Content uploaded by Chunxia Su
Author content
All content in this area was uploaded by Chunxia Su on Jul 17, 2021
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=tfac20
Food Additives & Contaminants: Part A
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tfac20
Development of a GC-FID method for the
quantitative determination of polyglycerol
polyricinoleate (PGPR) in foods
Chunxia Su, Paul Van der Meeren & Bruno De Meulenaer
To cite this article: Chunxia Su, Paul Van der Meeren & Bruno De Meulenaer (2021):
Development of a GC-FID method for the quantitative determination of polyglycerol polyricinoleate
(PGPR) in foods, Food Additives & Contaminants: Part A, DOI: 10.1080/19440049.2021.1951850
To link to this article: https://doi.org/10.1080/19440049.2021.1951850
Published online: 15 Jul 2021.
Submit your article to this journal
View related articles
View Crossmark data
Development of a GC-FID method for the quantitative determination of
polyglycerol polyricinoleate (PGPR) in foods
Chunxia Su
a,b
, Paul Van der Meeren
a
, and Bruno De Meulenaer
b
a
Particle and Interfacial Technology Group, Department of Green Chemistry & Technology, Faculty of Bioscience Engineering, Ghent University,
Ghent, Belgium;
b
NutriFOODchem Unit, Department of Food Technology, Safety and Health, Faculty of Bioscience Engineering, Ghent
University, Ghent, Belgium
ABSTRACT
Polyglycerol polyricinoleate (PGPR) is a powerful lipophilic emulsier used in low-fat spreads and
chocolate. It should be used at the lowest level at which the desired technological eect is
achieved, not exceeding the specied maxima according to Annexe II to Regulation (EC) No
1333/2008. A gas chromatography–ame ionisation detection (GC-FID) method was developed
for quantication of PGPR. This method is based on estimating the content of ricinoleic acid using
12-hydroxyoctadecanoic acid as an internal standard, from which the PGPR concentration was
deduced. The method involved saponication, methylation, a two-step solid phase extraction (SPE)
separation of the fatty acid methyl esters (FAMEs), silylation, and GC-FID analysis. The limits of
detection and quantication of ricinoleic acid were 2.2 and 6.7 μg/mL, respectively, at 0.1 µL
injection volume. Considering the average content of ricinoleic acid in PGPR (i.e. 86.63 ± 2.0 wt%)
and the amount of food product that is used in the proposed protocol (i.e. 20 mg), this resulted in
a LOD and LOQ of 0.76 and 2.32 μg PGPR per mg of food product, respectively. The developed
method was validated by determining PGPR recovery from a high oleic sunower oil (HOSO)
solution, from chocolate spiked with a commercially available PGPR, and from commercially
available low fat spreads with a known PGPR content. The actual recovery was more than 95%
for all matrices, indicating the accuracy of the developed analytical technique. Moreover, the
method proved to be very reproducible, with RSD < 4% for concentrations ranging from 0.2 to
5 wt%. The results showed that our proposed GC-FID method enables the reliable and quantitative
determination of the PGPR concentration in commercial food products with various fat contents.
ARTICLE HISTORY
Received 6 May 2021
Accepted 27 June 2021
KEYWORDS
PGPR; ricinoleic acid; FAMEs;
GC-FID; low fat spreads;
chocolate
Introduction
The food additive polyglycerol polyricinoleate (PGPR,
E 476) is manufactured by esterification of condensed
castor oil fatty acids with polyglycerol (Wilson et al.
1998; Dedinaite and Campbell 2000; Bastida-
Rodríguez 2013). The main fatty acid in castor oil is
ricinoleic acid ((9Z,12 R)-12-hydroxyoctadec-9-enoic
acid, C18:1–12OH), which because of the presence of
a hydroxyl-group is prone to polymerisation with
other ricinoleic acid molecules. In addition to ricino-
leic acid, castor oil and thus PGPR can contain oleic
(C18:1: 3–8%), linoleic (C18:2: 3–7%), and stearic acid
(C18:0: 0–2%) (Bastida-Rodríguez 2013). Polyglycerol
is mainly composed of linear di-, tri- and tetraglycer-
ols but may contain also branched and cyclic isomers
(De Meulenaer et al. 2000). A typical structure of
PGPR is shown in Figure 1. PGPR typically consists
of a complex mixture of various esters with a varying
polymerisation degree (both on the level of glycerol
and ricinoleic acid) and a varying esterification degree
(Orfanakis et al. 2013). In addition, a number of
minor products can be produced during production
of PGPR (Bastida-Rodríguez 2013).
It has been proven that PGPR is tolerated by ani-
mals at high doses without adverse effects and an
acceptable daily intake (ADI) of 25 mg/kg has been
agreed upon (Mortensen et al. 2017). As a food addi-
tive, legal restrictions with respect to the application of
PGPR in foods apply within the European Union.
These specify the types of food products in which
PGPR can be used and its maximum concentration
(European Commission 2011). It is used especially as
low-hydrophilic-lipophilic balance emulsifier in the
preparation of W/O emulsions (Vermeir et al. 2016;
Okuro et al. 2019). These emulsions are either used as
such, for example as low fat spreads or are used in the
CONTACT Chunxia Su chunxia.su@ugent.be Particle and Interfacial Technology Group, Department of Green Chemistry & Technology, Faculty of
Bioscience Engineering, Ghent University, Coupure Links 653, Ghent 9000, Belgium
FOOD ADDITIVES & CONTAMINANTS: PART A
https://doi.org/10.1080/19440049.2021.1951850
© 2021 Taylor & Francis Group, LLC
production of double emulsions (W/O/W emulsions)
for potential food applications (Balcaen et al. 2016;
Clegg et al. 1996; Balcaen et al. 2017; Rebry et al.
2020). In addition, PGPR can be applied in chocolate
products, particularly for improving the moulding
properties, increasing the tolerance to the thickening
effect and limiting fat bloom (Lonchampt and Hartel
2004; Peschar et al. 2004; Schenk and Peschar 2004;
Da Cunha et al. 2010). In order to enforce the speci-
fied legal restrictions, analytical methods should be
available enabling to determine the actual PGPR con-
tent in foods.
Currently, however, no methods are described
which allow the quantitation of PGPR in rele-
vant food matrices. There are only a few reports
focussing on the compositional characterisation
of PGPR as such.
1
H and
13
C spectra of PGPR
proved the presence of polyglycerol moieties of
various lengths being esterified by ricinoleic acid
and oligomeric ricinoleic acid and to a lesser
extent by oleic and linoleic acids (Orfanakis
et al. 2013). Because of the complex composition
of commercial PGPR products, it is impossible
to determine the individual PGPR species pre-
sent and consider their sum as total PGPR con-
tent. Therefore, an indirect approach is
presented based on the fact that ricinoleic acid
constitutes approximately 90% of the fatty acid
content of castor oil, whereas it is not present in
food products as such. Based on the average
ricinoleic acid content of PGPR products, the
PGPR content of a food can thus be estimated.
Ricinoleic acid was determined as its methyl
ester via gas chromatography coupled to flame
ionisation detection after silylating its hydroxyl
group to a trimethylsilylether. In order to avoid
interference with other fatty acids present in the
sample, fractionation of the fatty acid methyl esters
was required over silica using SPE.
Materials and methods
Chemicals and materials
PGPR 4175 A, 4150, 4125, and 4110 A were kindly
provided by Palsgaard A/S (Juelsminde, Denmark),
whereas PGPR 4175 B and 4110 B were provided by
Vandemoortele (Belgium), where A and B represent
different production batches. Admul WOL 1403 K
originated from Kerry (Ireland), whereas Radiamuls
Poly 2251 K was produced by Oleon (Belgium). High
oleic sunflower oil (HOSO; Iodine Value = 87; 82%
C18:1) was acquired from Contined BV
(Wageningen, the Netherlands). Silica gel 60 for col-
umn chromatography (particle size = 63–100 µm),
boron trifluoride-methanol solution (BF
3
/MeOH)
(20%), trimethylchlorosilane (TMCS), N,
O-bis(trimethylsilyl) trifluoroacetamide (BSTFA),
nonadecanoic acid (C19:0, ˃99%), ricinoleic acid
(C18:1–12OH, ˃99%) and 12-hydroxyoctadecanoic
acid (C18:0–12OH, 99%) were purchased from
Sigma-Aldrich (St. Louis, MO, USA). Palmitic acid
(C16:0, ˃99%), stearic acid (C18:0, ˃99%), linoleic
acid (C18:2, ˃99%), alpha linolenic acid (C18:3,
˃99%), arachidic acid (C22:0, ˃99%) and mixed
GLC 68D standard were obtained from Nu-Chek-
Prep, Inc. (USA). All other chemicals and reagents
were of analytical grade and obtained from local sup-
pliers. Four types of low fat spreads (brand A-D) were
kindly provided by Vandemoortele (Belgium); their
Figure 1. Chemical structure of polyricinoleate (PR) and polyglycerol polyricinoleate; the typical polymerisation degree of PR and
polyglycerol amount up to 5 and 4, respectively. Only the presence of linear polyglycerol isomers was considered.
2C. SU ET AL.
fat and PGPR content (according to the information
provided by the producer) are specified in Table 1. The
specific PGPR batches used to produce these products
were provided by Vandemoortele as well. A milk cho-
colate (Côte-d’Or, Mignonette) containing at least
33% of cocoa butter and soybean lecithin was pur-
chased in a local supermarket after careful inspection
of the label to confirm the absence of PGPR.
Preparation of high oleic sunower oil (HOSO)-PGPR
mixtures
Homogeneous mixtures were obtained by adding
1.0, 3.0 and 5.0 wt% PGPR 4175 A to HOSO after
which the mixture was heated to 60°C under mag-
netic stirring for 15 min (Zhu et al. 2017).
Incorporation of PGPR into commercial chocolate
The chocolate was purchased in a local store after
careful inspection of the label to confirm the absence
of PGPR. According to Schantz and Rohm (2005)
and Atik et al. (2020), PGPR 4150 and PGPR 4125
(Palsgaard, Denmark) are appropriate to be used in
chocolate. Considering the maximum level of PGPR
in chocolate (5 g/kg according to European Directive
95/2/EC), PGPR 4150 and PGPR 4125 were mixed at
two different concentrations (i.e. about 0.25 and
0.5 wt%) into the chocolate. After weighing an
appropriate amount (approximately 100 g) of melted
chocolate (50°C) into a beaker, PGPR preheated for
10 minutes at 50°C was added dropwise under con-
stant stirring (at speed ‘I’, using a Braun Multiquick
3 MQ3035 mixer (Germany)). After incorporating
the PGPR, the mixture was further stirred for 5 min-
utes, after which it was cooled at room temperature.
The concentration of added PGPR (C
added PGPR
(g/
kg)) in chocolate was calculated using Equation (1):
Cadded PGPR¼m1m2
m0þðm1m2Þ100 (1)
m
0
is the mass of accurately weighed melted
chocolate, m
1
is the total mass of PGPR and test
tube before PGPR addition, and m
2
is total mass of
remaining PGPR and test tube after adding PGPR.
Extraction of oil phase from low fat spread
In order to extract the oil phase from the low fat
spreads, 10 mL of petroleum ether (PE) was added
to about 1.0 g of sample, and the mixtures were
vortexed for 1 min. Centrifugation (Hettich Rotina
380 R, German) at 500 g was performed for 5 min
(at 4°C) to obtain a clear organic solvent phase. The
upper PE phase was collected; the three organic
solvent phases obtained by repeating the extraction
three times were combined. The obtained PE
phases were diluted to 50.0 mL, of which 1.0 mL
was dried under nitrogen in the presence of about
1.0 mg of IS (C18:0–12OH) which was previously
added from a diethyl ether solution by drying as
well under nitrogen. The overall dried residue was
converted to methyl esters as described below.
Preparation of fatty acid methyl esters (FAMEs)
In a glass test tube containing about 1 mg of C18:0–
12OH (for PGPR samples) or 0.5 mg of C18:0–
12OH (for PGPR-HOSO mixtures, chocolate
spiked with PGPR and low-fat spread samples)
(previously dissolved in diethyl ether which was
evaporated under nitrogen), 4.0–20.0 mg of sample
was added (4.0 mg for the commercial PGPR pro-
duct, which was first dissolved into diethyl ether,
and evaporated under nitrogen; 15.0 mg for the
PGPR-HOSO blend or chocolate; 20.0 mg for the
low fat spread). The samples were subsequently
saponified and methylated according to the AOCS
method (Ce 1b-89) (1989), with some modifica-
tions. Therefore, 2 mL NaOH-solution in methanol
(either 0.5 N or 1.0 N) was added and the test tube
was firmly closed with a screw cap. Then, the test
tube was placed in a boiling water bath for
a specified time (7, 10, or 15 min), and subsequently
transferred in cold water for cooling. Afterwards,
2 mL of BF
3
/MeOH-reagent (20%) was added and
Table 1. Low fat spread samples used in this study with their fat
and PGPR content according to the manufacturer’s specifications.
Sample PGPR type
a
fat content
a
(%) PGPR content
a
(%)
A Palsgaard PGPR 4175 B 25 0.34
B Palsgaard PGPR 4110 B 25 0.35
C Palsgaard PGPR 4110 B 35 0.20
D Palsgaard PGPR 4175 B 20 0.40
a
based on data provided by the producer of the low fat spreads.
FOOD ADDITIVES & CONTAMINANTS: PART A 3
the closed test tube was vortexed for 1 min, placed
in a boiling water bath for 8 min and cooled in cold
water again. Finally, 3 mL of isooctane was added
into the test tube, which was vortexed for 1 min,
and immediately afterwards, 5 mL of an aqueous
saturated NaCl-solution was added, followed by an
additional 1 min vortexing step. After phase separa-
tion, the isooctane phase was transferred into a test
tube with a Pasteur pipette. The residual aqueous
phase was extracted again with 3 mL of isooctane
and vortexed for 1 min. The combined isooctane
fractions (with the resultant FAMEs) were dried
under nitrogen and re-dissolved into respectively
3 mL (for chocolate and low fat spread samples) or
6 mL (for PGPR and spiked HOSO samples) of
n-hexane-diethyl ether (98:2, v/v) for SPE.
For some samples and during method develop-
ment C19:0 was used as a second internal standard
(about 1.0 mg), but for the final method, this was
not required.
SPE of hydroxylated FAMEs
The SPE steps followed the procedure proposed by
Mubiru et al. (2014), with some modifications.
Silica was dried in a muffle furnace at 450°C for
12 h and cooled in a desiccator. The moisture con-
tent was adjusted to 10% (w/w) and the silica was
equilibrated on a shaker for 1 h before use. An
empty SPE cartridge column (6 mL, 6.5 cm ×
1.3 cm) was filled with 1.0 g of activated silica and
5 mL of the elution solvent (n-hexane/diethyl ether
(98:2, v/v)) was added and thoroughly mixed in the
column to prepare a slurry. The column was tapped
carefully to ensure a uniform packing, and finally
a small amount of sea sand was added. The elution
solvent was passed from the column after which an
aliquot of 2.0 mL of FAMEs in n-hexane–diethyl
ether (98:2, v/v) was loaded onto the prepared silica
column. Unless stated differently, the non-polar
fraction, which comprises the non-OH FAMEs,
was eluted with 20 ml of n-hexane–diethyl ether
(98:2, v/v). When PGPR samples were analysed,
15 mL of n-hexane–diethyl ether (98:2, v/v) was
used instead. The polar compounds (OH-FAMEs)
were subsequently eluted with 15 mL of diethyl
ether. The two fractions were completely dried
under nitrogen and dissolved in 2 mL of isooctane
for further analysis.
Separation of the fractions was confirmed by
thin-layer chromatography (TLC) according to
Marmesat et al. (2008), using silica gel 60
(5 cm×10 cm plates, 0.25 mm thickness). The
plate was deliberately overloaded, developed with
hexane–diethyl ether–acetic acid (80:20:1, v/v/v)
and visualised with iodine vapour.
Silylation of hydroxylated FAMEs
The silylation of the polar FAMEs was performed
according to Mubiru et al. (2013) using 10% (v/v)
trimethylchlorosilane (TMCS) in N,O-bis (tri-
methylsilyl) trifluoroacetamide (BSTFA). Around
30 μL polar FAME of PGPR samples and 100 μL of
polar FAME of PGPR-HOSO mixtures, chocolate
spiked with PGPR, and low-fat spread samples, to
be silylated, were dried under nitrogen and 100 µL of
silylating reagent was added. After 20 min reaction at
room temperature, the reagent was evaporated
under nitrogen and the analytes were dissolved in
200 µL of isooctane prior to GC injection.
GC-FID analysis of FAMEs
The (silylated) FAMEs were analysed using an
Agilent 6890 N series gas chromatograph (Agilent,
USA): 0.1 μL was injected directly into the column
using a cold on column injector (COC). Separation
was performed on a CP-Sil 88TM (Agilent, USA) for
FAME (60 m × 0.25 mm i.d.) capillary column
coated with a 0.2 μm film. A deactivated fused silica
precolumn of (2–6) m × 0.25 mm i.d. (Agilent,
Belgium) was fitted to protect the column. The
oven temperature programme started at 50°C for
4 min, then increased to 225°C at 20°C/min, and
finally was held for 25 min. The flame ionisation
detector temperature was set at 300°C. The detector
flow rates for hydrogen, air, and helium (make-up)
were 40, 400 and 20 mL/min, respectively. The col-
umn flow rate of helium as a carrier gas was 1 mL/
min. For quantification of ricinoleic acid, the
response factor of C18:1–12OH towards C18:0–
12OH was determined.
Identification of the non-hydroxylated FAMEs in
PGPR was carried out by comparison of their reten-
tion times with those of authentic standards (GLC
68D, Nu-Chek Prep., Inc., USA). For their quantifica-
tion, calibration solutions containing C16:0, C18:0,
4C. SU ET AL.
C18:2, C18:3 and C20:0 as fatty acids (between about
20 and 80 μg/mL) and C19:0 (as fatty acid, 27 μg/mL)
were analysed by saponification, methylation and GC
analysis. The response factor of non-hydroxylated
FAMEs towards C19:0 was calculated by linear regres-
sion analysis of the peak area ratio of the non-
hydroxylated fatty acids to the internal standard versus
the concentration ratio of the non-hydroxylated fatty
acids to the C19:0 internal standard. The results indi-
cated that the p-value corresponding to the t test of the
intercept for each fatty acid (95% confidence interval)
was greater than 0.05. Hence, the linear regression was
repeated without intercept (calibration curves not
shown). The obtained linearity was excellent within
the studied concentration range: all the determination
coefficients R
2
were higher than 0.9998. The response
factors obtained are shown in Table 2, and these
response factors were used to calculate the corre-
sponding non-hydroxylated fatty acid content in the
PGPR samples. As, according to the report of Singh
et al. (2014), there is little difference in response factors
between unsaturated fatty acids and saturated fatty
acids with the same carbon chain length, the content
of C18:1 was estimated using the response factor of
C18:0, and the response factor of C20:0 was used to
estimate the content of C20:1 and C20:2. The non-OH
fatty acid composition of the PGPR samples consid-
ered in this study is shown in Table 4.
Method validation
The linearity of the instrument was determined using
six silylated ricinoleic acid solutions in iso-octane with
concentrations from 10 to 60 µg/mL. Mandel’s fitting
test was used to evaluate the linearity of the linear
regression model (Van Loco et al. 2002). For the
calibration curve, the concentration of ricinoleic acid
versus the peak area ratio of ricinoleic acid to C18:0–
12OH internal standard was used. The mean of the
slopes (S) and the standard deviation of the intercepts
(σ) was calculated from three calibration curves and
used for the calculation of LOD and LOQ according
to the International Conference on Harmonisation
(Guideline 2005) using the formulas below:
LOD ¼3:3σð Þ=S (2)
LOQ ¼10 σð Þ=S (3)
The accuracy of the method was determined by
analysing the PGPR recovery from PGPR 4175 A in
HOSO with concentrations of 1.0, 3.0, or 5.0 wt%. In
addition, method accuracy was tested by spiking
a commercial chocolate with two types of PGPR at
two concentration levels. Finally, the accuracy was
also measured by analysing four types of commercial
low fat spreads with known amounts of PGPR. The
accuracy and repeatability were also evaluated accord-
ing to the Commission Decision of 12 August 2002
implementing Council Directive 96/23/EC concern-
ing the performance of analytical methods and inter-
pretation of results (European Commission 2002).
Given the concentration levels applicable in this
study, the guideline ranges for the deviation of the
experimentally determined concentration from the
certified value should be within (−20%) – (+10%).
The relative standard deviation (RSD) for repeated
analyses shall not exceed the level calculated by the
Horwitz equation:
AcceptableRSDð%Þ ¼2ð10:5logCÞ(4)
where C is the mass fraction. Acceptable RSD values
obtained from the Horwitz equation according to the
concentration level of PGPR in HOSO and in the
two food matrices are 3.1% (for C = 0.050) and 4.3%
(for C = 0.0059), respectively.
Statistical analyses
All experiments were completely carried out in
triplicate and results were represented as means of
the three replicates. Normality was checked with
a Shapiro–Wilk test and the Levene’s test applied to
confirm the homogeneity of variances. Unless sta-
ted differently, the statistical comparison between
the results was made using one-way analysis of
variance (ANOVA). When differences were
detected, multiple comparisons were performed
by the Tukey-b test. When analysing the PGPR
Table 2. Response factor (Rf = Peak area ratio of FA to IS/Concentration ratio of FA to IS) relative to C19:0 as
internal standard for determination of non-hydroxylated fatty acids.
Fatty acid C16:0 C18:0 C18:2 18:3 C20:0
Response factor 0.97 ± 0.00 1.01 ± 0.00 1.05 ± 0.01 1.03 ± 0.00 1.03 ± 0.01
FOOD ADDITIVES & CONTAMINANTS: PART A 5
recovery (based on the actual and average ricinoleic
acid content of PGPR, respectively) from food
matrices, the t-test was used instead. Significance
was set at a p-value < 0.05 for all comparisons.
Statistical analyses were processed by Microsoft
Excel 2010 and SPSS 22 statistics package (IBM,
SPSS, Inc.).
Results and discussion
In this study, the objective was to develop a GC-
FID method for the quantitative determination of
the PGPR content in food matrices, such as low-fat
spreads and chocolate. The analytical strategy was
based on the quantitative gas chromatographic ana-
lysis of ricinoleic acid.
Quantitative gas chromatographic analysis of
ricinoleic acid
For the quantitative analysis of ricinoleic acid, the
analytical target was converted to methyl ricinoleate
via BF
3
catalysed methylation followed by silylation
of the hydroxyl group. In parallel, an internal stan-
dard (12-hydroxyoctadecanoic acid (C18:0–12OH))
was used. Both compounds returned clear, sym-
metric and base-line separated peaks in the GC
chromatogram (Figure 3(c)). Calibration curves (10
up to 60 μg/mL ricinoleic acid, 20 µg/mL IS)
(Figure 2) proved to be linear: the Mandel’s test
value obtained was 6.6, which was below the tabu-
lated F-value of the 99% confidence level (F99 = 8.5).
As the p-value corresponding to the t test of the
intercept (95% confidence interval) was 0.08, it fol-
lows that the intercept was not significantly different
from zero; hence, the linear regression was repeated
without intercept (Figure 2). The determination
coefficient R
2
of linear regression was 0.9995, with
a slope not significantly different from 1.00 (95%
confidence interval of 1.00 ± 0.01) and with an
excellent repeatability (RSD = 0.55%) and reprodu-
cibility (RSD = 1.54%). The LOD and LOQ values of
ricinoleic acid were 2.2 and 6.7 μg/mL, respectively.
Considering the injection volume used (i.e. 0.1 µL),
these values correspond to absolute amounts of 0.22
and 0.67 ng, respectively.
Ricinoleic acid content (and fatty acid composition)
in commercial PGPR samples
Firstly, the fatty acid composition of castor oil was
determined by using the AOCS method (Ce 1b-89)
to prepare fatty acid methyl esters and using SPE
and silylation, as it was considered as a simpler
matrix as compared to PGPR since the ricinoleic
acid is not polymerised in castor oil. The castor oil
was saponified with a methanolic NaOH solution.
Subsequently, the fatty acids were esterified with
Figure 2. Calibration curve for the gas chromatographic analysis of silylated methyl ricinoleate (RA) by using silylated methyl-12-
hydroxystearate (C18:0-OH) as internal standard. The concentration of the latter equalled 20 µg/ml, whereas the ricinoleic acid
concentration ranged between 10 and 60 µg/ml.
6C. SU ET AL.
BF
3
/MeOH-reagent to obtain fatty acid methyl
esters. Figure 3(a) indicates that methyl linoleate
eluted closely to the silylated IS (C18:0–12OH) and
the silylated methyl ricinoleate, which could hinder
accurate peak integration for samples containing
a high level of linoleic acid, as would typically be
the case in low-fat spreads. In order to prevent peak
overlap, the obtained FAMEs were separated into
polar (i.e. hydroxylated) and non-polar FAMEs
using a previously described (Marmesat et al.
2008) two steps silica SPE method. As the hydroxy
compounds do not elute easily from the GC
column, resulting in the formation of broad peaks,
the polar fraction was analysed after silylation
(Tonta et al. 2019). Figure 3(c) shows that all the
non-OH FAMEs were absent in the polar fraction
after the SPE procedure. The content of C16:0,
C18:0, C18:1, C18:2, C18:3 and C20:1 was
0.93 ± 0.06, 1.09 ± 0.08, 2.96 ± 0.20, 3.87 ± 0.27,
0.59 ± 0.02 and 0.36 ± 0.04 wt%, respectively, while
the ricinoleic acid content equalled 85.61 ± 1.32 wt
%. From the fatty acid content (95.41 ± 1.99 wt%),
the glycerol content in castor oil can be calculated
based on the fact that the total number of moles of
Figure 3. GC chromatogram of silylated (a, c) and non-silylated (b) FAMEs of castor oil before (a, b) and after (polar fraction: c) SPE
fractionation. The length of the precolumn was around 5 m.
FOOD ADDITIVES & CONTAMINANTS: PART A 7
fatty acids is equal to 3 times the number of moles
of glycerol, with one mole representing 92 g, which
gave rise to an expected glycerol content of
9.85 ± 0.21 wt%. Considering the fact that 3 moles
of water (MM = 18 g/mol) are lost upon esterifica-
tion per mole of glycerol, the experimentally deter-
mined sum of glycerol and fatty acids should be
diminished by (9.85/92*3*18 =) 5.78 ± 0.12 wt% to
estimate the total triglyceride content. As the esti-
mated value (99.48 ± 2.32 wt%) was very close to
100%, this indicates the accuracy of the fatty acid
analysis procedure proposed. In addition, the
experimentally determined fatty acid composition
of castor oil corresponded very well to previously
reported data (Goodrum and Geller 2005; Scholz
and Da Silva 2008; Knothe et al. 2012). Hence, this
method was subsequently used to determine the
ricinoleic acid content of PGPR.
As shown in Table 3 (I), the obtained ricinoleic
acid content in PGPR 4175 A was 52.52 ± 1.43 wt%
using the AOCS method (Ce 1b-89), which was
much lower than the ricinoleic acid content in
castor oil. A similar unexpected outcome was
observed using the same saponification conditions
on a 5 wt% PGPR 4175A in HOSO mixture for
which, based on the ricinoleic acid content speci-
fied in the technical information sheet of the PGPR
4175 A, a recovery of only about 50% was obtained.
It was therefore speculated that because of the poly-
merised nature of PGPR, the accessibility of the
ester bonds for the base and therefore their suscept-
ibility for saponification was hindered. As a further
result, the saponification conditions which were
suitable for triacylglycerols proved not sufficient
for PGPR analysis. Therefore, the saponification
conditions were intensified by doubling the amount
of NaOH, by increasing the saponification time, by
using less sample and/or by increasing the mixing
intensity of the two-phase system at the beginning
of the saponification step by including boiling
stones.
In Table 3 it can be observed that the saponifica-
tion conditions impacted the estimated ricinoleic
acid content of PGPR 4175 A. In the absence of
boiling stones, increasing the NaOH content and
the saponification time of about 10 mg of sample
Table 3. Impact of varying saponification conditions on the estimated ricinoleic acid content in PGPR 4175 A; the data are mean values
± standard deviations of three independent determinations.
Condition of
saponification
C
NaOH
(N)
t
saponification
(min) Boiling stones
d
M
sample
(mg)
C
ricinoleic acid
(wt%)
I 0.5 7 - 10.69 52.52 ± 1.43
a
II 1.0 10 - 10.87 79.09 ± 2.47
b
10 + 10.67 86.54 ± 0.8
c
10 - 4.20 85.58 ± 0.41
c
10 + 4.20 84.90 ± 1.20
c
15 - 11.20 86.33 ± 0.71
c
15 + 10.67 85.75 ± 0.86
c
15 - 4.20 85.05 ± 0.61
c
15 + 4.20 84.79 ± 0.48
c
I: original conditions described in the AOCS method Ce 1b-89; II: additional conditions considered in this study.
a, b
and
c
: different superscript letters mean significantly different results.
d
: -: without boiling stones; +: with boiling stones.
Table 4. Fatty acid composition (in wt%) of commercial PGPR samples; the data are mean values ± SD of three independent determinations.
Fatty Acid PGPR 4175 A PGPR 4175 B PGPR 4150 PGPR 4125 PGPR 4110 A PGPR 4110 B
Admul WOL
1403
Radiamuls Poly
2251 K
C16:0 0.85 ± 0.02
a
0.93 ± 0.00
b
0.94 ± 0.01
b
0.93 ± 0.01
b
0.95 ± 0.03
b
0.92 ± 0.01
b
1.05 ± 0.02
c
1.13 ± 0.01
d
C18:0 1.07 ± 0.01
a
1.39 ± 0.01
d
1.38 ± 0.00
cd
1.31 ± 0.02
b
1.33 ± 0.03
b
1.40 ± 0.01
d
1.34 ± 0.01
bc
1.49 ± 0.01
e
C18:1 2.85 ± 0.02
a
2.76 ± 0.02
a
2.77 ± 0.01
a
2.79 ± 0.04
a
2.86 ± 0.10
a
2.85 ± 0.03
a
2.86 ± 0.04
a
3.38 ± 0.03
b
C18:2 3.50 ± 0.00
ab
3.42 ± 0.01
a
3.46 ± 0.02
ad
3.42 ± 0.05
a
3.49 ± 0.12
ab
3.54 ± 0.01
ab
3.58 ± 0.04
b
3.50 ± 0.04
ab
C20:1 0.33 ± 0.01 0.30 ± 0.01 0.31 ± 0.01 0.31 ± 0.00 0.31 ± 0.01 0.30 ± 0.01 0.31 ± 0.00 0.35 ± 0.01
C18:3 0.29 ± 0.07
b
0.23 ± 0.00
ab
0.26 ± 0.00
ab
0.23 ± 0.00
ab
0.24 ± 0.01
ab
0.24 ± 0.0
ab
0.21 ± 0.00
a
0.22 ± 0.00
a
C20:2 0.26 ± 0.01
cd
0.27 ± 0.01
cd
0.22 ± 0.01
ab
0.21 ± 0.01
a
0.25 ± 0.01
bc
0.29 ± 0.02
d
0.29 ± 0.00
d
0.41 ± 0.01
e
C18:1–
12OH
85.58 ± 0.41
ab
83.48 ± 0.60
a
89.97 ± 0.35
d
86.68 ± 1.16
bc
88.75 ± 2.00
cd
85.59 ± 0.65
ab
88.34 ± 0.46
cd
87.99 ± 0.76
bcd
Other 2.80 ± 0.06
a
3.18 ± 0.02
b
2.90 ± 0.02
a
2.89 ± 0.02
a
2.90 ± 0.10
a
3.08 ± 0.03
b
4.12 ± 0.05
c
5.00 ± 0.06
d
a, b, c, d and e: different superscript letters in a row indicate significantly different results (p-value <.05).
A and B: different production batches.
8C. SU ET AL.
significantly increased the recovered ricinoleic acid
content. However, by ensuring better mixing (by
including boiling stones), an additional significant
increase in the experimentally determined ricino-
leic acid content was observed. On the other hand,
the impact of adding boiling stones was not signifi-
cant anymore if the saponification time was
increased to 15 minutes or if the amount of sample
was reduced to about 4 mg instead of 10 mg (always
using a double amount of NaOH). As added mate-
rials (such as boiling stones) may lead to unknown
effects on the experimental results (such as contam-
ination or adsorption), and as no added value was
observed, the following saponification conditions
were selected to determine the content of ricinoleic
acid in PGPR: around 4 mg of sample was boiled
for 10 min in 2 mL of 1 N NaOH in MeOH without
boiling stones. Thus, the quantitative method for
the profiling of all fatty acids in PGPR as shown in
Figure 4 was applied on eight different commercial
PGPR samples (Table 4). Although statistically sig-
nificant differences in the ricinoleic acid content of
the various PGPR samples were noticed, even for
samples of the same type but a different batch,
overall the variability in the ricinoleic acid content
of the various PGPR samples was small: it ranged
from 83.48 ± 0.60 to 89.98 ± 0.35 wt% sample and
equalled on average 86.63 ± 1.98 wt%. Combined
with the fact that PGPR is most likely the only
source of ricinoleic acid in foods, it was therefore
considered appropriate to determine the PGPR
content of food samples indirectly by determining
quantitatively their ricinoleic acid content, consid-
ering an average ricinoleic acid content in PGPR of
86.63 wt%. The data presented in Table 4 also
illustrate the good repeatability (RSD = 0.45–
2.25%) of the method applied.
Recovery of PGPR
The accuracy of the method was determined on
three different food matrices, i.e. HOSO and cho-
colate spiked with PGPR and commercially avail-
able low fat spreads containing known amounts of
PGPR. In order to illustrate the need to fractionate
the fatty acid methyl esters into an apolar (non-OH
FAME) and polar (OH FAME) fraction prior to GC
analysis, the obtained fatty acid methyl esters of
Figure 4. Proposed analytical scheme for the total fatty acid profiling of PGPR.
FOOD ADDITIVES & CONTAMINANTS: PART A 9
a spiked chocolate and low fat spread sample were
analysed without SPE fractionation (but with silyla-
tion to prevent tailing of the OH-fatty acid methyl
esters). As can be noted in Figure 5(a,b), the methyl
arachidate was not baseline separated from the
silylated methyl ricinoleate. Moreover, the choco-
late samples contained at least 33% cocoa butter,
which incorporates 1.7–3% of linoleic acid (Naik
and Kumar 2014), the interference peak of the IS
(Figure 3(a)). Moreover, the IS peak in Figure 5(a,
b) was wider than in the inserts, especially for the
low fat spread. This indicated that the methyl
linoleate peak interfered with the IS peak with the
GC column used, for both the chocolate and low fat
spread samples. Hence, a preliminary fractionation
between the apolar and polar FAMEs was required
to enable the accurate quantification of the RA
content and hence the concentration of PGPR in
foods. The impact of the non-polar elution volume
(10–30 mL) was evaluated by checking the GC
chromatograms: 20 mL was shown to be enough
to remove the interference peaks, as shown by the
inserts in Figure 5. For completeness, it has to be
mentioned that the retention times in Figure 5 are
slightly lower than in Figure 3 (e.g. about 16.8
versus 15.9 min for RA). The latter was due to
a slightly shorter precolumn in Figure 5.
Firstly, the recovery of PGPR 4175 A from spiked
HOSO was determined. The calculation was based
on the determined ricinoleic acid content of PGPR
4175 A (i.e. 85.58 ± 0.41 wt% according to Table 4).
In order to make the saponification reaction com-
plete, about 15 mg of the 5 wt% PGPR 4175
A solution in HOSO was boiled in 2 ml of 1 N
NaOH/MeOH during 10 min and 15 min, respec-
tively, without boiling stones. The results showed
that the recovery after 10 min of saponification was
92.57 ± 0.06 %, whereas it increased to 96.62 ± 0.48%
after 15 min of saponification. These results clearly
show that a reaction time of 10 minutes was not
Figure 5. GC chromatogram of the silylated FAMEs of low-fat spread B (a) and of PGPR-supplemented chocolate I (b) before and after
(polar fraction: inserts) SPE fractionation. The length of the precolumn was around 3 m.
10 C. SU ET AL.
sufficient to complete the saponification reaction
when PGPR was present in an excess of triglycerides.
Therefore, the saponification time was extended to
15 min in order to ensure complete liberation of the
fatty acids. Recoveries of PGPR added to HOSO in
a concentration range from 1 to 5 wt% were shown
to be at least 95% (Table 5), and within the (−20%) –
(+10%) range as specified in the Commission
Decision of 12 August 2002 implementing Council
Directive 96/23/EC concerning the performance of
analytical methods and interpretation of results
(European Commission 2002). Besides, there was
no significant effect of the amount of PGPR added
to the oil on its recovery (p-value = 0.09). Moreover,
the method applied showed an excellent repeatability
(RSD < 1%) and fulfilled the Horwitz criterium (RSD
should be < 3.1%). Hence, the method developed
proved to be robust and repeatable even in the pre-
sence of a large excess of triglycerides.
Additional recovery tests were performed on
chocolate spiked with PGPR at two concentration
levels and on commercial low-fat spread samples
with a known PGPR content (on basis of the data of
the producer). These tests included two different
types of PGPR for the low-fat spread samples and
another set of two different PGPR samples for the
spiked chocolate.
Besides direct saponification of the low-fat spread,
a preliminary oil phase extraction was also evaluated,
as this may prevent undesired effects of components
such as additional emulsifiers. After fat extraction by
petroleum ether, the actual recovery (based on the
ricinoleic acid content of the specific type of PGPR)
of the four low fat spreads was 60.92 ± 0.75%,
64.25 ± 0.75%, 87.45 ± 0.53, and 67.17 ± 1.72% for
brands A-D, respectively. These low results indicated
that a pre-treatment was undesirable for accurate
determination of the PGPR content in commercial
low fat spread products, because only part of the
PGPR species was recovered during the pre-
treatment. It is speculated that the partial loss upon
preliminary fat phase extraction may be due to the
adsorption of PGPR to the oil–water interfaces in the
low-fat spreads.
The actual recovery of PGPR as calculated based
on the experimentally determined ricinoleic acid
content of the specific PGPR sample was also deter-
mined for the low fat spread samples without pre-
liminary fat phase extraction. The actual recoveries
of all samples were excellent and fell within the legal
requirements irrespective of the type of PGPR used
(Table 6). In addition, the method was proved to be
highly repeatable as the RSD fulfilled the Horwitz
criterium (RSD should be < 4.3%). As, in reality, the
specific type of PGPR used in a commercial food
Table 5. Recovery of PGPR 4175 A from solutions with different
concentrations in HOSO; the data are mean values ± SD based on
three independent determinations. The accuracy indicates the
deviation of the experimentally determined concentration from
the real value, whereas the relative standard deviation (RSD)
represents the ratio of the standard deviation over the mean
value.
C
PGPR 4175 A
(wt%) Recovery (%)* Accuracy (%) RSD (%)
1 98.27 ± 0.80 −1.74 0.81
3 97.05 ± 0.94 −2.95 0.97
5 96.62 ± 0.48 −3.38 0.50
Table 6. Actual PGPR recovery (based on the ricinoleic acid content of the specific PGPR sample used) and practical recovery (based on
the average ricinoleic acid content of all PGPR samples considered) from different types of low fat spreads and PGPR-spiked chocolate;
the data are mean values ± SD based on three independent determinations. The accuracy indicates the deviation of the experimentally
determined concentration from the real value, whereas the relative standard deviation (RSD) represents the ratio of the standard
deviation over the mean value.
Matrix Sample
a
Based on actual RA content Based on average RA content RSD
(%) p-value
b
(-)Actual recovery (%) Accuracy (%) Practical recovery (%) Accuracy (%)
Low fat spread A 95.65 ± 2.74 −4.35 92.18 ± 2.64 −7.82 2.86 .19
B 97.80 ± 2.79 −2.20 96.63 ± 2.75 −3.37 2.85 .63
C 95.55 ± 3.54 −4.45 94.41 ± 3.50 −5.59 3.70 .71
D 96.39 ± 1.67 −3.61 92.90 ± 1.61 −7.10 1.74 .06
Chocolate I 96.78 ± 0.86 −3.22 100.53 ± 0.90 0.53 0.89 .01
II 98.22 ± 3.02 −1.78 102.03 ± 3.13 2.03 3.07 .20
III 101.28 ± 2.96 1.28 101.35 ± 2.96 1.35 2.92 .97
IV 96.45 ± 2.13 −3.55 96.52 ± 2.13 −3.48 2.21 .98
a
Chocolate I and II were spiked with 0.59 wt % and 0.27 wt% of Palsgaard PGPR 4150, whereas chocolate III and Ⅳ were spiked with 0.59 wt % and 0.25 wt% of
Palsgaard PGPR 4125, respectively.
b
the p-value originates from a t-test to evaluate significant differences between the actual recovery and practical recovery.
FOOD ADDITIVES & CONTAMINANTS: PART A 11
product, and hence the actual ricinoleic acid content
of the PGPR used, is not known, the recovery
(referred to as ‘practical recovery’) was recalculated
considering the observed average ricinoleic acid con-
tent of all types of PGPR analysed in this study.
These practical recoveries also showed a good
repeatability and fell within the legal requirements.
Moreover, the observed difference between the
actual and practical recovery for each sample was
very small (< 4%). In fact, no significant differences
between the actual and practical recoveries were
obtained, except for chocolate I (p-value = 0.01). In
conclusion, the systematic error introduced by
assuming an overall ricinoleic acid content of
PGPR of 86.63 wt%, irrespective of the specific
PGPR type actually present in the analysed sample,
is small and can be considered as acceptable.
Whereas the validation was done on only two
types of food products, we expect that this
method can be also used for other relevant
food matrices, such as sauces, particular confec-
tionary products, and cocoa products. Thus,
a quantitative determination method of the
PGPR concentration in a variety of relevant
foods as shown in Figure 6 can be proposed.
The actual sensitivity for PGPR depends on the
content of ricinoleic acid in PGPR and the
amount of food product analysed: the average
content of ricinoleic acid in PGPR (i.e.
86.63 ± 1.98 wt%) and the amount of food
product proposed in the method (i.e. 20 mg)
resulted in a LOD and LOQ of 0.76 and
2.32 μg PGPR per mg food product, respectively,
based on the protocol as shown in Figure 6. If
needed, the sensitivity can be further increased
by using all FAME for SPE (instead of only 2
out of 3 ml, hence 50% increase in sensitivity)
and using a higher amount of the polar FAME
fraction for silylation (instead of only 100 µl out
of 2 ml, hence up to 20 times increase in
sensitivity).
Conclusion
In conclusion, an accurate and precise method for
quantifying the PGPR concentration was successfully
developed based on determination of the ricinoleic
acid content via GC-FID after saponification of the
total sample, methylation, fractionation and silylation.
The results obtained showed that the average content
of ricinoleic acid in a representative set of commercial
PGPR products was 86.63 ± 2.0 wt%. Furthermore, the
Figure 6. Proposed analytical scheme for the determination of the ricinoleic acid content in commercial food products.
12 C. SU ET AL.
high PGPR recoveries from the three different food
matrices considered indicated that the developed
method was applicable in relevant foods with
a varying fat content.
Acknowledgments
We would like to thank Nathalie De Muer, Margot
Vansteenland, and An Maes for technical assistance, and
Vandemoortele for providing the commercial low-fat spread
products.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
Chunxia Su was supported by the China Scholarship Council
(CSC 201808420300).
References
[AOCS] American Oil Chemists’ Society Official Method Ce
1b-89. 1989. Official method for marine oil fatty acid com-
position by GLC. In: Official methods and recommended
practices of the American Oil Chemists’ society. 4th ed.
Campaign (IL): AOCS Press.
Atik DS, Bölük E, Toker OS, Palabiyik I, Konar N. 2020.
Investigating the effects of Lecithin-PGPR mixture on phy-
sical properties of milk chocolate. LWT-Food Sci Technol.
129:109548. doi:10.1016/j.lwt.2020.109548.
Balcaen M, Vermeir L, Declerck A, Van der Meeren P. 2016.
Influence of internal water phase gelation on the shear- and
osmotic sensitivity of W/O/W-type double emulsions. Food
Hydrocoll. 58:356–363. doi:10.1016/j.foodhyd.2016.03.011.
Balcaen M, Vermeir L, Van der Meeren P. 2017. Influence of
protein type on polyglycerol polyricinoleate replacement in
W/O/W (water-in-oil-in-water) double emulsions for food
applications. Colloid Surf A. 535:105–113. doi:10.1016/j.
colsurfa.2017.09.034.
Bastida-Rodríguez J. 2013. The food additive polyglycerol
polyricinoleate (E-476): structure, applications, and pro-
duction methods. Int Sch Res Notices. 2013:124767.
Clegg SM, Moore AK, Jones SA. 1996. Low-fat margarine
spreads as affected by aqueous phase hydrocolloids. J Food
Sci. 61(5):1073–1079. doi:10.1111/j.1365-2621.1996.
tb10935.x.
Da Cunha KM, Quast LB, Luccas V. 2010. Influence of the
addition of soy lecithin and polyricinoleate polyglycerol on
chocolate rheology. Food Process Res Center Bull. 28
(2):321–330.
De Meulenaer B, Vanhoutte B, Huyghebaert A. 2000.
Development of chromatographic method for the determi-
nation of degree of polymerisation of polyglycerols and
polyglycerol fatty acid esters. Chromatographia. 51(1–
2):44–52. doi:10.1007/BF02490694.
Dedinaite A, Campbell B. 2000. Interactions between mica
surfaces across triglyceride solution containing phospholi-
pid and polyglycerol polyricinoleate. Langmuir. 16
(5):2248–2253. doi:10.1021/la991018u.
[EC] European Commission. 2002. Commission Decision
2002/657/EC of 12 August 2002 implementing Council
Directive 96/23/EC concerning the performance of analyti-
cal methods and the interpretation of results. Off J Eur
Commun. 50:8–36.
[EC] European Commission. 2011. Commission Regulation
(EU) No 1129/2011 of 11 November 2011 amending Annex
II to Regulation (EC) No 1333/2008 of the European
Parliament and of the Council by establishing a Union list
of food additives. Off J Eur Commun. L295/1. 45:131–307.
Goodrum JW, Geller DP. 2005. Influence of fatty acid methyl
esters from hydroxylated vegetable oils on diesel fuel
lubricity. Bioresource Technol. 96(7):851–855.
doi:10.1016/j.biortech.2004.07.006.
[ICHHT] International Conference on Harmonisation
Harmonised Tripartite Guideline. 2005. Validation of ana-
lytical procedures: text and methodology. Q2 (R1). 1(20):05.
Knothe G, Cermak SC, Evangelista RL. 2012. Methyl esters
from vegetable oils with hydroxy fatty acids: comparison of
lesquerella and castor methyl esters. Fuel. 96:535–540.
doi:10.1016/j.fuel.2012.01.012.
Lonchampt P, Hartel RW. 2004. Fat bloom in chocolate and
compound coatings. Eur J Lipid Sci Tech. 106(4):241–274.
doi:10.1002/ejlt.200400938.
Marmesat S, Velasco J, Dobarganes MC. 2008. Quantitative
determination of epoxy acids, keto acids and hydroxy acids
formed in fats and oils at frying temperatures. J Chromatogr
A. 1211(1–2):129–134. doi:10.1016/j.chroma.2008.09.077.
Mortensen A, Aguilar F, Crebelli R, Di Domenico A,
Dusemund B, Frutos MJ, Galtier P, Gott D, Gundert-
Remy U, Leblanc JC; EFSA Panel on Food Additives and
Nutrient Sources added to Food (ANS), et al. 2017. Re-
evaluation of polyglycerol polyricinoleate (E 476) as
a food additive. EFSA J. 15(3):e04743.
Mubiru E, Shrestha K, Papastergiadis A, De Meulenaer B.
2013. Improved gas chromatography-flame ionization
detector analytical method for the analysis of epoxy fatty
acids. J Chromatogr A. 1318:217–225. doi:10.1016/j.
chroma.2013.10.025.
Mubiru E, Shrestha K, Papastergiadis A, De Meulenaer B.
2014. Development and validation of a gas chromatogra-
phy–flame ionization detection method for the determina-
tion of epoxy fatty acids in food matrices. J Agric Food
Chem. 62(13):2982–2988.
Naik B, Kumar V. 2014. Cocoa butter and its alternatives-a
review. J Bioresource Eng Technol. 2:1–11.
Okuro PK, Gomes A, Costa ALR, Adame MA, Cunha RL.
2019. Formation and stability of W/O-high internal phase
FOOD ADDITIVES & CONTAMINANTS: PART A 13
emulsions (HIPEs) and derived O/W emulsions stabilized
by PGPR and lecithin. Food Res Int. 122:252–262.
Orfanakis A, Hatzakis E, Kanaki K, Pergantis SA, Rizos A,
Dais P. 2013. Characterization of polyglycerol polyricinole-
ate formulations using NMR spectroscopy, mass spectro-
metry and dynamic light scattering. J Am Oil Chem Soc. 90
(1):39–51.
Peschar R, Pop MM, De Ridder DJ, van Mechelen JB,
Driessen RA, Schenk H. 2004. Crystal structures of 1, 3-dis-
tearoyl-2-oleoylglycerol and cocoa butter in the β (V) phase
reveal the driving force behind the occurrence of fat bloom
on chocolate. J Phys Chem B. 108(40):15450–15453.
Rebry F, Nelis V, Moens K, Dewettinck K, Van der Meeren P.
2020. Production of reduced-fat whipped toppings by solid
fat-based W/O/W double emulsions: proof of concept.
Int J Food Sci Tech. 55(5):1950–1961.
Schantz B, Rohm H. 2005. Influence of lecithin–PGPR blends
on the rheological properties of chocolate. LWT-Food Sci
Technol. 38(1):41–45.
Schenk H, Peschar R. 2004. Understanding the structure of
chocolate. Radiat Phys Chem. 71(3–4):829–835.
Scholz V, Da Silva JN. 2008. Prospects and risks of the use of
castor oil as a fuel. Biomass Bioenergy. 32(2):95–100.
Singh D, Chopra A, Kumar R, Sastry MIS, Patel MB,
Basu B. 2014. Response factor correction for estimation
of ester content in biodiesel. Chromatographia. 77(1–
2):165–169.
Tonta MM, Aydemir Sezer U, Olmez H, Gurek AG, Sezer S.
2019. Cost-effective synthesis of polyricinoleate: investiga-
tion of coating characteristics, in vitro degradation, and
antibacterial activity. J Appl Polym Sci. 136(44):48172.
Van Loco J, Elskens M, Croux C, Beernaert H. 2002. Linearity
of calibration curves: use and misuse of the correlation
coefficient. Accredit Qual Assur. 7(7):281–285.
Vermeir L, Sabatino P, Balcaen M, Declerck A, Dewettinck K,
Martins JC, Van der Meeren P. 2016. Effect of molecular
exchange on water droplet size analysis in W/O emulsions
as determined by diffusion NMR. J Colloid Interf Sci.
463:128–136.
Wilson R, Van Schie BJ, Howes D. 1998. Overview of the
preparation use and biological studies on polyglycerol poly-
ricinoleate (PGPR). Food Chem Toxicol. 36(9–10):711–718.
Zhu Q, Wang C, Khalid N, Qiu S, Yin L. 2017. Effect of protein
molecules and MgCl2 in the water phase on the dilational
rheology of polyglycerol polyricinoleate molecules adsorbed
at the soy oil-water interface. Food Hydrocoll. 73:194–202.
14 C. SU ET AL.
