Monitoring Chemical Interesterification1

Linsen Liua,* and Dan Lampertb

a
Central Research and bOilseed Division, Cargill, Inc., Wayzata, Minnesota 55391

ABSTRACT: Chemical interesterification has long been used melting points of the components. Under such conditions, in-
to modify oils and fats into functional products. Many chemical teresterification will rearrange all of the fatty acids into a ran-
compounds can be used as the catalysts, such as sodium metal dom pattern on the glycerol backbone after thermodynamic
and sodium methoxide. With addition of the catalyst, the oil equilibrium is achieved. If interesterification is conducted at
turns a well documented distinctive reddish brown. Many be- low temperature so that some fats with high melting points
lieve that this color compound may be the real catalyst in inter-
are crystallized, the oils in the liquid phase will then reach a
esterification. The spectral changes of an oil undergoing inter-
esterification were examined in the visible and ultraviolet wave-
new thermodynamic equilibrium. The crystallized fats will
length ranges. An absorbance peak was found from 320–450 not be involved in interesterification because they are sepa-
nm. Because this range is the wavelength of blue light, this ab- rated from the reaction phase. This practice is known as di-
sorption gives the oil a reddish brown color, the compensated rect interesterification. Random interesterification is most
color of blue light. The peak increases as the amount of sodium commonly used in food industry to prepare functional plastic
methoxide increases. To initiate and complete interesterifica- fats by interesterifying liquid oils such as soybean or canola
tion, the oil absorbance must reach levels of 0.4 and 1.0 at 374 oil with hard fats such as fully hydrogenated soybean oil or
nm, respectively. Controlled partial interesterification is now palm stearine.
also possible by controlling the absorbance between 0.4 and Chemical interesterification can be induced by chemical
1.0. A novel patent-pending technology, based on this observa- catalysts including alkali metals, alcoholates, and sodium hy-
tion, was developed to monitor the progress of interesterifica-
droxide with glycerol (3). One of the most commonly used
tion on-line using fiber optic technology. The reaction time and
the dosage of sodium methoxide used for randomization have
catalysts is sodium methoxide. These catalysts have been
been significantly reduced on bench-scale when using this new shown actually to be precatalysts that react with triglycerides
tool. Furthermore, the oils interesterified with monitoring have to produce the real catalyst intermediate. With the addition of
higher oxidative stability and tocopherol contents compared to these precatalysts, a reddish-brown color appears indicating
those by conventional randomization. the successful initiation of interesterification. Two types of
Paper no. J9025 in JAOCS 76, 783–787 (July 1999). catalytic reaction mechanisms have been suggested to explain
chemical interesterification. Weiss et al. (1) suggested that the
KEY WORDS: Color, fiber optic, interesterification, mechanism, reaction starts when sodium methoxide attacks the α-hydro-
monitor, partial interesterification, plastic fats, randomization, gen of an acyl group to form the enolate ion. The enolate ion
sodium methoxide. then reacts with another ester to form a β-keto ester. This is
also referred to as the Claisen mechanism.
Another hypothesis (4) suggests that the methoxide ion at-
Chemical interesterification is an important process for the tacks the carbonyl group directly and forms a diglyderide
oils and fats industry. This reaction modifies the physical anion that functions as the real catalyst and transfers acyl
properties of oils by rearranging the distribution of fatty acids groups around the glyeride backbones. Both mechanisms re-
on the glycerol backbone without changing their chemical sult in the formation of diglyceride ions. The difference be-
composition. This method was widely used in the United tween these two mechanisms is the donor of an acyl group for
States in the 1950s and 1960s for lard crystal modification transesterification of the diglyceride ion. The first mechanism
(1,2). With the rising concerns over the nutritional impact of suggests that both β-keto and acyl esters are the donors for
trans fatty acids on health, interesterification may become acyl transesterification. The second one proposes that acyl es-
more popular for preparation of low or zero trans functional ters are the donor. Nonetheless, neither mechanism explains
plastic fats in the future. why the oils change color to reddish brown with addition of
There are two basic types of chemical interesterification: precatalysts during interesterification (3,5).
random and direct (3). In random interesterification, the reac- Interesterification is normally initiated at 70–120°C by
tion is initiated and completed at temperatures above the 0.05–0.5% sodium methoxide after the oil has been com-
*To whom correspondence should be addressed at 2301 Crosby Rd., pletely dried. The oil quality and the efficiency of the drying
Wayzata, MN 55391.
E-mail: [email protected] system used play critical roles in determining the dosage of
1
Presented at the 89th American Oil Chemists’ Society Annual Meeting, sodium methoxide. This catalyst is very reactive to moisture,
Chicago, IL, May 10–13, 1998. free fatty acids, and peroxide substances in the oil. Typically,

Copyright © 1999 by AOCS Press 783 JAOCS, Vol. 76, no. 7 (1999)
784 L. LIU AND D. LAMPERT

TABLE 1
The Fatty Acid Compositions of Experimental Oils
Fatty acids High-oleic canola oil Palm oil Soybean oil Soy stearine Tallow
16:0 3.8 44.0 10.3 10.6 23.0
18:0 2.7 4.4 4.1 86.4 15.8
18:1 76.6 39.3 23.1 1.0 44.6
18:2 11.0 9.7 53.1 0.3 3.7
18:3 2.7 0.3 7.8 0.0 0.3
Others 3.2 2.4 1.6 1.8 3.5
Total trans 1.1 0.5 0.3 0.5 9.4

interesterification is conducted with high enough doses of cat- stainless steel pans and then measured with a Perkin-Elmer
alyst and sufficient time to ensure complete reaction. Various DSC-7 (Norwalk, CT 06859). The programs of heating and
postreaction tests including solid fat index, melting point, and cooling are listed in Table 2. The melting curves of these
sterol profile are often used to verify the completion of reac- products from −30 to 70°C were used to calculate the SFC.
tion. These off-line tests are both time- and labor-consuming. High-performance liquid chromatography (LC). Triglyc-
An on-line method would eliminate the need for such testing eride profiles were analyzed under these conditions:
and certainly increase the productivity and reduce the cost of Spherisorb C18 column, 15 cm × 4.6 mm, S3 OD2; Waters Al-
interesterification. liance 2690 pump; ELSD IIA detector (Varex, MD). The mo-
This paper describes a novel patent-pending method that bile phase was a mixture of dichloromethane and acetonitrile
is based on a quantitative measurement of oil spectral change at 0.7 mL/min. The initial ratio of dichloromethane and ace-
after the addition of sodium methoxide during interesterifica- tonitrile was 20:80 (vol/vol); it was changed to 30:70, 60:40,
tion. This technology can monitor interesterification on-line and 20:80 at 1, 30, and 31 min, respectively, after injection of
with a fiber optic spectrometer, track the progress of inter- samples. The peaks were identified by comparison with
esterification reaction, and minimize the dosage of sodium triglyceride standards and oils with known triglyceride com-
methoxide and reaction time. Partial interesterification can positions. The area percentage was used to quantitate the com-
also be carried out using this on-line monitoring technology. position of triglycerides. The column temperature was 40°C.
Partially interesterified products have unique functional prop-
erties (melting point and solid fat index/content) compared to
the physical and randomized blends. Some of these products TABLE 2
can be used directly as functional plastic fat ingredients. Differential Scanning Calorimetry Programming Method
Temperature (°C) Time (min) Rate (°C/min)
30.0 0.0 50.0
EXPERIMENTAL PROCEDURES 75.0 1.0 20.0
Sodium methoxide. Sodium methoxide (25 wt%) in methanol −30.0 10.0 10.0
70.0 — —
was purchased from Agros Organics (Pittsburgh, PA). Solid
sodium methoxide was freshly prepared from this solution by
evaporating methanol and grinding the solids to a fine pow-
der before adding them to the oil.
Oils and fats. High-oleic canola oil, palm oil, soybean oil,
and fully hydrogenated soybean oil (soy stearine, iodine value
<5) were obtained from Cargill Oilseeds Division. Tallow was
a product of Cargill Excel Division. Table 1 shows the fatty
acid compositions of these oils used for interesterification.
Interesterification. Specific blends of oils were mixed and
dried at 110°C under 10 mm Hg vacuum for 60 min, then up
to 0.5% (w/w) of sodium methoxide fine powder was added to
the dried oil blends to initiate interesterification. Samples were
taken at various intervals and their spectra were recorded. The
sampled oils were mixed in a 1:10 ratio with hexane contain-
ing 1% acetic acid. The solutions were then filtered through
Whatman #4 filter paper (Whatman International Ltd., Maid-
stone, England). The solvents in the filtrates were completely
evaporated on a steam bath with a nitrogen stream.
Differential scanning calorimetry (DSC) and solid fat con- FIG. 1. All the oils tested—canola, soybean, palm, and tallow—gave the
tent (SFC). Ten milligrams of oil samples were loaded on similar unique absorbance peak with the addition of sodium methoxide.

JAOCS, Vol. 76, no. 7 (1999)

 MONITORING CHEMICAL INTERESTERIFICATION 785

FIG. 4. The melting curve of the mixture of canola and soy stearin
(50:50, w/w) interesterified at an absorbance of 0.8 stayed the same
from 15 to 120 min at 120°C. (A) 15 min after the absorbance at 374
FIG. 2. The absorbance of oil mixture (canola oil/soy stearin, 70:30, nm reached 0.8; (B) 60 min after the absorbance at 374 nm reached
w/w) increases with the addition of sodium methoxide (0.05–0.3%). 0.8; (C) 120 min after the absorbance at 374 nm reached 0.8.

ful interesterification. This phenomenon has been known for

a long time, but no spectrum has been reported. We found that
absorbance by oil of visible and ultraviolet light increases
after addition of sodium methoxide. One peak in particular
appeared between 300 and 450 nm (the border between the
ultraviolet and visible ranges) after the addition of a certain
amount of sodium methoxide. This range represents the
wavelength of blue color and therefore gives the oil a reddish
brown color. Figure 1 shows the spectral peaks observed with
canola-, palm-, soybean-, and tallow-based interesterifica-
tions. All the oils tested had the same unique peak regardless
of their origin, fatty acid composition, unsaponifiable matter,
and initial color. Therefore, the color substance is very likely
involved in the mechanism of interesterification although its
FIG. 3. The melting curve of the mixture of canola and soy stearin identity is still unknown.
(70:30, w/w) started to change when the absorbance of 374 nm reached
The absorbance of oil increases as more sodium methox-
0.4. (A) The physical blend without the addition of sodium methoxide;
(B) the blend with addition of sodium methoxide and the absorbance of ide is added, as shown in Figure 2. The peak increases with
374 nm is 0.4. the addition of more sodium methoxide. Figure 3 shows that
the DSC melting curve of tested blends starts to change as the
absorbance at 374 nm reaches 0.4. Below this critical ab-
RESULTS AND DISCUSSION
sorbance, the melting curves did not change even when the
During interesterification, the oil changes color to a reddish reaction was run overnight. Interesterification equilibrated
brown after a chemical catalyst such as sodium methoxide is (randomized) when this absorbance reached 1.0 as shown by
added. Color change is often used as an indicator for success- their SFC and triacylglycerol composition analysis (Table 3).

TABLE 3
The Solid Fat Content and Triacylglycerol Composition of Canola/Soy Stearine (70:30, w/w) Before and After Interesterification
Solid fat content (%)
at temperature (°C) Triacylglycerol composition
Absorbance NaOMe (%) 10.0 21.1 26.7 33.3 40.0 LOO LOP OOO LOS OOP OOS OPS PPS OSS PSS SSS
a
0.0 0.0 42 38 38 37 35 11.6 1.1 45.0 — 3.7 1.6 — — — 6.8 24.1
1.0 0.3 47 25 22 15 9 5.3 2.4 17.8 6.9 6.0 33.3 4.5 0.7 15.0 1.1 2.0
1.1 0.4 47 25 22 15 9 6.2 2.7 19.9 7.1 5.2 31.7 4.6 0.8 13.8 1.1 2.0
1.3 0.5 47 25 22 15 9 5.7 3.0 16.6 7.8 5.0 32.9 4.9 0.9 13.9 1.0 1.4
Theoretical random calculation 5.5 1.0 21.8 5.3 5.8 31.4 5.6 0.2 15.0 1.3 2.4
a
—, not detected; L, linoleic acid; O, oleic acid; P, palmitic acid; S, stearic acid.

JAOCS, Vol. 76, no. 7 (1999)

786 L. LIU AND D. LAMPERT

TABLE 4
The Solid Fat Content and Triacylglycerol Composition of Interesterified Canola/Soy Stearinea (70:30, w/w)
Solid fat content (%)
at temperature (°C) Triacylglycerol composition
# Absorbance 10.0 21.1 26.7 33.3 40.0 LOO LOP OOO LOS OOP OOS OPS PPS OSS PSS SSS
a
1 0.00 42 38 38 37 35 11.6 1.1 45.0 — 3.7 1.6 — — — 6.8 24.1
2 0.36 42 38 38 37 35 11.6 1.3 46.5 — 4.1 0.9 — — 3.2 5.9 20.7
3 0.43 42 39 39 37 34 11.30 1.2 47.8 — 4.1 0.8 — — 3.7 5.9 18.8
4 0.53 42 37 37 34 30 11.6 1.4 44.6 — 4.6 5.3 0.6 — 1.6 5.8 18.2
5 0.68 43 36 36 31 26 10.1 1.5 44.6 — 4.3 11.6 1.5 — 5.9 4.5 12.9
6 0.77 43 35 35 29 24 9.91 2.0 37.6 — 5.4 14.7 2.2 — 7.3 3.7 10.9
7 1.0 47 25 22 15 9 5.3 2.4 17.8 6.9 6.0 33.3 4.5 0.7 15.0 1.1 2.0
a
For abbreviations see Table 3.

The same results are observed for the other blends of all the teresterification, and these are different from both the physi-
experimental oils listed in Table 1 (data not shown). cal mixture (Sample 1) and completely randomized product
Interesterification was only partly achieved when the ab- (Sample 7). By partial interesterification, the oil’s physical
sorbance was between 0.4–1.0, and the degree of this partial in- properties such as melting point, SFC, and crystal form can
teresterification depends on the absorbance. Figure 4 shows that be modified, providing an additional dimension to formula-
the melting curves of partially interesterified test blends did not tion options in product development. This method can give
change up to 180 min at 120°C after the peak absorbance excellent repeatability as shown in Table 5.
reached 0.8 with a reduced amount of sodium methoxide. Simi- This technology has also been applied to optimize the re-
lar results were found when partial interesterification was con- action conditions of randomization. Randomization has been
ducted at various temperatures (data not shown). These results widely used in industry to prepare functional products such
suggest that interesterification needs a minimum dosage of cat- as margarine oils and specialty shortenings. Because of the
alyst to initiate and complete (randomization) and the color lack of on-line monitoring capability, randomization has been
compound is very likely the active catalyst. conducted with excess dosages of catalyst and varying reac-
A series of partial interesterifications was conducted by tion times. When processing conditions change, these condi-
controlling absorbance. As shown in Table 4, partially inter- tions need to be readjusted. Therefore, excess time and/or cat-
esterified oils (Samples 2–6) have solid fat contents and alyst are used to ensure complete randomization. This on-line
triglyceride compositions that depend on their degree of in- monitoring system provides a novel way to monitor inter-
esterification and significantly reduce reaction time and/or
TABLE 5 catalyst dosage, as demonstrated in our laboratory. With re-
The Solid Fat Contents of Canola/Soy Stearine Mixtures (50:50, w/w)
duced catalyst and reaction time, the soap, the main byprod-
Interesterified at the Specific Absorbances
uct in interesterification, will be reduced so the yield of prod-
Solid fat content (%)
uct will be increased. Therefore, the cost of interesterification
at temperature (°C)
is significantly reduced and the capacity of interesterification
Absorbance 21.1 33.3 40.0
can be increased without the expansion of processing equip-
0.57 44 44 36 ment. Furthermore, the time for randomization is reduced to
0.57 46 46 38
as short as 5 min by on-line monitoring technology, which is
0.57 45 45 39
0.78 37 36 28 significantly shorter than the 15–30 min commonly used in
0.79 37 36 27 conventional interesterification. With significantly reduced
0.78 39 39 30 reaction time, on-line monitored interesterification produced

TABLE 6
The Total Tocopherol Contents and Oxidative Stability of Interesterified Oils
Reaction conditions RBD interesterified oilsb
Time Interesterification Degree of Tocopherol
Samples NaOMe (%) (min) method interesterificationa AOM (h) (ppm)
Canola oil/soy stearin (50:50) 0.15 15 Conventionalc Randomized 27 100
0.15 5 On-line monitoringd Randomized 35 220
Soybean oil/soy stearin (50:50) 0.15 15 Conventional Randomized 21 260
0.20 5 On-line monitoring Randomized 26 490
a
Determined by differential scanning calorimetry melting curves.
b
Refined, bleached, and deodorized (RBD).
c
Conventional: the reaction was conducted at 95–100°C for 15 min after addition of sodium methoxide.
d
On-line monitoring: the reaction was stopped once the absorbance at 374 nm reached 1.0; the reaction time was reported.

JAOCS, Vol. 76, no. 7 (1999)

 MONITORING CHEMICAL INTERESTERIFICATION 787

the products with higher oxidative stability and tocopherol 2. Wiedermann, L., T.J. Weiss, G.A. Jacobson, and K.F. Mattil, A
content. Table 6 lists some of the results for the randomized Comparison of Sodium Methoxide-Treated Lard, Ibid.,
38:389–395 (1961).
products of canola and soybean oils with fully hydrogenated
3. Sreenivasan, B., Interesterification of Fats, Ibid. 55:796–805
soybean oil. These data show that the total tocopherol con- (1978)
tents are significantly higher (1.9–2.2 times) in on-line moni- 4. Coenen, J.W.E., Fractionation and Interesterification of Fats:
tored interesterified products than that in the conventional in- I—Interesterification, Rev. Fr. Corps Gras 21:403–413 (1974).
teresterified ones. As a result, the former products have 5–8 h 5. Ramamurthi, S., and A.A. McCurdy, Interesterification—Cur-
rent Status and Future Prospects, in Development and Process-
longer active oxygen method than the later ones.
ing of Vegetable Oils for Human Nutrition, AOCS Press, Cham-
paign, 1996, pp. 62–86.
REFERENCES
1. Weiss, T.J., G.A. Jacobson, and L.H. Wiedermann, Reaction
Mechanics of Sodium Methoxide Treatment of Lard, J. Am. Oil
Chem. Soc. 38:396–399 (1961). [Received September 21, 1998; accepted March 6, 1999]

JAOCS, Vol. 76, no. 7 (1999)