1138 Liu et al.: Journal of AOAC International Vol. 95, No.
4, 2012
FOOD COMPOSITION AND ADDITIVES
Determination of Trigonelline, Nicotinic Acid, and Caffeine in
Yunnan Arabica Coffee by Microwave-Assisted Extraction and
HPLC with Two Columns in Series
Hongcheng Liu, Jinliang Shao, Qiwan Li1, Yangang Li, Hong Mei Yan, and Lizhong He
Supervision and Testing Center for Farm Product Quality, Ministry of Agriculture (Kunming), Institute of Quality Standard and
Testing Technology, Yunnan Academy of Agriculture Science, 650223, Kunming, People’s Republic of China
A simple, rapid method was developed for
simultaneous extraction of trigonelline, nicotinic
acid, and caffeine from coffee, and separation
by two chromatographic columns in series. The
trigonelline, nicotinic acid, and caffeine were
extracted with microwave-assisted extraction (MAE).
The optimal conditions selected were 3 min, 200 psi,
and 120°C. The chromatographic separation was
performed with two columns in series, polyaromatic
hydrocarbon C18 (250 × 4.6 mm id, 5 µm particle size)
and Bondapak NH2 (300 × 3.9 mm id, 5 µm particle
size). Isocratic elution was with 0.02 M phosphoric
acid–methanol (70 + 30, v/v) mobile phase at a flow
rate of 0.8 mL/min. Good recoveries and RSD values
were found for all analytes in the matrix. The LOD of
the three compounds was 0.02 mg/L, and the LOQ
was 0.005% in the matrix. The concentrations of
trigonelline, nicotinic acid, and caffeine in instant
coffee, roasted coffee, and raw coffee (Yunnan
Arabica coffee) were assessed by MAE and hot water
extraction; the correlation coefficients between
concentrations of the three compounds obtained were
close to 1.
C
offee is a very popular beverage all over the world, with
numerous species in use. Good coffee has characteristic
taste and aroma. Typical flavor compounds in coffee are
caffeine, trigonelline, and nicotinic acid (1).
The contents of caffeine, trigonelline, and nicotinic acid
in commercial coffee may be influenced by coffee species,
variety, geographical origin, and roasting conditions (2).
In addition to their impact on coffee quality and flavor, the
trigonelline/ nicotinic acid ratio can also be used as an indicator
of roasting degree by the coffee industry. Therefore, the
simultaneous determination of these components should be a
useful tool for QC of raw materials and monitoring of coffee
roasting condition (3).
There are several chromatographic methods (1, 4–6) for
simultaneous determination of trigonelline, nicotinic acid, and
caffeine, all of which use simple extraction of coffee by hot
water for 2 h or more. Microwave-assisted extraction (MAE)
has been used by analysts in the past few years as a refined
Received June 23, 2011. Accepted by SG August 10, 2011.
1
Corresponding author’s e-mail: cmliu_0@sina.com
DOI: 10.5740/jaoacint.11-275
technique to facilitate extraction of analytes from food (7–9).
However, there are no reports on MAE for coffee analysis. The
goal of this research was to develop a method based on MAE
and separation of trigonelline, nicotinic acid, and caffeine by
HPLC with two chromatographic columns in series.
Experimental
Chemicals and Plant Samples
All reagents were of analytical grade, except for methanol
(HPLC grade) from Merck (Darmstadt, Germany); trigonelline
and nicotinic acid (chemically pure, 99%) from Sigopharm
Chemical Reagent Co. Ltd (Beijing, China), and caffeine
(chemically pure, 99%) from Farco Chemical Supplies (Hong
Kong, China).
Raw and roasted coffee (Yunnan Arabica coffee) were
obtained from Baoshan, Yunnan Province, China. They were
ground to a fine powder using a Waring (Hunan, China) HD100
blender at high speed (20 000 rpm) for a short time to avoid loss
of sample components. The powder, particle size <1 mm, was
stored frozen at 4°C until use. Instant coffee (from industry of
Yunnan Province, China) was bought from a market.
Apparatus
A closed-vessel microwave system, Model Mars-5, with a
rotating tray with 14 tube holes was purchased from CEM Corp.
(Matthews, NC). A maximum control pressure at 300 psi and
maximum control temperature of 200°C were attained.
MAE
A 200 mg sample was accurately weighed into a Teflon
pressure vessel, and 20 mL water was added. The vessel was
tightly sealed and placed in the microwave system.
After MAE, the sample was cooled to room temperature. The
mixture was filtered through 18 cm quantified filter paper (from
Special Paper Co. Ltd, Hangzhou, China) and the final volume
was made up with water to 50 mL. The solution was filtered
with a 0.45 µm membrane (Jinteng, TianJing, China), and 10 µL
was injected for HPLC.
Traditional Extraction
Extraction was performed by a reported method (1, 3–5) with
a small modification. Initially, a 200 mg sample was mixed with
 Liu et al.: Journal of AOAC International Vol. 95, No. 4, 2012 1139

Figure 1. Chromatogram of the three compounds

with PAH C18 (250 × 4.6 mm id, 5 µm partice size) and
Bondapak NH2 (300 × 3.9 mm id, 5 µm particle size)
columns in series, and the mobile phase 0.02 mol/L
phosphoric acid–methanol (70 + 30, v/v) at a flow
rate of 0.8 mL/min. (1) trigonelline; (2) nicotinic acid;
(3) caffeine.

50 mL water in a flask, and the solution was heated at 80°C for

2 h.

HPLC Analysis Conditions

An Alliance-2695 HPLC system (Waters Corp., Milford,

MA) equipped with an autosampler, quaternary pump system,
UV detector, thermostatted column compartment, degasser, Figure 2. Chromatograms of hot water:
and Empower II software was used. The analytical columns (a) extraction and (b) MAE and extracts. 1, trigonelline;
were polyaromatic hydrocarbon (PAH) C18 (250 × 4.6 mm 3, caffeine; 4, 5, unknown compounds.
id, 5 µm particle size) and Bondapak NH2 (300 × 3.9 mm id,
5 µm particle size) from Waters Corp. The mobile phase
was 0.02 M phosphoric acid–methanol (70 + 30, v/v) and was compounds such as caffeine, and the column with NH2 functional
pumped at 0.8 mL/min. The column temperature was 30°C, and groups can separate trigonelline. It was an unexpected result to
the detector wavelength was 260 nm. obtain the resolution of trigonelline, nicotinic acid, and caffeine
by use of PAH C18 and NH2 columns in series. Figure 1 shows
a typical chromatogram for separation of trigonelline, nicotinic
Quantification and Data Analysis
acid, and caffeine. Fast retention times (<12 min) and resolution
obtained by the two columns in series with isocratic elution
Standard solutions of trigonelline, nicotinic acid, and
using 0.02 mol/L phosphoric acid–methanol (70 + 30, v/v)
caffeine were prepared in water (1000 mg/L) and kept in the
mobile phase were better than values reported in the literature
dark under refrigeration at 4°C. For the calibration curve, stock
(1, 4).
solutions were diluted to five series of concentrations. The
The calibration curves showed good linearity with the
chromatographic system determined the following parameters:
following equations and relative coefficients:
repeatability and reproducibility of retention time (Rt) and peak
area, and sensitivity. y = 1.71 e4x + 2.62 e3 (R = 0.9952),
The contents of trigonelline, nicotinic acid, and caffeine in
20 samples were compared for the two extraction procedures
with analysis by HPLC.
y = 1.21e4x – 0.62 e4 (R = 0.9936), and
Results and Discussion

The original mobile phase, methanol–water (82 + 18, v/v),
had been successfully used for trigonelline in an NH2 column,
but the resolution of nicotinic acid and caffeine was different in
NH2 or the resolution of trigonelline and nicotinic acid in PAH
columns.
Here, properties of stationary phases were different between
the two columns; contrary to octadecylsilane C18, the stationary
phase with PAH functional groups can easily separate aromatic
y = 2.43e4x – 6.67 e3 (R = 0.9968)
for trigonelline, nicotinic acid, and caffeine, respectively;
y = concentration and x = peak area. The LOD value for each
compound was 0.02 mg/L, and the LOQ values were all 0.005%
in the matrix. Samples with concentrations lower than the LOQ
were considered not quantifiable. Reproducibility of Rt and
peak area was good for 1 month; all values were 4% (Rt) and
1140 Liu et al.: Journal of AOAC International Vol. 95, No. 4, 2012

7% (peak area) or less. Furthermore, repeatability of Rt and peak

area, which was evaluated on 1 day, proved more satisfactory,
trigonelline content with

1.2 with values of 2.5% (Rt and peak area) or less.

microwave assisted

y = 1.0009x - 0.0088
1
extraction (%)

R2 = 0.9914
0.8 Optimization of MAE Conditions
0.6
0.4 MAE was performed for 0 to 30 min at 120°C, 300 W and
0.2 150°C, 600 W. The MAE setting at 120°C, 300 W was preferred
0 because the higher power setting often caused rapid boiling
0 0.2 0.4 0.6 0.8 1 1.2 that resulted in a break in the seal of the lid. The time required
trigonelline content with hot water extraction
(%)
to achieve quantitative recoveries was chosen as 3 min, and
the results were comparable with those obtained at 10 min or
longer. Under the experimental conditions tested, no breakdown
with microwave assisted

0.12
nicotinic acid content

y = 0.9291x + 0.001
of trigonelline, nicotinic acid, or caffeine was observed. It
0.1
extraction (%)

2
R = 0.9679 should be emphasized that these settings are specific for the
0.08
CEM MARS-5 microwave oven.
0.06
Figure 2 shows chromatograms of the three compounds
0.04
from a hot water extraction and MAE of coffee matrix. No
0.02
interfering peaks were observed for the quantification of the
0
three compounds. In the case of MAE, two peaks (unknown
0 0.02 0.04 0.06 0.08 0.1 0.12
nicotinic acid content with hot water compounds, relatively lower than the peak of caffeine) were
extraction (%) found to the left of caffeine, while only one peak (unknown
compound, relatively higher than the peak of caffeine) was
1.4 found for the hot water extraction.
caffeine content with

1.2 The concentrations of trigonelline, nicotinic acid, and caffeine

microwave assisted

y = 0.9758x - 0.0024
extraction (%)

2
1 R = 0.9855 in five instant, five roasted, and five raw coffee samples were
0.8 assessed with traditional extraction and MAE. Figure 3 presents
0.6 a comparison between values obtained with the two methods for
0.4 the three compounds. MAE was performed in a shorter time (3
0.2
min) compared to the conventional method (2 h).
0
Nicotinic acid was not found in the most analyzed raw
0 0.2 0.4 0.6 0.8 1 1.2
caffeine content with hot water extraction (%) coffee sample. This was expected, since it was produced as
a consequence of trigonelline thermal degradation during the
Figure 3. Regressions analysis of trigonelline, roasting process. Among commercial instant coffee samples,
nicotinic acid, and caffeine for the two extraction the values of nicotinic acid were lower than 0.01%. These levels
methods. were in agreement with those reported elsewhere (3).
Caffeine values ranged from 0.875 to 1.121% in raw and
roasted coffee. But caffeine amounts in instant coffee were
lower than those reported (3).

Table 1. Recovery test results

Recovery, % RSD, %

Sample Spiking value, g/100 g Trigonelline Nicotinic acid Caffeine Trigonelline Nicotinic acid Caffeine

Instant coffee 0.02 98.2 93.2 92.4 3.02 3.19 4.17

0.04 95.7 90.7 96.4 1.29 2.29 2.11

0.08 101.3 89.3 93.2 2.28 3.23 4.65

Roast coffee 0.1 96.5 98.5 99.1 2.03 1.17 3.21

0.2 92.3 100.3 93.6 3.21 4.19 3.11

0.4 95.1 94.1 101.7 3.02 3.14 2.48

Raw coffee 0.4 103.2 101.2 103.4 2.09 3.05 4.82

0.8 94.6 97.6 93.8 2.02 2.92 1.35

1.2 97.3 93.3 99.1 4.17 3.92 1.43

 Liu et al.: Journal of AOAC International Vol. 95, No. 4, 2012
Recovery Study
The accuracy of the whole HPLC method was evaluated
by development of a recovery study carried out at three
concentration levels. Trigonelline, nicotine acid, and caffeine
were determined five times at each concentration. Table 1
shows that the mean recovery was 92.3–103.2%, and RSD was
4.2% or less.
References
(1) Casal, S., Oliverira, M.B., Alves, M.R., & Ferreira, M.A. (2000)
J. Agric. Food Chem. 48, 3420–3424. http://dx.doi.org/10.1021/
jf990702b
(2) Ky, C.L., Dussert, S., Guyot, B., Hamon, S., & Noirot, M.
(2001) Food Chem. 75, 223–230. http://dx.doi.org/10.1016/
S0308-8146(01)00204-7
1141
(3) Perrone, D., Donangelo, C.M., & Farah, A. (2008) Food
Chem. 110, 1030–1035. http://dx.doi.org/10.1016/j.
foodchem.2008.03.012
(4) Casal, S., Oliveira, M.B., & Ferreira, M.A. (1998) J. Liq.
Chromatogr. Relat. Technol, 21, 3187–3195. http://dx.doi.
org/10.1080/10826079808001267
(5) Casal, S., Oliveira, M.B., & Ferreira, M.A. (2000) Food Chem.
68, 481–485. http://dx.doi.org/10.1016/S0308-8146(99)00228-9
(6) De Maria, C.A.B., Trugo, L.C., Moreira, R.F.A., &
Petracco, M. (1995) Food Chem. 52, 447–449. http://dx.doi.
org/10.1016/0308-8146(95)93298-6
(7) Singh, S.H.B., Foster, G.D., & Khan, S.H.U. (2004) J. Agric.
Food Chem. 52, 105–109. http://dx.doi.org/10.1021/jf030358p
(8) Huang, J.L., & Zhang, Z.X. (2004) Anal. Sci. 20, 395–397.
http://dx.doi.org/10.2116/analsci.20.395
(9) Groener, J.E.M., Poorthuis, B.J.H.M., Kuiper, S., Helmond,
M.T.J., Hollak, C.E.M., & Aerts, J.M.F.G. (2007) Clin. Chem.
53, 742–747. http://dx.doi.org/10.1373/clinchem.2006.079012
Copyright of Journal of AOAC International is the property of AOAC International and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.