Effect of Heating and Homogenization on the Stability of Coconut Milk Emulsions

Effect of Heating and Homogenization
on the Stability of Coconut Milk Emulsions
NATTAPOL TANGSUPHOOM AND JOHN N. COUPLAND
ABSTRACT: The effects of homogenization and heat treatment on the colloidal stability of coconut milk were studied.
Fresh coconut milk (15% to 17% fat, 1.5% to 2% protein) was extracted and stored at 30 °C before homogenization at
40/4 MPa (stage I/stage II). Both homogenized and non-homogenized samples were heated at 50 °C, 60 °C, 70 °C, 80 °C,
and 90 °C for 1 h. Homogenization reduced the size of the primary emulsion droplets from 10.9 to 3.0 m, but
increased the degree of flocculation, presumably via a bridging mechanism. This flocculation was also responsible for
increased viscosity of the homogenized samples. Heating increased the degree of flocculation in both non-homog-
enized and homogenized samples. A slight amount of coalescence was also observed after heating above 80 °C. All
samples creamed after 24 h of storage, but the heated samples formed a larger cream layer, presumably because the
flocculated droplets packed together less efficiently. Optical microscopy was used to confirm the combination of
flocculation and creaming responsible for changes in coconut milk quality. The information obtained from this study
provides a better understanding of the emulsion science important in controlling coconut milk functionality.
Keywords: coconut milk, emulsion, emulsion stability, heating, homogenization
E: Food Engineering & Physical Properties
Introduction
The physical properties of coconut emulsions have not been well
Coconut milk is the white opaque liquid obtained from shredded studied. It is common practical knowledge that heating and homog-
coconut (Cocos nucifera L.) meat made by comminuting or grat-
enization affect the stability of coconut milk emulsions. However,
ing the flesh of the nut (with or without the addition of water) and
the mechanisms of such stability alterations and the underlying
pressing or dewatering the comminuted pulp. It is an important in-
emulsion science are still unclear. In this work we determine the
gredient for Asian cuisine as well as in other parts of the world. The
effects and mechanisms of homogenization and heat treatment on
composition of coconut milk varies according to variety, age, grow-
the colloidal stability of freshly-manufactured coconut milk.
ing environment of the coconut, cultural practices, method of prep-
aration, and the process conditions used in extraction, for example,
Materials and Methods
the amount of added water and the temperature used for extraction
(Cancel 1979; Gonzalez 1990). Typical compositions of the coconut
Sample preparation
milk directly expelled from coconut kernel (without added water)
Coconuts were purchased from a local retailer, deshelled, and
are protein, 2.6% to 4.4%; water, 50% to 54%; lipids, 32% to 40%; and
shredded using a traditional coconut grater. Coconut milk was pro-
ash, 1% to 1.5% (Seow and Gwee 1997).
duced by mixing the shredded pulp with an equal weight of warm
Coconut milk is essentially an oil-in-water emulsion, stabilized by
distilled water (60 °C) in a Waring blender (Waring, 1120, Winstel,
the naturally occurring proteins (globulins and albumins) and phos-
Conn., U.S.A.), filtered through a double-layered cheese cloth, and
pholipids (for example, lecithin and cephalin) (Birosel and others
manually squeezed with a twisting motion to extract most of the
1963). As with all emulsions, coconut milk is not physically stable and
milk. Thimerosal (0.01 wt%; Sigma Chemical, St. Louis, Mo., U.S.A.)
is prone to phase separation. Natural coconut milk will separate into
was added as an antimicrobial agent. The extracted emulsion was
a cream and serum layer within 5 to 10 h of manufacture.
stored at 30 °C before analysis and used within 24 h of manufac-
Thermal processing is an effective means of extending the shelf
ture.
life of coconut milk. The processing of canned coconut milk starts
The crude protein content of coconut milk was measured using
with the extraction of coconut milk, which is then heated to a tem-
the nitrogen combustion method by an automatic nitrogen analyz-
perature of about 92 °C to 95 °C for 5 to 20 min (a process often re-
er (Leco, FP-528, St. Joseph, Mich., U.S.A.). The fat content was
ferred to in the coconut industry as pasteurization) and often
determined using solvent extraction with petroleum ether (Sigma
mixed with emulsifiers and/or stabilizers before a homogenization
Chemical) at a sample-solvent ratio of 1:3 in a Majonnier flask
process. The homogenized milk is either hot-filled in cans or passed
(FMC, Chicago, Ill., U.S.A.) and weighing of the fat after the solvent
through an exhaust box before can sealing. Because the pH of co-
was evaporated in a Soxtec extraction unit (Foss Tecator, Eden Prai-
conut milk is about 6, it is considered as a low-acid food and the
rie, Minn., U.S.A.).
cans must be retorted (Timmins and Kramer 1977; Arumugan and
Homogenized samples were prepared by recirculating fresh coco-
others 1993).
nut milk through a twin-stage valve homogenizer (GEA Niro Soavi,
Panda, Hudson, Wis., U.S.A.) at a pressure of 40/4 MPa for 4 min to
achieve multiple passes through the valves. Both non-homogenized
MS 20050246 Submitted 4/26/05, Revised 5/10/05, Accepted 7/11/05. The
authors are with Dept. of Food Science, The Pennsylvania State Univ., 126
and homogenized samples were heated in a temperature-controlled
Borland Laboratory, Univ. Park, PA 16802. Direct inquiries to author
water bath set at 50 °C, 60 °C, 70 °C, 80 °C, and 90 °C for 1 h and then
Coupland (E-mail: coupland@psu.edu).
cooled to 30 °C in another water bath before analysis.

---

Stability of coconut milk emulsions . . .
Particle size analysis
Melville, N.Y., U.S.A.) equipped with a color video camera (DXC-
The weight-average diameters (d
970MD, Sony, New York, N.Y., U.S.A.).
43) of the coconut milk emulsion
droplets and their size distribution (volume fraction as a function
of particle size) were measured using a laser diffraction particle
Rheological studies
analyzer (Horiba, LA-920, Irvine, Calif., U.S.A.) assuming a relative
Rheological measurements were carried out using a controlled
refractive index of 1.15. Coconut milk samples were diluted to ap-
strain rheometer (TA Instruments, Ares, New Castle, Del., U.S.A.)
proximately 0.001% fat before analysis to minimize multiple scat-
operating with a cone and plate geometry (cone dia, 50 mm; cone
tering effects. In some experiments, the coconut milk was diluted
angle, 0.04 radian). Samples were equilibrated at 30 °C, gently
in 1 wt% sodium dodecyl sulfate (SDS, Sigma Chemical) solution
mixed, and then portions (about 1.5 mL) were transferred to the
rather than water. SDS is an anionic surfactant that effectively dis-
instrument. The instrument had previously been equilibrated at
places protein from the surface of emulsion droplets and disrupts
30 °C and the test was run immediately. The shear rate was in-
droplet flocs formed due to interdroplet protein-protein interac-
creased from 0/s to 100/s over 7 min and the required stress used
tions. Particle size measured in water is referred to in this work as an
to calculate the apparent viscosity.
“effective” particle size and includes the presence of flocs, where-
as measurement in SDS solution is called the “primary” particle
Creaming stability measurements
size. Emulsion coalescence will be seen as a change in both primary
Portions (10 g) of coconut milk samples were transferred into
and effective particle diameter, whereas flocculation will increase
glass tubes, covered, and allowed to stand for 24 h at room temper-
the effective diameter but the primary diameter will remain un-
ature. All samples separated into the opaque layer at the top and
changed.
the transparent aqueous phase at the bottom during storage. The
extent of the phase separation was assessed by creaming index,
Microscopy
which is the percentage ratio between the height of the transpar-
Samples of coconut milk (about 25 L) were placed on a micro-
ent layer (HT) and total height of the emulsion (HE) in the test tube.
scope slide, gently covered with a cover slip, and observed at 200×
magnification using an optical microscope (BX40,Olympus,
Determination of free fat
The degree of emulsion destabilization was measured as the
amount of solvent-extractable oil. Samples of coconut milk (10 g)
were transferred to Majonnier flasks and extracted with petroleum
ether (Sigma Chemical) at the volume ratio of 3:1. The organic ex-
tracts were evaporated to dryness in a Soxtec extraction unit (Foss
Tecator) and the extractable oil weighed. The extraction was repeated
5 times, and the cumulative value of extracted fat was calculated.
Statistical analyses
Most experiments were conducted in triplicate with freshly pre-
pared coconut milk used on each occasion. Data were analyzed
using SPSS for Windows, release 11.5.0 (SPSS, Chicago, Ill., U.S.A.).
One-way analysis of variance (ANOVA) and Duncan’s multiple
range tests were used to evaluate the significance of differences
E: Food Engineering & Physical Properties
(P < 0.05) between the samples. Only significantly different
(P < 0.05) results are discussed in the text. Data are presented as the
mean and standard deviation.
Results and Discussion
The coconut milk prepared by the described method contained
15% to 17% fat and 1.5% to 2% protein. The droplet size distri-
bution in the fresh milk had an approximately log-normal form (Fig-
ure 1a). The effective droplet diameter (d43) of the fresh coconut
milks was 13.1 m with a standard deviation of the distributions of
approximately 2 m. Surprisingly, the homogenized milk had only
slightly different effective particle size from the non-homogenized
milk (Figure 1b). However, when the milk was dispersed in SDS
solution rather than distilled water before laser diffraction particle
sizing, the homogenized samples were much smaller than the non-
homogenized samples. SDS displaces the protein from the oil-water
interface, thus disrupting any flocculation caused by interdroplet
protein-protein interactions and allows the instrument to measure
the primary particle size rather than the apparent size of flocs
present. The non-homogenized milk particle size was not markedly
Figure 1—Representative droplet size distribution of non-
affected by dilution in SDS, suggesting the particles present were
homogenized (a) and homogenized (b) coconut milks dis-
less significantly flocculated.
persed in distilled water (———= effective particle size dis-
This suggests that homogenization significantly reduces the
tribution including the presence of flocs) or sodium dode-
mean diameter of the droplets, but the fine droplets formed quickly
cyl sulfate (SDS) (— = primary particle size distribution) be-
fore analysis.
flocculate to approximately the same effective size as was present

---

Stability of coconut milk emulsions . . .
before homogenization. The non-homogenized milk has large
to 1 droplet ends up simultaneously adsorbing to the surface of
droplets but these are largely non-flocculated. Optical micrographs
2 droplets, leading to bridging flocculation (McClements 1999).
of the homogenized and non-homogenized samples reveal more
In other experimental work, the effect of homogenization pres-
large droplets in the former and more flocculation in the latter (Fig-
sure on the effective and primary particle diameter of coconut milk
ure 2). A likely explanation for this is the amount of protein capable
was measured (Figure 3). Samples of freshly extracted coconut
of stabilizing the emulsion is limited in coconut milk (del Rosario
milks were homogenized at 20/2, 40/4, and 60/6 MPa (stage I pres-
and Punzalan 1977). As the particle size is reduced, the interfacial
sure/stage II pressure) for 1 to 5 passes through the homogenizer.
area increases and a single protein molecules originally adsorbed
Homogenization reduced both the effective and primary particle
sizes of coconut milk by about 50% to 75%. Increasing the homog-
enization pressure marginally decreased the effective particle size,
and after the 3rd pass through the homogenizer, subsequent pass-
es had no further effect on particle size. This supports our hypoth-
esis that the amount and quality of protein limits the effectiveness
of homogenization on coconut milk.
Figure 4 shows the mean droplet diameter of both non-homog-
enized and homogenized coconut milks heated at different tem-
peratures. The effective particle size of homogenized coconut milk
increased dramatically (from about 10 m to more than 22.7 m)
after heating at above 70 °C for 1 h, whereas the primary particle
size increased only after heating at 90 °C. For non-homogenized
samples, effective particle size changed from 12.2 to 30.5 m when
the heating temperature increased. Significant changes in the pri-
mary particle size are also detected at higher heating temperatures.
Both the effective and primary particle size increased in non-ho-
E: Food Engineering & Physical Properties
mogenized and homogenized coconut milks heated at tempera-
tures above 70 °C, suggesting that both flocculation and possibly a
slight degree of coalescence occurred in heated coconut milk emul-
sions. This is supported by observations of the microstructure of
the heated emulsion samples (Figure 2). Just as solutions of glob-
ular proteins sometimes gel if thermally denatured, protein-stabi-
lized emulsions have been shown to flocculate after heating as the
protein-protein associations formed bind the droplets together in
a network (Sliwinski and others 2003). Coconut proteins have been
shown to denature and coagulate at 80 °C and higher (Gonzalez
1990; Kwon and others 1996); it seems likely that the denaturation
and aggregation of surface-bound proteins is responsible for the
thermally-induced flocculation seen here. Coalescence (seen as a
Figure 2—Micrographs taken of non-homogenized (a, c,
change in primary particle size) may then result due to the break-
e) and homogenized (at 40/4 MPa) (b, d, f) coconut milks
down of the lamella separating the flocculated droplets as the ef-
either unheated (A, B) or heated to heated to 50 °C (C, D)
fectiveness of the proteins as stabilizing agents is reduced.
or 90 °C (E, F). Scale bar is 50 m.
Figure 3—Effect of homogenization on the mean particle
Figure 4—Effect of heating temperature on the mean par-
diameter of coconut milks homogenized at (
,
) 20/2,
ticle diameter of ( , ) non-homogenized and ( , ) homog-
(
,
) 40/4, and (
,
) 60/6 MPa. Filled points represent
enized (at 40/4 MPa) coconut milks. Filled points repre-
emulsions dispersed in water; open points represent emul-
sent emulsions dispersed in water; open points represent
sions dispersed in sodium dodecyl sulfate (SDS).
emulsion dispersed in sodium dodecyl sulfate (SDS).

---

Stability of coconut milk emulsions . . .
Rheological measurements showed that both the non-homoge-
the shear rate, K is the consistency index, and n is the flow behavior
nized and homogenized coconut milk samples were shear-thinning
index the shear stress. Power-law equations are frequently used to
fluids whose apparent viscosity decreased with increasing shear rate.
describe emulsion rheology, and the K and n parameters are often
Similar flow behavior was reported in previous studies (Vitali and
used to describe the inherent structure of whatever weak network is
others 1986; Simuang and others 2004). The flow curves were mod-
present and how readily it is disrupted by shear, respectively (Mc-
eled using a power law equation ( = K. n, where is the shear stress,
Clements 1999). In all cases, the power-law equation described the
data well (r2 > 0.98), and values of K and n are reported in Figure 5.
The homogenized samples were more viscous than the non-
homogenized coconut milks, consistent with the presence of floc-
culated droplets. Emulsion flocculation leads to a higher effective
particle volume fraction and thus higher viscosity (McClements
1999). The consistency index increased with temperature for both
homogenized and unhomogenized samples probably due to addi-
tional thermally induced flocculation. The flow behavior index
decreased with temperature as the structures formed could be
readily disrupted by applied shear.
The creaming indices for the non-homogenized and homoge-
nized coconut milk samples are shown as a function of thermal his-
tory in Figure 6. All samples creamed after 24 h storage, but the
heated samples separated less as indicated by the lower serum
heights and, thus, the lower creaming indices. Non-homogenized
coconut milk is more prone to creaming than homogenized coconut
milk because of its larger globule size (Monera and del Rosario
1982). The large but flocculated particles present in homogenized
milk cream more slowly because the density contrast of a floc is
smaller than that of a droplet and secondly because large flocs can
be extensively interconnected and trap the droplets in a network
(Parker and others 1995). The large network necessary to inhibit
creaming may not be seen in light scattering measurements from
a diluted emulsion as the process of sample preparation disrupts
the fragile structures. Better creaming stability was found in both
non-homogenized and homogenized coconut milks after heating
at temperatures above 80 °C for 1 h, probably due to the higher
viscosity in these samples slowing the creaming rate (McClements
1999) or the differences in the structure of the creamed layer (Cha-
namai and McClements 2000).
Fink and Kessler (1985) argued that oil extraction by organic sol-
vent (in native milk fat globules) is a measure of the ability of the
E: Food Engineering & Physical Properties
interfacial layer to protect the fat against the destabilization pro-
cess. They also argued that degradation of this interfacial layer
Figure 5—Effect of heating temperature on the power-law
could be inferred from changes in the amount of extractable oil. In
flow behavior index, n (a) and consistency index, K (b) of
( ) non-homogenized and ( ) homogenized (at 40/4 MPa)
our experiments, more than 80% of oil in fresh coconut milk could
coconut milks
Figure 6—Effect of heating temperature on the creaming
Figure 7—Effect of heating temperature on the free oil
index of ( ) non-homogenized and ( ) homogenized (at 40/
solvent-extracted from ( ) non-homogenized and (
) ho-
4 MPa) coconut milks
mogenized (at 40/4 MPa) coconut milks

---

Stability of coconut milk emulsions . . .
be extracted by petroleum ether, whereas only about 10% of the oil
ples increased because of the differences in accessibility of the ex-
from similar homogenized milks could be extracted (Figure 7). Ho-
tracting solvent to the fat globules. The information obtained from
mogenization reduces the primary particle size of the coconut milk
the study provides a better understanding of the changes in stabil-
and leads to extensive flocculation (Figures 1b, 2, and 4) and it
ity of coconut milk emulsion during processing important in control-
seems likely that the droplets in the core of the flocs are more pro-
ling coconut milk functionality.
tected from the extracting solvent.
Thermal processing progressively decreased the amount of sol-
Acknowledgments
vent-extractable oil in the non-homogenized samples while in-
This work was supported by the Royal Thai Government scholar-
creasing it in the homogenized samples. The markedly changes
ship under the Committee Staff Development Project of the Com-
were found at the heating temperature between 60 °C and 70 °C.
mission on Higher Education.
When heated at 90 °C, the amount of oil can be extracted from the
coconut milks was independent of the homogenization process.
References
Heating the emulsion leads to protein denaturation. We hypothe-
Arumughan C, Balachandran C, Sundaresan A. 1993. Development of a process
for coconut cream on commercial scale. J Food Sci Technol 30(6):408–12.
size that, on heating the homogenized samples, the protein in-
Birosel DM, Gonzalez AL, Santos MP. 1963. The nature and properties of the
volved in bridging flocculation rearranges and is pulled away from
emulsifier system of oil globules in coconut milk and cream. Phil J Sci 92(1):1–
15.
1 of the bridged droplets to expose some lipid surface and allow
Cancel LE. 1979. Coconut food products and bases. In: Woodroof JG, editor. Coco-
more fat to be solvent extracted. In the non-homogenized samples,
nuts: production, processing, products. 2nd ed. Westport, Conn.: AVI Publishing.
the protein is not involved in bridging flocculation so it is not nec-
p 162–87.
Chanamai R, McClements DJ. 2000. Creaming stability of flocculated monodis-
essarily pulled away from the droplet surface on denaturation. The
perse oil-in-water emulsions. J Colloid Interface Sci 225(1):214–8.
resultant flocculation of the non-homogenized samples is induced
del Rosario RR, Punzalan GC. 1977. Quality control of coconut milk processing:
emulsion stability studies. Phil J Coconut Stud 2(4):9–14.
by protein-protein hydrophobic attractions and may serve to de-
Fink A, Kessler HG. 1985. Changes in the fat globule membrane produced by
crease the amount of solvent-extractable oil.
heating. Milchwissenschaft 40(5):261–4.
Gonzalez ON. 1990. Coconut milk. In: Banzon JA, Gonzalez ON, de Leon SY,
Sanchez PC, editors. Coconut as food. Quezon City, Philippines: Philippine Co-
Conclusions
conut Research and Development Foundation. p 15–48.
Kwon K, Park KH, Rhee KC. 1996. Fractionation and characterization of proteins
E: Food Engineering & Physical Properties
Coconut milk is a natural emulsion extracted with minimal pro from coconut (Cocos nucifera L.). J Agric Food Chem 44(7):1741–5.
cessing from fresh coconuts. The emulsion particle size is nat-
McClements DJ. 1999. Food emulsions: principles, practices and techniques. Boca
Raton, Fla.: CRC Press. 378 p.
urally of the order of 13.1 m and can only be slightly reduced on
Monera OD, del Rosario EJ. 1982. Physico-chemical evaluation of the natural
homogenization as the quality and quantity of emulsifiers (prob-
stability of coconut milk emulsion. Ann Tropical Res 4(1):47–54.
ably protein) naturally present is low. The homogenized emulsions
Parker A, Gunning PA, Ng K, Robins MM. 1995. How does xanthan stabilize salad
dressing? Food Hydrocoll 9(4):333–42.
tend to be highly flocculated, probably by bridging flocculation,
Seow CC, Gwee CN. 1997. Coconut milk: chemistry and technology. Intl J Food Sci
and this increases the product viscosity. Thermal treatment increas-
Technol 32(3):189–201.
Simuang J, Chiewchan N, Tansakul A. 2004. Effects of fat content and tempera-
es the degree of flocculation, probably because of protein-protein
ture on the apparent viscosity of coconut milk. J Food Eng 64(2):193–7.
hydrophobic attractions following denaturation, and leads to in-
Sliwinski EL, Roubos PJ, Zoet FD, van Boekel MAJS, Wouters JTM. 2003. Effects of
heat on physicochemical properties of whey protein-stabilised emulsions.
creased effective particle size and apparent viscosity. The creaming
Colloids Surf B Biointerfaces 31(1-4):231–42.
stability of the emulsion improved on heating because the larger
Timmins WH, Kramer EC. 1977. The canning of coconut cream. Phil J Coconut
Stud 2(4):15–25.
flocs can form a network in the more viscous emulsion. Solvent-ex-
Vitali AA, Soler MP, Rao MA. 1986. Rheological behavior of coconut milk. In:
tractable oil of non-homogenized coconut milk decreased with in-
LeMaguer M, Jelen P, editors. Food engineering and process applications. Vol.
1. transport phenomena. New York: Elsevier Applied Science. p 33T:–8.
creasing heating temperature, whereas those of homogenized sam-

---