Journal of Electroanalytical Chemistry 468 (1999) 53 – 63

A review of a new voltammetric method for determining acids

Kiyoko Takamura, Tetsuo Fuse, Kensuke Arai, Fumiyo Kusu *
School of Pharmacy, Tokyo Uni6ersity of Pharmacy and Life Science, 1432 -1, Horinouchi, Hachioji, Tokyo 192 -0392, Japan

Received 22 October 1998; received in revised form 15 January 1999; accepted 22 January 1999

Abstract

A new method for determining acids based on the voltammetric reduction of quinone was developed with the objective of
providing a sensitive and rapid means for acid assay superior to the conventional titration method. An assessment of this method
was made for determining the free fatty acid content in fats and oils, the total acidity of beverages, and the esterase activity to
demonstrate its usefulness for a wide range of applications. © 1999 Elsevier Science S.A. All rights reserved.

Keywords: Acids; Flow injection analysis; Electrochemical detection; Quinone

1. Introduction solvents [5–7]. Quinone itself gave a well-defined reduc-

tion peak on a potential sweep voltammogram using a
Fats and oils tend to decompose slowly upon storage glassy carbon working electrode in an ethanol solution
in contact with the atmosphere and to release their fatty containing LiClO4. Added acid in the solution was
acid constituents. Therefore, a quality and freshness found to give rise to a new peak (termed a prepeak) at
assessment of fats and oils can be made based on the a potential more positive than that for the reduction
determination of the free fatty acid content. In bever- peak of quinone. The half-peak potential of the pre-
ages such as coffee, fruit juice and wine, various or- peak shifted to a negative value along with an increase
ganic acids are present. Even though the content of any in pKa of the added acid, and its height was virtually
acid is small, the total acid amount has a significant independent of pKa but proportional to the acid
effect on the taste and aroma of the beverages. Acid concentration.
assay of beverages is thus important for quality control. On the basis of these facts, we developed a new
A commonly used method for determining the acid voltammetric method for determining the total acid
content is alkali titration using an appropriate indicator content in various samples such as fats and oils, and
[1 – 4]. However, in such a method, there is the possibil- beverages. Following the addition of such a sample to
ity of error due to an inability to detect a subtle color an ethanol solution of quinone, only one prepeak at a
change of the indicator in the transition range. The signal potential with a height proportional to the total
titration method is not sufficiently sensitive to detect a acid concentration was observed, provided the acid
small acid content, and accordingly it requires large strengths of the acids in the sample were nearly the
amounts of sample and much time to conduct. same.
We previously studied the voltammetric behavior of To avoid the problems of conventional titration, the
quinone in the presence of acids in unbuffered protic method of flow injection analysis with electrochemical
detection was developed as a sensitive and rapid means
for acid assay. This article reviews the assessment of

Review lecture presented at the International Symposium on this method for determining the free fatty acid content
New Trends in Electroanalytical Chemistry, Seoul, South Korea,
10–12 Sjeptember, 1998.
in fats and oils, the total acidity of beverages, and the
* Corresponding author. Fax: +81-426-76-4570. esterase activity in order to demonstrate the usefulness
E-mail address: [email protected] (F. Kusu) of the method for a wide range of applications.

0022-0728/99/$ - see front matter © 1999 Elsevier Science S.A. All rights reserved.
PII: S 0 0 2 2 - 0 7 2 8 ( 9 9 ) 0 0 0 4 9 - 2
54 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63

2. Voltammetric reduction of quinone in the presence of

acids

In general, the electrode redox process of quinone

is known to involve a two electron transfer coupled
with a two proton transfer represented as the follow-
ing reversible redox reaction [8,9]:
Q +2H + + 2e − X QH2 (1)
In our earlier studies of the proton donor effects of
the polarographic reduction of quinone in unbuffered
protic solvents such as water and alcohols, the pres- Fig. 2. Relation between half-wave potential of prewave and pKa
ence of a small amount of acid in a solution was obtained in aqueous (A) and 2-methoxyethanol (B) solutions of
[HA]t =7 × 10 − 4 M. (a) CCl3COOH; (b) C6H5SO3H; (c)
found to cause a new peak (termed a prepeak) at a
Cl2CHOOH; (d) ClCH2COOH; (e) C6H5COOH; (f)
more positive potential than the original reduction C6H5CH2COOH; (g) CH3COOH; (h) 2,6-Cl2C6H3OH; (i) 2,5-
wave of quinone. An example is shown in Fig. 1, in Cl2C6H3OH [5].
which methyl-p-benzoquinone gives a well-defined re-
duction peak at − 0.2 V (curve a). Following the
addition of a small amount of perchloric acid, a pre- more positive than the normal reduction potential.
peak appears at +0.2 V (curve b and c). The current Based on this fact, the occurrence of the prepeak can
height of the prepeak (IH) increases with increasing be ascribed to the increased availability of protons
amounts of perchloric acid in proportion to the acid from the added acid compared to the solvent
concentration. Similar results were also obtained by molecules and has resulted in a lowering of the Gibbs
the addition of organic acids. The half-peak potential energy for the reduction process, leading to the reduc-
of the prepeak shifted to a more negative value ac- tion potential shift in the positive direction depending
companied by an increase in the pKa of the acid (Fig. on the acid strength [5,10,11].
2). The prepeak height was found to be virtually in- Based on the above, the determination of the acid
dependent of the pKa but proportional to the acid content in fats and oils was examined recently by
concentration up to twice the quinone concentration potential sweep voltammetry using a glassy carbon
[5]. From these results, the prepeak was assigned to electrode in an ethanol solution containing LiClO4
the acid added to the solution. [6,7]. As a quinone species, vitamin K3 (VK3, 2-
The prepeak current was found to be controlled by methyl-1,4-naphthoquinone) was normally used be-
the diffusion of the acid in the solution to the elec- cause it facilitated this determination due to its
trode surface [5]. The electrolysis of quinone at the solubility and stability in ethanol. A well-defined re-
potential of the prepeak demonstrated that the two duction peak was observed on the voltammogram of
electrons transferred per molecule produced hy- VK3 as shown in Fig. 3 (curve a). Following the
droquinone although the electrolysis potential was addition of palmitic acid, a prepeak also appeared at
a more positive potential than the original peak (Fig.
3, curve b). Its height (IH) was in proportion to the
palmitic acid concentration from about 10 − 6 to 10 − 3
M with a coefficient of variation of 0.980. Essentially
the same was observed for various kinds of car-
boxylic acids including higher fatty acids examined,
from which one might expect a novel approach to the
determination of total acids in food and biological
samples by measuring the prepeak of quinone, pro-
vided their acid strengths and diffusion coefficients
are nearly the same. In such cases, only one prepeak
at an identical potential with a height proportional to
the total acid concentration was actually observed.
The voltammetric method stated above was found
to be far superior to the conventional titration
method on the points of sensitivity and accuracy
Fig. 1. Polarograms of 2.20 ×10 − 3 M methyl-p-benzoquinone in
[6,7]; however, the measurement procedure was some-
aqueous solutions containing 0.25 M NaClO4. Concentration of what time-consuming for practical use. A simple and
HClO4: (a) 0; (b) 2.50× 10 − 4; (c) 4.67× 10 − 4 M [5]. more rapid method thus became desirable. This de-
 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63 55

mand prompted us to conduct a flow injection analysis

with electrochemical detection (FIA/ECD) of acids [12].
High performance liquid chromatography with ECD
(HPLC/ECD) of fatty acids in oils was also carried out.

3. Experimental

3.1. Voltammetry [6,7]

Fig. 4. Potential dependence of the flow signal for 1.0 × 10 − 4 M
Potential sweep voltammetry was conducted using a palmitic acid. Injection volume, 5 ml; Flow rate, 0.6 ml min − 1;
potentiostat (Model HA-301, Hokuto Denko, Tokyo, Carrier solution, ethanol solution containing 3 mM VK3 and 38 mM
Japan) in conjunction with a function generator LiClO4 [12].
(Model HB-104, Hokuto Denko, Tokyo, Japan) and
an X–Y recorder (Model 3077, Yokogawa Electric
Corporation, Tokyo, Japan) at 25°C in nitrogen-satu-
rated ethanol containing the quinone (usually 3 mM chemical cell was fabricated from a glassy carbon (di-
VK3) and 38 mM LiClO4. A glassy carbon disk (di- ameter of 6 mm, Tokai Carbon, Tokyo, Japan)
ameter of 6 mm, Tokai Carbon, Tokyo, Japan) work- working electrode, an Ag AgCl KCl sat. reference
ing electrode, a saturated calomel reference electrode electrode, a stainless-steel auxiliary electrode and a
(SCE), a platinum auxiliary electrode and a beaker- polychlorotrifluoroethylene cell body. The cell volume
type electrochemical cell were used. Following deaera- was 2.4 ml. The flow lines were made of polyte-
tion of the test solutions by bubbling with trafluoroethylene tubing (0.5 mm i.d.) shielded with
oxygen-free nitrogen, voltammograms were recorded aluminum foil. The length of the tubing from the
at a sweep rate of 5 mV s − 1. sample injector to the detector was 1 m.
In FIA/ECD, an ethanol solution containing 3 mM
VK3 and 38 mM LiClO4 served as a carrier solution.
3.2. Flow injection analysis with electrochemical detec-
The dissolved oxygen in this solution was removed by
tion [12]
the degasser. The flow rate of the carrier solution was
The flow-injection (FI) system [12] consisted of a
0.6 ml min − 1. To determine the detection potential
degasser (Model DG-980-50, JASCO, Tokyo, Japan),
for monitoring the acid content in a test solution,
a pump (Model DMX-2200-T, SNK. Ind., Tokyo,
hydrodynamic voltammograms for each acid exam-
Japan), a sample injector (Model 7125, Rheodyne,
ined were obtained using FIA/ECD. An example is
Cotati, CA, USA), an electrochemical cell, a poten-
shown in Fig. 4, in which the peak height of the flow
tiostat (Model 312, Huso Electrochemical System,
signal obtained for each injection of a 5 ml aliquot of
Kawasaki, Japan) and a recorder (Model 807-IT,
the test solution containing 1.0× 10 − 4 M palmitic
JASCO, Tokyo, Japan). The wall-jet-type electro-
acid is plotted against the applied potential. The de-
tection potential for monitoring the acid content was
then determined at − 0.33 V. A flow signal appeared
for each amount of palmitic acid (Fig. 5). The re-
sponse in concentration was linear between 5.0×
10 − 6 and 3.0× 10 − 4 M (25–1500 pmol per test).
Palmitic acid at 2.0× 10 − 4 M was determined ten
times with a relative standard deviation (R.S.D.) of
1.4%. Linearity and reproducibility were basically the
same for other higher fatty acids and carboxylic acids
examined in this work. The detection system was sta-
ble for about 6 months when processing test solu-
tions.

3.3. HPLC with electrochemical detection [13]

Fig. 3. Voltammograms of 3.0 mM VK3 (a) without and (b) 3.2 mM The system of HPLC with electrochemical detection
palmitic acid in an ethanol solution containing 38 mM LiClO4. WE, (HPLC/ECD) is shown in Fig. 6 and consists of a
glassy carbon; RE, SCE; CE, platinum wire. Scan rate: 5 mV s − 1. degasser (Model DG-980-50, JASCO, Tokyo, Japan),
56 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63

% %

Fig. 7. Retention time (A) and peak height (B) as functions of solvent
ratio of the mobile phase of ethanol +acetonitrile mixture: a, linoleic
acid; b, oleic acid; c, palmitic acid; d, stearic acid. Amount of
acid= 200 pmol. HPLC conditions: Quinone solution, 6 mM VK3 +
76 mM LiClO4 in ethanol +acetonitrile mixture; sample volume, 20
ml; column, LiChrospher 100 RP-18 (250 mm ×4 mm i.d., 5 mm);
column temperature, room temperature; flow rate, 1.1 ml min − 1;
Fig. 5. Flow signals obtained with (a) 5.0 × 10 − 5; (b) 1.0 × 10 − 4; (c) applied potential, − 415 mV versus SCE [13].
1.5× 10 − 4 and (d) 2.0 × 10 − 4 M palmitic acid. Injection volume, 5
ml; Flow rate, 0.6 ml min − 1; Applied potential, −0.33 V versus A 20 ml sample aliquot and standard acid solutions
Ag AgCl; Carrier solution, ethanol solution containing 3 mM VK3 were injected into an ODS column maintained at room
and 38 mM LiClO4 [12].
temperature. The deaerated mobile phase and the
quinone solution were made to flow at 1.1 ml min − 1
with P1 and P2. The detection potential for monitoring
pumps P1 and P2 (Model PU-980, JASCO, Tokyo, free fatty acids was maintained at −415 mV versus
Japan), a sample injector (Model 7125, Rheodyne, Co- SCE. Each fatty acid in the sample solution was deter-
tati, CA, USA), an octadecylsilica (ODS) column mined based on the signal peak height.
(LiChrospher 100 RP-18, 250 mm ×4 mm i.d., 5 mm, For HPLC/ECD, reversed-phase separation of higher
Cica –MERCK, Tokyo, Japan), an electrochemical cell fatty acids was carried out using an ODS column and a
(Model EC-840, JASCO, Tokyo, Japan), a potentiostat mobile phase of an ethanol+ acetonitrile mixture. The
(Model 311B, Huso Electrochemical System, Kawasaki, dissolved oxygen in the mobile phase and the quinone
Japan) and a recorder (Model 807-IT, JASCO, Tokyo, solution was removed by the degasser. A 20 ml aliquot
Japan). The electrochemical cell was made from a of the solution containing fatty acids was injected into
glassy carbon working electrode, an SCE reference the column; the eluate was mixed with the quinone
electrode and a stainless-steel auxiliary electrode. Flow solution and the fatty acids were detected with ECD.
lines were made from stainless-steel tubing (0.5 mm i.d.) The influence of the ratio of acetonitrile to ethanol in
and polytetrafluoroethylene tubing (0.5 mm i.d.) cov- the mobile phase on the separation characteristics and
ered with aluminum foil to shield the sample from the the sensitivity of the acid determination was examined.
light. In Fig. 7, the retention time (A) and the peak height (B)
To prepare a sample solution for injection into the of signals for 200 pmol linoleic, oleic, palmitic and
separation column, the proper amount of the sample oil stearic acids were plotted against the ratio of the two
was mixed with an ethanol+acetonitrile (10:90) mix- solvents. The larger the content of acetonitrile, the
ture. When the oil was not soluble in the ethanol +ace- greater was the separation of the acid peaks. The peak
tonitrile (10:90) mixture, the mixture was centrifuged height was maximum at about 90% acetonitrile for all
after mixing, and the supernatant was used as a sample fatty acids examined. The 10:90 ethanol+ acetonitrile
solution. mixture was concluded to be the most suitable mobile
phase.
Based on hydrodynamic voltammograms for free
fatty acids, the potential for fatty acid detection was
determined to be −415 mV versus SCE.
A typical chromatogram for a mixture of standard
linoleic, oleic, palmitic and stearic acids is shown in
Fig. 8. The acids were well separated within 15 min.
The peak height was linear versus the acid amount
injected in the range of 20–1200 pmol. Linoleic, oleic,
Fig. 6. HPLC system: MP, mobile phase, ethanol + acetonitrile mix- palmitic and stearic acid at 200 pmol were determined
ture; Quinone solution, 6 mM VK3 + 76 mM LiClO4 in ethanol+ 10 times with relative standard deviations (R.S.D.) of
acetonitrile mixture; DG, degasser; P1, P2: pumps; S, sample injector
(20 ml); Column (LiChrospher 100 RP-18, 250 mm × 4 mm i.d., 5
1.3, 1.2, 0.96 and 0.79%, respectively. The detection
mm); MC, mixing coil (50 cm); D, electrochemical detector, electro- limit (signal-to-noise ratio=2) of acids per injection
chemical cell and potentiostat; R, recorder [13]. was 20 pmol.
 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63 57

peanut oil (Kozakai Seiyaku, Tokyo, Japan), cacao

butter (Fujisawa–Astra, Osaka, Japan), soy bean oil
(Kanto Chemical, Tokyo Japan) and glyceryl monos-
tearate (Wako Pure Chemical Industries, Osaka, Japan)
were tested.
The FIA/ECD method was applied to a determina-
tion of the free fatty acid content in nine kinds of fat
and oil samples. Content values obtained were of the
order of 10 − 6 to 10 − 5 mol g − 1 for each sample.
Recovery tests were made with the addition of standard
linoleic acid in appropriate amounts, in each case being
nearly the same as the acid amount in each sample. The
recoveries of added linoleic acid were in the range from
Fig. 8. Typical chromatogram of a standard mixture containing (a) 96 to 102%.
linoleic acid (200 pmol); (b) oleic acid (200 pmol); (c) palmitic acid The results obtained by the present method were
(200 pmol); (d) stearic acid (200 pmol); other HPLC conditions are
the same as those in Fig. 7 [13].
compared with those by voltammetric and conventional
titration (using KOH) methods (Table 1). In titration,
4. Applications the end points were determined based on the color
change of phenolphthalein and potentiometry.The con-
tent values obtained by the present method showed a
4.1. Determination of free fatty acid content in fats and good correlation with those obtained by voltammetry
oils and the potentiometric titration method (both R=
0.999). In Table 1, the R.S.D. of the content by the
An assessment was made of the present FIA/ECD as present method is less than 2.0%, the best among the
a rapid means for monitoring higher free fatty acids in four methods. The R.S.D. was greatest in conventional
fats and oils [12]. Standard acid solutions were pre- titration with phenolphthalein. Such a decreased repro-
pared by dissolving palmitic acid (99.5%, Wako Pure ducibility is considered to be due to an indicator error
Chemical Industries, Osaka, Japan) or linoleic acid and/or undesirable progress of the hydrolysis of fats
(99%, Wako Pure Chemical Industries, Osaka, Japan) and oils by the catalytic effect of KOH. However, no
in ethanol solution containing 3 mM VK3 and 38 mM marked differences in R.S.D. could be detected for the
LiClO4. To prepare the test solutions, an appropriate two methods and the potentiometric titration. Potentio-
amount of sample (4 – 400 mg) was completely dissolved metric titration does not show any indicator error. The
in 5 ml ethanol solution containing 3 mM VK3 and 38 greatest R.S.D. in conventional titration is thus shown
mM LiClO4. When the sample was not soluble in to be due to an indicator error.
ethanol, it was first dissolved in ether and then trans- The use of the FIA/ECD method was found to be
ferred to ethanol. empirically valid for samples containing acids, the dif-
Camellia oil (Shiseido Seiyaku, Tokyo, Japan), corn fusion coefficient values of which are nearly the same.
oil (Hayashi Chemicals, Tokyo, Japan), mentha, olive The application of this method to the determination of
and sesame oils (Miyazawa Yakuhin, Tokyo, Japan), total amounts of free fatty acids in fats and oils for

Table 1
Comparison of the data of free fatty acid content in various samples obtained by our methods with those by titration methods [12]

Sample Titration

FIA Voltammetry Potentiometry Color changea

105×content R.S.D.b 105×content R.S.D.b 105×content R.S.D.b 105×content R.S.D.b

/mol g−1 /% /mol g−1 /% /mol g−1 /% /mol g−1 /%

Camellia oil 4.0 1.9 3.9 1.9 4.1 2.0 4.1 2.3
Cacao butter 4.2 2.0 4.1 2.0 4.1 2.7 4.2 4.7
Glyceryl 1.5 1.0 1.6 1.2 1.6 1.5 1.8 3.7
monostearate
Mentha oil 0.36 1.3 0.41 2.4 0.36 2.5 0.39 3.4
Corn oil 0.26 0.54 0.23 1.3 0.23 2.4 0.29 4.8

a
Using phenolphthalein indicator.
b
n= 5.
58 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63

food seems a favorable choice because their constituent

acids are almost restricted to higher monocarboxylic
acids (C16 –C20) with a straight chain of carbon atoms.

4.2. Determination of total acidity of be6erages

The acid assay of coffee beans and coffee is impor-

tant for quality control. A sensory test, i.e. taste assess-
ment by human gestation, is usually conducted for
quality assessment but is not adequate due to a lack of
objectivity and efficiency. Common methods of acid
assay, involving alkali titration and pH measurement,
are not sufficiently sensitive to follow slight acidity
Fig. 9. Correlation of FIA signal and titratable acidity for the 21
changes. Thus we examined the possibility of using the coffee [14].
same FIA/ECD method for the acid assay of coffee
[14].
For the coffee preparation, 250 ml hot water was coffees prepared using commercially available coffee
poured onto 20 g coffee powder on a filter paper in a powders. For the sour sensory test, 17 volunteers were
filter (Kalita model EX-102, Tokyo, Japan). Coffee available. A tray of randomly coded hot coffees, a
from 21 different commercially available coffee pow- sensory test ballot with a 5-point linear scale and a cup
ders was used for the determination of acidity by FIA of water were used for evaluating. Each coffee was
and titration and the sensory test by panelists. Roasted assessed for sourness intensity using the 5-point scale
coffee beans were purchased from a coffee grain shop from 1 to 5 (scoring test). Mean scores were obtained
(Green Bean, Hachioji, Japan), and the shop was asked from the data, compared to the FIA signals and good
to roast the green coffee beans by hot air heating for correlation (R= 0.847) was noted, as shown in Fig. 10.
0 – 20 min. The beans were subsequently powdered in a From Figs. 9 and 10, the present method, as well as the
coffee mill (National model MK-52M, Tokyo, Japan) titratable acidity measurement, appear to be reliable
at our laboratory. For FIA measurements, the coffee methods for assessing sour taste. The pH values for
was diluted with an ethanol solution containing 3 mM roasted Kilimanjaro (K), Guatemala (G), and Hawaii
VK3 and 38 mM LiClO4 to prepare the test solutions. Kona (H) were basically the same, from 4.9 to 5.0. On
In the present method, the slope of the FIA signal the other hand, even though the test solution for the
current versus concentration depends on the number of present method was highly diluted compared to those
acid functional groups [15]. Thus the FIA signal, I, for titration and pH measurement, FIA revealed dis-
given by tinct current signals of 6.2–9.0 nA.
Sour taste depends on the roasting of the beans. The
I= % ki ci (2) acid content attains a maximum and then decreases
i
with roasting time. The change in sour taste was exam-
where ci is the concentration of acid i, and ki, the ined during roasting. The FIA signal current was plot-
proportionality constant of an individual FIA signal ted against roasting time. As seen in Fig. 11, B(1), the
current, is related to the concentration of acid i. current essentially remained constant and low up to 11
Although titration has a problem with sensitivity, -
titratable acidity commonly serves as a measure of the
sourness of coffee. Comparison was thus made of FIA
signals with the titratable acidity of coffee. To deter-
mine the titratable acidity, 50 ml of coffee were neutral-
ized with 0.100 M NaOH up to pH 8.2 [16] using the
automatic potentiometric titrator. The volume of 0.100
M NaOH required was defined as titratable acidity.
The FIA signal showed good correlation (correlation
factor, R= 0.814) with the titratable acidity of 21 coffee
samples (Fig. 9), suggesting this parameter to be an
alternate means for coffee quality assessment.
A comparison of FIA signals with sourness sensory
test data was made to show the usefulness of the
present method for assessing coffee sourness. The 5- Fig. 10. Relation between FIA signal and sourness intensity scores
point sensory test for sourness intensity was done for 21 [14].
 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63 59

Although the FIA signals proved to be practically

useful as an alternative to the titration method for
determining the total acidity of beverages, the present
method can still be applied to other suitable areas after
examination. When the precise values of acid content are
required, the preseparation of acids is needed prior to the
electrochemical detection (cf. 3.3).

4.3. Determination of esterase acti6ity

In our study, an assessment of the FIA/ECD method

was further made for the determination of the esterase
activity [17,18]. The amounts of acids released from the
substrate esters through enzyme reaction with esterase
were determined by this method, and the method was
applied to the determination of enzyme activity values in
human serum.
Fig. 11. Effects of roasting time on acidity of coffee measured by FIA
(1), titration (2) and a pH meter (3). Coffee beans: (A) Kilimanjaro, 4.3.1. Lipase
(B) Guatemala, (C) Hawaii Kona [14]. Lipase, a fat digestive enzyme, catalyzes hydrolysis
reactions of higher fatty acid glycerides to release con-
min, then increased markedly, attained a maximum at stituent fatty acids, as follows:
about 13 min and then decreased gradually. Sensory test
results from seven participants for the same coffee as
noted above clearly showed that the acid taste intensity
changed with the roasting time in the order, 13\ 15\
(3)
16.5 \ 0 min. Comparison of the results showed excellent
agreement, suggesting that the current of the FIA signal A unit activity of lipase is defined as that amount of
serves as an accurate indication of sour taste. The change enzyme capable of producing 1 mmol of fatty acid per 1
in titratable acidity and pH of coffees extracted from K, min. Alkali titration [19], turbidimetry [20] and enzyme
G, and H coffee beans was also followed during roasting, cycling reaction/UV absorption spectroscopy [21] are
and the results are shown in Fig. 11 A – C, as (2) and (3), conventionally used for determining lipase activity.
respectively. The pH versus time curve showed a mini- Solution A was a buffer solution (pH 7) containing
mum value at 13–16 min (Fig. 11, (3)). In the pH 1/15 M KH2PO4 and 1/15 M Na2HPO4, solution B, a
measurement, only a slight change was noted throughout buffer solution (pH 9) containing 0.1 M KH2PO4 and
the roasting period in contrast to Fig. 11, (1), because the 0.05 M Na2B4O7. A standard lipase solution was pre-
potential of the glass electrode cell responds only to the pared by dissolving lipase (EC 3.1.1.3) from porcine
logarithm of the concentration of dissociated protons in pancreas (Wako Pure Chemical Industries, Osaka,
the sample solution. In Fig. 11, (2), the titratable acidity Japan) in solution B. A 15 ml sample aliquot containing
is plotted against roasting time giving a curve similar to lipase was added to a mixed solution of 15 ml olive oil,
that in Fig. 11, (1) as expected from Fig. 10. Titration 40 ml water and 35 ml solution B in a glass tube. The
to pH 8.2 may be accompanied by a small error due to mixture in the tube was incubated for 10 min at 37°C,
endpoint mismatch, not to mention the large amount of followed by the addition of 3 ml of diethylether. Free
sample and time required: 50 ml coffee and about 10 min fatty acids produced by enzyme reactions of lipase in the
for each plot in Fig. 11, (2). Only 5 ml of 40-fold diluted tube were extracted three times with 3 ml each of ether.
coffee and about 1.5 min are required for each plot in Following evaporation of the ether, the oily precipitate
Fig. 11, (1). was dissolved in 300 ml ethanol solution containing 3 mM
The measurement of the prepeak in a voltammogram VK3 and 38 mM LiClO4 for the preparation of a test
of quinone was also useful for the determination of total solution. In order to obtain a test solution from a serum
acidity in other beverages such as fruit juice and wine. sample, solution A was used instead of solution B. The
In the case of orange juice, one prepeak appeared on the blank test solution was prepared by omitting incubation.
voltammogram following the addition of the juice to A 5 ml aliquot each of the test and blank test solutions
quinone ethanol solution. The total acid content was was injected into the FI system. The acid content in the
calculated as citric acid using the prepeak. The results sample and blank solutions was determined based on the
showed good correlation to those obtained by conven- current height of the flow signals. Lipase activity was
tional titration methods. calculated as follows:
60 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63

(IS −IB)V complex reaction pathway.

Lipase activity (U l − 1) = (4)
IA6t
4.3.2. Cholinesterase
where IS is the peak height for the test solution (mA), IB
Cholinesterase (ChE) is an enzyme for the hydrolysis
the peak height for the blank test solution (mA), IA the
reaction of the choline esters with the release of their
peak height per 1 mM of standard oleic acid, V the total
constituent acids. Its activity in serum is important in
volume of the test solution (ml), 6 the added amount of
diagnosing liver diseases and organophosphorus com-
the substrate (ml), and t the enzyme reaction time (min).
pound poisoning [24–26]. Various methods for serum
A test solution was prepared with extracts from the
ChE activity determination, such as the Takahashi–Shi-
reaction mixture of porcine pancreatic lipase. The main
bata method [24], the DTNB method [25,26] and the
constituent fatty acids of olive oil should be oleic, linoleic
4-aminoantipyrine –phenol–peroxidase method (4 AA)
and palmitic acids [22]. Values determined for lipase
[27], are available but have not sufficiently proved
activity were shown to be linearly related to the lipase
adequate in all cases.
amount of 10–1000 U l − 1 (R = 0.989). The R.S.D. was
The present method was used to determine the activity
1.2% for 70 U l − 1 (n= 5). Data accuracy was confirmed
of true and pseudo ChE. True ChE catalyzes the hydrol-
by close agreement with values obtained by conventional
ysis reaction of acetylcholine, with consequent release of
titration using NaOH in the range of 10 – 1000 U l − 1
the constituent acetic acid. Pseudo ChE catalyzes the
(R = 0.999). Lipase activity in the control serum (com-
hydrolysis reaction not only of acetylcholine but also of
mercially available) was determined by the present
benzoylcholine, to release the constituent acid, i.e. acetic
method, and the data were correlated with those of the
or benzoic acid. The reactions are as follows:
enzyme cycling method using a Nescauto lipase kit-VE
(Nihonshoji, Osaka, Japan) (Table 2). The R.S.D. of
lipase activity in human serum was 2.8% (n = 5).
Titration [19] is generally used in industrial processes,
in which fatty acids released from substrate oil are
(5)
neutralized with KOH or NaOH. This method requires
a large sample and is not suitable for serum lipase
determination. In turbidimetry [20], serum lipase can be
determined by following the transmittance change during
specified time intervals after adding serum to the sub-
strate oil emulsion. This method is simple, but it is
difficult to prepare a homogeneous and stable oil emul- (6)
sion, with a possible compromise of the precision and
A unit activity of ChE is defined as that amount of
reproducibility of the data. The enzyme cycling method
enzyme capable of producing 1 mmol of benzoic or acetic
using diglycerides as substrates by coupling with the
acid per 1 min.
enzyme cycling reaction involving the b-oxidation of
Test solutions were prepared from enzyme reactions of
fatty acids is sensitive and specific for lipase assay and
pseudo ChE and acetylcholine, and of pseudo ChE and
is suitable for clinical use [21]. The rate assay of the
benzoylcholine. Standard pseudo and true ChE solutions
resulting NADH2 is used, but data are sometimes less
were prepared by dissolving pseudo ChE (EC 3.1.1.8)
reproducible (R.S.D.B10%) [23], possibly due to the
from horse serum and true ChE (EC 3.1.1.7) from eel
Table 2 (Sigma, St. Louis, MO), respectively, in solution C, a
Lipase activity in serum samples obtained by the present and enzyme buffer solution (pH 7.5) containing 0.01 M KH2PO4 and
cycling methods [18] 0.01 M Na2HPO4. A 15 ml sample aliquot containing ChE
Control serum Enzyme cycling method/ Present method/
was added to a mixed solution of 10 ml of benzoylcholine
U l−1 U l−1 or acetylcholine (0.1 M), and 10 ml of solution C in a glass
tube. The mixture in the tube was incubated for 15 min
A 57 198 at 37°C, followed by the addition of 5 ml of ethanol
B 45 77 containing 3 mM VK3 and 38 mM LiClO4 for test
C 49 73
D 61 137
solution preparation. The blank test solution was pre-
E 93 292 pared by omitting incubation. The ChE activity was
F 85 290 calculated as follows:
G 97 471
H 65 292 (IS − IB)V
ChE activity (U l − 1)= (7)
I 110 584 IA6t
J 49 65
K 61 168 where, IA is the peak height per 1 mM of standard acid
(mA mM − 1) (benzoic acid for pseudo ChE activity
 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63 61

Table 3
Cholinesterase activity in serum samples obtained by the present and
4-aminoantipyrine–phenol–peroxidase methods [18]

Control serum 4-AA method/U l−1 Present method/U l−1

A 976 638
B 1200 873
C 762 504
D 952 605
E 903 571
F 869 571
G 776 503
H 1055 806
I 693 573
Fig. 12. Effects of pH on the enzyme reaction of pseudo J 1024 738
cholinesterase. Substrate: a (“) acetylcholine, b () benzoylcholine. K 652 436
Temperature: 37°C, Reaction time: 15 min [18]. L 948 638
M 1052 672
N 790 571
measurement; acetic acid for pseudo and true ChE
activity measurement) and other parameters are as
before.
an indication that pseudo ChE is a constituent in
Fig. 12 shows the effects of pH on the enzyme
serum.
reactions of ChE; the pH for monitoring the enzyme The Takahashi–Shibata method [24] is used for
reaction was maintained at 8.3 and 7.5 for acetylcholine pseudo ChE, in which the acetic acid released from
and benzoylcholine as substrate, respectively. The en- acetylcholine is monitored by a phenol red color change
zyme reaction time was determined to be 15 min. When (570 nm), but this method requires a large sample and
acetylcholine and benzoylcholine were used as the sub- much time. The DTNB method [25,26] frequently gives
strates, pseudo ChE showed virtually the same activity, erroneous data due to the presence of sulfhydryl com-
and a linear relation between its concentration and pounds. In the 4 AA method, red quinone produced
activity was found when the amount of ChE was in the from 4-aminoantipyrine, phenol and hydrogen peroxide
range 10–1500 U l − 1 (R = 0.999). On the other hand, by the peroxidase enzyme reaction can be determined
the true ChE activity using only acetylcholine as the by the absorption measurement at 505 nm, in which
substrate was linearly related to a true ChE amount hydrogen peroxide is produced from choline by the
from 10–1500 U l − 1 (R = 0.996, Fig. 13). The ChE choline oxidase enzyme reaction [27]. Choline is re-
activity was determined in the control serum. The val- leased by the pseudo ChE enzyme reaction from ben-
ues were fairly well correlated with those of the usual 4 zoylcholine. Thus redox substances present in serum
AA method using cholinesterase B-testwako (Wako easily affect the reaction, which may give rise to erro-
Pure Chemical Industries, Tokyo, Japan) (Table 3). The neous data. Furthermore, this method is very sensitive
R.S.D. of ChE activity in human serum was 2% (n= 5). and suitable for clinical use but not for hemolysis
The ChE activity values in the control serums obtained samples.
by using acetylcholine and benzoylcholine as the sub- In applying the present method, it is preferable to
remove acid components contained in a sample (such as
strates were virtually the same (R =0.985), this being
phosphoric acid in a biological sample) during test
solution preparation to keep the background current
low. However, they exist in the buffer solution in which
the enzyme reaction is carried out. It should be noted
that relatively large amounts of the acid released from
the substrate is needed to detect currents in a high
signal to noise ratio.

4.4. Determination of free fatty acids in oils

The analysis of olive oil, camellia oil, corn oil, rape-

seed oil and soy bean oil was carried out. Chro-
matograms were readily obtained following the
Fig. 13. Relation between true cholinesterase concentration and activ- injection of sample solutions. A typical chromatogram
ity. Substrate: a (“) acetylcholine, b () benzoylcholine [18]. for corn oil is shown in Fig. 14. The peak height ratio
62 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63

for the free fatty acids was palmitic acid: stearic acid: nonpolar constituent hardly came into contact with the
oleic acid: linoleic acid at 5:1:13:25, this being essen- working electrode, as well as the ODS in the column
tially the same as the constituent ratio reported for used in the present method. The working glassy carbon
fatty acids in corn oil [28]. The content values of each electrode surface remains stable for periods of more
free higher fatty acid in oil determined are of the order than 3 months for analysis of more than 10 samples a
of 10 − 8 to 10 − 5 mol g − 1 of oil. The relative standard day using the HPLC system. The applied potential for
deviation (n=5) was between 0.8 and 2.8% [13]. monitoring fatty acids is considered negative enough
For recovery assessment, a mixture of standard not to cause carbon surface oxidation and the organic
palmitic acid, stearic acid, oleic acid and linoleic acid eluent solvent may prevent the adhesion of hydropho-
was added to each sample oil, olive, camellia, corn, bic contaminants on the electrode surface. These fac-
rapeseed and soy bean oils, and then analyzed by the tors also lead to greater reproducibility. A detection
present method. The recoveries of each fatty acid were limit as low as 20 fmol was possible with HPLC/FL
in the range from 90 to 109%. [29–33]. The detection limit was as low as 20 pmol by
A comparison with HPLC with fluorescent detection the present method. Because of the simple procedure of
(HPLC/FL) using 9-anthryldiazomethane as a fluores- the sample solution preparation without derivatization
cent derivatization reagent indicated that the present using labeling agents, the amount of sample required is
HPLC/ECD method made possible the simple determi- less and the time for determination is reduced. Mea-
nation of higher fatty acids within a short time. The surements can be made only through injection of the
former method required ca. 100 min for sample pre- sample solution directly into the HPLC system. The
treatment and HPLC procedure, while the latter, only present method with ECD is thus shown to be quite
20 min. Detection limits of the former and the latter effective for the determination of higher fatty acids.
methods were 2 and 20 pmol per one injection. How-
ever, the relative standard deviation by fluorescent de-
tection is less than 5%, even though an internal 5. Advantages of the present method
standard is used, while that by the present method for
oil samples is less than 3% (n = 10). Although acid–base titration is commonly used to
The good reproducibility is considered due to the test the acid content in fats and oils and beverages in
simple pretreatment procedure without derivatization their quality assessment, the inability to detect a subtle
of sample fatty acids. When the oil contained a highly change in the indicator color frequently leads to an
nonpolar constituent, it was not soluble in an error, especially in cases of colored samples, such as
ethanol +acetonitrile (10:90) mixture, and thus only plant oil and coffee. For such cases, a potentiometric
the supernatant of the oil+ethanol +acetonitrile mix- measurement is recommended for end-point matching
ture was injected into the column. Thus, the highly [1]. However, in the case of the fats and oils in our
study, the adhesion of insoluble fatty acid salts (soaps)
sometimes occurred during titration on the glass elec-
trode surface along with a consequent lowering of the
response rate.
In the voltammetric method [6], although the sensi-
tivity as well as the reliability of the data were greatly
improved compared to the conventional titration
method, the measurement procedure is not always suffi-
ciently rapid to be suitable for practical use. The
present FIA/ECD method was found not only to be
highly sensitive (detection limit: 25 pmol acid per one
test) with high reproducibility but also quite simple and
rapid. The FIA operating conditions made possible the
processing of 60 samples h − 1. The sample amount was
only several thousandths of that required for titration,
thus making test solution preparation much easier.
Titration requires larger sample amounts and some-
times longer times are required, especially for the sam-
ples insoluble in ethanol. Use of a small sample amount
is also favorable for the test solution preparation of
esterases. In addition, the fact that the sample color
Fig. 14. Typical chromatograms of corn oil. HPLC conditions:
applied potential, −415 mV versus SCE; other HPLC conditions are causes no interference in the detection of the analyte
the same as those in Fig. 7 [13]. permits the present method to be applied to the deter-
 K. Takamura et al. / Journal of Electroanalytical Chemistry 468 (1999) 53–63 63

mination of ChE activity in forensic samples because of [13] T. Fuse, F. Kusu, K. Takamura, J. Chromatgr. A 764 (1997)
177.
its possible use for hemolysis blood.
[14] T. Fuse, F. Kusu, K. Takamura, J. Agric. Food Chem. 45
The present method is thus shown to be practically (1997) 2124.
useful for acid assay of food and biological samples. It [15] F. Kusu, T. Fuse, K. Takamura, Colloq. Sci. Int. Cafe, [C.R.]
should find a broad range of applications such as 16th 1 (1995) 351.
checking the quality of materials in industrial fields. [16] N. Suzuki, Japanese Standard Methods for the Examination of
Foods, Japan Food Hygiene Association, chapter 23, Tokyo,
1978.
[17] K. Takamura, T. Fuse, F. Kusu, J. Electroanal. Chem. 396
References (1995) 507.
[18] K. Takamura, T. Fuse, F. Kusu, Anal. Sci. 12 (1996) 343.
[1] Official and Tentative Methods of the American Oil Chemists’ [19] I.S. Cherry, L.A. Crandall, Am. J. Physiol. 100 (1932) 266.
Society, Method Cd 3d–63, American Oil Chemists’ Society, [20] Z.K. Shihabi, C. Bishop, Clin. Chem. 17 (1971) 1150.
Champaign, 1989. [21] S. Imamura, H. Misaki, Selected Topics in Clinical Enzymol-
[2] Official and Tentative Methods of the Japan Oil Chemists’ ogy 2 (1984) 73.
Society, Method 2.4.1–83, Japan Oil Chemists’ Society, [22] Japan Pharmacopoeia 13th, The Pharmaceutical Society of
Tokyo, 1990. Japan, Tokyo, 1996, p. D166.
[3] US Pharmacopoeia XXIII, United States Pharmacopeial Con- [23] Nihonshoji: Account of Nescauto lipase kit-VE.
vention, New York, 1995, p. 738. [24] H. Takahashi, S. Shibata, Med. Biol. 20 (1951) 96.
[4] Japan Pharmacopoeia 13th, The Pharmaceutical Society of [25] G.L. Ellman, K.D. Courtny Jr, V.A. Featherstone, R.M.
Japan, Tokyo, 1996, p. B522. Featherstone, Biochem. Pharmacol. 7 (1961) 88.
[5] K. Takamura, Y. Hayakawa, J. Electroanal. Chem. 31 (1971) [26] A.A. Dietz, H.M. Rubinstein, T. Lubrano, P.J. Garry, F.M.
225. Orsini, B.C. Strever, Clin. Chem. 19 (1973) 1309.
[6] F. Kusu, T. Fuse, K. Takamura, J. AOAC Int. 77 (1994) [27] H. Okabe, K. Sagesaka, A. Noma, N. Nakajima, Jpn. J. Clin.
1686. Pathol. 25 (1971) 751.
[7] T. Fuse, F. Kusu, K. Takamura, Bunseki Kagaku 44 (1995) [28] D. Swern, Bailey’s Industrial Oil and Fat Products, 4th edi-
29. tion, Wiley, NY, 1979, p. 389.
[8] G.W. Milner, Principles and Applications of Polarography, [29] S. Lam, E. Grushka, J. Chromatogr. 158 (1978) 207.
Longmans, London, 1962, p. 515. [30] N. Nimura, T. Kinoshita, Anal. Lett. 13 (1980) 191.
[9] W. Mansfield Clark, Oxidation–reduction Potentials of Or- [31] H. Tsuchiya, T. Hayashi, H. Naruse, N. Takagi, J. Chro-
ganic Systems, Bailliere, Tindall, and Cox, London, 1960. matogr. 234 (1982) 121.
[10] O.H. Muller, J. Am. Chem. Soc. 62 (1940) 2434. [32] M. Yamaguchi, S. Hara, R. Matsunaga, M. Nakamura, Y.
[11] B.R. Eggins, J.Q. Chambers, J. Electrochem. Soc. 117 (1970) Ohkura, J. Chromatogr. 346 (1985) 227.
186. [33] Y. Yasaka, M. Tanaka, T. Shono, T. Tetsumi, J. Katakawa,
[12] K. Takamura, T. Fuse, F. Kusu, Anal. Sci. 11 (1995) 979. J. Chromatogr. 508 (1990) 133.