Boron Trifluoride a s Catalyst to Prepare Methyl Esters From Oils
Containing Unusual Acyl Groups
R. KLEIMAN, G. F; SPENCER and F. R. EARLE
Northern Regional Research Laboratory,1 'Peoria, Illinois '61604
ABSTRACT
The procedure of Metcalfe et al. (3) for
the preparation of fatty acid methyl esters,
using boron trifluoride as catalyst, is shown
to be suitable for use with oils containing
fatty acids of unusual structures, such as
conjugated unsaturation, hydroxyl or
epoxy groups, and cyclopropenes in addi-
tion to oils with only the common acids.
In some cases, boron trifluoride was less
destructive to unusual groups than con-
ventional mineral acid catalysts; in others,
derivatives were formed that were suit-
able for quantitation in subsequent gas
chromatographic analysis.
INTRODUCTION
Extensive chemical investigation of a variety
of vegetable oils requires, for efficiency in
operation, a rapid, generally applicable pro-
cedure for conversion Of the oils to methyl
esters. Conventional acid- or base-catalyzed
methanolysis involves extended reaction time
and recovery of the esters by extraction with
solvent. Early procedures employing boron tri-
fluoride (1) or boron trifluoride etherate as
catalysts included a solvent extraction step
and, if the catalysts were used in too high
concentrations, artifacts (polymers or unidenti-
fied derivatives) were formed that interfered
with subsequent analysis (2). The later meth-
od of Metcalfe et al. (3) using 12.5% ( w / v )
boron trifiuoride in methanol eliminated ex-
traction and was found suitable for volatile
fatty acids. We have found it to be suitable
not only for the usual oils but also for those
containing unusual structures such as cyclopro-
pene rings, vicinal hydroxyls, oxirane oxygen,
conjugated dienols and conjugated trienes.
A recent report of the Instrumental Tech-
niques Committee (4) indicates that the boron
trifluoride method is applicable to the common
oils, to dehydrated castor oil and to fats con-
taining isolated t r a n s unsaturation, but prohibits
its use with some of the structures mentioned
above. Results presented here indicate that,
if the methyl esters are intended for gas
~No. Utiliz. Res. Dev. Div., ARS, USDA.
118
chromatography, the scope of the method may
be extended.
M ET H,ODS
Ester preparations were catalyzed by hydro-
chloric acid, sodium methylate, or boron tri-
fluoride. In most preparations catalyzed by
mineral acid, oils were refluxed with 5% an-
hydrous hydrochloric acid or 1% sulfuric acid
in methanol for 3 hr, diluted with water, ex-
tracted with ether, washed and recovered by
removal of the ether under nitrogen on a
steam bath. With 0.5N methanolic sodium
methylate (5), samples were shaken 3V2 hr at
room temperature and recovered as above.
Boron trifluoride was used essentially as di-
rected by Metcalfe et al. (3) except that Bab-
cock milk test bottles were used for the reac-
tions. 'Catalyst concentration in the reagent was
5% or 12.5% and reaction times ranged from
2 to 10 min. Hydrochloric acid was used with
the Metcalfe technique at the same concen-
trations and reaction times. Ten millimeters of
boron trifluoride etherate (Eastman) was used
as catalyst for one sample of soybean oil
refluxed for 1 hr. After the heating period,
saturated sodium chloride solution was added
until the bottle was almost full and the mixture
was well shaken. Additional salt solution was
added to bring the liquid level into the neck
of the bottle. The bottle was centrifuged at
1800 rpm (International Centrifuge head no.
233, carrier no. 362) and the ester layer was
drawn off with a narrow tipped pipette.
I n one such preparation by this procedure
with anhydrous hydrochloric acid-methanol the
resulting salt precipitate, from the hydrochloric
acid-sodium hydroxide neutralization, was re-
covered by decantation and washed three times
with ether. The salt was redissolved in water
and the solution was extracted with ether. The
extract was washed with water and the solvent
was removed on a steam bath.
All esters were analyzed by GLC on a
Burrell K-5 instrument equipped with a thermal
conductivity detector (6).
When oil samples of less than 50 mg were
esterified, Babcock skim milk test bottles (Kim-
ble no. 530) with a capillary neck were nsed.
TLC was carried out on plates spread with
 ESTERS FROM UNUSUAL SEED OILS
a 275 /t layer of Silica Gel G impregnated
with boric acid. The plates were developed in
a 1 mm "sandwich" open chamber (7), with
hexane-ether (70:30) as the developing solvent.
The front was allowed to move 15 cm. The
spots were visualized by iodine vapor.
Infrared ( I R ) analyses were performed by
use of a Perkin-Elmer Infracord Model 137
with liquid films on sodium chloride disks.
Ultraviolet (UV) absorption was measured
with a Beckmann Model D K - 2 A spectropho-
tometer in 1 cm silica cells with 95% ethanol
as solvent.
Methoxyl groups were determined by the
Vieb/Sck and Schwappach method (8).
RESULTS AND DISCUSSION
Substitution of Babcock bottles for volu-
metric flasks as reaction vessels provides two
major advantages: they can be easily centri-
fuged, thereby causing the esters to separate
quickly from the saline solution; and in their
slender (or capillary) necks, the esters form
in longer columns more easily removable from
the aqueous layer.
Use of a capillary-necked flask has been
reported (9) but it must be specially made,
whereas the Babcock bottle is commercially
available.
Esters From Oils With Normal Unsaturation
G L C analyses of esters prepared by the
Metcalfe procedure and by the traditional
methods show little differences in relative
amounts of those acids found in common seed
oils. Analysis of Crambe oil, in which more
than 60% of the acids have chain lengths
greater than C1~, also gives almost identical
results for the two types of ester preparations.
The boron trifluoride-etherate reagent, used
with 5 rain or 1 hr heating time, produced
the previously reported artifact (2) (Table I).
The esters prepared from soybean oil, using
this catalyst for the longer time, were markedly
low in methyl linoleate and the methyl linole-
nate was completely removed (Table I). Such
loss was never observed when boron trifluoride
in methanol was used in the amounts and con-
centrations indicated.
TLC of the reaction products showed only
methyl esters except when 5% boron trifluoride
reagent was used. In this case, a component
was found which corresponded to free acid,
indicative of incomplete esterification, but the
esters formed were representative of the oil.
When hydrochloric acid was used as catalyst
in the Metcalfe procedure, a precipitate of
119
potassium chloride or sodium chloride was
formed and yields of ester were low because
of occlusion or absorption by the salt precipi-
tate. Esters recovered from the washed pre-
cipitate were enriched in polyunsaturates,
whereas those recovered by flotation were
comparable to esters prepared by the usual
methods. The error caused by this effect is
negligible for soybean oil but might be serious
with other oils.
Esters From Oils With Unusual Acids
UV analysis of the oil of Maytenus illici-
/olia indicated the presence of an aromatic
component (10) that, if calculated as benzoic
acid (Xm~x 227 m/x, e 10,960), constituted
7.9% of the oil. G L C analyses of esters pre-
pared by the hydrochloric acid transesterifica-
tion procedure with ether extraction failed to
show any benzoate. G L C of the esters made by
the Metcalfe procedure revealed 2.8% of a
component with retention characteristics iden-
tical to those of methyl benzoate. No attempt
was made to resolve the difference between
the UV and G L C results. Acetic acid is also
present in Maytenus oil (10), but the condi-
tions used for analysis precluded detection of
methyl acetate in either preparation.
G L C (Table I) and TLC analysis of the
ester preparations, catalyzed by boron trifluor-
ide, from castor and Lesquerella lescurii oils
showed no alteration of the ricinoleic, densi-
polic (12-hydroxy-cis-9-cis-15 octadecadieno-
ic), hydroxypalmitic, and hydroxystearic acids
contained in these oils. These results agree with
previous work (11) which showed that the
hydroxyl group of 2-hydroxymyristic acid was
not affected by boron trifluoride.
When boron trifluoride was used as a catalyst
for ester preparations of oils containing ver-
nolle acid (Vernonia anthelmintica and Eu-
phorbia lagascae), almost all of the vernolic
acid was converted to a derivative with equiva-
lent chain lengths (12) of 20.2 on Apiezon L
columns and 25.1 on LAC-2-R 446 columns
(Fig. 1). Only 2 - 3 % of the oil was con-
verted to methyl vernolate (Table I). The
mixed esters from the oils had strong IR ab-
sorption bands at 2.77 and 9.13tJ., the wave-
lengths associated with hydroxyl and ether
groups, respectively (13). Methoxyl analysis
of mixed esters from V. anthehnintica oil gave
13.9%-OCH 3. The methoxyl analysis supports
the assumption that the derivative is the methyl
methoxy-hydroxy-oleate. The positions of the
substituents are not indicated. The derivative
can be used for the quantitative analysis of
the esters because its weight per cent can be
LIPIDS, VOl.. 4, No. 2
120 R. KLEIMAN, G. F. SPENCER AND F. R. EARLE

TABLE I
C o m p a r i s o n of Esterification Procedures on Some Oils C o n t a i n i n g U n u s u a l F a t t y Acids

Normal components
U n o s u a l components
by G L C (area % )
Satu- M o n o - E q u i v a l e n t chain length
rated enoic Apiezon LAC-
Oil Catalyst esters esters 18:2 18:3 L 2-R 446 % Identity

Soybean oil HCI 17 26 52 5 -- --

BF3 17 26 52 5 -- -- __ m
BF3
etherate 22 31 24 ~ 18.1 -- 12 a
18.7 8
19.1 20.7 3
Maytenus illici]oliab BFa 20 42 34 0.1 8.6 11.8 3 Benzoate
Mart. 4.0 4.0 0.3 4:0
8.0 8.0 Trace 8:0
HC1 20 43 35 1 16.7 -- 0.4 Unknown
19.5 0.3 Unknown
Castor oil BF3 2 3 5 -- 19.6 24.3 90 18:1 O H
HC1 2 4 5 -- 19.6 24.3 89 18:1 O H
Lesquerella lescurii BF~ 12 38 2 15 17.7 23.0 1 16:0 OH
( G r a y ) S. Wats. 19.8 24.6 1 18:0 OH
19.6 24.8 4 18:1 OH
19.6 25.0 27 18:2 OH
H'.,SO4 12 35 3 14 17.7 23.0 2 16:0 OH
19.8 24.6 2 18:0 OH
19.6 24.8 4 18:1 OH
19.6 25.0 28 18:2 OH
Vernonia anthelmintica BF~ 6 4 14 0.4 19.2 23.0 3 Vernolate
(L.) Willd. 20.1 25.1 66 Vernolate
derivative
20.7 29.5 6 Dihydroxyoleate
NaOCH3 7 5 13 0.1 19.2 23.0 68 Vernolate
20.7 28.9 7 Dihydroxyoleate
Euphorbia lagascae BFs 7 23 9 0.3 19.2 23.0 2 Vernolate
Spreng. 20.1 25.1 58 Vernolate
derivative
NaOCH3 7 22 9 0.4 19.2 23.0 62 Vernolate
Tung oil e BFs 7 7 7 -- 19.3 22.2 79
19.6 22.6 Trace Conj. 18:3
HCI 7 10 7 -- 19.3-19.7 22.2 76
Conj. 18:3
22.6
Dimorphetheca BF3 5 13 14 0.2 19.1 21.8 68
sinuata DC.d Conj. 18:3
19.9 22.6
HCI 5 13 15 0.2 19.5 21.7 66
Conj. 18:3
19.9 22.6

a Probably the artifact reported by L o u g h (2).

bUW analysis: 7.9% as benzoate.
eUV analysis: 83% as a-eleostearic.
d U V analysis: 70% as dimorphecolic.

related to the weight per cent of the original
methyl vernolate. Under the conditions of
analysis used, the response of the derivative
seems equal to the response of methyl ver-
nolate. The V. anthelmintica oil used in this
study contained 7% dihydroxyoleic acid as well
as the vernolic acid. Since the G L C analysis
of the boron trifluoride-catalyzed esters gave
6% dihydroxyoleic ester, the dihydroxy func-
tional grou0s were not altered during esterifi-
cation (Table I).
IR analysis of the boron trifluoride-catalyzed
esters from Sterculia Joetida showed a strong
band at 9.92 /~. This band is characteristic of
LIPIDS, VOL. 4, NO. 2
the cyclopropene structure (14). The hydro-
gen bromide equivalent of the oil at 55 C
was 60% (as sterculic acid) and that of the
esters was 66%. While the titrations are not
in good agreement, these results indicate that
the cyclopropenoid functional group in the
acids of Sterculia oil was undisturbed during
the esterification reaction. Conventional acid
catalysts altered this functional group, pre-
sumably by a reaction analogous to the addi-
tion of hydrogen bromide to cyclopropenes
(15).
UV analysis of tung oil indicated 84% ccn-
jugated triene calculated as ~-eleostearic [?'r~a~
 ESTERS FROM UNUSUAL SEED OILS 121

Vernolate Ct8
NaOMe F.B_23 CI8 5%HCI
Dis
Derivative

I ?
Air

C16 I
-~ /~' ~CtA6i r C1~
~ DIG 12 ?. ~ ~?

I I I J I I I I I I
1416 18 19.2 1416 18 20 1416 18 20
EquivalentChainLength
FIG. 1. Gas-liquid chromatography of methyl esters of Euphorbia lagascae oil prepared with
different catalysts as shown. Column: 125 cm, 1% Apiezon L on Chromosorb G, isothermal
at 258C.

270 m~, e 47,000 (16)] acid. Esters prepared conjugated triene in the esters was presum-
by refluxing with the hydrochloric acid-meth- ably formed from dimorphecolic acid by heat-
anol reagent have maxima at 258, 268, and ing under acidic conditions as described by
278 m/~ while the esters prepared by the Smith et al. (20). In the G L C analysis, the
boron trifluoride method retained the original methoxy diene is converted to conjugated
chromophore [)~m~x 261, 270, and 282 m/~ trienes much as acetylated methyl dimorpheco-
(16)]. In addition, the chromatogram result- late is converted under the same conditions to
ing from the G L C of the hydrochloric acid conjugated trienes (21).
esters showed several components in the con-
jugated triene region while the boron trifluoride
esters showed basically one peak (Fig. 2).
Quantitation of these chromatograms agreed
well: 76% combined conjugated triene in the
hydrochloric acid esters and 79% conjugated
triene in the boron trifluoride esters. It appears
that the prolonged heating required with the
hydrochloric acid method, caused isomerization
while the boron trifluoride method caused little
or none.
The oil of Dirnorphotheca sinuata contained
70% dimorphecolic acid by UV analysis [Xm~x
231 m/~, e 33,600 (17)]. The boron trifluoride-
catalyzed ester preparation had UV absorp-
tion (X.... 258, 268, and 278 m/~ equivalent to
19% calculated as fi-eleostearic acid and a
broad band at 231 m~ (44% as methyl di-
morphecolate). The IR analyses of these
esters revealed a strong band at 10.1 /~, indica-
tive of conjugated trans unsaturation (18) and
showed no evidence of hydroxyl. A band at
9.18 t~ indicated the presence of an ether
linkage (13). It is asssumed that the absorp- 22.2 22.6
tion at 231 m/z is due to methoxy dienes re- Equivalent Chain Length
ported by Powell et al. (19) to be formed FIG. 2. Conjugated triene region from gas-liquid
from hydroxy dienols in the presence of acid. chromatography of Tung methyl esters. Column:
The methoxyl content of the esters was 14.1%, 200 cm, 15% LAC-2-R 446 on Chromosorb W,
which is in good agreement with the 14.5% isothermal at 197C; esterification catalyst as indi-
expected of the methoxy diene derivative. The cated.
LIPIDS, VOL. 4, No. 2
122 R . KLEIMAN, G . F . SPENCER AND F . R . EARLE
REFERENCES
1. Metcalfe, L. D., and A. A. Schmitz, Anal. Chem. 33,
363-364 (1961).
2. Lough, A. K., Biochem. J. 90, 4C (1964).
3. Metcalfe, L. D., A. A. Schmitz and R. J. Pelka, Anal.
Chem. 38, 514-515 (1966).
4. O'Connor, R. T., R. R. Allen, J. R. Chipault, S. S.
Herb and C. W. Hoerr, JAOCS 45, 103-106 (1968).
5. Miwa, T. K., F. R. Earle, G. C. Miwa and I. A.
Wolff, Ibid. 40, 225-229 (1963).
6. Kleiman, R., F. R. Earle, I. A. Wolff and Q. Jones,
Ibid. 41, 459-460 (1964).
7. Honegger, C. G., Helv. Chim. Acta 46, 1730-1734
(1963).
8. Clark, E. P., J. Ass. Offie. Agr. Chem. 15, 136-140
(1932).
9. Endres, ~I. G., JAOCS 43, 600 (1966).
10. Kleiman, R., R. W. Miller, F. R. Earle and I. A.
Wolff, Lipids 2, 473-478 (1967).
11. Morrison, W. R., and L. M. Smith, J. Lipid Res. 5,
600~608 (1964).
LIPIDS, VOL. 4, N o . 2
12. Miwa, T. K., JAOCS 40, 309-313 (1963).
13. Bellamy, L. J., in " T h e Infrared Spectra of Complex
Molecules," 2nd ed., Methuen and Company Ltd.,
London, 1958, p. 96, 116.
14. Bailey, A. V., F. C. Magne, G. J. Boudreaux and
E. L. Skau, JAOCS 40, 69-70 (1963).
15. Smith, C. R., Jr., M. C. Burnett, T. L. Wilson, R. L.
L o h m a r and I. A. Wolff, Ibid. 37, 320-323 (1960).
16. Crombie, L., and J. L. Tayler, J. Chem. Soc. 2816
(1954).
17. Binder, R. G., T. H. Applewhite, M. J. Diamond and
L. A. Goldblatt, JAOCS 41, 42--44 (1964).
18. O'Connor, R. T., Ibid. 33, 1-15 (1956).
19. Powell, R. G., C. R. Smith, Jr. and L A. Wolff, J.
Org. Chem. 32, 1442-1446 (1967).
20. Smith, C. R., Jr., T. L. Wilson, E. H. Melvin and
I. A. Wolff, J. Amer. Chem. Soc. 82, 1417-1421 (1960).
21. Morris, L. J., R. T. Holman and K. FonteI1, J. Lipid
Res. 1, 412-420 (1960).
[ R e c e i v e d J u l y 11, 1 9 6 8 ]