25905071

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 1/12

ORIGINAL PAPER 

Comparison of two different solvents employed

for pressurised fluid extraction of stevioside

from   Stevia rebaudiana: methanol versus water

Jaroslav Pól   & Elena Varad ová Ostrá   & Pavel Karásek   &

Michal Roth   & Karolínka Benešová   &

Pavla Kotlaříková   & Josef  Čáslavský

Received: 15 October 2006 /Revised: 23 May 2007 /Accepted: 25 May 2007 / Published online: 27 June 2007# Springer-Verlag 2007

Abstract  Pressurised fluid extraction using water or metha-nol was employed for the extraction of stevioside from Stevia

rebaudiana Bertoni. The extraction method was optimised in

terms of temperature and duration of the static or the dynamic

step. Extracts were analysed by liquid chromatography

followed by UV and mass-spectrometric (MS) detections.

Thermal degradation of stevioside was the same in both

solvents within the range 70 – 

160 °C. Methanol showed better extraction ability for isolation of stevioside from   Stevia

rebaudiana  leaves than water within the range 110 – 160 °C.

However, water represents the green alternative to methanol.

The limit of detection of stevioside in the extract analysed

was 30 ng for UV detection and 2 ng for MS detection.

Keywords   Pressurised fluid extraction .

Liquid chromatography . Mass spectrometry.

Stevia rebaudiana . Stevioside

Introduction

Recently, an increasing demand has been noted for new

natural substitute sweeteners for sucrose and synthetic

sweeteners such as saccharine and aspartame. Stevioside

is a sweetener of plant origin possessing a 200 – 350 times

higher sweetening property than sucrose. Stevioside comes

from the South American perennial shrub  Stevia rebaudiana

Bertoni, and it is the most abundant among the nine known

sweet glycosides of the plant [1]. Application of stevioside

varies among countries according to their laws and

traditions; however, in the USA and Japan it has been a

 popular sweetener for years because of its low caloric

intake and no side effects. The reason for not permitting the

use of stevioside in some countries (e.g. EU countries) is

steviol, a possible harmful aglycone of stevioside. Steviol

has been proved to be a metabolite of stevioside produced

 by rat [2] and human [3, 4] digestion. On the other hand, it 

has been demonstrated that stevioside is not considerably

metabolically converted to steviol by chicken [5] and,

moreover, causes no harm to chicken embryos or to the

individual’s development [6]. Konoshima and Takasaki [7]

Anal Bioanal Chem (2007) 388:1847 – 1857

DOI 10.1007/s00216-007-1404-y

J. Pól : E. Varadová Ostrá : P. Karásek : M. Roth : K. Benešová :

P. Kotlař íková : J.  Čáslavský

Institute of Analytical Chemistry, Academy of Sciences

of the Czech Republic,

Veveř í 97,

611 42 Brno, Czech Republic

 Present address:

J. Pól (*)

Division of Pharmaceutical Chemistry, Faculty of Pharmacy,

University of Helsinki,

P.O. Box 56, 00014 Helsinki, Finland

e-mail: [email protected]

 Present address:

K. Benešová

State Phytosanitary Administration,

Zemědělská 1a,

613 00 Brno, Czech Republic

 Present address:

P. Kotlař íková

Pliva-Lachema,

Karásek 1,

621 33 Brno, Czech Republic

 Present address:

J.  Čáslavský

Faculty of Chemistry, Brno University of Technology,

Purkyňova 118,

612 00 Brno, Czech Republic

e-mail: [email protected]

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 2/12

have shown that stevioside has cancer-chemopreventive

effects in mice. As reviewed by Geuns [8], stevioside is

safe when used as a sweetener and there are no side effects

like mutagenicity, carcinogenicity or teratogenicity.

Stevioside is usually determined in Stevia rebaudiana by

hot water leaching or supercritical fluid extraction (SFE)

followed by liquid-chromatographic analysis of the extract.

Water leaching has been performed by multiple mixing of leaves with boiling water [9, 10], followed by filtration and

cleaning by passing the aqueous extract through an SPE

cartridge prior to liquid-chromatographic analysis. The low

cost of the instruments required has been countered by the

time and labour consumption. When water is replaced with

ethanol [11], the extraction step can be done more rapidly

while keeping the same extraction recovery. SFE employ-

ing CO2   as a medium for extraction is faster than the

 previous method. It benefits from the physical-chemical

 properties of supercritical CO2, which possesses a higher 

diffusivity and lower viscosity than conventional liquid

solvents. However, pure CO2   does not have sufficient solvation power for polar stevioside and therefore a polar 

cosolvent has to be added. Investigated cosolvents have

included methanol [12], water and/or ethanol [13], a mixture

of methanol and water [14], and water [15]. In most of the

 published papers, the highest extraction recoveries reported

were obtained at a pressure of 200 bar and a subcritical

temperature of 30 °C.

The procedures for isolation of stevioside from   Stevia

rebaudiana   leaves on a pilot scale have been summarised

 by Pasquel et al. [13]. The methods include liquid

extraction with different solvents — supercritical CO2   and

cosolvent, hot water or hot alcohols both employed below

their respective boiling points at ambient pressure. The

extracts have been purified by an additional adsorption, or 

 precipitation with a salt or alkali.

Stevioside is made up of the terpenic steviol with several

glycosidic substitutents attached. From this point of view,

the chromatographic separation of steviol glycosides

resembles that of sugars. A wide range of various chro-

matographic systems have been employed for the purpose.

The two systems used most frequently [16] are described

 below. Both of them are reversed-phase systems using an

acetonitrile – water mixture as a mobile phase. The first one

is based on the amino-substituted stationary phase [11, 17],

whilst the second one uses an octadecyl silica stationary

 phase [9, 18]. UV photometry has been used quite often for 

the detection [11,   18]. UV-spectra of stevioside are very

simple, with a maximum close to 200 nm. Unfortunately,

the detection close to 200 nm causes problems during

gradient elution in acetonitrile – water mixtures. Pulsed

amperometric detection [17] offers much better sensitivity,

 but there is the necessity of postcolumn addition of 0.1 M

 NaOH solution to the acetonitrile – water eluent. Mass-

spectrometric detection offers better sensitivity and unbeat-

en selectivity compared with UV detection. Electrospray

ionisation and negative ion detection have mostly been

applied [3,   14,   19]. Recently, two-dimensional liquid

chromatography with time-of-flight mass spectrometry

(MS) has been employed for analysis of aqueous   Stevia

extract; it has been used to separate stevioside and other 

sweet glycosides from the plant matrix within a singlechromatographic run [20].

Pressurised fluid extraction (PFE) has become a popular 

alternative to Soxhlet extraction or sonication because of its

multiple rate and lower solvent consumption. PFE operates

at temperatures above the boiling point of the extraction

solvent and therefore elevated pressure is used to keep the

solvent in the liquid state. Since high temperatures are used,

the method is limited only to thermally stable analytes. The

 principal benefits of the solvent properties in PFE include

enhanced mass transfer and increased solvation power. PFE

has been employed for the extraction of different target 

analytes from various matrices. Applications of PFE toenvironmental samples were reviewed in [21].

If the organic solvent in PFE is replaced with water, the

method is often termed   “ pressurised hot water extraction”

(PHWE) or subcritical water extraction when the extraction

temperature employed is below the critical temperature of 

water. The relative permittivity of water in the liquid state

decreases noticeably as the temperature increases, which

makes it possible to employ water as an extraction solvent 

for nonpolar compounds [22]. Last but not least, water is

also an environmentally friendly and economically beneficial

alternative to harmful organic solvents. Many papers have

reported hot water extraction of both environmental and food

samples and they were summarised in a recent review [23].

This paper presents a comparative study of two

extraction methods, PFE employing methanol and PHWE,

for   Stevia rebaudiana   leaves. Extraction efficiency and

degradation of solutes were studied. Extracts were analysed

 by liquid chromatography (LC) with both UV and mass-

spectrometric detection.

Experimental

Chemicals

Dried leaves of Stevia rebaudiana were obtained from Sigma

(Prague, Czech Republic). Stevioside (purity 97 mol%) was

 purchased from Latoxan (Valence, France). Deionised water 

used as an extraction medium for PHWE and as a mobile-

 phase component for LC was prepared in our laboratory

using an Ultra Clear UV device (SG – Wasseraufbereitung

und Regenerierstation, Hamburg, Germany). Acetonitrile

(SupraGradient Grade) used in the LC mobile phase and

1848 Anal Bioanal Chem (2007) 388:1847 – 1857

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 3/12

methanol used for PFE were purchased from Riedel-de Haën

(Prague, Czech Republic).

Extraction apparatus

One single laboratory-made apparatus was designed and

constructed for both PFE and PHWE (Fig.  1). The solvent 

reservoir was equipped with a purging unit for removingresidual oxygen by gently stripping with helium. The

extraction solvent (water or methanol) was delivered to

the extraction vessel using an LC1120 high-performance

LC (HPLC) pump (GBC Scientific Equipment, Dandenong,

Victoria, Australia) working in constant flow rate mode.

The solvent was preheated prior to it entering the extraction

vessel by an electronically controlled heater. Three types of 

extraction vessels were manufactured from a noncorrosive

alloy (INCONEL 625, Special Metals Wiggin, Hereford,

UK), and they could accommodate different volumes of 

samples: 11, 22, and 33 ml. They were equipped with a

removable 10-μ m frit on their outlets to avoid clogging of the outlet capillaries. The extraction vessel was placed in a

thermostatted block that allowed heating up to 350 °C. The

temperature of the thermostatted block was regulated with

an accuracy of ±0.1 °C by a microprocessor unit with

 proportional-integrated-derivative (PID) regulation (3116

temperature and process controller, Eurotherm, Durrington,

Worthing, UK). Further, the extract was led from the

extraction vessel to a pair of two independently thermo-

statted decompression units that either heated or cooled the

extract. Each thermostatted decompression unit consisted of 

stainless steel tubing (0.03-in. inner diameter) tightly

wound on a thermostatted aluminium block. The temper-

atures of the decompression units depended on the ex-

traction temperature and the boiling point of the solvent at 

ambient pressure; details are discussed in   “Results and

discussion”. All connections were made via stainless steelcapillaries (0.0625-in. outer diamter, 0.03-in. inner diameter).

Valve V1 operated in three positions allowing (1)

 purging of the extraction vessel with nitrogen, (2) intro-

duction of the extraction solvent into the extraction vessel,

and (3) closing of both previously described inputs. Valve

V2 was open during all experiments. Valve V3 had two

 positions and it controlled the duration of the static period

of extraction by the closing time. The function of the two-

 position valve V4 was to switch between the stainless steel

capillary outlet and the silica capillary outlet, and it was

used to control the dynamic mode of extraction.

The extraction apparatus can operate in three modes:(1) static, (2) dynamic, and (3) semidynamic. In this work,

only the modes 1 and 2 were employed and they are de-

scribed in   “Experimental procedure”.

Liquid chromatography – UV detection

The screening studies were performed using a laboratory-

assembled HPLC setup that included an LC1150 HPLC

Fig. 1   High-pressure extraction

apparatus for pressurised fluid

extraction (PFE) and pressurisedhot water extraction.  1  solvent 

reservoir, 2  high-performance

liquid chromatography pump,

3   nitrogen tank,  4  preheating

unit,  5  pressure sensor,  6  ex-

traction vessel placed in a ther-

mostatted block,  7 ,

8  thermostatted decompression

units,  9  collection vial for ex-

tract, 10  fused-silica restrictor,

11 proportional-integrated-de-

rivative control units,  V2,

V3   two-way valves,  V1,  V4

three-way SSI valves,  T  ther-

moinsulation covers

Anal Bioanal Chem (2007) 388:1847 – 1857 1849

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 4/12

quaternary pump (GBC Scientific Equipment, Dandenong,

VIC, Australia), a C-model sampling valve fitted with a 20-μ l

sampling loop (Ecom, Prague, Czech Republic), a Luna NH2

column (250 mm×4.6-mm inner diameter, particle size

5   μ m, Phenomenex, Aschaffenburg, Germany) housed in

an LCO 101 column oven (Ecom), and a SpectraFocus UV

detector (Thermo Separation Products, Fremont, CA, USA)

running in the scan mode. The column was kept at 40 °C.The mobile phase flow rate was 1 ml/min with gradient 

elution by acetonitrile (solvent A) and water (solvent B): 0

15 min 10% solvent A, 15 – 19 min 100% solvent A, and 19 – 

24 min 100% solvent A.

Liquid chromatography – mass spectrometry

LC with both UV and ion-trap MS detectors used for 

identification was performed using an Esquire-LC system

(Bruker Daltonics, Bremen, Germany). The liquid chro-

matograph was an Agilent 1100 series instrument consist-

ing of a vacuum degasser, a binary gradient pump, anautosampler, thermostatted a column compartment, and a

diode-array UV – vis detector. The mass spectrometer was

equipped with an electrospray ion source and spherical ion

trap analyser operated in negative ion detection mode. The

column employed was a Discovery RP Amide C16,

15 cm×2.1 mm, particles 5   μ m and precolumn 2 cm×

2.1 mm (both Supelco), flow rate 0.25 ml/min, binary

gradient acetonitrile – water: 0 min 30% acetonitrile , 10 min

60% acetonitrile, 13 min 60% acetonitrile, 15 min 30%

acetonitrile, stabilisation period 5 min. The column tem-

 perature employed was 35 °C and UV detection was carried

out at a wavelength of 210 nm. The injection volume was

2  μ l. Electrospray conditions were as follows: drying tem-

 perature 365 °C; drying gas nitrogen, flow 9 l/min; nebuliser 

gas nitrogen, pressure 40 psi; capillary voltage 4 kV.

Experimental procedure

The extraction step was started by heating the extractor to a

selected temperature. In all experiments, the block for 

accommodating the extraction vessel and the preheating

unit were heated to the same temperature. Meanwhile, dried

and ground   Stevia rebaudiana   leaves were weighed

(approximately 100 mg), mixed with sand, and directly

 put into the extraction cartridge. Then, the extraction

cartridge was inserted into the extraction device and the

whole system was immediately purged with nitrogen for 

2 min (switching by valve V1) to remove residual air. This

 prevented potential oxidation of a matrix, target analytes,

and materials used for setup of the apparatus.

The static mode of extraction was controlled only by

valve V3, valve V2 was open permanently and valve V4

was switched to the stainless steel capillary outlet. At the

 beginning of this mode, valve V3 was closed and the pump

was initiated to deliver the solvent. A pressure and tem-

 perature programme was applied during this step and the

details are in   “Extraction”. When the pressure reached the

set value at the set temperature, the pump automatically

turned off. This point was considered as the start of the

static extraction. The static extraction mode was ended by

careful opening of valve V3, causing slow depressurisationand release of the extract into the vial by a stream of nitrogen.

In the dynamic mode of extraction, valves V2 and V3

were permanently open and valve V4 was switched to the

fused silica capillary outlet (length 70 cm, inner diameter 

50  μ m). The pump started to deliver the solvent employing

a flow rate of 8 ml/min until the set pressure was reached.

Then the flow rate was changed to be the same as the outlet 

flow rate from the fused silica capillary. Similarly to the

static mode of extraction, this point was defined as the start 

of the extraction step. The dynamic mode was terminated

 by turning the pump off and switching valve V4 to the

stainless steel capillary outlet. The extract was transferredinto the vial by a stream of nitrogen.

The extract was made up to 50-ml volume and

subsequently injected into the liquid chromatograph.

The experiments to test the degradation of stevioside at 

elevated temperatures were performed in a similar manner.

The extraction cartridge was filled with glass balls

(diameter 1 mm) and then an aqueous standard solution of 

stevioside was dosed in. The extraction cartridge was

 placed in the extraction device and the extraction step was

initiated as described above.

Results and discussion

In the development of this comparative study of PFE and

PHWE, liquid-chromatographic separation with UV and

mass-spectrometric detection was optimised first by analy-

sis of the aqueous solution of the stevioside standard. The

extraction methods were optimised in terms of the

extraction time and temperature. The extraction perfor-

mances of water and methanol for the isolation of stevio-

side were compared as well as the influence of the

extraction media employed, and the effects on the thermal

degradation of stevioside were also compared.

Liquid-chromatograhic analysis

LC was used with both UV and ion-trap MS detectors. The

calibration curve was constructed from five points (0.1031 – 

1.0317 mg/ml) according to the typical concentration of 

stevioside in the extract. For UV detection, a linear 

calibration line was used ( R=0.9999). For MS a nonlinear 

(quadratic) dependence was employed ( R=0.9999). The

1850 Anal Bioanal Chem (2007) 388:1847 – 1857

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 5/12

limit of detection defined for a signal-to-noise ratio of 3

was 30 ng for UV detection and 2 ng for ion-trap MS.

Stevioside has two absorption maxima in the UV region,

210 and 405 nm. The more intense maximum of the two – 

210 nm – was used for UV detection. The reproducibility of 

five-times repeated injection was 2.1%. The mass spec-

trometer was operated in the negative ion mode with

“universal”   detection of the characteristic fragment ion of stevioside (m/  z  641) and it was used for quantification. The

details of the stevioside mass spectra are discussed in the

next section.

LC-MS analysis of stevioside standard solution

The standard of stevioside that is available on the market 

has only a 97 mol% purity. Other sweet glycosides present 

in Stevia, the source of stevioside, were also observed in the

LC-MS chromatogram (Fig.   2). Stevioside was the most 

abundant peak in both UV and total ion chromatography

traces (6.2 – 6.3 min). MS/MS of the deprotonated molecular ion of stevioside (m/z  803) resulted in three fragments (m/z 

641, m/z  479, and  m/z  317) that referred to the loss of one,

two, and three molecules of glucose, respectively (Fig.  3).

m/z   641 and   m/z   479 were also observed as fragments

coming from the ion source in single MS mode. The peak 

at 7.5 min with intense   m/z    6 11 migh t refer to a

deprotonated structure consisting of a molecule of steviol

with two molecules of sucrose. Rebaudioside C was

identified as a deprotonated molecular ion (m/z   949) at 

7.2 min. MS/MS of this precursor produced fragment  m/z  787

(observed also as an ion source fragment in single MS

mode), indicating the loss of one molecule of glucose, and

MS3 showed the steviol fragment, thus proving rebaudioside

C was present. Rebaudioside A was eluted at 6.1 min and it 

was characterised by a deprotonated molecular ion at   m/z 

965. Rebaudioside B has the same molecular formula as and

a very similar structure to stevioside, and therefore it is

difficult to separate the two glycosides employing single-column LC-MS. In addition, we suppose that the MS/MS

spectra of both compounds would have very similar patterns

owing to their structural similarity (three glucose units on the

steviol skeleton). The percentage of sweet glycosides in the

 plant depends on the growing conditions, and it has been

reported to be 43 – 80% for stevioside and 0 – 0.02% for 

rebaudioside B [20]. The content of rebaudioside B is

negligible in comparison with that of stevioside, and

therefore the peak at 6.2 – 6.3 min was considered to pertain

to stevioside only. Structures of the sweet glycosides present 

in  Stevia  are shown in Fig.  4.

In general, the MS/MS experiments are known toimprove the sensitivity and selectivity. Unfortunately, in

the present case the MS/MS mode did not result in any

significant advantage over the simple MS mode. Under 

the low-energy fragmentation on the spherical ion trap,

the isobaric compounds stevioside and rebaudioside B

which are coeluted under the conditions used in our 

experiments would have similar fragmentation patterns

consisting of sequential cleavage of three glucose units.

However, this is only a hypothesis that could not be

verified because of the absence of the rebaudioside B

4.8

6.2 004-0401.D: TIC ±All

004-0401.D: UV Chromatogram, 210.4 nm

6.3

8.0 004-0401.D: EIC 479 -All, Smoothed (3.2,1, GA)

7.5004-0401.D: EIC 611 -All, Smoothed (3.2,1, GA)

4.8

6.3 004-0401.D: EIC 641 -All, Smoothed (3.2,1, GA)

7.3

11.9

004-0401.D: EIC 787 -All, Smoothed (3.2,1, GA)

6.1 004-0401.D: EIC 803 -All, Smoothed (3.2,1, GA)

7.2004-0401.D: EIC 949 -All, Smoothed (3.2,1, GA)

6.1

004-0401.D: EIC 965 -All, Smoothed (3.2,1, GA)

0

2

4

7x10

Intens.

0

2

0

1

2

5x10

0.0

0.5

1.0

1.55x10

0

1

2

7x10

0.0

0.5

1.0

5x10

0

2

4

5x10

0

2

4

6

4x10

0

2

4

5x10

2 4 6 8 10 12 Time [min]

Fig. 2   Chromatogram traces of 

stevioside standard; total

ion chromatography (TIC ), UV

at 210 nm, electrostatic ion

chromatography ( EIC )  m/  z  479,

EIC  m/  z  611, EIC  m/  z  641,

EIC  m/  z  787, EIC  m/  z  803, EIC

m/  z  949, and EIC  m/  z  965.

Stevioside is presented at 6.2 – 

6.3 min as [M-H]- at  m/  z  803

and [M-glucose]- at  m/  z  641

Anal Bioanal Chem (2007) 388:1847 – 1857 1851

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 6/12

standard on the market. Since the experimental setup and

optimisation of the MS/MS experiments is much more

complicated and time-consuming than that of simple MS,

we finally decided to use the simple MS mode for the

quantitation.

Extraction

In some static PFE experiments described in the literature,

the extraction cell was heated only after it had been filled

with the sample and pressurised with the (cold) solvent.

This procedure can result in severe overpressurising of the

cell during heating. To relieve the pressure, it may be

necessary to release some solvent from the cell to the

collection vial even before the cell has reached the

operating temperature. Obviously, this results in an uncer-

tain composition of the final extract because different 

 portions of the extract correspond to different extraction

conditions. Such an uncertainty is highly undesirable,

especially if analyte degradation processes are to bestudied. In this work, therefore, a different procedure was

employed. After filling the cell with the sample and purging

it with nitrogen, the cell was pressurised to about a quarter 

of the operating pressure, then it was heated to the

operating temperature for 2 min and carefully pressurised

to the operating pressure. In this way, the composition

uncertainty mentioned above was avoided.

Since large differences exist between PFE and PHWE as

regards the solvent properties, special attention was paid to

decompression of the extracts. In PFE with organic

solvents, the solubility of extracted nonpolar analytes is

usually sufficient even after decompression and cooling of 

the extract to ambient pressure and ambient temperature. In

this case, the analytes are quantitatively transferred into the

collection vial. In PHWE, however, the decompression

 process is more complicated because the aqueous solubil-

ities of nonpolar analytes fall dramatically as the temper-ature decreases. After decompression and cooling of the

aqueous extract, the analytes can precipitate on their way

from the extraction vessel to the collection vial. This

usually results in nonquantitative transfer of the analytes

from the extraction vessel or, in the worst case, in plugging

of the transfer capillaries. Therefore, as mentioned in

“Extraction apparatus”, a series of two thermostatted

decompression units were placed between the outlet of the

extraction vessel and the collection vial (Fig.   1). The

 purpose of the units was to avoid possible precipitation of 

the analytes and to prevent boiling of the extract at the

extractor outlet. The temperature of each unit was electron-ically controlled and could be adjusted according to the

extraction solvent employed. The temperatures of the two

units were 110 and 95 °C for PHWE and 110 and 60 °C for 

PFE with methanol, respectively. Therefore, the down-

stream unit (no. 8 in Fig.   1) was always kept at a

temperature below the normal boiling point of the extrac-

tion solvent. Also, precipitation was thereby prevented even

in nonpolar constituents of the matrix that could be

extracted from the sample. After cooling to room temper-

317.1

413.3

479.2

641.3

803.2

161.9

162.1

162.1

18.1

18.1

65.9

100 200 300 400 500 600 700 800 m/z

0.0

0.2

0.4

0.6

0.8

1.0

4x10

Intens.Fig. 3   Tandem mass spectrom-

etry fragmentation spectra of the

deprotonated molecular ion of 

stevioside [M-H]- m/  z  803.2

indicating the loss of three

glucose units (m/  z  162)

1852 Anal Bioanal Chem (2007) 388:1847 – 1857

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 7/12

ature, no precipitation was observed in the   Stevia   extracts

from both PFE and PHWE.

The static and the dynamic mode, both lasting 10 min,

were compared for both solvents at 50 bar and 110 °C. The

same extraction yield was achieved in both modes. Further,

we decided to use the simpler static mode in the inves-

tigations described below.

For both PFE and PHWE, several approaches to the preparation of   Stevia   leaves before the extraction were

tested. The leaves were ground and mixed with sand or 

glass balls or were just untreated and then inserted into the

extraction cartridge. Extractions performed under the same

conditions (50 bar, 110 °C, 10 min) resulted in very similar 

extraction yields for all the pretreatments. In the subsequent 

experiments, the untreated   Stevia   leaves were extracted

directly; this procedure shortened the total time of the

analysis and also eliminated the source of possiblecontamination of the sample.

Fig. 4   Structures of steviol,

stevioside, and other glycosides.

The constituents attached to the

 base structure of steviol are

glucose (Glc), rhamnose ( Rha),

and xylose ( Xyl )

Anal Bioanal Chem (2007) 388:1847 – 1857 1853

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 8/12

Fig. 5   Degradation of stevio-

side in water (circles) and

methanol ( squares) as a function

of temperature. The experiment 

was repeated three times with a

relative standard deviation of 

3 – 5%. Relative recovery of 

100% is equal to the concentra-

tion of stevioside in the standard

solution

6.2

005-0501.D: EIC 479 -All, Smoothed (3.2,1, GA)

4.8

6.3005-0501.D: EIC 641 -All, Smoothed (3.2,1, GA)

6.1

005-0501.D: EIC 803 -All, Smoothed (3.2,1, GA)

6.2

10.1

005-0501.D: EIC 965 -All, Smoothed (3.2,1, GA)

0.0

0.5

1.0

1.5

4x10

Intens.

0.0

0.5

1.0

1.5

6x10

0

1

2

3

4x10

0.0

0.5

1.0

1.5

2.0

4x10

2 4 6 8 10 12 Time [min]  Fig. 6   EIC traces of stevioside standard treated at 110 °C: m/  z  479,  m/  z  641,  m/  z  803, and  m/  z  965

1854 Anal Bioanal Chem (2007) 388:1847 – 1857

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 9/12

Degradation of stevioside

Thermal degradation of stevioside in aqueous solution has

 been reported; the maximum temperatures employed were

80 °C [24] and 100 °C [25] and the degradation products

have not been identified. The study of stevioside degrada-

tion at elevated temperatures and the determination of the

degradation products provides important guidelines to carry

out the PFE and PHWE experiments.

A 1-ml aliquot of stevioside standard solution (1 mg/ml)

in water was dosed into an empty extraction cartridge.

Then, the normal extraction procedure followed. The water 

4.8

006-0601.D: EIC 479 -All, Smoothed (3.2,1, GA)

4.8

6.2

006-0601.D: EIC 641 -All, Smoothed (3.2,1, GA)

4.8

6.3

006-0601.D: EIC 803 -All, Smoothed (3.2,1, GA)

006-0601.D: EIC 965 -All, Smoothed (3.2,1, GA)

0.0

0.5

1.0

1.5

4x10

Intens.

0.0

0.5

1.0

1.5

6x10

0

1

2

3

4

5

4x10

-1

0

1

2

4x10

2 4 6 8 10 12 Time [min]  

Fig. 7   EIC traces of stevioside

standard treated at 160 °C:

m/  z  479,  m/  z  641,  m/  z  803, and

m/  z  965

Fig. 8   Extraction efficiency

of stevioside from  Steviarebaudiana leaves as a function

of temperature and extraction

solvent; water (circles) and

methanol ( squares)

Anal Bioanal Chem (2007) 388:1847 – 1857 1855

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 10/12

effluent was analysed by LC-MS for stevioside and

degradation products. A very similar decrease in the con-

centration of stevioside with increasing temperature was

observed in both solvents (both in water and in methanol)

(Fig. 5). On the other hand, there was an increase in the area

of the peak at a retention time of 4.8 min starting at 110 °C

(Fig.   6), and this continuously increased up to 160 °C

(Fig. 7). The molecular mass and the MS/MS spectra of thischromatographic peak were the same as those of stevioside.

A rearrangement of the stevioside molecule probably

occurred under the high-temperature treatment, so the

molecular mass remained the same but the retention in LC

changed owing to different interaction with the stationary

 phase. Unfortunately, the changed structure could not be

described with the current instrumentation.

Extraction with methanol

The extraction temperatures tested ranged from 70 to 160 °C,

and the extraction was run in static mode lasting 10 min. The pressure was kept at 50 bar. Figure   8   shows the gradual

increase of extraction yield with temperature within the

whole profile.

The suggested temperature for carrying out further 

experiments was 110 °C because (1) the extraction yield of 

stevioside reached an almost quantitative relative recovery

 plateau and (2) the degradation of stevioside was not yet 

significant. A higher temperature would result in a slightly

higher extraction yield but also in significant degradation of 

stevioside, and the presence of degradation products could

result in errors and interferences in the analysis.

The influence of the duration of the extraction step on

the extraction yield was tested at 110 °C and 50 bar for 10,

20, and 30 min. The extraction yield was same for all three

times tested. Enhanced solvation power and diffusivity of 

subcritical methanol were probably sufficient for dissolving

and releasing stevioside from the matrix within 10 min.The proposed extraction conditions for PFE with

methanol are 110 °C, 50 bar, and 10 min of static extraction

time. Fragmentograms showing the distribution of main

components in the extract obtained are presented in Fig. 9.

Five-times repeated extraction of   Stevia rebaudiana   leaves

was accomplished with a reproducibility of 5.2% for 

stevioside.

Extraction with water 

Extraction with water was performed in a similar manner to

the extraction with methanol. The extraction temperaturestested were from 70 to 160 °C at 50 bar and the static

extraction period lasted 10 min. The extraction yield of 

stevioside increased continuously up to 110 °C and then a

linear decrease was observed (Fig.   8). Stevioside has a

 polar character and its solubility probably decreases at 

temperatures above 110 °C; therefore, the extraction yield

is lower at temperatures higher than 110 °C. Together with

thermal degradation of stevioside, the extraction yield

affects the shape of the extraction curve. The proposed

8.0

6.3

007-0701.D: EIC 479 -All, Smoothed (2.2,1, GA)

7.5

007-0701.D: EIC 611 -All, Smoothed (2.2,1, GA)

7.4

007-0701.D: EIC 625 -All, Smoothed (2.2,1, GA)

6.2

4.8 7.6

007-0701.D: EIC 641 -All, Smoothed (2.2,1, GA)

7.2

007-0701.D: EIC 787 -All, Smoothed (2.2,1, GA)

6.1

4.8

007-0701.D: EIC 803 -All, Smoothed (2.2,1, GA)

7.3

007-0701.D: EIC 949 -All, Smoothed (2.2,1, GA)

6.1

007-0701.D: EIC 965 -All, Smoothed (2.2,1, GA)

0

2

5x10

Intens.

0

1

2

5x10

0.0

0.5

1.0

6x10

0.0

0.5

1.0

7x10

0

2

5x10

0.0

0.5

6x10

0

1

5x10

0

2

4

5x10

2 4 6 8 10 12 Time [min]

Fig. 9   Liquid-chromatographicanalysis of PFE extract.  m/  z  479:

6.3 min — fragment of stevio-

side, 8.0 min — unknown com-

 pound;  m/  z  611: 7.5 min — 

steviol with two sucrose mole-

cules; m/  z  625: unknown com-

 pound;  m/  z  641: 4.8 min — 

compound with molecular mass

804 (possible structural isomer 

of stevioside), 6.2 min — frag-

ment of stevioside, 7.6 min — 

unknown compound;  m/  z  787:

7.2 min — fragment of rebaudio-

side C;  m/  z  803: 4.8 min — 

compound with molecular mass

804 (possible structural isomer 

of stevioside, it gives fragment 

m/  z  641), 6.1 min — stevioside

(molecular peak);  m/  z  949:

7.3 min — rebaudioside

C (molecular peak); m/  z  965:

6.1 min — rebaudioside A

1856 Anal Bioanal Chem (2007) 388:1847 – 1857

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 11/12

temperature for extraction of stevioside from   Stevia

rebaudiana  leaves with water is 110 °C.

In the same way as for methanol, a static extraction step

of 10-min duration proved to be sufficient for quantitative

relative extraction recovery.

The optimised parameters for PHWE of stevioside from

Stevia rebaudiana leaves are 110 °C, 50 bar, and 10 min of 

static extraction time, the same as for methanol. Thereproducibility of PHWE evaluated from five-times repeat-

ed extraction was 4.7%.

Conclusion

A temperature of 110 °C was determined to be optimal for 

extraction of stevioside from Stevia rebaudiana  leaves using

either water or methanol. An increased temperature resulted

in significant degradation of stevioside in the environment of 

 both solvents or in a decline in the extraction yield in water.

Both solvents demonstrated stevioside extraction with verysimilar reproducibility and the proposed extraction parame-

ters are the same for both methods. The total analysis time,

including sample handling, extraction, and analysis, was

50 min. Water represents a more environmentally conscious

and cheaper alternative to methanol. Compared with current 

extraction methods, our method is faster owing to the accel-

erated mass transfer in subcritical solvents. Extraction with

 pressurised hot water also presents a possibility for upscaling

the extraction process and establishing a   “green” method for 

isolation of stevioside sweetener from Stevia rebaudiana.

Acknowledgements   Financial support from the Grant Agency of the

Academy of Sciences of the Czech Republic (project no. B4031405)

and from the Czech Science Foundation (project no. 203/05/2106) is

gratefully acknowledged.

References

1. Crammer B, Ikan R (1986) Chem Br 22:915 – 916

2. Pezzuto JM, Compadre CM, Swanson SM, Nanayakkara NPD,

Kinghorn AD (1985) Proc Natl Acad Sci USA 82:2478 – 2482

3. Koyama E, Kitazawa K, Ohori Y, Izawa O, Kakegawa K, Fujino

A, Ui M (2003) Food Chem Toxicol 41:359 – 374

4. Gardana C, Simonetti P, Canzi E, Zanchi R, Pietta P (2003) J

Agric Food Chem 51:6618 – 66225. Geuns JMC, Malheiros RD, Moraes VMB, Decuypere EMP,

Compernolle F, Buyse JG (2003) J Agric Food Chem 51:1095 – 1101

6. Geuns JMC, Bruggeman V, Buyse JG (2003) J Agric Food Chem

51:5162 – 5167

7. Konoshima T, Takasaki M (2002) Pure Appl Chem 74:1309 – 1316

8. Geuns JMC (2003) Phytochemistry 64:913 – 921

9. Bovanová L, Brandšteterová E, Baxa S (1998) Z Lebensm Unters

Forsch A 207:352 – 355

10. Vaněk T, Nepovím A, Valíček P (2001) J Food Compos Anal

14:383 – 388

11. Kolb N, Herrera JL, Ferreyra DJ, Uliana RF (2001) J Agric Food

Chem 49:4538 – 4541

12. Liu J, Ong CP, Li SFY (1997) J Chromatogr Sci 35:446 – 450

13. Pasquel A, Meireles MAA, Marques MOM, Petenate AJ (2000)

Braz J Chem Eng 17:271 – 28214. Choi YH, Kim I, Yoon KD, Kee D, Lee SJ, Kim CY, Yoo KP,

Choi YH, Kim J (2002) Chromatographia 55:617 – 620

15. Yoda SK, Marques MOM, Petenate AJ, Meireles MAA (2003) J

Food Eng 57:125 – 134

16. Herbreteau B (1992) Analusis 20:355 – 374

17. Ahmed MJ, Smith RM (2002) J Sep Sci 25:170 – 172

18. Hutapea AM, Toskulkao C, Wilairat P, Buddhasukh D (1999) J

Liq Chromatogr Relat Technol 22:1161 – 1170

19. Koyama M, Yoshida K, Uchibori N, Wada I, Akiyama K, Sasaki

T (2005) J Food Hyg Soc Jpn 46:72 – 78

20. Pól J, Hohnová B, Hyötyläinen T (2006) J Chromatogr A

1150:85 – 92

21. Morales-Muñ S, Luque-García JL, Luque de Castro MD (2003)

Crit Rev Environ Sci Technol 33(4):391 – 421

22. Smith RM (2002) J Chromatogr A 975:31 – 4623. Ong ES, Cheong JSH, Goh D (2006) J Chromatogr A 1112:

92 – 102

24. Kroyer GT (1999) Lebensm -Wiss Technol 32:509 – 512

25. Chang SS, Cook JM (1983) J Agric Food Chem 31:409  – 412

Anal Bioanal Chem (2007) 388:1847 – 1857 1857

8/12/2019 25905071

http://slidepdf.com/reader/full/25905071 12/12