5.0 mass% ester. tors set at 1.00 and the ester response factor set at 1.19.
The described mixtures were injected onto the SFC and Figure 2 shows a chromatogram of 10% (w/w) methyloleate
response factors calculated for each ester as follows: spiked into a diesel fuel. The ester eluted in the backflush mode
[gravimetric mass % ester × total area of hydrocarbons]
as an approximately symmetrical peak, and the hydrocarbons
[gravimetric mass% total hydrocarbons × area ester]
RF = were separated into non-aromatics (“saturates”), monoaro-
matics, and ploynuclear aromatics, as in conventional D5186.
where: total hydrocarbons = sum of hexadecane + toluene + “Polynuclear aromatics” is defined as diaromatics + hydrocar-
tetralin + naphthalene bons. Figure 2 indicates that the addition of the backflush capa-
bility to the silica column in D5186 leads to the normal
determination of the aromatic hydrocarbon types according to
D5186 specifications in the forward elution, and also to the
Results and Discussion simultaneous determination of the biodiesel ester components
by backflushing the column after the elution of the polynuclear
Initial studies using D5186 indicated that the biodiesel esters aromatics. For instruments performing only conventional
were not eluted from the silica gel separation column specified in D5186, the single simplest modification required would be the
D5186. The esters, which are more “polar” than the diesel hydro- installation of a four port instead of a six port switching valve to
carbons, are strongly adsorbed on the silica gel, and their elution backflush the silica gel column.
times in the forward elution from the column were extremely Figure 3 shows the 1.0 to 50% (w/v) external standard calibra-
long. As a result, a backflush of the silica column after the elu- tion curve (Area of Ester vs. Concentration of Ester) for methy-
tion of the hydrocarbons was evaluated. For convenience, a com- loleate. The curve was linear with a correlation coefficient
bination commercially available D5186//D6550 instrument > 0.999. The signal-to-noise ratio at 0.1% was 25.
Table II shows the repeated normalized per-
cent analysis of 10% methyloleate in diesel
Table IV. Aromatics Analyses of the Ester Test Solutions in Table III* with the 1.19 ester response factor. The short-
term relative standard deviation (n – 1, n = 10
Ester Spiked Det.† Calc. Det. Calc. Det. Calc. over two days) was 0.4%. For methyloleate,
Fuel Mono-Arom. Mono-Arom. PNAs PNAs Total Arom. Total Arom. the mean recovery is was 100%.
The calibration was further tested with
Diesel base fuel 29.5 30.4 4.5 4.3 34.0 34.7
solutions of pure ester compounds gravimet-
(no esters present) (D5186) (D5186) (D5186) rically spiked into a diesel fuel. These results
Methyl-Oleate 25.6 26.5 4.2 4.1 29.8 30.6 are summarized in Table III. The average rela-
Ethyl-Oleate 25.3 25.5 4.2 4.2 29.5 29.7 tive error of recovery was 3.9%. Table III indi-
Ethyl-Myristate 25.4 26.5 4.3 4.0 29.7 30.5 cates a similar pattern noted with the RFs of
Methyl-Decanoate 25.6 26.5 4.4 4.0 30.0 30.5 the individual esters in Table I. Generally,
Methyl-Undecanoate 25.7 26.5 4.3 4.1 30.0 30.6 esters with a carbon number of C14–C20
Methyl-Pentadecanoate 25.8 26.6 4.4 4.1 30.2 30.7 yielded reasonable results using the average
Methyl-Tridecanoate 25.4 26.3 4.3 4.0 29.7 30.3 RF = 1.19. An error of 3–4% is acceptable for
Methyl-Dodecanoate 25.5 26.4 4.3 4.0 29.5 30.4 most applications. Only esters with carbon
Methyl-Arachidate 25.7 26.6 4.3 4.1 30.0 30.7
number < C14 exhibited the greatest devia-
* Results are in mass % on a total sample basis.
tion. Most current biodiesel esters used in
† Det. = Determined. Calculated from the relative amounts of aromatics and the determined data for the diesel blending are in the carbon range of
mass % of esters (See Experimental section). C14–C20, and reasonable accuracy by this SFC
technique is expected. Accuracy could be

692
Journal of Chromatographic Science, Vol. 45, November/December 2007

Table V. Recovery Studies with Various Types of

Conclusion
Biodiesels Spiked in Several Types of Conventional
Hydrocarbon-Only Diesels A precise and accurate SFC method for determining total
methyl and ethyl esters in blended diesel fuels has been devel-
Sample/Origin Determined Actual* % Relative Error† oped, with the simultaneous determination of hydrocarbon
types. Such a technique should be useful in determining the final
Diesel 1/Rapeseed 10.8 ± 0.1 10.6 1.9 blend composition of blended commercial diesels. Furthermore,
Diesel 2/Tallow 10.4 ± 0.1 11.0 6.4 this technique can be implemented on existing instrumentation
Diesel 3/Tallow 11.2 ± 0.1 11.1 0.9
used for D5186 and/or D6550 analyses with simple modifications
Diesel 4/Soy Bean 10.8 ± 0.1 10.8 0.0
Diesel 5/ Rapeseed 10.7 ± 0.1 10.7 0.0
and calibration.

* Actual = Calculated from the relative amounts of aromatics and the gravimetric data
for the mass % of esters (See Experimental section).
† Average % Relative Error = 1.8%.
Acknowledgment
improved further by calibrating with the actual esters or We acknowledge Jennifer McCulley for assistance in per-
biodiesel used in blended diesel if this information were avail- forming some of the analyses.
able.
Table IV summarizes the aromatic hydrocarbon analyses of the
synthetic blends prepared with individual esters listed in Table
III. The analysis of the ester-free “unspiked” starting diesel fuel References
using the backflush SFC system used for these solutions is
included in the top line, and compared against results obtained 1. ASTM D5186-03, “Standard Test Method for Determination of
from a second instrument with no backflush which met D5186 Aromatic Content and Polynuclear Aromatic Content of Diesel
specifications. The overall data demonstrates that the deter- Fuels by Supercritical Fluid Chromatography“. For ASTM standards,
mined values using the modified backflush SFC system and visit the ASTM website, [Link], or ASTM Customer Service
at service@[Link].
average RF are close to those calculated from the unspiked diesel 2. T.A. Norris and M.G. Rawdon. Determination of hydrocarbon types
corrected for dilution by the addition of the esters. in petroleum liquids by supercritical fluid chromatography with
Table V summarizes accuracy data for a single diesel fuel flame ionization detection. Anal. Chem. 56: 1967 (1984).
spiked several times with different sources of “commercial” 3. R.M. Campbell, N.M. Djordjevic, K.E. Markides and M.L. Lee.
biodiesel esters. Generally, the results are reasonable, with a Supercritical fluid chromatographic determination of hydrocarbon
groups in gasolines and middle distillate fuels. Anal. Chem. 60: 356
recovery relative average error of 4.2%. The diesel 2/tallow rela- (1988).
tive error is slightly higher, most likely due to preparation of 4. F.P. DiSanzo and R.E. Yoder. Determination of Aromatics in Jet and
blended diesel, because similar tallow biodiesel blended in diesel Diesel Fuels by Supercritical Fluid Chromatography with Flame
3 yielded a lower error. Ionization Detection (SFC-FID); A Quantitative Study.
Tests with a variety of glycosides (e.g., mono-, di-, and tri-gly- J. Chromatogr. Sci. 29: 4 (1991).
5. ASTM D6550, “Standard Test Method for Determination of Olefin
cosides) of oils such as those derived from corn, soy bean, sun- Content of Gasolines by Supercritical-Fluid Chromatography” For
flower, pork (lard), and fish showed that these ASTM standards, refer to the ASTM website, [Link], or
pre-trans-esterification feedstock materials did not interfere ASTM Customer Service at service@[Link].
with the analysis. The latter pre-esterification materials are not
likely to be present in significant amounts in the final esters used
for blending, due to industry biodiesel specifications limiting the Manuscript received July 12, 2007;
level of such components in the final ester product for blending. Revision received September 3, 2007.

693