Food Chemistry 211 (2016) 654660

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Determination of furan levels in commercial orange juice products and

its correlation to the sensory and quality characteristics
Mina K. Kim, Min Yeop Kim, Kwang-Geun Lee
Department of Food Science and Biotechnology, Dongguk University-Seoul, 32, Dongguk-ro, Ilsandong-gu, Goyang-si, Gyeonggi-do, Republic of Korea

a r t i c l e

i n f o

Article history:
Received 16 March 2016
Received in revised form 16 May 2016
Accepted 16 May 2016
Available online 17 May 2016
Keywords:
Orange juice
Furan
Vitamin C
Sensory analysis

a b s t r a c t
The objective of current study was to determine the furan levels in commercial orange juices (OJs) and
relate to the sensory and quality characteristics of OJs. The factors among sensory and quality characteristics that showed high correlation to furan were identified. The furan levels found in 18 commercial OJs
ranged from 0.59 to 27.39 ng/mL. Freshly-squeezed type OJs (n = 4) had significantly lower furan levels
(4.68 ng/mL) than other OJs treated with heat processing (p < 0.05). Vitamin C content, specifically,
dehydroascorbic acid showed higher correlation to the furan level in OJs (r = 0.833). A descriptive sensory
analysis result revealed the different flavor profile of commercial OJs according to the processing method
and added ingredient in OJs. Current approach of using sensory analysis for prediction of furan level in
food products can be applied to future studies in many other food commodities.
2016 Elsevier Ltd. All rights reserved.

1. Introduction
Furan is a colorless, heterocyclic aromatic compound (C4H4O)
and is highly volatile. The International Agency for Research on
Cancer classified this compound as Group 2B, a potential human
carcinogen (Kromhout, 1995). According to the National Toxicology
Program, furan is hepatoxic and results in cancer in mice and rats
(NTP, 1993). Numerous studies have previously been conducted
to determine the formation mechanisms of furan in food products.
These have shown that the furan formation is not limited to one
specific type of processing, and a variety of precursors exist, including amino acids, carbohydrates, polyunsaturated fatty acids, ascorbic acids, and carotenoids (Anese & Suman, 2013; Mariotti et al.,
2012; Van Lancker, Adams, Owczarek-Fendor, Meulenaer, & De
Kimpe, 2011; Kim, Her, Kim, & Lee, 2015) To date, thermal processing has been identified as the most likely cause in commercial food
products, such as canned and jarred foods (4.948.5 ng/mL), coffee
(67.81476 ng/mL), soy sauce (44.32178 ng/mL), and baby foods
(Kim, Lee, Kim, Park, & Lee, 2009; Kim, Kim, & Lee, 2010; Sijia,
Enting, & Yuan, 2014; Zoller, Sager, & Reinhard, 2007; Feng et al.,
2013; Limacher, Kerler, Davidek, Schmalzried, & Blank, 2008).
Orange juice (OJ) is a highly popular fruit juice, due to its unique
taste and nutritional benefits, which include high vitamin C content
(Kim, Lee, Kwak, & Kang, 2013). In 2012, South Korea was the 13th
highest OJ-consuming country worldwide, consuming 37 metric
Corresponding author.
E-mail address: kwglee@dongguk.edu (K.-G. Lee).
http://dx.doi.org/10.1016/j.foodchem.2016.05.114
0308-8146/ 2016 Elsevier Ltd. All rights reserved.

tons, which equates to a market share of approximately $240 million (CITRUSBR. Brazilian association of citrus exporters., 2015;
Gil-Izquierdo, Gil, & Ferreres, 2002). Commercial OJ-processing
involves squeezing fresh orange, followed by a centrifuging step,
and subsequent heat processing. The typical pasteurization temperature of fresh juice is set at 75 C for 30 s, whereas frozen, concentrated juice is pasteurized at 95 C for 30 s, with a subsequent
freezing and thawing process (Gil-Izquierdo et al., 2002). The sensory characteristics and physiochemical properties of OJ can be
affected by different thermal processing conditions (Gil-Izquierdo
et al., 2002; Kim et al., 2013; Bettini, Shaw, & Lancas, 1998).
Although the majority of commercial OJs undergo the thermal
processing step, limited research has been conducted to investigate
the levels of furan in OJ products. A recent study (Wegener &
Lopez-Sanchez, 2010) evaluated the levels of furan in fruit and vegetable juices, but OJ was not included in the fruit juice category.
Only one previous study has reported the furan levels in commercial OJ products, and the average amount of furan found ranged
from 3.33 to 7.59 ng/mL (Kim et al., 2010). Since commercial OJ
products undergo heat treatment, and contain a relatively high
level of vitamin C, one of the known precursors for furan formation,
a high level of furan is to be expected. To date, very few studies
have investigated furan levels in OJ products and related the result
to sensory and quality properties. In the present study, the furan
levels were determined in 18 commercial OJ products. The descriptive sensory analysis and quality characteristics, including vitamin
C content, were further investigated to determine the factors
affecting the formation of furan.

655

M.K. Kim et al. / Food Chemistry 211 (2016) 654660

2. Materials and methods

2.3. Quality characteristics of commercial OJ samples

2.1. OJ samples for testing

The pH of the 18 OJ samples was determined using a pH meter

(S20 SevenEasyTM pH meter, Mettler Toledo GmbH 8603, Switzerland). The sugar content was measured by a Refractometer
(Master-T series, Cat. No. 2312, Atago , Japan), with a range of
Brix from 0.0 to 33.0%. The color of the OJ products was measured
by a color-difference meter (NE 4000, Nippon Denshoku, Tokyo,
Japan), according to the methods specified in the equipment manual. The results were reported as L (lightness), a (redness), and b
p
(yellowness) values, and 4E (the formula of the 4E = [(DL)2 +

Eighteen OJ samples were selected on the basis of their market

share and processing types (Table 1). The selected OJs are readily
available in grocery stores in Korea, and are widely consumed by
Korean people. Of the 18 samples, 13 were made via diluted from
frozen concentrate (DFC)processing, and four were made by
freshly-squeezed (FS-R) processing. Seven of the 13 DFC juices
required refrigeration during storage (DFC-R; samples 17), and
six required room temperature storage (DFC-RT; samples 813).
The four FS-R OJs had not been thermally processed, so they had a
short shelf-life (samples 1417). One OJ sample was prepared in
the lab by squeezing an orange (sample 18), and this served as a control. A detailed description of each OJ can be found in Table 2. All OJs
were purchased at a local grocery store, with the exception of sample 18, and were stored at 4 C until further analysis was carried out.
2.2. Chemical reagents and materials
Furan and d4-furan with purity higher than 99% were purchased from the Sigma-Aldrich Corporation (St. Louis, MO, USA).
Water and methanol (high-performance liquid chromatography
[HPLC] grade) were bought from J. T. Baker (Phillipsburg, NJ,
USA) for the analysis of furan concentration. Chemical reagents
for measuring the vitamin C content of commercial OJs were purchased from Sigma-Aldrich Corporation (St. Louis, MO, USA), that
included metaphosphoric acid (65.0% purity), 2,4-dinitrophenyl
hydrazine (97.0% purity), and sulfuric acid (99.9% purity), thiourea
(99.0% purity) and L-ascorbic acid (Vitamin C, 99.0% purity) and 2,
6- dichloroindophenol sodium salt hydrate (25 g, indicator grade).

(4a)2 + (4b)2] was calculated.

2.4. Determination of Vitamin C content of commercial orange juice
samples
A total of 5 mL of OJ and the same amount of 5% metaphosphoric
acid solution were mixed. The 5% metaphosphoric acid solution was
prepared by mixing 15 g of metaphosphoric acid (HPO3, 35% purity)
and 40 mL of citric acid in 100 mL of water, after which the volume
was increased by adding distilled water in a 250 mL volumetric
flask. Once thoroughly mixed, 5 mL were aliquoted into a 100 mL
volumetric flask and increased to 100 mL via the addition of weak
metaphosphoric acid solution for extraction processing. Following
the extraction, the extraction solution was filtered using filter paper
(Whatman No. 2) ahead of vitamin C analysis. The 4-dinitrophenyl
hydrazine (DNP) method was chosen for this analysis, in order to
gauge the overall vitamin C content, including total ascorbic acid,
ascorbic acid, and dehydroascorbic acid. The procedure for manufacturing via the DNP method was used in accordance with the Korean Food Standards Codex for analyzing vitamin C content (MFDS,
2012) and involves oxidation, formation of osazone, and dissolution

Table 1
Sample description and quality characteristics of 18 commercial OJs included in this study.
Sample

No 1.
No 2.

Storage
condition

DFC-R

No 3.
No 4.
No 5.
No 6.
No 7.
No 8.

DFC-RT

No 9.
No 10.
No 11.
No 12.
No 13.
No
No
No
No
No

14.
15.
16.
17.
18.

FS-R

Ingredients

Quality characteristics
Sugar
Content
(%, Brix)

Color measurements
L

4.46 0.02a
4.43 0.06a

10.27 0.12fg
10.00 0.00h

61.07
54.58

3.13
2.79

44.85
43.37

0
6.27

4.38 0.01ab
4.32 0.05b

10.13 0.12gh
11.80 0.00b

57.02
53.87

3.35
2.52

41.39
47.76

5.33
7.79

4.12 0.05ef

10.47 0.12de

52.23

3.68

44.16

8.89

4.18 0.01e

10.73 0.12c

49.15

0.17

33.54

16.76

fg

13.2 0.20

62.01

3.51

47.58

2.91

4.20 0.01de

10.6 0.00cd

47.68

2.55

31.85

18.67

4.12 0.02ef

10.4 0.00ef

52.19

2.17

48.13

10.84

4.29 0.02bcd

10.73 0.12c

52.77

1.99

36.95

11.52

4.30 0.01bc

8.73 0.12k

52.68

4.13

36.03

12.21

cde

gh

53.01

1.11

35.99

12.15

3.97 0.06g

10.53 0.12de

51.99

1.47

35.79

12.94

3.83 0.12h
4.39 0.08ab
4.37 0.09ab
3.99 0.11g
3.88 0.06

9.07 0.12j
10.53 0.12de
10.47 0.12de
9.80 0.00i
9.73 0.12

49.49
53.29
53.14
50.26
43.84

2.34
1.03
2.78
0.28
0.74

52.85
42.98
34.83
48.32
43.62

15.10
8.27
12.78
11.71
17.44

pH

Concentrated orange juice (100%), Purified water

Concentrated orange juice, Purified water, Natural flavoring substances
(orange)
Orange squash (100%), Purified water, Vitamin C
Concentrated orange juice (organic), Natural flavoring substance (orange),
Purified water
Concentrated orange juice (100%), Synthesized flavoring substances (orange),
Purified water
Concentrated orange juice (100%), Orange juice extraction, Natural flavoring
substances (orange), purified water
Orange squash (100%), Purified water, Natural flavoring substances (orange)
Orange concentrate, Purified water, Natural flavoring substances (orange),
Vitamin C
Concentrated orange juice, Purified water, High fructose corn syrup, Vitamin
C, White sugar, Calcium gluconate, Calcium lactate, maize starch, Citric acid
Concentrated orange juice, Orange juice extraction, Purified water, Natural
flavoring substances (orange)
Concentrated orange juice, Purified water, High fructose corn syrup, Calcium
lactate, Citric acid, Vitamin C
Concentrated orange juice, Purified water, Natural flavoring substances
(orange)
Orange squash concentrate, Purified water, High fructose corn syrup, milk,
Synthesized flavoring substances (orange), Citric acid, Vitamin C
Orange
Orange
Orange
Orange
Orange

(100%)
squash (92%), Orange flesh (8%)
squash (98.3%), Orange flesh (1.7%)
(100%)
(100%)

4.02 0.03

4.21 0.02

10.13 0.12

DE

Numbers in a column followed by different letters represents significant difference at p < 0.05 using Duncans multiple range tests DFC-R represents diluted from frozen
concentrate-Refrigerated storage, DFC-RT represents Diluted from frozen concentrate-room temperature storage, FS-R represents freshly squeezed-refrigerated storage.

656

M.K. Kim et al. / Food Chemistry 211 (2016) 654660

Table 2
Vitamin C analysis results of 18 commercial OJ samples using 2,4 dinitrophenyl hydrazine (DNP) method.
Sample

Vitamin C (mg/100 mL)

Total ascorbic acid

Ascorbic acid

Dehydroascorbic acid

Conc. (mg/100 g)

Group Avg.

Conc. (mg/100 g)

Group Avg.

Conc. (mg/100 g)

Group Avg.

No
No
No
No
No
No
No

1.
2.
3.
4.
5.
6.
7.

100.31 0.88ef
119.07 1.87c
119.34 3.45c
100.45 4.93ef
94.70 2.10g
105.78 3.19de
93.57 2.80gh

84.90

62.71 2.95i
59.34 1.52j
32.55 0.49l
76.60 4.27gh
73.37 2.86h
80.25 1.29f
74.78 0.88h

65.66

37.60 2.96c
59.73 3.39b
86.79 3.21a
23.85 6.52efg
21.32 1.06fg
25.53 1.94def
18.80 2.08g

19.24

No
No
No
No
No
No

8.
9.
10.
11.
12.
13.

108.73 5.49d
96.94 2.80fg
108.44 0.46d
54.92 0.47i
89.09 2.57h
102.69 2.95e

75.45

79.69 2.71fg
74.21 1.2h
68.88 3.58i
43.56 0.82k
66.22 1.06h
76.32 1.35gh

68.15

29.04 3.74de
22.73 2.34fg
39.56 3.22c
11.36 0.95h
22.87 2.32fg
26.37 1.75def

7.30

No
No
No
No
No

14.
15.
16.
17.
18.

133.98 1.48b
105.74 5.61de
136.21 0.55b
122.47 3.84c
144.50 1.35a

98.83

96.52 1.35cd
93.43 0.99de
98.80 0.27b
92.70 0.96e
139.31 0.84a

90.12

37.46 2.23c
12.31 5.38h
37.41 0.47c
29.78 2.98d
5.19 0.64i

8.71

Numbers represent the mean standard deviation of triplicate analyses of vitamin C content.
Numbers in a column followed by different letters represents significant difference at p < 0.05 using Duncans multiple range tests.
Group average values were calculated for each Vitamin (total ascorbic acid, ascorbic acid, and dehydroascorbic acid) within the orange juice category according to its
processing method (DFC-R, DFC-RT, FS-R).
Total ascorbic acid, ascorbic acid, and dehydroascorbic acid contents were analyzed using a 2,4-dinitrophenylhydrazine (DNP) method.

of osazone. The vitamin C content of each sample was calculated

using the formula described in the Korean Food Standard Codex
(MFDS, 2012) which calculated the type of reduction of vitamin C
(ascorbic acid) content from the difference between the total vitamin C content and oxidation of vitamin C presented in the original.
The calibration curve for measuring vitamin C content was created
using 0.5, 1, 2, 3, 4, and 5 mg/100 mL using the ascorbic acid solution. The drawing of a standard vitamin C curve was progressed
in the same way using the DNP method. Analysis was conducted
using the same method as for vitamin C analysis, and it was also
performed in triplicate.
2.4.1. Furan analysis of commercial OJ samples
Furan analysis methods, including stock, working solutions, and
validation methods used, were similar to those that have previously been published (Kim et al., 2015, and Kim et al., 2010). In
brief, stock solutions of furan and d4-furan were prepared in
20 mL headspace vials by adding a 1 lL aliquot of native furan
and d-4 furan to cold methanol. Working solutions were prepared
daily, using 100 lL refrigerated stock solutions and 10 mL HPLCgrade water. The prepared stock and working solutions were
sealed with silicone-PTFE septa and aluminum seals and stored
at -18 C and 4 C, respectively. The furan validation parameters
were as follows: linearity, correlation coefficient (r2), repeatability
(water and orange juices, respectively), limit of detection (LOD),
limit of quantification (LOQ), and percentage recovery.
2.4.2. Orange juice samples preparation
The OJ samples were prepared according to the liquid sample
analysis method used in previous studies (Kim et al., 2015; Sijia
et al., 2014). In brief, 5 mL OJ sample and 5 mL HPLC-grade water
were mixed in a 20 mL headspace vial (Agilent Santa Clara, CA,
USA). Vials containing the mixtures were sealed with siliconePTFE septa and aluminum seals and were stored at 4 C until analysis. Internal standard (10 ppb D-4 furan) was added to each vial
before solid-phase micro extraction (SPME) analysis.

2.4.3. Solid Phase Micro Extraction (SPME) followed by Gas

chromatography-mass spectrometry (GC/MS)
A carboxen/polydimethylsiloxane fiber (75 lm Supelco, Bellefonte, PA, USA) was selected for furan analysis, as suggested in previous studies. The SPME fiber was conditioned in the injector of an
Agilent Technologies 6890 N network gas chromatograph (GC) system (Agilent, Santa Clara, CA, USA) at 300 C for 1 h before optimizing furan analysis. The fiber was exposed to the headspace at a
depth of 1.5 cm at an optimized extraction temperature (50 C)
for 20 min. After extraction, the fiber was separated from the analysis vials and inserted into the injection port at a depth of 4 cm for
the analyte desorption. In order to minimize/eliminate any residual analytes, the fiber was baked for 15 min at 250 C until the next
sample analysis.
Furan analysis was carried out using the GC system coupled to
an Agilent Technologies 5977 inert mass selective detector (Agilent, Santa Clara, CA, USA), and the columns used were silica capillary columns (15 m  0.32 mm I.D., 20 lm film thickness, J & W
Scientific, Folsom, CA, USA). The initial temperature of the GC oven
was set at 50 C for 5 min, and was then increased at a speed of
25 C/min until it reached 230 C, where it was held for 2 min.
Total run time was 14.2 min, and the post run was set at 230 C
for 10 min. High-purity helium was used for carrier gas, and a flow
rate was set at 1.5 mL/min under the constant flow mode. The
injection port was heated at 250 C under the splitless injection
mode. The mass spectrometer was progressed positive electron
impact ionization mode (EI+, 70 eV) and selective-ion monitoring
was represented as m/z 68 [M]+ as a quantifier and m/z 39 [MCHO]+ as a qualifier for furan. When applied in the same way to
d-4 furan, m/z 72 [M]+ was a quantifier and m/z 42 [M-CHO]+
was a qualifier.
2.5. Descriptive sensory analysis of OJ samples
A trained descriptive panel evaluated the 18 OJ samples. This
panel consisted of six highly trained individuals (four females
and two males, aged 2336 years), each of whom had greater than

657

M.K. Kim et al. / Food Chemistry 211 (2016) 654660

100 h of experience in evaluation of OJ flavor via the SpectrumTM

method, using an established OJ flavor language that had been
developed in previous research (Kim et al., 2013). A sensory lexicon
and references for each term can be found in Table 5. The panelists
were trained on flavor attributes during five 1 h calibration sessions before the actual evaluation. On a given day, the panelists
were presented with OJ samples in 210 mL clear plastic cups (made
of a polypropylene material, cleanwrap, Ghimhae, South Korea)
labeled with a 3-digit random code, along with sensory references
for flavor attributes. The flavor profile of the OJs was created by
each panelist (n = 6) in triplicate in a randomized balanced design.
All samples were stored at 4 C until the day of sample evaluation,
and were served to panelists at 8 C. During the evaluation, the
panelists were provided with water for palate cleansing, and a period of 2 min rest was enforced between samples, in order to minimize the carry-over effect. The panelists entered their responses
using a 15-pt universal scale, and a paper ballot was utilized for
data collection. All panel participants were invited to an appreciation dinner as compensation for their time.

Group average of FS-R (sample 1418) for total ascorbic acid was
98.83 mg/100 mL of OJ. Of the DFC juices (samples 113), DFC-R
(samples 17) had a higher total ascorbic acid content than DFCRT (samples 813). The group means of DFC-R and DFC-RT for total
ascorbic acid were 84.90 mg/100 mL (samples 17) and
75.45 mg/100 mL (samples 813), respectively. The amount of
dehydroascorbic acid showed an inverse relationship to the
amount of total ascorbic acid, in that the level of dehydroascorbic
acid was lowest in freshly-squeezed OJ (sample 18). Previous studies analyzing vitamin C in orange products have shown that dehydroascorbic acid gradually increases while total ascorbic acid
decreases during the extended storage of the products (Yaylayan,
2006). This explains the high total ascorbic acid content and lowest
dehydroascorbic acid content of sample 18, which was analyzed
immediately after the orange was squeezed. Interestingly, the total
ascorbic acid content of OJs with fortified vitamin C (samples 3, 8,
9, 11, and 13) was still significantly lower than that found in the
FS-R OJs (samples 1417).
3.2. Analysis of Furan in 18 orange juice samples

2.6. Statistical analysis

All data analyses, including quantification of furan, quality characteristics (pH, sugar contents, color parameters), and vitamin C
content, were conducted using IBM SPSS Statistics (ver. 21.0,
Armonk, NY, USA). Analysis of variance, followed by Duncans multiple comparison test, was carried out at the a = 0.05 level. All
results are presented as averages standard deviations (SD) from
the triplicate measurements of analyses. Principal component
analysis (PCA) was conducted to determine how the OJ samples
were located in the quality and sensory characteristics map, and
XLSTAT (v.2014, Addinsoft, Paris, France).
3. Results and discussion
3.1. Quality characteristics of 18 orange juices
3.1.1. Quality characteristics of 18 OJ samples
Quality characteristics of the OJ samples including pH, sugar
content in% (Brix), and color measurements, can be found in
Table 1. With regard to the pH conditions, seven of the 13 DFC-R
OJs (samples 17), had values ranging from 4.02 to 4.46. Sample
7 had the lowest (4.02) and Sample 1 had highest (4.46) pH value.
Six of the 13 DFC-RT OJs (samples 813) had pH values ranging
from 3.97 (sample 13) to 4.30 (sample 11). Five FS-R OJs (samples
1418) had pH values ranging from 3.83 to 4.39, and juices with
100% orange (samples 14, 17, and 18) had lower pH values than
the juices with pulp (samples 15 and 16). The Brix value of the
OJ products ranged from 8.73 (sample 11) to 13.2 (sample 7) in
the DFC samples (samples 113). The FS-R samples (samples 14
18) had Brix values ranging from 9.07 to 10.53, and the OJs with
100% orange (samples 14, 17 and 18) had lower Brix values than
the OJs with added pulp (samples 15 and 16). With regard to the
color, the value of L(lightness) in sample 7 was higher than in
the other OJs, and sample 17 had the highest b(brightness) value
of all the OJ products.
3.1.2. Vitamin C contents of 18 OJ samples
The results of the vitamin C analysis can be found in Table 2. In
foods, vitamin C has been used as an indicator of quality change in
storage, processing, and distribution. FS-R OJs (samples 1418) had
significantly higher total ascorbic acid than any other types of OJs,
and the amounts ranged from 105.74 to 144.5 mg/100 mL of juice.
Freshly-squeezed OJ (sample 18), which was made in the lab, had
the highest total ascorbic acid content with 144.5 mg/100 mL.

3.2.1. Validation specification for the furan analysis

Table 3 shows the validation specification for the method of
furan analysis used in the present study. A calibration curve was
used to plot the peak area ratio of furan/d-4 furan. This involves
linearity, correlation coefficient, repeatability, LOD, LOQ, and
recovery. Six concentrations were analyzed in triplicate to create
the calibration curve. To examine specifications for the furan analysis, linearity was represented by the formula (y = 0.101X + 0.2329,
R2 = 0.9997). Repeatability in water was expressed at three concentrations (1, 50, and 100 ng/mL). A relative standard deviation (RSD,
%) was calculated in OJ from all of the categories: RFD-R, RFC-RT,
and FS-R. The RSD values were as follows: 9.50%, 10.54%, and
4.80% in water, with low, medium, and high concentrations,
9.50% (RFC-R), 12.67% (RFC-RT), and 11.47% (FS-R). The measured
recovery was 104.79 3.81% from the furan standard in water.
LOD and LOQ were based on signal-to-noise ratios of 3:1 and
10:1, (Sijia et al., 2014; Van Lancker et al., 2011) and the values
of LOD and LOQ were 0.079 ng/mL and 0.26 ng/mL, respectively.
3.2.2. Furan concentrations found in the 18 OJ samples
Table 4 shows the furan concentrations obtained from the 18
commercial OJ samples. Overall, the furan levels found ranged
from 0.59 to 27.39 ng/mL. These were slightly higher than those

Table 3
Validation specification of furan analysis.
Specification

Experimental value

Linearity

Y = 0.101X + 0.2329

Correlation coefficient (r2)

0.9997

Repeatability (RSD, %)
Water

Orange Juice

1 ng/mL
50 ng/mL
100 ng/mL
Refrigeration
(Diluted Conc.)
Room temperature
(Diluted Conc.)
Refrigeration
(Squeeze)

Limit of detection (LOD)a

Limit of quantification (LOQ)b

0.079 ng/mL
0.26 ng/mL

Recovery (%)c

104.79 3.81

7.76
10.54
4.80
9.50
12.67
11.47

LOD = 3  SD/Slope (ng/mL).

b
LOQ = 10  SD/Slope (ng/mL).
c
Recovery (%) = (Volume of furan in spiked standard solution  Volume of furan
in standard solution)  100/Volume of spiking furan.

658

M.K. Kim et al. / Food Chemistry 211 (2016) 654660

Table 4
Furan concentrations found in 18 OJ samples using SPME-GCMS.
Sample

Furan conc. (ng/mL)

Group Avg.

No.1
No.2
No.3
No.4
No.5
No.6
No.7

ef

7.39 0.21
27.39 1.24a
20.68 2.11b
2.90 0.19ij
0.59 0.11j
2.58 0.26ij
2.26 0.19ij

9.12

No.8
No.9
No.10
No.11
No.12
No.13

2.85 0.15ij
8.84 0.38e
13.06 0.69d
3.37 0.24ghij
17.39 3.00c
5.93 0.06efgh

8.57

No.14
No.15
No.16
No.17
No.18

4.41 0.19ghi
4.81 0.29fghi
6.29 0.48efg
3.21 0.51hij
1.18 0.16j

4.68

1.18

Means that does not share the same alphabetical letter represent the significant
differences at p < 0.05.

observed in the previous furan-monitoring study on heatprocessed foods consumed in Korea, in which furan levels in
canned OJ ranged from 3.33 to 7.59 ng/mL and 3.24 to 3.60 ng/
mL (Kim et al., 2015). When the furan levels of OJ with different
processing and storage requirements were directly compared,
FS-R OJs (samples 1417) had lower furan levels than the other
OJ groups (DFC-R and DFC-RT). The average furan level found in
FS-R OJs was 4.68 ng/mL, whereas DFC-R OJs (samples 17) had
an average level of 9.12 ng/mL, and DFC-RT OJs (samples 813)
contained an average of 8.57 ng/mL. A freshly squeezed OJ that
was prepared in the lab (sample 18) had the lowest level of furan
with 1.18 ng/mL. Samples 2 and 3 had higher furan concentrations
than the other OJ samples. It is worth noting that these two samples also had higher levels of dehydroascorbic acid. According to
previous study (Yaylayan, 2006), formation of furan is attributed
from the presence of glycolaldehyde and acetaldehyde during
heating process. For example, the formation mechanism of furan
with carbohydrate as a source of precursor included four carbon
precursors such as aldotetrose and 2-deoxy-aldotetrose, which in
turn, be converted into furan. Ascorbic acid also play a similar role
in terms of furan formation pathway. Quick oxidation and hydrolysis of ascorbic acid in food during formation pathway, can lead to
subsequent conversion to 2,3-diketogulonic acid (DKG), and this
DKG is converted to aldotetrose and finally form furan (Yaylayan,
2006). Based on previously reported formation mechanism of
furan, the presence of Vitamin C, previously known precursor of
furan, with subsequent heating process as part of sterilization
process, influences the formation of furan in OJs. The higher
concentrations of furan in DFC samples (DFC-R and DFC-RT) than
FS-R samples, may have been derived from oxidation and hydrolysis of ascorbic acid during heat treatment of OJs.

3.3. Descriptive sensory analysis of 18 OJ samples

Sensory attributes utilized in the flavor profile of the 18 OJ samples included eight flavor-related attributes and three basic tastes
(Table 5). These sensory lexicons were modified from the previously reported sensory lexicon for OJ products in Korea (Kim
et al., 2013). Among 11 sensory attributes, six were differentiating
for 18 commercial OJ samples, as follows: orange flesh, orange
peel, cooked orange, burnt, citrus, and medicinal flavors and a
sweet taste (p < 0.05). Sample 18, freshly-squeezed OJ, had a distinctively strong orange-like flavor characteristic, including being

high in orange flesh and orange peel flavor, which was significantly
higher than that of the other OJ products (p < 0.05). In addition,
sample 18 had greater sweet and sour basic taste intensities than
the other samples (p < 0.05), and had a relatively strong bitter
taste. It has previously been reported that the essential oil
squeezed from the outer layer of the orange can be a source of a
bitter taste in OJ products (Kimball, 1987) and a recent report
revealed a high correlation between bitter taste and higher pulp
(orange-peel flavor) content in freshly squeezed OJ (Kim et al.,
2013). However, the correlation coefficient of bitter taste and
orange peel flavor attribute was r = 0.522 in the present study,
which was not necessarily in accordance with these results. Medicinal flavor is defined as a characteristic aromatic that is associated
with cough syrup, and it has been reported as an off-flavor in OJ. A
previous study reported that 2-methyl-3-furanthiol is responsible
for this medicinal (vitamin-like) off-flavor (Bezman, Rouseff, &
Naim, 2001). In the present study, sample 17 had a significantly
higher medicinal flavor, as well as a bitter taste and strong orange
peel flavor (p < 0.05), than the other samples. As a result of an
open-ended question, the panelists noted a distinct vitamin-like
medicinal flavor, as well as a slight oxidized note, in sample 17.
OJs with added flavoring substances (samples 28, 10, 12, and
13) had greater intensities of orange-like flavor attributes, including orange flesh, orange peel, and cooked orange, and FS-R OJs
(samples 1417) had relatively lower intensities of orangerelated flavor (p < 0.05). A distinct flavor difference between two
groups of OJs has previously been reported, in that refrigerated
from-concentrate OJs had strong orange-like and candy-(Kim
et al., 2013; Lotong, Chambers, & Chambers, 2003). The present
study demonstrated that the distinct differences between
orange-like flavors were based on added orange flavoring substances, rather than on processing condition.
3.4. Correlating quality and sensory characteristics to the furan levels
in OJs
The PCA result showed the multivariate statistical analysis of the
18 OJ samples with a varying degree of quality and sensory characteristics, and provided an overview of how these samples were
located in quality and sensory space (Fig. 1). In addition, the quality
and sensory attribute that showed a high correlation with the level
of furan was identified in a PCA biplot. Fig. 1 explains 39.4% of the
total variability in the dataset, with PC1, the horizontal axis being
the primary axis, explaining 22.7% of variability, and PC2 the secondary axis, explaining 16.7% of total variability. All datasets,
including quality characteristics, descriptive sensory profile, and
concentration of furan found in all 18 OJ samples were loaded on
each PC. Vitamin C content was separated by PC1, in that both total
ascorbic acid and ascorbic acid were located on the positive PC1, and
dehydroascorbic acid was located on the negative PC1. On the negative PC1, furan was located near dehydroascorbic acid, which
means a stronger association between the formation of furan and
the presence of dehydroascorbic acid. In fact, correlation analysis
revealed a high correlation between dehydroascorbic acid and
furan, with r = 0.833. The correlation between total ascorbic acid
content and furan was relatively low (r = 0.080). Ascorbic acid has
been known to affect the browning reaction of OJ (Vranov,
Bednrikov, & Ciesarov, 2007). In addition, a previous study
reported the formation of furan precursors, including furfural and
2-furoic acid, by a browning reaction in the OJ and other juice model
systems (Yaylayan, 2006). As such, the L value of OJ was also
located close to furan and dehydroascorbic acid, on the negative
PC1 axis. The PC2 axis primarily separated sensory characteristics,
in that medicinal, burnt, and bitter sensory attributes were located
on the positive PC2 axis, and the orange flesh attribute was located
on the negative PC2 axis. Medicinal, bitter, burnt flavor attributes

Table 5
Descriptive sensory analysis result of 18 OJ samples using a trained panel (N = 6).

Grape
Fruit
Cooked

Burnt
Citrus
Yogurt

Medicinal
Sweet
Sour
Bitter

No. 1

No. 2

No. 3

No. 4

No. 5

No.
6

No. 7

No. 8

No. 9

No.
10

No.
11

No.
12

No.
13

No.
14

No.
15

No.
16

No.
17

No.
18

p-Value

Characteristic aromatics associated with orange

flesh Reference: freshly squeezed orange
Characteristic aromatics associated with inner
layer (white lining) of orange Reference: a slice of
orange peel
Characteristic aromatics associated with
grapefruits Reference: a slice of grapefruit
Orange
Characteristic aromatics associated with cooked
orange (Syrup-like flavor), Reference: canned
orange juice concentrate
Characteristic aromatics associated with burnt
woods Reference: burnt match
Characteristic aromatics associated with citrus,
fruits (that is, orange, tangerine, clementine)
Characteristic aromatics associated with plain
Yoplait (that is, acetaldehyde) Reference: Yoplait
plain yogurt
Characteristic aromatics associated with various
cough syrup Reference: cough syrup
Basic taste elicited by sugar
Basic taste elicited by acid
Basic taste associated with caffeine

2.5abc

2.8abc

2.8abc

2.8abc

3.3abc

3.5ab

2.7abc

3.3abc

2.3abc

2.7abc

3.3abc

3.0abc

2.6abc

2.1bc

3.3abc

2.0c

2.0c

3.7a

0.001

1.8ab

1.6ab

1.5ab

2.1ab

1.8ab

1.0b

2.2ab

1.7ab

1.3ab

1.0b

1.4ab

1.5ab

1.9ab

1.3ab

1.0b

1.3ab

3.0a

3.0a

0.018

1.6a

1.4a

1.1a

1.7a

2.1a

2.0a

1.8a

2.1a

0.9a

1.7a

0.6a

1.8a

1.1a

1.4a

2.3a

1.7a

1.5a

N/D

0.084

1.5ab

1.9ab

2.2ab

2.3ab

1.8ab

0.5b

0.5b

2.1ab

4.3a

4.2a

3.5ab

2.6ab

3.5a

1.8ab

1.4ab

2.2ab

2.0ab

2.0ab

0.000

N/D

N/D

N/D

N/D

N/D

N/D

N/D

N/D

1.3ab

N/D

0.8b

N/D

N/D

0.6b

0.8ab

2.2ab

2.0ab

N/D

<0.0001

1.6ab

1.6ab

2.3ab

2.1ab

2.4a

2.3a

2.0ab

2.3a

2.3a

2.5a

2.3a

2.2ab

2.1ab

1.8ab

3.3a

1.5ab

1.3ab

3.0a

N/D

0.5a

1.9a

1.6a

0.9a

1.5a

1.0a

0.8a

1.8a

2.3a

1.0a

1.9a

1.8a

1.5a

1.5a

2.5a

1.2a

2.0a

N/D

0.067

0.8ab

0.6b

0.5b

0.6b

N/D

1.0a

N/D

0.5b

0.6b

N/D

0.6b

0.7ab

0.9ab

0.9ab

1.0ab

N/D

2.0a

N/D

0.135

4.5bc
5.0a
0.7a

4.4bc
4.3a
0.5a

4.5bc
4.0a
N/D

4.3bc
4.5a
1.0a

4.0c
5.0a
1.0a

4.0c
4.5a
N/D

4.2c
4.6a
N/D

5.0abc
4.8a
0.5a

5.0abc
4.3a
0.3a

5.5ab
3.7a
N/D

5.3abc
4.6a
N/D

4.6bc
4.6a
0.8a

5.5bc
4.0a
1.0a

4.4bc
4.5a
0.4a

4.3bc
5.0a
0.5a

4.0c
4.8a
N/D

5.0abc
4.0a
1.5a

6.0a
3.5a
N/D

0.045
0.381
0.334

M.K. Kim et al. / Food Chemistry 211 (2016) 654660

Orange
Flesh
Orange
Peel

Definition1

Means in a row that does not share a same alphabetical letter represent significant differences (p < 0.05).
1
Sensory lexicon and definition was modified from Kim et al. (2013).

659

660

M.K. Kim et al. / Food Chemistry 211 (2016) 654660

PC2 (16.7%)

Yogurt

References

Medicinal

Citrus

Burnt
Bitter

No. 17

b*

a*

No. 14
No. 9

Dehydroascorb
Grape Fruitic acid
SourFuran No. 15
No. 2No. 12
PC1 (22.7 %)
L*
pH

Cooked
Orange

No. 16
No. 4
No. 3
No. 1No. 5 No. 6
No. 8
No. 7 No. 11 No. 10

Total ascorbic
acid
No. 13
Orange Peel
Ascorbic acid
Sweet

Brix
No. 18

Orange Flesh
Fig. 1. Principal Component Analysis biplot of 18 OJ products with different sensory
and quality characteristics and furan level. PC1 and PC2 explains a total of 39.4% of
variability of total dataset. Numbers followed by No. represents each orange juice
products. Words in italicized font represents the sensory attributes generated from
descriptive analysis of orange juices.

have previously been reported as off-flavor characteristics in OJ

(Kim et al., 2013). These off-flavor related sensory characteristics
are considered as oxidation-related flavor characteristics in OJs.
High in b value and these off-flavor related flavor attributes may
have been related to vitamin C (ascorbic acid) oxidation in OJs
(Ayhan, Yeom, Zhang, & Min, 2001). The sensory characteristics
showed a relative high correlation to the b value of color measurement, so the samples with greater yellowness may have had a stronger correlation with these off-flavor-like sensory characteristics.
Vitamin C content has been reported as a major contributor in
the formation of furan in food products. The present study showed
a high correlation of dehydroascorbic acid with high furan level in
OJs. In addition, L, which indicates a darker color, attributed to the
Maillard reaction, showed a correlation with the level of furan in
OJ. The present study confirmed the previous findings on the formation of furan and possible contributors to furan formation. In
addition, the current study demonstrated a new approach to the
prediction of furan levels in OJ by using sensory characteristics.
This method of using sensory analysis to predict the amount of a
potentially toxic compound in OJ can be applied to numerous other
food commodities in future studies.
Acknowledgement
This research was supported by basic Science Research Program
through the National Research Foundation of Korea (NRF,
NRF-2014R1A1A1002093), and further funded by basic Science
Research Program through the National Research Foundation of
Korea (NRF-2015R1A2A2A01005772), both funded by the Ministry
of Science, ICT & Future Planning.

Anese, M., & Suman, M. (2013). Mitigation strategies of furan and 5hydroxymethylfurfural in food. Food Research International, 51(1), 257264.
Ayhan, Z., Yeom, H. W., Zhang, Q. H., & Min, D. B. (2001). Flavor, color, and vitamin C
retention of pulsed electric filed processed orange juice in different packaging
materials. Journal of Agriculture and Food Chemistry, 49, 669674.
Bettini, M. F. M., Shaw, P. E., & Lancas, F. M. (1998). Sensory and analytical
evaluations of Brazilian orange juice and aromas. Fruit Processing, 8, 283289.
Bezman, Y., Rouseff, R. L., & Naim, M. (2001). 2-Methyl-3-furanthiol and methional
are possible off-flavors in stored orange juice: Aroma-similarity, NIF/SNIF GC-O,
and GC analyses. Journal of Agricultural and Food Chemistry, 49(11), 54255432.
CITRUSBR. Brazilian association of citrus exporters. (2015). http://www.
citrusbr.com/en/citric-exporters/consumption/consumption-statistics2352061.asp. (Accessed on March 2015).
Feng, Y., Cui, C., Zhao, H., Gao, X., Zhao, M., & Sun, W. (2013). Effect of koji
fermentation on generation of volatile compounds in soy sauce production.
International Journal of Food Science & Technology, 48(3), 609619.
Gil-Izquierdo, A., Gil, M. I., & Ferreres, F. (2002). Effect of processing techniques at
industrial scale on orange juice antioxidant and beneficial health compounds.
Journal of Agricultural and Food Chemistry, 50(18), 51075114.
Kim, M. Y., Her, J. Y., Kim, M. K., & Lee, K. G. (2015). Formation and reduction of furan
in a soy sauce model system. Food Chemistry, 189, 114119.
Kim, T. K., Kim, S., & Lee, K. G. (2010). Analysis of furan in heat-processed foods
consumed
in
Korea
using
solid
phase
microextractiongas
chromatography/mass spectrometry (SPMEGC/MS). Food Chemistry, 123(4),
13281333.
Kim, T. K., Lee, Y. K., Kim, S., Park, Y. S., & Lee, K. G. (2009). Furan in commercially
processed foods: Four-year field monitoring and risk assessment study in Korea.
Journal of Toxicology and Environmental Health, Part A, 72(2122), 13041310.
Kim, M. K., Lee, Y. J., Kwak, H. S., & Kang, M. W. (2013). Identification of sensory
attributes that drive consumer liking of commercial orange juice products in
Korea. Journal of Food Science, 78(9), S1451S1458.
Kimball, D. A. (1987). Debittering of citrus juices using supercritical carbon dioxide.
Journal of Food Science, 52(2), 481482.
Kromhout, H. (1995). IARC Monographs on the evaluation of carcinogenic risks to
humans; dry cleaning, some chlorinated solvents and other industrial
chemicals.
Limacher, A., Kerler, J., Davidek, T., Schmalzried, F., & Blank, I. (2008). Formation of
furan and methylfuran by Maillard-type reactions in model systems and food.
Journal of Agricultural and Food Chemistry, 56(10), 36393647.
Lotong, V., Chambers, E., & Chambers, D. H. (2003). Categorization of commercial
orange juices based on flavor characteristics. Journal of Food Science, 68(2),
722725.
Mariotti, M., Granby, K., Fromberg, A., Risum, J., Agosin, E., & Pedreschi, F. (2012).
Furan occurrence in starch food model systems processed at high temperatures:
Effect of ascorbic acid and heating conditions. Journal of Agriculture and Food
Chemistry, 60, 1016210169.
Ministy of Food and Drug Safety (MFDS). Korean Food Standard Codex 2012 http://
fse.foodnara.go.kr/residue/mobile/menu_01_03.jsp?idx=317. (Accessed on
April, 2015).
National Toxicology Program (1993). Toxicology and Carcinogenesis Studies of
Furan (CAS No. 11000-9) in F344 Rats and B6C3F1 Mice (Gavage Studies).
National Toxicology Program Technical Report Series, 402, 1.
Sijia, W., Enting, W., & Yuan, Y. (2014). Detection of furan levels in select Chinese
foods by solid phase microextractiongas chromatography/mass spectrometry
method and dietary exposure estimation of furan in the Chinese population.
Food and Chemical Toxicology, 64, 3440.
Van Lancker, J., Adams, A., Owczarek-Fendor, A., De Meulenaer, B., & De Kimpe, N.
(2011). Mechanistic insights into Furan formation in Maillard model systems.
Journal of Agriculture and Food Chemistry, 59, 229235.
Vranov, J., Bednrikov, A., & Ciesarov, Z. (2007). In-house validation of a simple
headspace gas chromatography-mass spectrometry method for determination
of furan levels in food. Journal of Food and Nutrition Research, 46, 123127.
Wegener, J. W., & Lpez-Snchez, P. (2010). Furan levels in fruit and vegetables
juices, nutrition drinks and bakery products. Analytica Chimica Acta, 672(1),
5560.
Yaylayan, V. A. (2006). Precursors, formation and determination of furan in food.
Journal fr Verbraucherschutz und Lebensmittelsicherheit, 1(1), 59.
Zoller, O., Sager, F., & Reinhard, H. (2007). Furan in food: Headspace method and
product survey. Food Additives and Contaminants, 24(sup1), 91107.