Analysis of Herbal Medicines

and Healthcare Products

Application Compendium
Introduction This compendium is a collection of applications on the analysis of traditional
herbal medicines and healthcare products. The applications are grouped
according to the types of substances being analyzed. These include pesticide
residues, heavy metals, residual solvents, bioactive compounds and metabolites.
All applications featured in this compendium have been developed with
complete solutions using Agilent analytical products, software and services.
The analytical techniques deployed include gas chromatography (GC), liquid
chromatography (LC) and mass spectrometry (MS) as well as hybrid techniques
such as GC/MS, LC/MS and ICP-MS. Further details about each solution can be
found on the Agilent web site. Simply visit the web address given at the end of
this compendium.

2
Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2
Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-4
Development of Analytical Methods for Quality Control . . . . . . . . . . . . . . . .5-44
Extraction and HPLC Analysis of Alkaloids in Goldenseal . . . . . . . . . . . .7
The High-Resolution Reversed-Phase HPLC Separation of Licorice
Root Extracts using Long Rapid resolution HT 1.8-µm Columns . . . . .13
Fast, high-resolution analysis of notoginseng by
rapid resolution liquid chromatography . . . . . . . . . . . . . . . . . . . . . . . . . . .19
Analysis of ginseng and American ginseng using the
Agilent 1120 Compact LC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
Development of reliable quality control methods for TCM
preparations using rapid resolution LC with UV and MS detection . . .29
Analysis of Traditional Chinese Medicines with the
Agilent 1200 Series evaporative light scattering detector . . . . . . . . . . .23
Analysis of TCMs with the Agilent 1200 Series
evaporative light scattering detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39
Monitoring of Pesticide Residues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45-58
Screening for 430 Pesticide Residues in Traditional Chinese
Medicine Using GC/MS: From Sample Preparation to
Report Generation in One Hour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
Low Part-per-Billion Level Pesticides Screening in Traditional
Chinese Medicine Using the Agilent 7000A GC/MS/MS . . . . . . . . . . .53
Determination of Heavy Metal Content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .59-70
Determination of Toxic Elements in Traditional Chinese
Medicine Using Inductively Coupled Plasma Mass Spectrometry . . . .61
Fast determination of five toxic elements in Traditional
Chinese Medicine (TCM) by ICP-MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . .67
Quality Control for Residual Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71-80
Better precision, sensitivity, and higher sample throughput
for the analysis of residual solvents in pharmaceuticals –
Using 7890A GC with G1888 headspace sample . . . . . . . . . . . . . . . . . . .73

3
Purification and Profiling of Bioactive Compounds . . . . . . . . . . . . . . . . . . .81-122
Isolation of formononetin and other phytoestrogens from
red clover with the Agilent 1100 Series purification system . . . . . . . . .83
Analysis of a complex natural product extract from ginseng –
Part I: Structure elucidation of ginsenosides by rapid resolution
LC-ESI TOF with accurate mass measurement . . . . . . . . . . . . . . . . . . . .89
Analysis of a complex natural product extract from ginseng –
Part II: Structure elucidation of ginsenosides by high resolution
ion trap LC/MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .97
Analysis of a complex natural product extract from ginseng –
Part III: Species differentiation of ginseng plants and authentication
of ginseng products by LC/MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .103
Agilent MassHunter – Fast computer aided analysis of
LC/ESI-TOF data from complex natural product extracts – Part 1:
Analysis of 6210 data with the Molecular Feature Extractor in
MassHunter Workstation software . . . . . . . . . . . . . . . . . . . . . . . . . . . . .111
Agilent MassHunter – Fast computer aided analysis of
LC/ESI-TOF data from complex natural product extracts – Part 2
Comparison of 6210 TOF data from different biological origin
with the Mass Profiler in MassHunter software . . . . . . . . . . . . . . . . . .117
Metabolic Profiling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .123-132
An interwoven multi-algorithm approach for computer-assisted
identification of drug metabolites - Rapid identification of
drug metabolites from accurate QTOF MS and MS/MS data by
MassHunter metabolite identification software . . . . . . . . . . . . . . . . . .125

4
 Palmatine

2 3 4 5 6 Development of Analytical Methods

7 8

for Quality Control

5
6
 Extraction and HPLC Analysis of Alkaloids
in Goldenseal
Application

Consumer Products and Drug Manufacturing/QA/QC

Author Introduction
Holly A. Weber Goldenseal, Hydrastis canadensis L., is one of the oldest
Midwest Research Institute herbal medicinal plants and is of current interest as a nat-
425 Volker Blvd ural medicine. There are two alkaloids present in gold-
Kansas City, MO 64110 enseal that are the expected active components, berberine
USA and hydrastine. Canadine, hydrastinine, berberastine, and
canadaline are minor alkaoloids. Palmatine, which is closely
Maureen Joseph
related to berberine, is not found in H. canadensis, but is
Agilent Technologies, Inc
found in Coptis, another berberine-containing plant [1].
2850 Centerville Road
Wilmington, DE 19808-1610 Goldenseal has been used as an anti-inflammatory and
USA antibiotic. It has also been used to treat nasal congestion,
cold, flu, and a variety of intestinal disorders. The whole
root of the plant is used and is currently available in bulk
Abstract (dried or powdered roots), tablets, capsules, and tinctures.
Goldenseal plants have been overharvested and many are
An ambient extraction of goldenseal root powder fol- now grown on farms for use as herbal supplements.
lowed by HPLC analysis of the alkaloids on a Zorbax Inconsistent quantities of alkaloids are present in the prod-
Rapid Resolution Eclipse XDB-C18 column provides an ucts sold as herbal supplements. A simple process for
accurate method for the determination of key alkaloids in extracting and analyzing the alkaloids is highly desirable to
goldenseal, including berberine and hydrastine. The evaluate product quality. Figure 1 shows the structures of
extraction and HPLC analysis can be applied to several the alkaloids in goldenseal.
other alkaloids, including canadine, hydrastinine, and
palmatine, and may be applicable to other berberine-con-
taining plant roots. The Rapid Resolution Eclipse XDB-C18
column is used for an isocratic separation with high res-
olution of all components in under 15 minutes.

7 5988-7136EN
 O
O H 3C N
O

H3CO O
N+ O
H3CO
H3CO O
OCH3
Berberine Hydrastine
O
O

N O N
H3C H3CO
OH OCH3

Hydrastinine Canadine

OCH3
OCH3

N+
H3CO
OCH3 Palmatine

Figure 1. Structures of key alkaloids in goldenseal and related plants.

Experimental 4. Dilute extracts 1/5

Extraction Procedure
Literature reports by Betz and Anderson [1] and Burney [2]
indicate that ambient extraction of alkaloids from H.
canadensis is possible. This is followed by an HPLC deter-
mination for accurate quantitation of the alkaloid extracts.
The optimized ambient extraction conditions used for the
analysis of neat goldenseal root powder are summarized in
the steps below.
1. Weigh ~ 0.5 g of root powder
2. Mix with 100 mL of acetonitrile:water:H3PO4 (70:30:0.1,
v/v/v)
3. Sonicate 5 min, shake (wrist-action shaker)
10 min, centrifuge 5 min
5. Direct HPLC analysis of diluted extracts
HPLC Analysis
The HPLC analysis of alkaloids present in goldenseal needs
to resolve the major alkaloids. In addition, it is desirable to
resolve palmatine, because it is present in other berberine-
containing plants. An HPLC method was developed to
resolve all of these components. Optimum resolution and
peak shape were obtained using the Zorbax Eclipse
XDB-C18 column with an ammonium acetate buffer and
acetonitrile. The separation is shown in Figure 2 with a
complete list of the optimized conditions used. For consis-
tent retention times, temperature control was required at
30°C [3].

5988-7136EN 8
 Column: Zorbax Eclipse XDB-C18 (3.5 µm, 4.6 × 150 mm)
Guard column: Zorbax Eclipse XDB-C18 (5 µm, 4.6 × 12.5 mm)
Solvent: 68% 30 mM ammonium acetate, 14 mM TEA, pH ~ 4.85
32% acetonitrile
Wavelength: 230 nm
Flow rate: 1 mL/min
Column temperature: 30 ˚C
Injection volumn: 10 µL
Run time: 17 min
Berberine

Hydrastine
Response

Palmatine

Canadine
Spiked sample

Sample

Mixed standard

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Time (min)

Figure 2. HPLC separation of goldenseal extract on Eclipse XDB-C18 column.

Method Validation

This HPLC method was applied in the validation of the

ambient extraction method. Linearity, precision, and alka-
loid recovery were investigated. Table 1 shows these
results. The precision of the method is excellent, and
recoveries of the different alkaloids ranged from
92%–102%, excellent for a quantitative method. The linear-
ity was very good (Figure 3) and sensitivity was good. From
the standard data, the calculated LOD (limit of detection)
for berberine was 0.50 mg/mL and the LOQ (limit of quan-
titation) was 1.65 mg/mL, so that accurate quantitation
down to low levels is possible, making it easy to test the
quality of different goldenseal products.

9 5988-7136EN
Table 1
Validation Results of the Ambient Extraction Method
Palmatine Berberine Hydrastine Canadine
Precision (n = 10) 0.18 ±0.002(s)% 3.06 ±0.05(s)% 2.04 ±0.01(s)% 0.08 ±0.001(s)%
Alkaloid recovery 0.18 ±0.003(s) 3.10 ±0.06(s)% 2.05 ±0.02(s)% 0.08 ±0.001(s)%
(~ 0.3–2 g of GS)
(~ 0.6–1 mg/mL) (n = 12)
Spike and recovery Spike level = ~ 0.15% Spike level = ~ 2.0% Spike level = ~ 2.0% Spike level = ~ 0.10%
(n = 3) 92.2 ±5.5(s)% 101.5 ±0.2(s)% 101.9 ±0.2(s)% 101.9 ±7.9(s)%

(s) = standard deviation

6000000
Linearity of Berberine and Hydrastine Standards
5000000

4000000
Area

3000000
Berberine
Hydrastine
2000000

1000000

0
0.000 20.000 40.000 60.000 80.000 100.000 120.000

µg/mL

1800000 Goldenseal Linearity

Berberine and Hydrastine
1600000

1400000
Berberine
1200000

1000000
Area

800000

600000 Hydrastine
400000

200000

0
0.5000 0.6000 0.7000 0.8000 0.9000 1.0000 1.1000
mg goldenseal/mL diluent

Figure 3
Linearity of berberine and hydrastine as standards and from goldenseal extracts.

5988-7136EN 10
Results
Goldenseal Testing

The ambient extraction and HPLC analysis method were

applied to six lots of goldenseal root powder from three dif-
ferent vendors to determine the alkaloid content. The
results are summarized in Table 2. These results demon-
strate variability among vendors and the reason a good
quantitative HPLC method is desirable.

Table 2
Results of Testing Goldenseal Root Powder from Multiple Lots and Multiple Vendors with the
Ambient Extraction Method and HPLC

Total weight % of
Vendor Lot number % Palmatine % Berberine % Hydrastine % Canadine known alkaloids
1 A nd 3.27 2.36 0.09 5.94
B nd 3.29 2.40 0.07 5.98
2 C 0.19 3.01 1.99 0.09 5.53
D 0.18 3.06 2.04 0.08 5.36
3 E nd 4.60 4.06 0.12 8.99
F nd 3.93 2.67 0.20 6.93

Conclusions Acknowledgements
This ambient extraction method of goldenseal is simple This work was performed by Holly A. Weber, Matthew K.
and reliable. This is followed by an isocratic HPLC analysis Zart, Andrew E. Hodges, Kellie D. White, Roger K. Harris,
with a Zorbax Rapid Resolution Eclipse XDB-C18 column, and Alice P. Clark of Midwest Research Institute, 425
which provides high resolution and excellent peak shape Volker Blvd, Kansas City, MO 64110.
of six alkaloids in 15 minutes. This analysis provides reli-
able quantitative results of the alkaloids in goldenseal, Diane Overstreet and Cynthia Smith of the National
including berberine and hydrastine. The method was Toxicology Program, 111 Alexander Drive, Research
applied to goldenseal from three different vendors and may Triangle Park, NC 27709.
be applicable to other berberine-containing plants.
This work was funded by the National Institute of
Environmental Health Sciences, Contract Nos. N01-ES-
References 55385 and N01-ES-05457.
1. J. M. Betz, S. M. Musser, and G. M. Larkine,
“Differentiation between Goldenseal (Hydrastis
canadensis L.) and Possible Adulterants by LC/MS,”
presented at the 39th Annual Meeting of the American
Society of Pharmacognosy, Orlando, FL, July 19–23,
1998. For More Information
2. M. L. Anderson and D. P. Burney, J. AOAC Intern.,
For more information on our products and services, visit our
81:1005–1110 (1998).
Web site at www.agilent.com/chem.
3. H. A. Weber, et. al., J. Liq. Chrom. and Rel. Tech. 24(1)
87–95 (2001).

11 5988-7136EN
5988-7136EN 12
 The High-Resolution Reversed-Phase HPLC
Separation of Licorice Root Extracts Using
Long Rapid Resolution HT 1.8-µm Columns
Application
Food Additive, Natural Products, and Pharmaceuticals

Author COOH
Bernard Permar and Ronald E. Majors
Agilent Technologies, Inc.
2850 Centerville Road
Wilmington, DE 19808-1610
COOH
USA O
O

Abstract
High-resolution reversed-phase HPLC analytical studies
using licorice, a licorice hydrolysis product, and commer-
cial licorice samples, showed that resolution and
throughput using a ZORBAX 1.8-µm column greatly
exceeded that obtained using the conventional
5.0-µm column.
Introduction
Licorice is derived from the root of the Glycyrrhiza glabra
plant, a 4- to 5-foot woody shrub that grows in Europe, the
Middle East and Western Asia. The root of the plant is
known to contain about 4% glycyrrhizin, the potassium or
calcium salt of glycyrrhizinic acid. The latter is a glycoside
of a pentacyclic triterpine carboxylic acid (18-ß-glycyrrhetic
acid) with two molecules of glucuronic acid (Figure 1).
Glycyrrhizin is about 50 times sweeter than sucrose
(cane sugar) but at high dosage is known to have toxicity.
Upon hydrolysis, the glycoside loses its sweet taste and
is converted to the aglycone glycyrrhetinic acid plus two
molecules of glucuronic acid.
COOH O
O
Figure 1
Structure of glycyrrhizinic acid.
Extractions from any of the many species of this plant will
yield a complex mixture containing more than 100 com-
pounds. Several of these compounds are used as additives
in candy as sweeteners, in cough syrup as flavoring
agents, and in drugs to mask a bitter taste, or for their
therapeutic qualities, mainly in Traditional Chinese
Medicines (TCMs). The medicinal properties of licorice
have been known for several centuries in China, as well as
India, Egypt, Greece, and Rome. Uses included cough sup-
pressant, laxative, and treatments for gastric ulcer, early
Addison disease, and liver disease. Most recently, gly-
cyrrhizin has been shown to have antiviral activity against
DNA and RNA viruses (influenza A and B, HIV, VZV,
Hepatitis B and C) [1]. Licorice has also been used in topi-
cal cosmetic applications.

13 5989-4907EN
The abundance of certain compounds of interest will vary Standards: Purchased from Sigma Aldrich
greatly according to the species of the plant, the time of • (G) 0.1-mg glycyrrhizic acid ammonium salt, ~75 %,
harvest, and the method of extraction. Thus, analytical dissolved in 0.5-mL mobile phase B, then brought to 1.0 mL
methods to follow the active ingredients are required. by adding 0.5-mL mobile phase A
Gradient elution reversed-phase HPLC has been found to • (GA) 0.1-mg 18-beta-glycyrrhetinic acid, 97%, dissolved in
be an effective method for separating some of the impor- 0.5-mL mobile phase B, then brought to 1.0 mL by adding
tant compounds in licorice [2]. This application note com- 0.5-mL mobile phase A
pares the traditional HPLC methodology and the newer Samples:
Rapid Resolution high-throughput (RRHT) columns. We
will apply these HPLC techniques to investigate the differ- • Licorice root extract A (HERB FARM brand)
ences between two commercially available licorice root • Licorice root extract B (Newark Natural Foods)
extracts. Both extracts should be vortexed, then filtered (0.2 micron) prior
to injection.
Experimental Injection volume: 5 µL of extract

Two reversed-phase (RP) columns were used in this study:

The most important compound found in a typical licorice
• Conventional ZORBAX StableBond (SB)-C18,
extract is G and to a lesser extent, its hydrolysis product,
4.6 mm x 250 mm, 5 µm
GA. These substances can be purchased commercially.
• ZORBAX SB-C18 RRHT, 4.6 mm x 150 mm, Although some of the other components of licorice
1.8 µm have been identified and are available commercially, they
are quite expensive. Since our main objective was to
The smaller particle size of the RRHT column allows use of demonstrate the advantage of using shorter, high-resolu-
a shorter column to achieve the same resolution as the tion HPLC columns, we used only two standards (G and
longer conventional column, and also allows more rapid GA) to develop the initial method. Figure 2a shows the
separations. gradient separation of G and GA on the conventional col-
umn (ZORBAX SB-C18, 4.6 mm x 250 mm, 5 µm) using gra-
Results dient elution. Since the licorice extract to be examined was
quite complex, isocratic conditions were not usable to sep-
arate all of the components. The G being more polar by
virtue of the additional sugar moieties eluted first while the
HPLC conditions GA came off the column much later. Using this gradient,
the GA eluted in just under 42 minutes. By switching to the
Instrument: Agilent 1200 Series Rapid Resolution System
shorter ZORBAX SB-C18 RRHT column (4.6 x 150 mm,
Detector: Multiple wavelength detector (MWD), 1.8 µm), the separation was virtually the same, as can be
254 nm/100 BW, 450 nm reference seen in Figure 2b. However, the separation time was now
Mobile phase: A = 1% Acetic acid in water
just over 25 min, a savings of about 40% in time.
B = 1% Acetic acid in acetonitrile
Gradient conditions for ZORBAX SB-C18 columns:
Conventional: 4.6 mm x 250 mm, 5 µm
5% to 100% B in 50 minutes
RRHT: 4.6 mm x 150 mm, 1.8 µm
5% to 100% B in 30 minutes
Flow: 1.0 mL/min
Temperature: Ambient

5989-4907EN 14
 mAU
ZORBAX SB-C18, 4.6- × 250-mm, 5 µm
80 GA 41.4 min
2a
60 G 22.9 min

40

20

0 10 20 30 40 50 min

mAU ZORBAX SB-C18 RRHT, 4.6- × 150-mm, 1.8 µm

160 GA 25.5 min
2b G 14.6 min
120

80

40

0
0 5 10 15 20 25 30 min

Figure 2
Gradient reversed-phase separation of G and GA on the test 5.0- and 1.8-µm columns.

To investigate the use of these columns for the separation
of actual licorice root extracts, Figures 3 and 4 depict the
complex chromatograms observed by injection of filtered
extracts, identified in the Experimental section. Figure 3a
shows the complex chromatogram obtained using the 5-µm
250-mm column. The cut-away shows the small amount of
GA that was present in the extract. Since GA is a hydroly-
sis product of G, it should be at a much lower concentra-
tion in a licorice extract, unless the extract was treated to
enhance the concentration of the hydrolyzed product.
Based on the area count, the GA concentration was less
than 0.5% of the G concentration in extract A.
Although the actual peaks were not counted, the calculat-
ed peak capacity (3) for the 5-µm column was determined
to be 290 (resolution: 1.0). Running the same extract A on
the 1.8-µm 150-mm column, one can see the finer struc-
tured (that is, higher resolution) chromatogram that results
(Figure 3b). The calculated peak capacity for this higher
efficiency column was determined to be 442, over 50%
higher than by using the longer 5-µm column. Thus, it
would be easier to determine minor components on this
shorter rapid resolution column. The peaks per unit time
(Resolution = 1.0) was calculated to be 17.7 peaks/min for
the 1.8-µm column versus 7.1 for the 5-µm column.
Figures 4a and 4b show similar runs using
extract B. This extract was even more complex than
extract A which is borne out by comparing the high resolu-
tion chromatograms of Figure 3b versus Figure 4b. Again,
the calculated peaks per minute for the 1.8-µm column
greatly exceeded that of the 5-µm column (17 versus 7.5
respectively). Based on the peak area counts for GA, it was
roughly 1% of the concentration of G in
extract B.

15 5989-4907EN
(A)

mAU

400

G 22.8 min Peak capacity (@ Rs=1.0) = 290

300

200

100
GA 41.4 min

0
0 10 20 30 40 50 min

mAU
50

40
Analysis of licorice root
extract A using a ZORBAX 30

SB-C18, 4.6 × 250 mm, 20 GA

5-µm column 10

-10

-20
25 30 35 40 45 50 min

(B)

mAU
450
400 Peak capacity (@ Rs=1.0) = 442
350
300
250
G 14.6 min
200
150
GA 25.5 min
100
50
0
0 5 10 15 20 25 30 min

mAU
70

60
Analysis of licorice root extract A
using a ZORBAX SB-C18 50
GA
RRHT 4.6 × 150 mm, 1.8-µm column 40
For conditions: see Experimental 30

20
15.896

10

0
16 18 20 22 24 26 28 30 32 34 min

Figure 3a and 3b
The gradient reversed-phase separation of licorice root extract A on the 5.0-µm column (A) and on the 1.8-µm column (B).

5989-4907EN 16
(A)

mAU

400
G 22.8 min Peak capacity (@ Rs 1.0) = 306

300

200

GA 41.40 min
100

0
0 10 20 30 40 50 min

mAU
70
Analysis of licorice root
22.016
60
extract B using a ZORBAX
SB-C18 4.6 × 250-mm column, 50
GA
5 µm 40

30

20

10

0
25 30 35 40 45 50 min
(B)

mAU
450
400
350
G 14.6 min
300 Peak capacity (@ Rs=1.0) = 426
250
200
150 GA 25.5 min
100
50
0
min
0 5 10 15 20 25 30

mAU
90

80
Analysis of licorice root extract B GA
using a ZORBAX SB-C18 70
RRHT 4.6 × 150-mm column 60

50

40

30

20
16 18 20 22 24 26 28 30 32 34 min

Figures 4a and 4b
Gradient reversed-phase separation of licorice root extract B on the 5.0-µm column (A) and on the 1.8-µm column (B).

17 5989-4907EN
Conclusions
No attempt was made to perform quantitative analysis on
the components of the licorice extracts. From our studies,
it was obvious that resolution and throughput using the
1.8-µm column greatly exceeded that obtained using the
5.0-µm conventional column. As more complex samples of
natural products are encountered and researchers require
more detailed component analyses, the use of high resolu-
tion, small particle columns should grow. In the investiga-
tion of licorice root, other natural products, and TCMs, it is
necessary to have gradient capability and sensitive detec-
tion.

References
1. S. Fanali, Z. Aturki, G. D’Orazio, M. A. Raggi,
M. G. Quaglia, C. Sabbioni, and A. Rocco, (2005) J. Sep.
Sci., 28, 982–986.
2. I. Kitagawa, (2002) Pure Appl. Chem. 74 (7), 1189–1198.

For More Information

For more information on our products and services, visit our
Web site at www.agilent.com/chem.

5989-4907EN 18
 Fast, high-resolution analysis of
notoginseng by rapid resolution
liquid chromatography

Application Note Zhi-xiu Xu

Absorbance [mAU]

1000

800

600

400

200

0.5 1.0 1.5 2.0 2.5 3.0

Time [min]

Abstract

This Application Note describes:

• The development of rapid resolution liquid chromatography (RRLC) method for
the analysis of notoginseng.
• The results of method transfer from conventional HPLC to RRLC.
• The use of the RRLC method to shorten run times while maintaining good
Agilent Equipment: resolution of complex components, thereby increasing sample throughput and
1200 Series Rapid Resolution LC system
ZORBAX XDB C18 RRHT column lowering costs.
• The chromatograms obtained with optimized methods for the different parts of
Application Area:
the notoginseng plant, which show different peak profiles and different concen-
Traditional Chinese Medicine
trations of certain saponins.

19 5989-6757EN
Introduction
Traditional Chinese Medicines (TCMs)
have a long history of use in China and
their therapeutic effects are well
known in China and other Asia coun-
tries such as Korea and Japan. In
Western countries the use of TCMs as
food supplements or nutriments is
becoming more and more popular.
More than 11,000 kinds of TCMs have
been used over time. Research and
quality control of TCMs rely heavily on
instrumental separations and the per-
formance of these
separations.
In this study a rapid resolution liquid
chromatography (RRLC) method for
the analysis of notoginseng was
developed. The conventional HPLC
analysis method was transferred
easily to RRLC using Agilent’s method
translator. Different extraction meth-
ods were used to produce different
samples from different parts of the
notoginseng plant. These samples
were separated with optimized meth-
ods and the resulting chromatograms
showed different peak profiles. This
Application Note also shows the
quantitative results. Using the faster
and better methods developed in this
study, quality control departments will
be able to reduce analysis times and
increase sample throughput. Using
these methods also reduced the cost
of solvent as well as improved the
overall analysis process.
Notoginseng, which is also known as
in Chinese as Sanqi or Tianqi, is an
important and highly valued traditional
medicine in China. It has been cultivat-
ed for about 400 years and more than
85 % of the notoginseng production
originates from the Yunnan Province,
China. Notoginseng is known for its
efficiency in promoting blood circula-
5989-6757EN
tion, removing blood stasis, inducing
blood clotting, relieving swelling and
alleviating pain. Current pharmacologi-
cal studies revealed that Panax noto-
ginseng possesses anticarcinogenic
and hepatoprotective properties, as
well as protective effects on cardio-
vascular and cerebrovascular
systems1. Compared with the well
known Panax ginseng and Panax quin-
quefolium (American Ginseng), the pro-
file of the saponins in Panax notogin-
seng is similar because they belong to
the same genus. Notoginseng is used
mostly in south China in different tradi-
tional Chinese formulations. The
famous traditional Chinese formula-
tions Yun Nan Bai Yao and Pien Tze
Huang also contain Panax notogin-
seng.
The complex matrixes of TCMs always
present major challenges for quality
control and research. The 2005 edition
of the China Pharma-copeia lists
Materia Medica Sanqi and Chinese
patent medicines such as Sanqi
Shangyao Pian as containing notogin-
seng saponin R1, which must be ana-
lyzed in these formulations. The typical
run time for the analysis of notogin-
seng saponin is longer than 60 min-
utes. With more and more research on
TCMs, scientists realized that control
of only one or two components is not
enough to determine the quality of a
TCM and that it is necessary to devel-
op other quality control methods that
separate more components.
Separating complex samples such as
TCMs requires longer analysis times
than synthetic drugs. When sample
throughput is also an issue, having
similar or better performance with
shorter analysis times would be an
ideal solution for this situation. With
RRLC it is possible to develop methods
with shorter run times and with better
performance. RRLC methods also give
20
users the benefits of time and solvent
savings. For manufacturing quality
control, using RRLC means batches of
products can be released faster com-
pared with using conventional HPLC
methods
Experimental
Equipment
For development of the RRLC method
an Agilent 1200 Series Rapid
Resolution LC system with the
following modules was used:
Agilent 1200 Series RRLC system
consisting of:
• Agilent 1200 Series binary pump SL
with vacuum degasser
• Agilent 1200 Series high-perfor-
mance autosampler SL
• Agilent 1200 Series thermostatted
column compartment SL
• Agilent 1200 Series diode array
detector SL with micro flow cell
(2 µL volume, 3 mm path length)
• Agilent ChemStation B.02.01 SR1 for
data acquisition and evaluation
• Agilent ZORBAX XDB C18 RRHT
column, 4.6 x 50 mm,
1.8 µm particle size
For comparison with conventional
HPLC a standard configuration of
an Agilent 1100 Series LC with the
following modules was used:
• Agilent 1100 Series binary pump with
degasser
• Agilent 1100 Series autosampler
• Agilent 1100 Series thermostatted
column compartment
• Agilent 1100 Series diode array
detector
• Agilent ZORBAX SB C18
column, 4.6 x 50 mm, 5 µm particle
size
 Samples and sample preparation the process easy and quick (figure 1). Results and discussion
• Notoginseng saponin R1, ginsenoside Depending on individual requirements
Rg1 and ginsenoside Rb1 standards it might be necessary to make further Chromatograms
were purchased from the National modifications to the transferred According to results of pharmacologi-
Institute for Control of method based on the different factors cal studies, different parts of the noto-
Pharmaceutical and Biological listed by the method translator. For ginseng plant have different therapeu-
Products (NICPBP), China. details of how to translate manually a tic effects as a result of different com-
Gensenoside Re and ginsenoside Rc conventional HPLC method to an RRLC ponents being present. Specific extrac-
standards were purchased from method, further Application Notes are tion methods were used with the
Sigma-Aldrich, USA. available from Agilent2,3,4. objective of obtaining more effective
• Notoginseng caudexes and leaf components. The notoginseng sam-
extracts (NCLE) were kindly provided
by a customer’s laboratory. The
caudexes and leaves were extracted
with 60% ethanol, the solvent was
evaporated to dryness and the
residue dissolved in n-butanol, the
solvent was then evaporated to dry-
ness again and the residual yellow
powder was dissolved in methanol
and used for injection.
• Notoginseng root extracts (NRE)
were kindly provided by a customer’s
laboratory. The raw components
were extracted with water, filtered
through a marcoporous membrane
and the filtrate evaporated to dry-
ness. The residual white powder was
dissolved in methanol and used for
injection.
• Notoginseng was purchased from a
local drug store. The brand was Tong Figure 1
Agilent’s method translator for transfer of standard HPLC methods to RRLC technology.
Ren Tang. The raw material was
extracted with water/methanol
(30/70, v/v), treated ultrasonically for 120
Absorbance [mAU]

30 minutes, filtered through a 0.22 µm

membrane and the clear filtrate was 100
used for injection.
• Water, acetonitrile and methanol 80
were purchased from Fisher, USA.
60
Method translation
The conventional HPLC method using 40
the column packed with 5 µm particles
needed to be transferred to an RRLC 20
method using a column packed with
1.8 µm particles. Agilent’s method 0
translator (available on CD for Agilent
0 10 20 30 40 50 60
1200 Series Rapid Resolution LC sys- Time [min]
tem, publication number 5989-5130EN) Figure 2
was used for this purpose and made Chromatogram of NCLE analyzed using a conventional HPLC method on an Agilent 1100 Series LC
system. See table 1 for chromatographic conditions.

21 5989-6757EN
ples were analyzed with conventional

Absorbance [mAU]
HPLC methods and RRLC methods to 120
reveal different peak profiles. Figure 2 100
shows the chromatogram of notogin-
80
seng caudexes and leaf extracts
(NCLE) analyzed using the convention- 60
al HPLC method. Typically, this type of 40
separation requires more than 60 min-
utes5. As described in the experimen- 20
tal section, the RRLC method used for 0
analyzing the notoginseng samples
-20
was developed based on the conven-
2.5 5 7.5 10 12.5 15 17.5 20
tional HPLC method. The same NCLE Time [min]
sample was analyzed with the conven- Figure 3
tional HPLC method and the RRLC Chromatogram of NCLE, analyzed using RRLC method 1 on an Agilent 1200 Series RRLC system.
method 1. The chromatograms are See table 1 for chromatographic conditions.
shown in figures 2 and 3. The peaks in
figure 3 were narrower than those in
Absorbance [mAU]

figure 2. This means that the peak

100
capacity increased for the complex
samples. The resolution or separation 80
performance of the RRLC analysis was
better than conventional HPLC. At the 60
same time the run time was reduced
40
dramatically from 60 to 20 minutes.
Another factor that should be consid- 20
ered is the selectivity of the columns
for conventional HPLC and RRLC, 0
which should be kept the same. If the
new RRLC column has totally different 2 4 6 8 10 12
selectivity properties than the conven- Time [min]
tional HPLC column, transferring the Figure 4
method makes no sense. In this study Chromatogram of NCLE, analyzed using RRLC method 2 on an Agilent 1200 Series RRLC system.
peak profiles and elution sequence See table 1 for chromatographic conditions.

Chromatographic conditions for Chromatographic conditions for Chromatographic conditions for Chromatographic conditions for
traditional method with Agilent RRLC method 1: RRLC method 2: RRLC method 3:
1100 Series LC:
Column: Agilent ZORBAX Column: Agilent ZORBAX Column: Agilent ZORBAX Column: Agilent ZORBAX
SB-C18, 4.6 x 250 mm, XDB C18, 4.6 x 50 mm, XDB C18, 4.6 x 50 mm, XDB C18, 4.6 x 50 mm,
5 µm particle size 1.8 µm particle size 1.8 µm particle size 1.8 µm particle size
Mobile phase: A: Water; B: ACN Mobile phase: A: Water; B: ACN Mobile phase: A: Water; B: ACN Mobile phase: A: Water; B: ACN
Gradient: 0 min, 20 %B; 10 min, Gradient: 0 min, 20 %B; 3 min, Gradient: 0 min, 20 %B; 1.5 min, Gradient: 0 min, 30 %B; 2.5 min,
30 %B; 20 min, 35 %B; 30 %B; 7 min, 35 %B; 30 %B; 5 min, 35 %B; 60 %B; 5 min,
30 min 40 %B; 40 min, 10 min 40 %B; 13 min, 7 min 40 %B; 10 min, 100 %B
60 %B; 50 min, 60 %B; 17 min, 60 %B; 12 min,
100 %B 100 %B 100 %B
Flow rate: 1.0 mL/min Flow rate: 1.0 mL/min Flow rate: 1.5 mL/min Flow rate: 2.5 mL/min
Injection volume: 5 µL Injection volume: 5 µL Injection volume: 5 µL Injection volume: 5 µL
Diode array Diode array Diode array Diode array
detection: 203 nm ±8 nm, detection: 203 nm ±8 nm, detection: 203 nm ±8 nm, detection: 203 nm ±8 nm,
Ref. 360 nm ±100 nm Ref. 360 nm ±100 nm Ref. 360 nm ±100 nm Ref. 360 nm ±100 nm
Table 1
Chromatographic conditions.

5989-6757EN 22
were the same so that it was not nec-

Absorbance [mAU]
essary to identify the peaks again in
the new chromatogram. Further, the 500
UV spectra from the diode array detec-
tor helped to do further confirmation. 400

Figure 4 shows the chromatogram of 300

the NCLE sample analyzed by RRLC
method 2. Increasing the flow rate to 200
1.5 mL/min reduced the run time. At
100
the same time it was possible to main-
tain the same or comparable perfor-
0
mance as the results shown in figure
3. For more complex samples, shorter 2.5 5 7.5 10 12.5 15 17.5 20
Time [min]
run times mean less peak capacity.
Hence users should choose a balance
between peak capacity and run time. Figure 5
Chromatogram of NRE, analyzed using RRLC method 1 on an Agilent 1200 Series RRLC system.
See table 1 for chromatographic conditions.
As mentioned in the experimental sec-
tion, three kinds of samples were
obtained from different origins. Figure
5 shows the chromatogram of the 1500
Absorbance [mAU]

notoginseng root extraction (NRE)

1400
sample analyzed with the RRLC
method 1. This sample was extracted 1200
from the notoginseng root using a dif- 1000
ferent procedure. The peak profiles 800
show that NRE and NCLE contain
different kinds of saponins and other 600
components.To test if the developed 400
RRLC method could be used with other 200
notoginseng samples, we purchased
0
notoginseng from a local TCM store.
The brand was Tong Ren Tang and the 2.5 5 7.5 10 12.5 15 17.5 20
notoginseng had already been ground Time [min]
to powder at TCM store. The chro-
Figure 6
matograms of the notoginseng analy- Chromatogram of notoginseng from Tong Ren Tang, analyzed using RRLC method 1 on an Agilent
ses are shown in figures 6 and 7. 1200 Series RRLC system. See table 1 for chromatographic conditions.
RRLC method 1 was used to obtain
the chromatogram in figure 6 and
RRLC method 3 was used in figure 7.
Compared with the conventional HPLC be used to obtain analysis times while
method, which needs more than 60 maintaining resolution. For less com-
minutes, the chromatogram in figure 7 plex samples such as Tong Ren Tang
demonstrates that the analysis can be notoginseng extractions, the faster
done in 4 minutes. Notoginseng method can be used to complete the
caudexes and leaf extracts are more analysis in a shorter time as shown in
complex than the other samples ana- figure 7.
lyzed and RRLC method 1 or 2 should

23 5989-6757EN
Quantitative results

Absorbance [mAU]
The standards listed in the experimen- 1000
tal section were analyzed using RRLC
method 2. Using different levels of 800
concentrations of the standards, a cor-
relation curve was created and used to 600
determine the concentrations of the
three samples. Table 2 shows the 400
quantitative results. NCLE contained
many kinds of saponins but did not
200
contain Rc or Re. NRE contained fewer
kinds of saponins but did contain the
0
five standards analyzed in this study.
The Tong Ren Tang notoginseng 0.5 1.0 1.5 2.0 2.5 3.0
extraction did not contain many kinds Time [min]
of saponins and did not contain Rc or
Figure 7
Re. The concentrations of R1, Rb1 and Chromatogram of notoginseng from Tong Ten Tang, analyzed using RRLC method 3 on an Agilent
Rg1 were higher than in NCLE and 1200 Series RRLC system. See table 1 for chromatographic conditions.
NRE. The concentration of notogin-
seng R1 ranged from 0.029 to 0.89 Standard Formula Correlation Concentration Concentration Concentration
µg/µL across all samples. The quanti- Y: peak area in NCLE in NRE in Tong Ren Tang
x: amount (µg/µL) (µg/µL) notoginseng
tative method developed in this study (0.1 µg/µL) powder (µg/µL)
can be deployed easily in the quality
R1 Y = 54.03850x + 0.395722 0.99995 0.029 0.48 0.89
control of notoginseng with different
Rb1 Y = 71.64751x – 0.159297 1.0000 0.18 4.64 2.136
extraction methods. Re Y = 93.85331x + 3.83488 0.99982 N/A 1.28 N/A
Rg1 Y = 151.66579x + 0.501417 1.0000 0.07 1.52 1.54
Rc Y = 11.12002x + 0.128832 0.99980 N/A 0.025 N/A
Conclusion Table 2
Quantitative results of the saponin standards.
Traditional Chinese Medicines are
complex natural products and their References
analysis by conventional HPLC
requires a high performance system 1. 4.
and long run times. The RRLC methods Wei J.X., Du Y.C., “Modern science Michael Frank, “Saving analysis time
developed in this study demonstrated research and application of panax and gaining resolution by simple
how to shorten run times while main- notoginseng”, Yunnan Science and means”, Agilent Application Note,
taining good resolution of complex Technology Press, Kunming, China, 1996. Publication Number 5989-6819EN,
components. The chromatograms 2007.
show the different peak profiles corre- 2.
sponding to the different notoginseng Angelika Gratzfeld-Huesgen, “Fast and 5.
extraction samples. Taking the require- Ultra-fast Analysis with the Agilent Lau A.J., Woo S.O., Koh H.L., “Analysis
ments of the application into consider- 1200 Series Rapid Resolution LC of saponins in raw and steamed Panax
ation, RRLC with a faster flow rate can System Compared to a Conventional notoginseng using high-performance
be deployed to achieve a faster analy- Agilent 1100 Series LC System Using liquid chromatography with diode array
sis method. The quantitative methods Sub 2-µm Particle Columns”, Agilent detection”, J. Chromatogr. A, 1011, 77-
developed in this study can be imple- Application Note, Publication Number 87, 2003.
mented easily as a quality control 5989-5672EN, 2006.
method for notoginseng products.
Using RRLC methods maintain or give 3.
even better performance while at the Michael Woodman, “Improving the
same time reducing the run time dra- Effectiveness of Method Translation
matically, thereby increasing sample for Fast and High Resolution
throughput and saving costs. Separations”, Agilent Application
Note, Publication Number 5989- Zhi-xiu Xu is an Application Chemist at
5177EN, 2006. Agilent Technologies, Shanghai, China.

5989-6757EN 24
 Development of reliable quality control
methods for TCM preparations using rapid
resolution LC with UV and MS detection

Application Note Zhixiu Xu

Abstract

This Application Note describes the development of reliable quality control

Agilent Equipment:
1200 Series Rapid Resolution LC system
methods for analysis of individual traditional Chinese medicines (TCMs) and
6140 Quadrupole MS TCM formulations using rapid resolution liquid chromatography with UV-visible
Application Area: and quadrupole mass spectrometry detection. The chromatograms obtained
Traditional Chinese Medicine with UV and MS detection for different individual TCMs are compared and the
combination of UV and MS spectra is used to qualify target compounds.

25 5989-7682EN
Introduction
China has a long history of using
TCMs and TCM preparations and has
vast experience of the therapeutic
effects. However, quality control is
based on appearance only. With the
development of new technologies, it
was recognized that TCMs contain
hundreds of compounds and that the
concentration levels are wide ranging.
Further, differences in composition
result from herbal TCMs being grown
in different regional areas and harvest-
ed at different seasonal times as well
as from preparation and manufactur-
ing processes. As a result, controlling
the quality of the compounds in TCMs
has become a tremendous challenge.
Today, tracking only one or two com-
ponents in TCM preparations is not
enough to control the quality because
TCM preparations with specific formu-
las are often made from more than
one TCM. For example, Qishenyiqi
dripping pills, a traditional patent
Chinese medicine used to treat coro-
nary diseases, include four kinds of
TCMs:
• Astragali (huangqi)
• Salviae miltiorrhizae (danshen)
• Notoginseng (sanqi)
• Lignum dalbergine ordoriferae
(jiangxiang)
Qishenyiqi dripping pills were used as
the sample in this study, in which a
new method for the quality control of
TCMs was developed using using
rapid resolution liquid chromatography
(RRLC) with UV-visible and MS detec-
tion. The main objective was to devel-
op a reliable method to determine the
potential target components that do
not undergo change during heating or
mixing.
5989-7682EN
Experimental
Equipment
• Agilent 1200 Series Rapid Resolution
LC system comprising binary pump
SL with vacuum degasser, high per-
formance autosampler SL, ther-
mostatted column compartment SL,
and diode array detector SL with
micro flow cell (2 µL volume,
3 mm path length)
• Agilent 6140 quadrupole MS with
ESI source
• Agilent ChemStation B.02.01 SR1 for
data acquisition and evaluation
• Agilent ZORBAX SB C18 RRHT col-
umn, 3.0 x 50 mm, 1.8 µm particle
size
Standards
Salvianolic acid B, tanshinone I, tan-
shinone IIA, cryptotansshinone, dan-
shensu, notoginsenoside R1 and astra-
galoside were purchased from
National Institute of Chemical
Pharmaceutical and Biological
Products (NICPBP), China.
Ginsenoside Rg1, ginsenoside Rb1,
3,4-dihydroxybenaldehyde, 3,4-dihy-
droxybenzoic acid, ginsenoside Re and
ginsenoside Rc were purchased from
Sigma-Aldrich, USA.
Solvents
Acetonitrile was purchased from
Fisher, USA. Water was prepared with
a Milli-Q pure water system.
Samples and sample preparation
Qishenyiqi dripping pills, intermediate
extractions of huangqi, danshen, sanqi
and essential oil of jiangxiang were
kindly provided by the TASLY Pharma-
ceutical Company, China. The raw
TCMs were purchased from a local
TCM store.
26
The dripping pills, TCM extractions and
raw TCMs were prepared by dissolving
in a 70 % methanol/water solution,
treating ultrasonically for 30 minutes,
and filtering through a 0.22 µm mem-
brane.
RRLC method
• Mobile phase:
A: Water with 0.1 % formic acid B:
ACN with 0.1 % formic acid
• Gradient: 0 min, 10 %B;
8 min, 38 %B; 12 min, 100 %B;
hold for 3 min, then 10 %B
• Flow rate: 1.0 mL/min
(passive splitter reduces flow to MS
to about 0.4 mL/min)
• Column temperature: 45 °C
• Detection wavelength: 203 nm
• Peak width: 0.5 s
• Slit width: 4 nm
• Spectra: 190-400 nm, step 2 nm
MS method
• Scan: 80-1400 (pos/neg)
• Fragmentor: 70 (pos/neg)
• Drying gas: 12 L/min
• Nebulizer pressure: 50 psi
• Drying gas temperature: 50 °C
• Cap. Voltage: 3200 V (pos/neg)
Results and discussion
The Agilent method translation tool
(available on CD pub. no. 5989-
5130EN) was used to convert the tradi-
tional LC method to a rapid resolution
method.
Comparison of danshen sanqi,
huangqi and jiangxaing extracts
Figure 1 shows that the chro-
matograms of danshen and sanqi were
very different from the combined
extracts when detected at 203 nm.
After mixing and heating, some peaks
disappeared and new peaks emerged.
This means that target compound
screening during quality control is an Danshen
and sanqi
important step to make sure the com- extracts
pounds do not undergo changes during
manufacturing. Most researchers use Sanqi
sample
detection at 203 nm because the main
active compounds are saponins, which Danshen
sample
have weak absorbance at 203 nm. But
large peaks appearing at 203 nm might
not be relevant for quality control and
small peaks might be important for the
research of active components. As a
consequence, a complementary detec- Figure 1
tor with higher sensitivity is needed to Chromatograms of individual TCMs danshen and sanqi, and of the combined extracts.
provide more information about the
components.

Comparison between different

detectors and conditions
Figure 2 shows the differences in the
results from UV and MS detection.
More peaks appeared in the MS total
ion chromatogram (TIC) because some
components had weak or no UV
absorbance. Using MS detection added
an extra dimension and gave more
information about the structures of the
different compounds. The negative
polarity mode gave more complete Figure 2
information about the peaks and Chromatograms of danshen sanqi combinations, showing the different results obtained by
UV and MS detection.
enabled the target compounds to be
identified for quality control.
N o rm .

300000
1
Comparison of qishenyiqi dripping
pills and different extracts
Figure 3 shows chromatograms of 250000

danshen sanqi, huangqi and jiangxiang

extracts, as well as the qishenyiqi drip- 200000
4
ping pills, which are made from these
three extracts and other additives. The
150000

numbered peaks are the target com- 2 5

6
pounds that were screened for quality 3
control of this TCM preparation. Based 100000

on the UV and MS data, different

7
peaks were studied to determine 5 00 0 0

whether the compounds had under-

gone changes. Table 1 shows the 0

results of the qualifying process. 2 4 6 8 10 12 m in

Figure 3
Comparison of the negative TICs for the three TCMs combinations and the dripping pills sample.

27 5989-7682EN
Conclusion Peak Compound MS UV lq m/(nm)
1 Danshensu 197 [M–H]–, 395 [2M–H]– 280
The RRLC method with UV and MS
detection described in this Application 2 Procatechualdehyde 137 [M–H]–, 275 [2M–H]– 225, 280, 310
Note is more reliable than the current
3 Salvianolic acid B 717 [M–H]– 260, 280
pharmacopeia quality control method
because the quality of several compo- 4 Ginsenoside Rg1 845 [M+HCOO]– 210
nents in the TCM can be tracked
based on the information provided by 5 Ginsenoside Rb1 11071[M–H]– 210
the UV and MS detectors. Tracking
6 Astragaloside IV 829 [M+HCOO]– 210
these components is important
because the components could under- 7 Tanshinone I 295137 [M+H2O]– 230
go changes during manufacture and
Table 1
preparation. The Agilent 6140 quadru- Target compounds with structural formulas and detection details (numbers refer to the peaks in
pole MS used for the mass spectrome- figure 3).
try measurements is easy to use and
integrates seamlessly with the Agilent
1200 Series RRLC system and Agilent
ChemStation sofware. Zhixiu Xu is an Application Chemist at
Agilent Technologies, Shanghai, China.
References

1.
Ai-hua Liu, et.al., Journal of
Chromatography B, 846, 32-41, 2007.

2.
Jin-huai Liu, et.al., Journal of Chinese
Pharmaceutical Science, 13 (4), 2004.

3.
Chinese Journal of Analytical
Chemistry, 1676-1680, 2005.

4.
Agilent Application Notes, publication
numbers 5989-4506EN, 5989-5493EN,
5989-6757EN

5989-7682EN 28

Agilent Equipment:
1120 Compact LC
EZChrom Elite Compact software
HC-C18(2) column
Application Area:
Traditional Chinese Medicine
Analysis of ginseng and American ginseng
using the Agilent 1120 Compact LC
Application Note Zhixiu Xu
Rb1
Re
Rg1
Abstract
The Agilent 1120 Compact LC is the system of choice for conventional, analytical
scale liquid chromatography. It is an intergrated LC designed for ease of use, per-
formance and reliability. It is ideally suited for the routine analysis of Traditional
Chinese Medicines (TCMs) on account of its capability to achieve highly precise
retention times and peak areas, and low detection limits for the analyzed com-
pounds. This Application Note shows the chromatograms obtained with optimizd
methods for the most well-known TCMs ginseng and American ginseng, which
show different peak profiles and different concentrations of certain saponins.
29 5989-7458EN
Introduction
Traditional Chinese Medicines (TCMs)
have a long history of use and their
therapeutic effects are well known in
China and other countries. Ginseng,
perhaps the most well-known TCM,
has long been used as a tonic, anti-
fatigue, sedative and anti-gastric ulcer
drug. It is also widely used in different
TCM preparations. Another well-
known TCM, American ginseng, has
similar therapeutic effects as ginseng
but there are also some differences
because of the different saponin
contents.
According to the method in the phar-
macopeia of the People's Republic of
China1, ginseng must be analyzed by
HPLC to determine the ginsenosides
Rg1, Re and Rb1. Similar requirements
exist for the determination of these
ginsenosides in American ginseng.
These requirements make the determi-
nation of the ginsenosides Rg1, Re
and Rb1 in ginseng and American
ginseng important for quality control
of TCM raw materials and final prepa-
rations.
In this study an HPLC analysis method
was developed using the Agilent 1120
Compact LC for the determiniation of
ginsenosides in ginseng and American
ginseng.
5989-7458EN
Experimental
Equipment
• Agilent 1120 Compact LC comprising
gradient pump with integrated
degasser, autosampler with vial tray,
column oven and variable wave-
length detector, see figure 1
• Agilent HC-C18(2), high carbon load,
150 x 4.6 mm, 5 µm particle size col-
umn
• Agilent EZChrom Elite Compact soft-
ware
Figure 1
Agilent 1120 Compact LC
Samples and sample preparation
Ginseng and American ginseng were
purchased from a local TCM store and
samples prepared as follows. 1 g of
powder was weighed and dissolved in
50 mL of water saturated n-butanol.
The solution was treated ultrasonically
for 30 minutes and centrifuged for 5
minutes at 300 rpm. The solvent was
evaporated and the residue dissolved
in 5 mL methanol. The final solution
was filtered through a 0.20 µm mem-
brane before injection.
30
Chromatographic conditions
• Mobile phase:
A: Water, B: ACN
• Gradient: 0 min, 19 %B;
35 min, 19 %B; 55 min, 29 %B;
70 min, 29 %B; 100 min, 40 %B
• Flow rate: 1.0 mL/min
• Injection volume: 10 µL
• Column temperature: 40 °C
• Detection wavelength: 203 nm
Results and discussion
The chromatogram of the ginseng
separation is shown in figure 2.
The method used to obtain this
chromatogram was the same as the
method specified in the pharmacopeia
of People’s Republic of China. The
chromatogram shows excellent sepa-
ration of all the target compounds.
The ginsenosides Rg1 and Re are
well separated, demonstrating that
the Agilent 1120 Compact LC is well
suited for this analysis.
The chromatogram of the American
ginseng separation is shown in figure
3. The chromatogram shows excellent
separation of the target ginsenosides.
From the chromatograms of ginseng
and American ginseng, it can be seen
that the profiles of the two samples as
well as the concentrations of the gin-
senosides are different. For complex
TCM samples such as ginseng and
American ginseng, the Agilent 1120
Compact LC is a reliable tool to obtain
good results in routine analysis work.
 [ "?9+
;s)
"?
70
Although Traditional Chinese
60 Medicines are complex natural prod-
ucts, this study demonstrated that the
50
Absorbance [mAU]

Agilent 1120 Compact LC was capable

40 of analyzing the active components
and achieving excellent separation
30 performance. The results proved that
the Agilent 1120 Compact LC is ideal
20
Rg1 Rb1 for routine quality control testing of
Re complex TCM samples.
10

0
!o1o®
o?9o
10 20 30 40 50 60 70 80 90 100
Time [min] 1.
Pharmacopeia of the People's Republic
Figure 2 of China, Volume I, 2005.
Chromatogram of ginseng analyzed with the CHP method on an Agilent 1120 Compact LC system.

300

250 Rb1
Absorbance [mAU]

200

150
Re
100

50
Rg1
0

0 20 40 60 80 100 120
Time [min]

Figure 3
Chromatogram of American ginseng analyzed on an Agilent 1120 Compact LC system.
Zhixiu Xu is an Application Chemist at
Agilent Technologies, Shanghai, China.

31 5989-7458EN
5989-7458EN 32
 Analysis of traditional Chinese
medicines with the Agilent 1200 Series
evaporative light scattering detector

Application Note Zhixiu Xu

mV
120
100
80
60 3
40 1 4
20 2
0
0 5 10 15 20 25 30 35 min

Abstract
Traditional Chinese medicines often contain components that must be detected
Agilent Equipment by HPLC, but that lack a chromophore and so do not produce signals with an ultra-
• Agilent 1200 Series Rapid Resolution LC violet (UV) detector. The Agilent 1200 Series evaporative light scattering detector
system
• Agilent 1200 Series evaporative light (ELSD) is an excellent alternative because it detects all solutes that are less
scattering detector volatile than the mobile phase. In the Chinese Pharmacopoeia 2005, several TCMs
Application area require the ELSD as the detection method, and this Application Note illustrates
• Traditional Chinese medicine (TCM) two examples – the flavonoids in Ginkgo biloba L. and astragaloside in Astragali.

33 5989-8922EN
Introduction
Mobile
The evaporative light scattering detec- phase
tor is increasingly being used as a
quasi-universal detector for non-UV- Gas
Nebulization
absorbing analytes in HPLC systems.
Especially in herbal medicines, an
increasing number of non-UV-absorb-
ing compounds need to be detected
without derivatization. The Agilent
1200 Series HPLC family now has a
new member – the evaporative light
Evaporation
scattering detector – to help with such
applications.

The Agilent 1200 Series ELSD can

detect all solutes that are less volatile
than the mobile phase. It produces a
signal only for the nonvolatile particles
that are generated from the sample. Detection
If the compounds have no chro-
mophores and are less volatile than
the LC solvents, the ELSD is a good
Photomultiplier
detector that can provide both ease
of use and good sensitivity. Gradient
Figure 1
mobile phases do not interfere with Cross-sectional view of the Agilent 1200 Series evaporative light scattering detector.
detection. When a mobile phase pro-
duces strong absorption under a spe-
Some components of TCMs have no ELSD as the detection method for cer-
cific UV wavelength, the ELSD is an
chromophores, so they need a univer- tain components. Examples include
excellent alternative that maintains a
sal detector for detection. Compared the flavonoids in Ginkgo and astragalo-
stable baseline and produces strong
with other general detectors, the side in Astragali.
signals.
Agilent 1200 Series ELSD provides sig-
nificant benefits. Mass spectrometry Ginkgo biloba L. and its extracts are
The ELSD principle of operation con-
(MS) is more expensive and requires not only very famous in China, but also
sists of three main successive
well-trained, knowledgeable operators. are used worldwide to treat cardiovas-
processes:
MS is not normally used for the rou- cular and cerebrovascular diseases.
a) Nebulization of the chromatograph-
tine analytical work that is done in The therapeutic effects are due to the
ic eluent using nitrogen or air,
quality control departments. The ginkgolides and bilobalide that are
b) Evaporation of mobile phase at
refractive index detector provides uni- present together with the flavonoids.
relatively low temperature, and
versal detection, but can be used only Quality control methods must deter-
c) Light scattering by the residual
for isocratic analyses, so is not fit for mine the amount of the flavonoids in
particles, which ideally consist of
separation of complex mixtures. A sys- Ginkgo and its products.
analyte molecules.
tem with a UV detector that is con-
Figure 1 shows a schematic of the
nected in series with an ELSD can be Radix Astragali, in Chinese Huangqi, is
various stages of detection.
used for simultaneous determination one of the most widely used TCMs in
of multiple components with various China. Pharmaceutical studies and
structures, with or without chro- clinical practice have demonstrated
mophores. that Radix Astragali possesses many
biological functions; therefore, it is
In the Chinese Pharmacopoeia 2005, used for the treatment of nephritis,
there are several TCMs that require diabetes, hypertension, and other dis-

5989-8922EN 34
eases. One of the effective compo-
nents, astragaloside, has no chro-
mophores and needs a universal
detector to determine the amount of
the compound.
In this Application Note, the separa-
tion and the quantitative analysis of
the Ginkgo flavonoids and astragalo-
side are studied. The note also pro-
vides some general advice on use of
the Agilent 1200 Series ELSD.
Experimental
Equipment
For development of the Rapid
Resolution LC (RRLC) method, an
Agilent 1200 Series RRLC system with
the following modules was used:
• Agilent 1200 Series binary pump SL
with vacuum degasser
• Agilent 1200 Series high-performance
autosampler SL
• Agilent 1200 Series thermostatted
column compartment SL
• Agilent 1200 Series diode array
detector SL with micro flow cell (2 µL
volume, 3 mm path length)
• Agilent 1200 Series evaporative light
scattering detector with standard
nebulizer
• Agilent ChemStation B.03.02 for data
acquisition and evaluation
• Agilent ZORBAX XDB-C18 Rapid
Resolution High Throughput (RRHT)
column, 3.0 x 50 mm, 1.8 µm particle
size
System setup
The HPLC modules are plumbed in the
usual ways, as shown in figure 2. The
Agilent 1200 Series ELSD is connected
with the computer through an RS-232
cable. The remote cable connects the
Agilent 1200 Series ELSD with any
module of the HPLC.
degasser
pump remote cable
auto-sampler
inlet
column
com-
partment
DAD
LC ChemStation
RS 232
waste
Figure 2
Agilent 1200 Series RRLC system with Agilent 1200 Series evaporative light scattering detector.
System checkout the Agilent 1200 Series RRLC system
Before starting an experiment, a test can be used as the standard system
run should be done to make sure that when conventional HPLC is needed.)
the Agilent 1200 Series ELSD is in
good condition and has sufficient per- After the method has been run, the
formance. response of the detector should be
recorded when the instrument is in
Test run conditions: good condition after the installation.
• Sample: caffeine at 250 µg/mL Then one should use the same method
(or similar concentra- every time the performance check is
tion) needed.
• Solvent: isocratic, 80 % water,
20 % acetonitrile Nonvolatile buffers cannot be used
• Flow rate: 1 mL/min with the ELSD. If nonvolatile salt goes
• Injection volume: 20 µL into the ELSD, the detector cannot dis-
• Column: Agilent ZORBAX tinguish whether the nonvolatile parti-
Eclipse XDB-C18, cles came from the sample or from the
4.6 x 150 mm, 5 µm buffer. Therefore, the baseline increas-
(or other column with es dramatically and the signal-to-noise
comparable size) is reduced for the sample.
• Thermostatted column compartment
(TCC) temperature: 40 °C Because ELSD can detect only com-
• ELSD temperature: 40 °C pounds that are less volatile than the
• ELSD pressure: 3.5 bar (51 psi) mobile phase, when new methods are
• ELSD gain: 7 developed, another detector should be
• ELSD filter: 1 second used to make sure all the compounds
• Typical system: Agilent 1200 Series can be eluted from the column.
standard LC system (Please note that
35 5989-8922EN
Standards and sample preparation
The following standards, samples, and
solvents were used for these experi-
ments:
• Standards were purchased from
NICPBP (National Institute for the
Control of Pharmaceutical and
Biological Products).
• Solvents (methanol, acetonitrile,
tetrahydrofuran) were purchased
from Thermo Fisher Scientific.
• Water was obtained from a Millipore
pure water system.
• The Astragali sample was purchased
from a TCM drug store.
• Ginkgo extracts were kindly provided
by a customer.
Samples were prepared by weighing 1
gram sample, dissolving it in 5 mL
methanol, mixing in an ultrasonic bath
for 30 minutes, then
filtering and saving the clear liquid for
injection.
• ELSD: temperature = 40 °C,
pressure = 50 psi,
gain = 5,
filter = 3 seconds
Results and Discussion
Analysis of Ginkgo
Ginkgo and its products have a very
complex composition and require a
good separation by HPLC. Isocratic
methods were used frequently in the
past because of limitations in the
instrumentation. The disadvantages of
isocratic methods include long analy-
sis time and lower resolution. A gradi-
ent method was used in this applica-
tion to get better separation in a short-
er time.
Figure 3 shows chromatograms of
standards of Ginkgo flavonoids, and
compares the signals on ELSD and
two channels of UV. The flavonoids
have very low response at a wave-
length of 210 nm and no response at
254 nm. Comparison of the sensitivity
between UV and ELSD shows why it is
necessary to use ELSD to determine
the flavonoids in Ginkgo. By checking
complementary UV signals, this
method can also be used to tell if the
peaks are pure flavonoids.
The chromatogram in figure 4 shows
that there are many peaks in the
Ginkgo sample that can be detected by
ELSD. The sample composition is very
complex and the ELSD information can
help to find more components, includ-
ing those that lack chromophores. The
four flavonoid peaks are well-separat-
ed from each other and from the other
components.
When running ELSD methods, even
when there are no more peaks in the

Method for Ginkgo analysis

• Mobile phase: A = water, mAU
B = tetrahydro- UV 254
20
furan/methanol 10
with a volume 0
ratio of 10/25 0 2.5 5 7.5 10 12.5 15 17.5 20 min
• Flow rate: 0.7 mL/min
mAU
• Gradient: 0 min, 12 %B; UV 210
20
10 min, 16 %B;
10
15 min, 22 %B;
20 min, 30 %B 0
• ELSD: temperature = 40 °C, 0 2.5 5 7.5 10 12.5 15 17.5 20 min
pressure = 50 psi, mV
ELSD
gain = 7, 20
filter = 3 seconds 10 1 2 3 4
0
Method for Astragali analysis 0 2.5 5 7.5 10 12.5 15 17.5 20 min
• Mobile phase: A = water,
B = acetonitrile
Figure 3
• Flow rate: 0.7 mL/min Comparison of chromatograms of Gingko flavonoids between ELSD and different channels of UV.
• Gradient: 0 min, 20 %B;
1 min, 30 %B;
5 min, 35 %B,
30 min, 100 %B

5989-8922EN 36
ELSD signal, one still needs to monitor
the UV channels to make sure all the
components are eluted out of the sys-
mV
tem. This avoids contamination of the 120
column and system. 100
80
60 3
Analysis of Astragali 40 1 4
When developing the HPLC conditions 20 2
0
for analysis of astragaloside in
0 5 10 15 20 25 30 35 min
Astragali, one must separate all the
standards and the other components
in the real TCM samples. To reduce the
separation time, a gradient analysis
Figure 4
can be used. Isocratic analysis can be ELSD chromatogram of Ginkgo sample, with the flavonoid peaks numbered.
used when the separation time is not
too long and peak capacity is suffi-
mV
cient. The chromatogram at the top of 120
figure 5 shows a gradient analysis of 100
an Astragali sample, while the one at 80
60
the bottom shows the astragaloside 40
standard. The top chromatogram 20
shows good separation of the astraga- 0
loside from the other sample compo- mV 1 1.5 2 2.5 3 3.5 4 min
nents. 120
100
80
60 Astragaloside
Quantitative analysis 40
The ELSD is not always a linear detec- 20
0
tor. For some compounds, the relation-
ship between the concentration and 1 1.5 2 2.5 3 3.5 4 min
peak area is linear only in a certain
range. In some situations, the nonlin- Figure 5
ear function must be calculated. ELSD chromatogram of Astragali sample (top) and astragaloside standard (bottom).

In this application, the results shown
in table 1 indicate a linear function
because all the samples tested exhibit
a good response in a linear range.
The detection limit is highly related to
the conditions developed on the ELSD.
For example, the ELSD parameters for
gain and filter influence the detection
limit.
The limits listed in table 2 were
obtained using methods that were
similar to those used for the sample
analyses. The numbers shown in table
2 give the basic idea of the amount of
each standard that can be detected
with the current methods.

37 5989-8922EN
Conclusion Compound name Function Linear range (µg/mL) Regression factor
Ginkgolide A y = 31.45689x – 4.18609 20 - 2000 0.99791
Two methods that used the Agilent Ginkgolide B y = 49.66089x – 2.95274 260 - 2600 0.99932
Ginkgolide C y = 75.64218 + 17.60311 20 - 2000 0.99468
1200 Series RRLC and the Agilent
Bilobalide y = 89.66172x – 13.33119 15 - 1500 0.99955
1200 Series ELSD were developed Astragaloside y = 15.8618x – 8.0644 11 - 1100 0.9933
in this application. The methods
Table 1
achieved good separation of the stan- Quantitative results.
dards and TCM components. The
detection limits of the TCM standards
Compound name Detection limit (ng)
also showed that the system is capa-
ble of detecting low levels of compo- Ginkgolide A 12.5
Ginkgolide B 19.5
nents. The quantitative analyses done Ginkgolide C 30
here give the holistic idea of how to Bilobalide 15
meet the requirements of the Chinese Astragaloside 22
Pharmacopoeia. The methods can be Table 2
used for the quality control of TCMs Detection limits of standards.
that contain related components.

References
5.
1.
B.T. Mathews et al. “Improving
Pharmacopoeia of the People’s
Quantitative measurements for the
Republic of China, 2005.
Evaporative Light Scattering Detector “
Chromatographia, 60 ,625-633, 2004.
2.
L.-W. Qi et al., “Quality evaluation of
6.
Radix Astragali through a simultane-
E.Oppenheimer et. al “Examination of
ous determination of six major active
the concentration response
isoflavonoids and four main saponins
Evaporative Light Scattering Mass
by high-performance liquid chromato-
Detector” J. Chrom 323, 297-304, 1985.
graphy coupled with diode array and
evaporative light scattering detectors”,
7.
J. Chromatogr. A, 1134, 162-169, 2006.
M, Dreux et al. “The Evaporative
Light Scattering Detector- a Universal
3.
Instrument for Non- Volatile Solutes in
Wang Qiao'e et al. “Progresses on
LC and SFC” LC-GC Int, March 1996.
Evaporative Light-scattering
Detection” J. of Instrumental analysis
8.
(Chinese), Vol. 25, No.6 ,126-132,
Su Hong et al. “Determination of
2006.
Jujubaside A in Ziziphi Spinosae Zhixiu Xu is an Application Chemist at
Mixture by HPLC-ELSD” J. of Modern Agilent Technologies, Shanghai, China.
4.
Food and Pharmaceuticals (Chinese),
Roger Trones et. al ”Modified laser
Vol. 17 No 6, 35-37, 2007.
light-scattering detector for use in
high temperature micro liquid
chromatography” J. of Chrom A,
814, 55-61, 1998.

5989-8922EN 38
 Analysis of TCMs with the
Agilent 1200 Series evaporative
light scattering detector

Application Note
Manufacturing QA/QC

Author
*DAD1 C, Sig=210,4 Ref=off (OCT27/NOV11_BEIMU1.D)
Zhixiu Xu mAU *ELS1 A, Voltage (OCT27/NOV11_BEIMU1.D)
Agilent Technologies
80
Shanghai, China
60
Peimine
40 Peiminine

20

0 UV 210 nm

0 2 4 6 8 10 12 min

Abstract
This Application Note describes the HPLC separation methods that were developed
for five traditional Chinese medicines (TCMs) – Bulbus Fritillariae Thunbergii (zhe bei
mu), Rhizoma Anemarrhenae (zhi mu), Sanjin Tablets, Fructus Kochiae (difuzi), and
Fructus Liquidambaris (lulutong). The HPLC methods were optimized to give better
separation for the maximum number of components. Sample preparation methods are
also described, to help readers who wish to follow the entire
application.

39 5990-3412EN
Introduction
The Chinese Pharmacopoeia 2005 pub-
lished several methods that use an
evaporative light scattering detector
(ELSD) for analysis of TCMs.1 Because
these TCMs contain some compounds
that have no chromphores or have very
weak UV absorbance, one needs a uni-
versal detector like the ELSD to quanti-
fy these compounds in TCMs. Another
Application Note demonstrated the
ELSD analysis of Ginkgo and Astragali,2
which are two of the TCMs on the
pharmacopoeia requirement list.
Bulbus Fritillariae Thunbergii
(zhe bei mu), Rhizoma Anemarrhenae
(zhi mu), Sanjin Tablets, Fructus Kochiae
(difuzi), and Fructus Liquidambaris
(lulutong) are the rest of TCMs listed in
the Chinese Pharmacopeia 2005 ver-
sion that need ELSD as the detector to
analyze certain active components.
The startup procedure and the quantita-
tive analysis were already introduced in
the former two Application Notes.2,3
The quantitative analysis is quite sim-
ple with the pharmacopoeia require-
ment, which needs only two data
points and uses a log-transform of the
two points to generate a curve for
quantitation. So the quantitative analy-
sis will not be shown in this
Application Note. A brief summary of
the results for the separations, as well
as gradients and instrument parame-
ters, will be introduced.
Because TCMs are quite complex, nei-
ther the diode array detector (DAD) nor
the ELSD alone can detect all of the
components. In this study, DAD and
ELSD were used in a serial arrange-
ment to get maximum information for
the samples and to monitor whether all
the peaks eluted from the system.
Achieving a good separation for the
5990-3412EN
maximum number of peaks is also help-
ful for use of fingerprint chro-
matograms for analysis of other
unknown components.
An Agilent 1200 Series Rapid
Resolution LC (RRLC) system was used
for the separation. With this system,
one can easily choose to use sub-two-
micron RRLC columns or conventional
HPLC columns for different analysis
requirements for TCM samples. If a nar-
row internal diameter (id) column is
used, one needs an RRLC nebulizer
(Agilent part number G4218-20004) for
narrow peak width. The examples
shown in this Application Note take
separations and sample throughput
into account to balance both. One can
easily use another column with the
same packing material but different col-
umn inner diameter and length to
speed up the separation or increase the
peak capacity to meet different require-
ments.
Experimental
Standards and materials
Liquidambaric acid, kochioside Ic,
madecassoside, sarsasapogenin,
peimine, and peiminine were purchased
from the National Institute for the
Control of Pharmaceutical and
Biological Products (Beijing, China).
The samples of TCMs were procured
from the TCM drug store in Shanghai.
Instrumentation
An Agilent 1200 Series Rapid Resolution
LC system was used for conventional
and RRLC method development. It
included the following modules:
• Agilent 1200 Series binary pump SL
with vacuum degasser
• Agilent 1200 Series high performance
autosampler SL
40
• Agilent 1200 Series thermostatted
column compartment SL
• Agilent 1200 Series diode array detec-
tor SL with micro flow cell
(2 µL volume, 3 mm path length)
• Agilent 1200 Series evaporative light
scattering detector with standard neb-
ulizer
• Agilent ChemStation B.03.02 for data
acquisition and evaluation
Sample preparation
Fructus Liquidambaris (lulutong):
The sample was prepared by weighing
accurately 0.6 g of the powder into
a vial, adding 20 mL of dehydrated
ethanol, weighing the mixture, and ultra-
sonicating for 15 minutes. The
mixture was cooled and weighed
accurately again. After replenishing the
lost solvent with dehydrated ethanol, the
sample was mixed well and filtered.
Then 10 mL of the filtrate was measured
accurately and evaporated to dryness.
The residue was dissolved with dehy-
drated ethanol to a final volume of 2 mL.
The solution was filtered through a 0.22
µm nylon membrane filter prior to direct
injection.
Fructus Kochiae (difuzi):
This sample was prepared by weighing
accurately 0.5 g of the powder into
a vial and adding exactly 10 mL of
methanol. The mixture was weighed
accurately, allowed to stand overnight,
ultrasonicated for 30 minutes, cooled,
and weighed again. The lost weight was
made up with methanol to the
initial weight, and the sample was mixed
well. The solution was filtered through a
0.22 µm nylon membrane
filter prior to direct injection.
Sanjin Tablets: Bulbus Fritillariae Thunbergii Flow rate: 0.8 mL/min
Two tablets were pulverized. The coats (zhe bei mu): UV detector: 210 nm, 254 nm
were removed and the tablets were The sample was prepared by weighing ELSD temperature: 40 °C
ground into a fine powder. Then 0.5 g of accurately 2 g of the powder into a vial ELSD pressure: 50 psi
the powder was weighed accurately into with a cap, adding 4 mL of strong ammo- ELSD gain: 9
a vial, and 10 mL of methanol was added nia, and macerating for 1 hour. Then 20 ELSD filter: 3 s
accurately. The solution was ultrasoni- mL of a mixture of chloroform and
cated for 30 minutes, cooled, mixed well, methanol (4:1) was added accurately. Sanjin Tablets:
and filtered. Then 5 mL of the solution The sample was weighed and then heat- Column: Agilent ZORBAX SB-C18,
was measured accurately. The solution ed under reflux on a water bath at 80 °C 3 x 150 mm, 5 µm
was washed with two 0.5 mL quantities of for two hours. The sample was then Solvents: A = water, B = methanol
ammonia. The extracts were evaporated cooled and weighed again, and the lost Flow rate: 0.6 mL/min
to dryness, and the residue was dis- solvent was replenished with the above Gradient: 0 min, 45 %B; 10 min,
solved in methanol and mixed well to mixture. The sample was filtered and 5 55 %B; 20 min, 100 %B
create the test solution. The solution mL of the successive filtrate was mea- UV detector: 210 nm, 254 nm
was filtered through a sured accurately and evaporated to dry- ELSD temperature: 40 °C
0.22 µm nylon membrane filter prior to ness in an evaporating dish. The residue ELSD pressure: 50 psi
direct injection. was dissolved with methanol, trans- ELSD gain: 9
ferred to a vial, diluted with methanol to ELSD filter: 3 s
Rhizoma Anemarrhenae (zhi mu): a final volume of 10 mL, and was mixed
This sample was prepared by weighing well. The solution was filtered through a Rhizoma Anemarrhenae (zhi mu):
accurately 0.5 g of the powder into a vial 0.22 µm nylon membrane filter prior to Column: Agilent ZORBAX Eclipse
with a cap, then accurately adding 10 mL direct injection. C18, 3 x 50 mm, 1.8 µm
of ethanol, weighing, and macerating Solvents: A = water, B = methanol
overnight. The sample was ultrasonicat- Flow rate: 0.6 mL/min
ed for 40 minutes, cooled, and weighed Methods Gradient: 0 min, 40 %B; 5 min,
again. The lost weight was replenished 45 %B; 10 min, 80 %B;
with ethanol and the sample was mixed Fructus Liquidambaris (lulutong): 15 min, 100 %B
well. Then 10 mL of water and 1 mL of Column: Agilent ZORBAX Eclipse UV detector: 210 nm, 254 nm
hydrochloric acid were added, and the C18, 4.6 x 150 mm, 1.8 µm ELSD temperature: 45 °C
mixture was heated under reflux for two Solvents: A = water (0.1 % acetic ELSD pressure: 50 psi
hours. The sample was cooled, and a 40 acid), B = methanol ELSD gain: 9
% solution of sodium hydroxide was Flow rate: 0.8 mL/min ELSD filter: 3 s
added dropwise with shaking until the Gradient: 0 min, 80 %B; 10 min,
color of the solution changed suddenly 85 %B; 30 min, 100 %B Bulbus Fritillariae Thunbergii
from orange-yellow to orange-red. The UV detector: 210 nm, 254 nm (zhe bei mu):
sample was extracted with two 5 mL ELSD temperature: 40 °C Column: Agilent ZORBAX Eclipse
quantities of chloroform. The extracts ELSD pressure: 50 psi C18, 3 x 50 mm, 1.8 µm
were combined and evaporated to dry- ELSD gain: 9 Solvents: A = water (0.03 % ammonia,
ness. The residue was dissolved in ELSD filter: 3 s pH 9), B = acetonitrile
methanol, diluted to a final volume of 5 Flow rate: 0.6 mL/min
mL with methanol, and was mixed well. Fructus Kochiae (difuzi): Gradient: 0 min, 45 %B; 5 min,
The solution was filtered through a 0.22 Column: Agilent ZORBAX Eclipse 45 %B; 10 min, 70 %B
µm nylon membrane filter prior to direct C18, 4.6 x 100 mm, 3.5 µm UV detector: 210 nm, 254 nm
injection. Solvents: A = water (0.1 % acetic ELSD temperature: 45 °C
acid), B = methanol, ELSD pressure: 50 psi
isocratic, 90 %B ELSD gain: 9
ELSD filter: 3 s

41 5990-3412EN
Results and discussion *DAD1 A, Sig=254,4 Ref=off
ELS1 A, Voltage
Fructus Liquidambaris (lulutong) mAU
Figure 1 shows the chromatograms of Liquidambaric acid
140 2
Fructus Liquidambaris, in Chinese lulu-
tong, from the UV-254 nm channel and 120
the ELSD channel. The liquidambaric 80
acid is labeled in figure 1. By comparing
the two channels, one can easily find 60 UV 254 nm
1
the differences. Liquidambaric acid has 40
no response in the UV-254 nm channel,
but responds well with the ELSD. Some 20
other components that are eluted at the 0
beginning of the run have no response 0 5 10 15 20 25 30 min
with the ELSD, but do have a response
with UV detection at 254 nm. The 210
nm channel is not shown here because Figure 1
Fructus Liquidambaris chromatograms with ELSD detection (red), overlaid with chromatogram from
it exhibits fewer peaks and lower abun- UV channel (blue).
dances.

There are two peak pairs marked as

1 and 2, respectively. The LC method
was optimized to separate the two
peak pairs because they are the compo- Kochioside lc
nents that have a better response in
ELSD than in UV at 210 nm and could
be potential components that need
ELSD for quantitative analysis in the
future, although we do not know the
exact structure of them yet.
UV 210 nm
Fructus Kochiae (difuzi)
Figure 2 shows the separation of
Fructus Kochiae components in the
ELSD and UV-210 nm channels. The
kochioside Ic, which has very weak UV
absorbance at 210 nm, is called out.
The ELSD detected more peaks than
Figure 2
the UV detector, which shows that Fructus Kochiae chromatograms from ELSD channel (red) and UV channel at 210 nm (blue).
Fructus Kochiae has some components
that have no chromophore and need
both detectors in series to achieve
information from both channels in one
run. Some early-eluting components
have no chromophores and showed
response in the ELSD channel only.

5990-3412EN 42
Sanjin Tablets
Figure 3 shows chromatograms from *DAD1 C, Sig=210,4 Ref=off (OCT27/NOV11_BEIMU1.D)
mAU *ELS1 A, Voltage (OCT27/NOV11_BEIMU1.D)
both the ELSD and the UV-210 nm
channels for the analysis of Sanjin 80
Tablets. The active component, made-
cassoside, is labeled. The UV signal at
60
210 nm has fewer peaks than the ELSD
Peimine
channel because most of the compo-
nents injected onto the column have no 40 Peiminine
chromophores. In addition, the baseline
for the UV signal at 210 nm cannot 20
remain flat because the absorbance of
the solvent used for the separation 0 UV 210 nm
changes during the gradient. Without
the ELSD, there would not be very 0 2 4 6 8 10 12 min
much information regarding the peaks
and the fingerprint of Sanjin Tablets for Figure 3
quality control. Sanjin Tablets chromatograms with UV-210 nm channel (blue) and ELSD channel (red).

Rhizoma Anemarrhenae (zhi mu)

The chromatogram from the ELSD
channel showed that the TCM pur-
chased from the TCM store was not of
very good quality, as the peak of
sarsasapogenin was small. In the UV
signal at 210 nm, sarsasapogenin
showed no response. For other compo-
nents of Rhizoma Anemarrhenae, the
ELSD channel detected more peaks
than the UV channel at 210 nm, and the
UV 210 nm
chromatograms showed different pro-
files. Rhizoma Anemarrhenae needs
these two complementary detectors to Sarsasapogenin
get more information about the compo-
nents with and without chromophores.

Figure 4
Rhizoma Anemarrhenae chromatograms from ELSD channel (red) and UV-210 nm channel (blue).

43 5990-3412EN
Bulbus Fritillariae Thunbergii
(zhe bei mu) *DAD1 C, Sig=210,4 Ref=off
mAU ELS1 A, Voltage
Peimine and peiminine are two most
important components of Bulbus
80
Fritillariae Thunbergii. The UV trace at
210 nm showed very few peaks. The
ELSD showed the good separation of 60
the two active components, and the Peimine
separation finished in 12 minutes with 40 Peiminine
the sub-two-micron column and the
Agilent 1200 Series RRLC system. With
the current method, the two compo- 20
nents can be easily separated. When
peak capacity is not an issue, one can 0 UV 210 nm
speed up the analysis to achieve higher
sample throughput with the Agilent 0 2 4 6 8 10 12 min
1200 Series RRLC system.

Figure 5
Bulbus Fritillariae Thunbergii chromatograms of ELSD channel (red) and UV channel at 210 nm
Conclusion (blue).

In this Application Note and a previous

one (Agilent publication number 5989-
8922EN), analysis methods were devel-
oped for TCMs that require detection
by ELSD, according to the Chinese
Pharmacopoeia 2005. The sample
preparation methods and the instru-
mentation parameters were
discussed for various samples.
ELSD is one of the best choices when
analyzing compounds that contain no
chromophore. ELSD can provide a
better signal when the compounds lack
a functional group that absorbs well in
the UV range. ELSD also can maintain a
better baseline when UV-absorbing sol-
vents are used.
It is necessary to use UV and ELSD in
series to have signals from both chan-
nels because neither of them can
detect all the peaks if used alone. The
methods described in this Application
Note can be used in quality control
departments because the detailed
information is provided here.
References
1.
Pharmacopoeia of the People’s
Republic of China, volume 1, China
Pharmacopoeia Commission, People’s
Medical Publishing House, 2005.
2.
Z. Xu, “Analysis of traditional Chinese
medicines with the Agilent 1200 Series
evaporative light scattering detector,”
Agilent Technologies Application Note,
publication number 5989-8922EN, 2008.
3.
Z. Xu, H. Huang, X. Zhang, “Analysis of
Radix Bupleuri (Chaihu) using Agilent
1200 Series LC Systems with
Evaporative Light Scattering Detector,”
Agilent Technologies Application Note,
publication number 5989-9834EN, 2008.

5990-3412EN 44
Monitoring of Pesticide Residues

45
46
 Screening for 430 Pesticide Residues in
Traditional Chinese Medicine Using
GC/MS: From Sample Preparation to
Report Generation in One Hour

Application

Traditional Chinese Medicine

Authors Introduction
Traditional Chinese Medicine (TCM), which has been used
Wei Luan and Zhixiu Xu in China for thousands of years, shows very good thera-
Agilent Technologies (Shanghai) Co., Ltd. peutic effects in China as well as other countries. More
TCM crops than ever are being cultivated in large-scale
412 Ying Lun Road
farming operations. The pesticides that are used on the
Waigaoqiao Free Trade Zone farm need to be monitored to make sure that the amount
Shanghai 200131 of residues is kept within the safe range. TCM have very
China complex matrices, several hundred kinds of pesticides are
often used, and the amount of residue is very small. All
these facts make pesticide residue screening a very diffi-
Abstract cult task to accomplish in a short time.

A total solution of screening for pesticide residues in
Traditional Chinese Medicine (TCM) is described in this
application, including sample preparation procedure,
instrument acquisition parameter, advanced mass spec-
trum data processing, and automatic qualification and
quantitation report generation. Several techniques are
used in this screening method. QuEChERS is employed to
shorten the sample pretreatment time, reduce its cost,
and environmental contamination. A significant instru-
mentation innovation is the capillary flow technology of
Agilent 7890A, which can perform the backflushing to
remove chemical noise caused by dirty matrices and save
cycle time. Agilent's Deconvolution Reporting Software
(DRS) uses the AMDIS deconvolution algorithm, which
can be used to screen for 430 GC-amenable pesticides
with the new Japanese Positive List RTL database in only
2 or 3 minutes. Overall, this screening method, from
preparing the sample to generating a report, can be
accomplished in about 1 hour.
There are several steps to solve this problem, including
sample pretreatment, instrument acquisition, data process-
ing, and report generation. The basic requirements for
using a gas chromatograph/mass spectrometer (GC/MS)
are listed here:
1. An effective and easy sample preparation method.
2. Repeatable and reproducible retention times.
3. A database containing the information about retention
time and mass spectra
QuEChERS, a sample preparation method for pesticide
analyses in foodstuff, is the acronym for Quick, Easy,
Cheap, Effective, Rugged, and Safe developed by Dr.
Steven J. Lehotay and his colleagues [1-5]. QuEChERS is
selected here because it has many advantages over tradi-
tional techniques, such as high recovery for pesticides’
wide range of polarity and volatility, high sample through-
put, and the use of a smaller amount of organic solvent
and no chlorinated solvents [1-5]. QuEChERS meets the
first basic requirement mentioned above.

47 5989-9341EN
Retention time locking (RTL) [6] is the ability to very close-
ly match chromatographic retention time in any Agilent
6890/7890 GC system to those in another 6890/7890 GC
system with the same nominal column. It is remarkable
that the repeatability of retention time in Agilent 7890GC is
better than 0.01 minute. RTL can precisely match the
retention time between different instruments, which meets
the second basic requirement mentioned above.
On November 29, 2005, the Japanese Ministry of Health,
Labour, and Welfare (MHLW) published a “Positive List”
system for the regulation of pesticides, feed additives, and
veterinary drugs. This system is one of the most restrictive
regulations on pesticide residues all over the world.
Referring to this regulation, Agilent developed a new data-
base called the Japanese Positive List RTL database,
which includes all GC-amenable pesticides in the
Japanese Positive List, together with other pesticides that
are monitored by the Japanese Quarantine Stations. The
information in the Agilent database comprises the reten-
tion time and spectra of each pesticide, which meets the
last basic requirement mentioned above [7].
In addition, the advanced requirement for screening is to
rapidly identify pesticides, with minimal false positives and
negatives.
Agilent’s Deconvolution Reporting Software (DRS) [7] is
designed to automatically deconvolute the spectra from
matrices and generate a qualification and quantitation
report. DRS integrates information from three processes:
MSD ChemStation, Automated Mass Spectral
Deconvolution and Identification System (AMDIS), and
National Institute of Standards and Technology (NIST)
search. DRS increases the confidence in results for com-
plex matrices; the typical data processing time is about 2
or 3 minutes, which perfectly meets the advanced require-
ment mentioned above.
This application describes how to use the techniques men-
tioned above to develop a whole process for screening
pesticides in TCM matrices. The investigated targets
include different officinal parts of TCM, such as Flos
Lonicerae Japonicae for flower, Radix Astragali for root
and stem, Folium Ginkgo for leaf, Radix et Rhizoma
Ginseng for root, and Fructus Lycii for fruit. After about a
20-minute pretreatment by QuEChERS, the extract is inject-
ed into an Agilent 7890A/5975C. The system then runs the
Japanese Positive List method with backflushing in about
37 minutes. At the end, after a 2- or 3-minute DRS process-
ing with the Japanese Positive List RTL database, a qualifi-
cation and quantitation report is automatically generated.
5989-9341EN
Experimental
TCM raw materials were purchased from a local medicine
store, including Flos Lonicerae Japonicae, Radix Astragali,
Folium Ginkgo, Radix et Rhizoma Ginseng, and Fructus
Lycii. All materials are finely ground.
Sample Preparation
• Weigh 1 g of ground TCM and add it into a
50-mL Teflon tube; add 6 mL distilled water to thor-
oughly soak the sample and store the sample at ambi-
ent temperature for 2 hours.
• Add 4 mL 0.1% HAc/acetonitrile, and shake
vigorously for 1 minute by hand, then add 1.2 g MgSO4,
0.6 g NaAc, and 4 g NaCl, and shake
vigorously for 1 minute by hand. Centrifuge for 3 min-
utes at 4,000 rpm.
• Add 500 mg PSA (p/n 5188-5364), 250 mg C18 (p/n
189-1302), and 250 mg MgSO4 in a 5-mL centrifuge
tube. Transfer an aliquot (1.5 mL) of the above extract’s
upper layer into the tube and then centrifuge for 1
minute at 3,000 rpm. If the solution is dark after cen-
trifugation, graphitized carbon black can be added to
adsorb the pigment (for example, chlorophyll).
• Transfer an aliquot (1.0 mL) of the extract into the GC
vials.
For dedicated quantitation analysis after screening, ana-
lytical protectants and internal standard can be added to
the final solution.
GC/MS
The Japanese Positive List method was published by
MHLW for pesticide residues on all Japanese agricultural
products and imports to Japan. Here, the method was
modified to handle the dirty TCM matrices by adding the
backflushing. The detail is displayed in Table 1.
System Software Requirement
Agilent MSD ChemStation (p/n G1701EA, Ver. E.02.00);
Agilent Deconvolution Reporting Software (p/n G1716AA,
Ver. A.04.00); NIST05b Mass Spectral Library (p/n
G1033A), includes NIST MS Search (Ver. 2.0d) and AMDIS
(Ver. 2.65); Agilent Japanese Positive List Pesticide RTL
Database (p/n G1675AA).
48
Results and Discussion in dealing with the acidic component in TCM extract. By
screening for different doses of PSA, 500 mg is enough for
QuEChERS Method Optimization for Versatile the analysis. A typical chromatogram of mixed TCM sam-
TCM Matrices ple after QuEChERS is shown in Figure 1.

In Traditional Chinese Medicine formulation, different edi-
ble parts of TCM plants can be chosen to make each dose.
Moreover, the TCM matrices are versatile in chemical con-
stituents, such as saponins, polysaccharids, flavones, phe-
nolic acid, and fatty acid. Therefore, it is a tough job to
determine the pesticide residues in such kinds of complex
matrices as TCM.
In our study, Flos Lonicerae Japonicae was investigated at
first, then we mixed different kinds of TCM together to be
more representative in respect to chemical constituents.
Flos Lonicerae Japonicae for flower, Radix Astragali for
root and stem, Folium Ginkgo for leaf, Radix et Rhizoma
Ginseng for root, and Fructus Lycii for fruit are the five
selected TCM.
Regarding the QuEChERS method, the amount of
absorbent, such as PSA and C18, is critical to the recovery
of pesticides. It is demonstrated that PSA is more effective
Table 1. Gas Chromatography and Mass Spectrometers Conditions
Backflushing Using Capillary Flow Technology for
Complex TCM Matrices
Backflushing is demonstrated to significantly remove
chemical noise from run to run by reversing the column
flow to push the higher boilers out of the inlet end of the
column. Traditionally, the higher boilers stay near the front
of the column until the oven temperature is high enough to
move them through the column. It is well known that bak-
ing the column at high temperature is inclined to cause
heavy column bleed and result in shorter column lifetime.
For a dirty matrix, even baking for a long time cannot thor-
oughly remove the higher boilers, which may result in a
retention time shift in the following injection. Backflushing
is a perfect solution to avoid high-temperature baking and
retention time shift from run to run. Backflushing mini-
mizes the contamination that goes into the detector, which
is especially valuable for the MSD because it reduces ion
source cleaning. Another positive attribute of backflushing

GC Agilent Technologies 7890A or 6890N Front Injector

Sample washes 0
Front Inlet Split/splitless
Sample pumps 6
Injection type Splitless
Injection volume 2 µL
Inlet temp 280 °C
Syringe size 10 µL
Inlet liner Helix double taper deactivated
Preinj solv A 0
(p/n 5188-5398)
Preinj solv B 0
Pressure 17.446 psi
Postinj solv A 6
Purge flow 50 mL/min
Postinj solv B 6
Purge time 0.75 min
Viscosity delay 1 second
Total flow 54.51 mL/min
Plunger speed Fast
Gas saver On
Preinjection dwell 0 minutes
Saver flow 20.0 mL/min
Helium 0 minutes
Saver time 1.00 min
Gas type Helium MSD Agilent Technologies 5975C with Triple-
Axis Detector
Oven
Acquistion mode Scan
Oven ramp °C/min Next °C Hold min
Solvent delay 3.30 min
Initial time 50 1
Low mass 35
Ramp rate 25 125 0
High mass 450
Ramp rate 10 300 10
Threshold 100 (or setting according to noise level)
Post run 300 5
Sampling 2
Total run time 36.5 min
Quad temp 150 °C
Equilibration time 0.5 min
Source temp 230 °C
Column J&W DB-5ms Ultra Inert (p/n 122-5532UI) Transfer line temp 280 °C
Length 30 m Tune file Atune.u
Diameter 0.25 mm Gain factor 10
Film thickness 0.25 µm Trace ion detection On
Mode Constanr flow
RTL System retention time locked to
Pressure 17.446 psi (2 psi for post run)
Chlorpyrifos methyl at 13.443 min
Nominal initial flow 1.51 mL/min (–4.05 mL/min for post run)
Inlet Front Inlet Backflush
Outlet PCM (p/n G1530-63309) Device 3-way splitter (p/n G3183B)
Outlet pressure 2 psi (60 psi for Post run) Restrictor Size 0.706 m ˘ 0.18 mm id

49 5989-9341EN
is that it shortens cycle time and improves lab productivity.
Details about backflushing, including a method develop-
ment procedure, can be found in the Agilent application in
reference 8.
In our experiment, a three-way splitter with makeup gas
was used to perform the backflushing. The device has
extra makeup gas supply tubing and four plug-in spots: one
for the analytical column and three for the restrictor tube
connecting to three available detectors. Since only MSD is
used as a detector in this application, the first two spots
are plugged in the direction of makeup flow. The third spot
is used for column in and the last spot is used for restrictor
in. The main purpose of this configuration is to avoid peak
broadening due to extra dead volume. The length and inter-
nal diameter of the restrictor tubing is 0.706 m and 0.18
mm, respectively. A photo of the device and the GC/MS
system configuration schematics are displayed in Figure 2.
Target Pesticides Identification in TCM Matrices by DRS
Another challenge for pesticides analysis is the identifica-
tion of trace-level target compounds in complex matrices,
which are often disturbed by high-level chemical noise and
result in poor library match factors. Background subtrac-
tion is the traditional way to improve the match factor;
however, it is both matrix- and operator-dependent and can
yield inconsistent results.
Figure 1 Typical Total Ion Chromatogram of mixed TCM sample after QuEChERS: Flos Lonicerae Japonicae, Radix
Astragali, Folium Ginkgo, Radix et Rhizoma Ginseng, and Fructus Lycii.
5989-9341EN
DRS, which Agilent introduced in 2004, is a software pack-
age that combines the information from three separate
processes into one easy-to-read report. The primary benefit
of DRS is significant time saving when interpreting results
from complex matrices analyses. Moreover, a suitable
database is critical to perform DRS successfully. The
Japanese Positive List RTL database is a new DRS-usable
library for pesticide analysis, and includes the retention
time and mass spectra for 430 GC-amenable pesticides.
Detailed information about the database can be found in
the Agilent application in reference 7.
Using DRS with the Japanese Positive List RTL database is
highly recommended for data processing and report gener-
ation for pesticide residues analysis in TCM matrices.
Figure 3 is part of a typical DRS report for a spiked mixed
TCM sample.
Figure 4 shows the benefit of using DRS in the analysis of
pesticides in TCM matrices. Pirimiphos-methyl was spiked
into mixed TCM sample at a level of 10 ppb. The compound
was missed by MSD ChemStation but successfully picked
up by DRS. The upper window is the Total Ion Chromato-
gram (TIC), the middle window is the raw or dirty spectrum
in the scan No. 2199 (13.981 min), and the lower window
is the comparison of the deconvoluted spectrum (the white
plot) with the spectrum of pirimiphos-methyl in the
Japanese Positive List RTL database (the black plot). After
deconvolution, the spectrum of the scan No. 2199 is
"clean," and the Pirimiphos-methyl can be easily identified
in the mixed TCM sample.
50
 7683
0.706 m × 180 µm restrictor
Auto-sampler
AUX EPC
5975C
MSD
Column
7890A
GC
30 m × 250 µm × 0.25 µm DB-5MS UI
Figure 2 Schematics of the GC/MS system configuration and tubing connection for three-way splitter.
Conclusions
The primary interference for trace-level pesticides analysis
is chemical noise in complex TCM matrices. This applica-
tion introduces several techniques to eliminate the inter-
ference. Well-developed sample preparation is the main
approach to remove as much of the chemical noise as pos-
sible, QuEChERS was employed to simplify the sample
preparation procedure and improve lab productivity, com-
pleting the process in only about 20 minutes. However,
even with a pretreatment method such as QuEChERS, the
chemical noise cannot be completely removed.
Backflushing is an alternative approach to eliminate the
chemical noise from run to run. Although backflushing is
not a new concept, capillary flow technology can perform a
backflushing much more effectively than traditional tech-
niques can. In this application, a three-way splitter with
makeup gas is used for backflushing, and the data-acquisi-
tion step with backflushing is almost 37 minutes. The last
step related to chemical noise in data processing is the uti-
lization of a mathematic algorithm called deconvolution.
Agilent's DRS can shorten the data processing time to only
2 or 3 minutes, which is very helpful for pesticide analysis
in TCM matrices with the new Japanese Positive List RTL
database. Altogether, only 1 hour is needed to screen 430
pesticide residues in TCM matrices from sample pretreat-
ment to screening report generation.
References
1. M. Anastassiades, S. J. Lehotay, D. Stajhbaher, and F.
J. Schenck, “Fast and Easy Multiresidue Method
Employing Acetonitrile Extraction/Partitioning and
'Dispersive Solid-Phase Extraction' for the
51
3-way splitter
with make gas
2 ports plugged
Restrictor out to MSD
Column in
Plugged
Plugged
Aux EPC in
Determination of Pesticide Residues in Produce,” J.
AOAC Int., 2003; 86(2): 412
2. S. J. Lehotay, K. Mastovska, and S. J. Yun, “Evaluation of
Two Fast and Easy Methods for Pesticide Residue Analysis
in Fatty Food Matrixes,” J. AOAC Int., 2005; 88(2): 630
3. S. J. Lehotay, K. Mastovska, and A. R. Lightfield, “Use
of Buffering and Other Means to Improve Results of
Problematic Pesticides in a Fast and Easy Method for
Residue Analysis of Fruits and Vegetables,” J. AOAC
Int., 2005; 88(2): 615
4. S. J. Lehotay, A. de Kok, M. Hiemstra, and P. van
Bodegraven, “Validation of a Fast and Easy Method for
the Determination of 229 Pesticide Residues in Fruits
and Vegetables Using Gas and Liquid Chromatography
and Mass Spectrometric Detection,” J. AOAC Int.,
2005; 88(2): 595
5. M. Anastassiades, K. Mastovska, S. J. Lehotay,
“Evaluation of Analyte Protectants to Improve Gas
Chromatographic Analysis of Pesticides,”
J. Chromatogr. A, 2003; 1015(1-2): 163
6. V. Giarocco, B. Quimby, and M. Klee, “Retention Time
Locking: Concepts and Applications,” Agilent
Technologies publication 5966-2496E
7. Philip L. Wylie, “Screening for Pesticides in Food Using
the Japanese Positive List Pesticide Method: Benefits
of Using GC/MS with Deconvolution Reporting
Software and a Retention Time Locked Mass Spectral
Database,” Agilent Technologies publication 5989-7436EN
8. Chin K. Meng, “Improving Productivity and Extending
Column Life with Backflush,” Agilent Technologies
publication 5989-6018EN.
5989-9341EN
Figure 3 Part of typical DRS report.

Figure 4 AMDIS display showing the total ion chromatogram of a mixed TCM sample (top window), the spectrum where pir-
imiphos-methyl elutes (middle window), and the deconvoluted spectrum (in white) juxtaposed to the library spec-
trum for pirimiphos-methyl (in black) (bottom window).

5989-9341EN 52
 Low Part-per-Billion Level Pesticides
Screening in Traditional Chinese
Medicine Using the Agilent 7000A
GC/MS/MS

Wei Luan, Melissa Churley, and Mike Szelewski

Application Brief

The control of pesticide residue levels is an increasing global concern. Several Highlights
international organizations and governments have established maximum residue
limits (MRLs) for a growing list of pesticides within an ever-widening scope of • Consistent ion ratios over a wide
commodities. The primary task for pesticide residue monitoring laboratories is to linearity range and peak area
develop analytical methodologies to screen for a large number of analytes at repeatability at ultra-low concentra-
trace levels within a limited time frame. Gas chromatography coupled with mass tions in matrix make pesticide
spectrometry (GC/MS) has been adopted as the standard instrumentation for this screening more reliable.
purpose. As regulations in Japan and the European Union require lower MRLs for • The Agilent collision cell allows for
pesticide residues, the latest challenge has been to reach part-per-billion level excellent repeatability at 2 ppb level
concentrations for hundreds of pesticides in complex matrices, which in turn has in matrix using just 5 ms dwell
required greater sensitivity and efficiency in pesticide screening. The Triple times. A single time segment can
Quadrupole mass spectrometer, when used in Multiple Reaction Monitoring accommodate more MRM transi-
(MRM) mode, can dramatically remove matrix interferences and significantly tions for greater productivity.
increase the effective signal-to-noise ratio (S/N). This paper describes the use of
gas chromatography coupled with triple quadrupole mass spectrometry • Reduction of matrix interference
(GC/MS/MS) to screen pesticides in Traditional Chinese Medicine (TCM) at lev- using MRM mode substantially
els as low as 1 ppb. The target pesticides and associated MRM conditions are decreases the detection limits that
listed in Table 1. are required in pesticide screening.

53 5990-3568EN
Table 1. Target Pesticides List and MRM Conditions
Collision Collision Collision
Segment Compound name R.T. Quant energy Qual1 energy Qual2 energy
1 Dichlorvos 6.00 185 & 93 15 185 & 109 15 109 & 79 5
1 Methamidophos 6.10 141 & 95 5 95 & 80 5
1 Acephate 7.82 136 & 94 10 142 & 96 5
1 Dimethoate 8.53 125 & 79 5 125 & 93 15
1 Omethoate 9.53 156 & 110 5 156 & 79 20
2 Dimethipin 11.12 118 & 90 5 118 & 73 5
2 Cyanophos 11.41 243 & 109 10 243 & 116 5
3 Phosfamidon-E 12.41 264 & 127 15 127 & 109 10
3 Parathion-methyl 12.58 263 & 109 10
3 Phosfamidon-Z 13.13 264 & 127 15 127 & 109 10
3 Quinoclamine 13.23 207 & 172 15 207 & 179 12 172 & 128 10
3 Chlorpyriphos 13.52 314 & 258 15 314 & 286 5 197 & 169 15
3 Cyanazin 13.52 225 & 189 15 240 & 225 5 198 & 91 10
3 Parathion 13.54 291 & 109 10 291 & 81 10
3 Fosthiazate 1-2 13.85 195 & 103 5 195 & 139 5
4 Fipronil 14.33 367 & 225 25 367 & 224 20
4 Quinalphos 14.36 146 & 118 10 157 & 129 15
4 Endosulfan-alpha 14.83 241 & 206 15 229 & 194 10 195 & 159 10
4 Chlorfenapyr 15.77 247 & 227 15 247 & 197 20 408 & 59 10
4 Endosulfan-beta 15.90 241 & 206 15 229 & 194 10 195 & 159 10
5 Triazophos 16.38 161 & 134 5 161 & 106 10 257 & 162 5
5 Bromopropylate 17.64 341 & 183 20 341 & 185 20
5 Azinphos-methyl 18.31 160 & 77 20 160 & 132 5 132 & 77 15
6 Cafenstrole 19.92 100 & 72 5 188 & 119 25 188 & 82 20
6 Cyfluthrin 1-4 20.03 163 & 127 5 163 & 91 15 206 & 151 25
7 Flucythrinate 1 20.46 199 & 157 5 199 & 107 30 157 & 107 15
7 Flucythrinate 2 20.64 199 & 157 5 199 & 107 30 157 & 107 15
8 Fenvalerate 1 21.14 167 & 125 10 225 & 119 15
8 Fenvalerate 2 21.33 167 & 125 10 225 & 119 15
8 Difenoconazole 1 21.53 323 & 265 15 265 & 202 20
8 Difenoconazole 2 21.60 323 & 265 15 265 & 202 20
8 Delthamethrin 21.85 181 & 152 30 253 & 172 5 253 & 93 20

Experimental
Quantitation of trace level compounds is complicated by matrix, resulting in qualifier
ion ratios out of range, or target ions buried in the complex background. With single
quadrupole mass spectrometry, selected ion monitoring (SIM) is often used to
improve the detection limit and quantitative reproducibility. In SIM mode, the MS
monitors only a few ions for each target compound within the retention time (RT)
range that the target elutes from the column. By monitoring only a few specific ions,
the signal-to-noise ratio (S/N) improves dramatically. SIM may not work well for
trace levels in matrix as the interferences in SIM are the same as scan. Triple
quadrupole mass spectrometry, through further fragmentation in a hexapole collision
cell of a selected precursor ion, allows for drastic reduction or elimination of matrix
interference. This process, referred to as Multiple Reaction Monitoring or MRM, is
based on acquisition of highly selective precursor to product ion transitions that are
5990-3568EN 54
not likely to result from fragmentation of matrix components. Precursor selectivity is
the same as in SIM but there is a high probability that at least one of the resultant
product ions will be a unique dissociation product of the precursor and not the inter-
ference.

The primary benefit is the improvement of S/N at ultra-low analyte levels with con-
sistent qualifier ion ratios over a wide concentration range, even in the most com-
plex matrices. Figure 1 illustrates the Agilent GC/MS/MS workstation-Quantitative
view. The numbers inside the red frame are the qualifier ion ratios of cyanophos over
the range from 0.1 ppb to approximately 1,000 ppb in TCM matrix, showing very good
accuracy in matrix. The numbers inside the green frame indicate the accuracy of the
calibration curve. The enlarged view of the low end of the calibration curve is
shown. Table 2 is the repeatability of the peak area of 1 ppb pesticides in TCM
matrix with six parallel injections.

Figure 1. Calibration curve and qualifier ion ratio of cyanophos in TCM matrix.

Table 2. Peak Area Repeatability of Spiked 1 ppb Pesticides in TCM Matrix

RSD of peak area (%)(n = 6)
Cyanophos 4.83
Bromopropylate 5.12

The Agilent hexapole collision cell, with its linear acceleration design, is optimized
for high-speed performance without ion ghosting or cross-talk. High-speed MRM
capability at 500 MRM/sec maximizes the number of allowable transitions as well as
minimizes the dwell time for each transition in one time segment so that more com-
pounds can be screened per run. The dwell time experiment described in Table 3 was
performed using TCM matrix spiked with 2 ppb bromopropylate. As dwell time is
decreased, the RSD for replicate run peak areas remains at < 5.0% until 5 ms is
reached. The results at 2 ms are less precise; however, the results are acceptable for
2 ppb analysis in complex matrix.

55 5990-3568EN
Table 3. Dwell Time Experiment Results: 2 ppb Bromopropylate in TCM Matrix

100 ms 50 ms 10 ms 5 ms 2 ms
Area 1 5849 5910 6265 6704 6747
Area 2 5712 6167 6189 6728 6279
Area 3 5895 5941 5966 6131 6523
Area 4 5921 6471 6551 6243 6397
Area 5 5999 6299 6119 6504 4831
Area 6 5999 6524 6415 6796 4737
Sum 35375 37312 37505 39106 35514
Avg 5895.83333 6218.667 6250.833 6517.667 5919
Std 107.618617 260.1528 209.5638 276.2496 893.2359
RSD 1.83 4.18 3.35 4.24 15.1

Figure 2 is the total ion chromatogram of the TCM extract in MRM mode by
GC/MS/MS with injection volume of 1 µL. Eleven pesticides are identified in ppb
level as shown in Table 4.

Figure 2. Total ion chromatogram of the TCM extract in MRM mode by GC/MS/MS.

5990-3568EN 56
Table 4. Screening Result of TCM Matrix Blank by GC/MS/MS

RT Name Quant transition Matrix areas Cal result

(not spiked)
5.995 Dichlorvos 185.0 & 93.0 68 —
6.100 Methamidophos 141.0 & 95.0 0 —
7.760 Acephate 136.0 & 94.0 0 —
8.526 Dimethoate 125.0 & 79.0 0 —
9.478 Omethoate 156.0 & 110.0 0 —
11.100 Dimethipin 118.0 & 90.0 0 —
11.414 Cyanophos 243.0 & 109.0 0 —
12.390 Phosfamidon-E 264.0 & 127.0 0 —
12.560 Parathion-methyl 263.0 & 109.0 1227 6.21*
13.200 Quinoclamine 207.0 & 172.0 38 —
13.500 Cyanazin 225.0 & 189.0 16 —
13.510 Chlorpyrifos 314.0 & 258.0 1224 1.99*
13.520 Parathion 291.0 & 109.0 3950 5.62*
13.840 Fosthiazate 195.0 & 103.0 0 —
14.320 Fipronil 367.0 & 255.0 0 —
14.340 Quinalphos 146.0 & 118.0 3970 1.11*
14.820 Endosulfan-a 241.0 & 206.0 156 1.89*
15.750 Chlorfenapyr 247.0 & 227.0 0 —
15.880 Endosulfan-b 241.0 & 206.0 208 1.61*
16.360 Triazophos 161.0 & 134.0 531 1.63*
17.620 Bromopropylate 341.0 & 183.0 63 —
18.290 Azinphos-methyl 160.0 & 77.0 0 —
19.900 Cafenstrole 100.0 & 72.0 1260 3.37*
20.026 Cyfluthrin 163.0 & 127.0 0 —
20.439 Flucytrinate-1 199.0 & 157.0 0 —
20.624 Flucytrinate-2 199.0 & 157.0 0 —
21.122 Fenvarelate-1 167.0 & 125.0 4188 1.79*
21.326 Fenvarelate-2 167.0 & 125.0 13070 1.79*
21.520 Difenoconazole-1 323.0 & 265.0 0 —
21.574 Difenoconazole-2 323.0 & 265.0 0 —
21.840 Delthamethrin 181.0 & 152.0 2078 3.53*

*Potential positive sample

Summary
Pesticide residue monitoring is required to develop a methodology to accomplish the
screening for hundreds of target compounds in a very limited time and ultra low con-
centration levels. Gas chromatography coupled with triple quadrupole mass spec-
trometry can perform very selective MRM, which can dramatically remove chemical
noise from matrix interferences and improve the S/N, as well as detection limit. This
application develops a method to screen for ppb level pesticide residue in TCM.
GC/MS/MS can really help to improve the capability of pesticide residues screening
in ultra low level.

References
1. Wei Luan, and Zhixiu Xu, "Screening for 430 Pesticide Residues in Traditional
Chinese Medicine Using GC/MS: From Sample Preparation to Report Generation
in One Hour," Agilent Technologies publication 5989-9341EN

57 5990-3568EN
Wei Luan is an application chemist
based at Agilent Technologies,
Shanghai, China; Melissa Churley is an
application chemist based at Agilent
Technologies, Santa Clara, California,
USA; and Mike Szelewski is an applica-
tion chemist based at Agilent
Technologies, Wilmington, Delaware,
USA.

For More Information

For more information on our products
and services, visit our Web site at
www.agilent.com/chem.

5990-3568EN 58
Determination of Heavy Metal
Content

59
60
Determination of Toxic Elements in
Traditional Chinese Medicine Using
Inductively Coupled Plasma Mass
Spectrometry
Application Note
Hua Zhang
Yan-zhi Shi
Yu-hong Chen
Ying-feng Wang

Abstract

This Application Note describes a method for the analysis of Be, Cr, Mn, Ni, Cu, Zn,
As, Ag, Cd, Ba, Hg, Tl and Pb in Traditional Chinese Medicines using Inductively
Coupled Plasma Mass Spectrometry (ICP-MS). The samples were dissolved by
microwave digestion. To validate the method, two certified reference materials
were digested and measured. The method detection limits for all target elements
were between 0.1-7.2 ng·g-1.

61 5989-5591EN
Introduction
Traditional Chinese Medicines (TCMs)
and their related products have been
widely used in China for centuries. The
Chinese government as well as acade-
mic research scientists are paying
greater attention to the safety issues
of TCM in clinical use. This is a key
issue which needs to be resolved to
improve the development of TCM to
meet international standards and
advance its worldwide acceptance.
With the recent developments in sci-
ence and technology, people are
becoming more aware of the risks
associated with heavy metals such as
mercury (Hg), arsenic (As), lead (Pb)
and cadmium (Cd), which are present
in some TCMs. After these elements
enter the human body, they will
severely damage the hemopoietic,
immune, nervous and reproductive
systems. Because these substances
cannot be completely excreted from
the body, they will accumulate and
finally have an impact on health.
Therefore the analysis of toxic heavy
metals is crucial in quality control of
TCMs and regulations have been
implemented to restrict their levels1.
The export of TCMs is under intense
pressure due to strict regulations, and
the high content of heavy metals in
traditional Chinese medicine has
resulted in international response.
Limiting other toxic or harmful ele-
ments, such as beryllium (Be), chromi-
um (Cr), nickel (Ni), silver (Ag), barium
(Ba) and thallium (Tl), have not been
clearly specified in current hygiene
rules and regulations yet. However,
with the improvement of living stan-
dards and a desire for good health,
people now care more about whether
these elements offer advantages or
disadvantages for their wellbeing.
Therefore, strict control of the content
of these elements in TCM is required.
5989-5591EN
In addition, manganese (Mn) is good
for the human endocrine, nervous, and
enzyme systems and is the key ele-
ment of most enzymes as well as sup-
porting normal myocardial metabolism.
Zinc is good for the immune system.
Lack of zinc (Zn) will harm normal cell
metabolism and lack of copper (Cu) is
one of the reasons for coronary heart
disease2. However, if the level of these
elements is too high, they will place
human health at risk.
By nature TCM is very complex and
contains many elements which have
different concentration levels – high
and low. In the past, they have been
analyzed by a combination of ASS, AFS
and AES technology. These procedures
are time-consuming and costly3. ICP-
MS is fast becoming the technique of
choice for the determination of ele-
ments in a wide range of samples.
Additionally, it has the widest linear
range (nine orders of magnitude, from
1 ng·mL-1 to 1000 µg·mL-1), the high-
est sensitivity and lowest detection
limit for metals, as well as the ability
to rapidly and accurately measure mul-
tiple elements simultaneously. The lat-
est revision of the Pharmacopoeia of
the People's Republic of China
(Chinese Pharmacopoeia 2005) now
includes ICP-MS as one of the stan-
dard methods for the determination of
heavy metals in herbal medicines4.
In this study, microwave digestion is
used for sample preparation. High tem-
perature and a closed system assure
fast and complete digestion and avoid
Nebulizer Babington nebulizer
Spray chamber Quartz scott-type, peltier-thermostatted to 2 ± 0.1 °C
Torch Quartz, 2.5 mm ID
Interface Ni cone
Table 1
Agilent 7500 c ICP-MS Instrument DetailsZ.
62
the loss of those elements which can
easily evaporate, such as arsenic and
mercury. Thirteen toxic elements
including Be, Cr, Mn, Ni, Cu, Zn, As,
Ag, Cd, Ba, Hg, Tl and Pb were ana-
lyzed using the microwave digestion-
ICP-MS method. The method is validat-
ed by using two certified reference
materials: shrub leaves and tea leaves.
The results obtained agree with the
certified values. The recovery experi-
ment of standard addition was applied
in the study to evaluate the proposed
method. The recovery percentages are
between 90.3 -109.7 %.
Experimental
Instrumentation
An Agilent 7500 Series c ICP-MS was
used in this study (see table 1 for the
instrument details). The samples were
prepared by microwave digestion with
a CEM MARS5 microwave digestion
system (CEM Co. Ltd, USA), including
microwave oven, PTFE-TFE high pres-
sure vessel and fixed tray. Deionized
water (Milli-Q ultra-pure water system)
was used throughout.
Reagents
Nitric acid (HNO3), trace metal grade
(Merck); Hydrogen peroxide (H2O2),
MOS grade;standard stock solution: 10
µg ng·mL-1 mixed standard solution
including Be, Cr, Mn, Ni, Cu, Zn, As,
Ag, Cd, Ba, Hg, Tl and Pb. Calibration
standards were prepared by using
applicable dilutions of standard stock
solution with 5 % HNO3; Hg stock
solution: 1000 µg·mL-1 (National
Analytical Center for Iron & Steel,
China). Calibration standards were pre-
pared by using applicable dilutions of
standard stock solution with 5 %
HNO3; Internal standard solution: 1.0
µg/mL. Diluted from 10 µg·mL-1 Li, Sc,
Ge, Y, Tb and Bi mixed standard solu-
tion. (Agilent part number 5183-4680)
with 5 % HNO3; Tune solution:
10 ng·mL-1 7Li, 59Co, 89Y, 140Ce and
205Tl mixed standard solution (2 %
HNO3) (Agilent part no.5184-3566);
Deionized water (18.2 M_) produced
with the Milli-Q® ultrapure water
purification system (Milllipore Corp.)
was used in all standard solution and
sample preparations. Certified refer-
ence materials: GBW07602 (shrub
leaves) and GBW08513 (tea leaves).
Sample pretreatment
Exactly 0.5000 g of sample were
weighed and placed into the PTFE ves-
sel; 6 µL HNO3 and 1 µL H2O2 were
added. Because of the large organic
content of TCM and the large amount of
sample, pre-digestion is recommended.
To prevent sample loss due to an explo-
sion caused by gas and pressure build-
up produced in the digestion process,
Stage Power
Max / W %
1 1200 100
2 1200 100
the sample was dissolved in the solution meters were automatically optimized
and placed in the micro-wave oven. First, by the instrument. All of the specifica-
the temperature was raised and set at tions meet the installation require-
120 °C for approximately 5 minutes, then ments including sensitivity, back-
the sample was cooled completely and ground, oxide, doubly charge, stability,
the pressure was allowed to drop to nor- etc. The parameters are listed in table 4.
mal. Subsequently, the digestion pro-
gram specified in table 2 was followed. Calibration curves
After digestion the vessel was cooled to The mixed stock solution and Hg stock
room temperature and the contents solution were diluted to
poured into a 50-mL PET bottle. The ves- 0.1, 0.5, 2,10 ng·mL-1 using 5 % HNO3
sel and cap were washed with a small respectively. The blank solution is 5 %
amount of distilled water several times HNO3. The blank and calibration solu-
and this mixture was added to the PET tions were measured under optimized
bottle. Water was added to bring the conditions. The calibration curve was
exact weight to 50 g. The procedure for automatically plotted by the instrument.
reagent blanks is identical to that for Linear correlation coefficients (r) in all
test samples and is carried out concur- calibration curves were
rently without a sample. better than 0.9999.
ICP-MS parameters and target ele-
ment isotopes
The sample solution was analyzed
under the optimized condition. The tar-
get element isotopes are summarized
in table 3; 72Ge was selected as the
Parameter Set value
internal standard element for Be, Cr,
Mn, Ni, Cu, Zn and As, 115In was Power 1350 W
Flow rate of plasma gas 15.0 L·min-1
selected as the internal standard ele-
Flow rate of auxiliary gas 1.0 L·min-1
ment for Ag, Cd and Ba; and 209Bi was Flow rate of carrier gas 1.12 L·min-1
selected as the internal standard ele- Sampling rate 0.4 mL·min-1
ment for Hg, Tl and Pb. ICP-MS para- Sampling depth 7 mm
Orifice of sampling cone 1.0 mm
Orifice of skimmer cone 0.4 mm
Data acquisition mode Quantitative
Ramp /min T /°C Hold/min
analysis
Integration time 0.3 s /isotope
5:00 120 5:00 Cerium oxide/Cerium <0.5%
6:00 180 20:002 Doubly charge <2%

Table 2 Table 4
Microwave digestion program. ICP-MS operating parameters.

Element Be Cr Mn Ni Cu Zn As Ag Cd Ba Hg Tl Pb
Isotope 9 52 55 60 63 66 75 107 114 137 202 205 208
Table 3
Target element isotopes.

63 5989-5591EN
Results and discussion According to the results in Table 8, not tions also require regulations to man-
Method detection limit all TCMs meet the minimum legal age and control the production and
The blank sample was analyzed 11 requirements. The concentrations of sale of TCMs. The study uses
times under optimized conditions. The some elements highly exceed the microwave digestion for sample prepa-
method detection limits (MDL) for each allowable limit. However, in most ration, With optimized working para-
element were calculated (table 5). TCMs, the amount of all toxic ele- meters all elements in different medi-
ments is low. cines are analyzed simultaneously
with an Agilent 7500 Series c ICP-MS.
Element MDL This method is validated with certified
Be 0.1 Conclusion reference materials by conducting
Cr 7.2 The evidence suggests that the cur- recovery experiments using known
Mn 1.1 rent TCM market is complex and strict standard additions. It offers many
Ni 1.9 management and quality controls are
Cu 2.5
advantages over other alternative
Zn 3.4
needed to ensure that the maximum techniques, such as precision and
As 3.5 allowable limits of toxic elements set accuracy, simplicity, rapidity, low limits
Ag 0.6 by regulations are not exceeded. of detection and multiple element
Cd 0.8 National and local drug administra- determination.
Ba 2.4
Hg 1.1
Tl 0.1 Element GBW07602 (shrub leaves) GBW08513 (tea leaves)
Pb 0.9 Certified value Found value Certified value Found value
ug·gPT ug·gPT ug·gPT ug·gPT
Table 5
Method detection limits (MDL) (Unit: ng·gPT). Be 0.056 ± 0.014 0.046 / 0.088
Cr 2.3 ± 0.3 2.0 / 2.2
Determination of standard materials Mn 58 ± 6 54 2170 ± 110 2074
Ni 1.7 ± 0.4 1.7 5.09 ± 0.76 4.94
To evaluate reliability and accuracy,
Cu 5.2 ± 0.5 4.6 8.96 ± 0.59 8.23
the microwave-ICP-MS method was Zn 20.6 ± 2.2 19.8 22.6 ± 1.5 22.4
applied to the determination of two As 0.95 ± 0.12 0.79 0.180 ± 0.049 0.134
certified reference materials: Ag 0.027 ± 0.006 0.024 / 0.022
GBW07602 (shrub leaves) and Cd 0.14 ± 0.06 0.18 0.023 ± 0.004 0.022
Ba 19 ± 3 17 120 ± 10 112
GBW08513 (tea leaves). The results
Hg / 42.5 0.017 0.018
are in strong agreement with the certi- Tl / 0.015 / 0.016
fied values, a comparison is shown in Pb 7.1 + 1.1 6.1 1.00 ± 0.05 0.91
table 6.
Table 6
Comparison of found value and certified value.
Recovery
The percent recovery for each
element was determined using Element Spiked value Found value Recovery percentage
(ng·mLPT) (ng·mLPT) (%)
the standard addition method to evalu-
ate the reliability and accuracy of the Be 10 9.37 93.7
method. Percent recoveries of all ele- Cr 10 10.63 106.3
Mn 10 10.97 109.7
ments were between 90.3 %-109.7 %. Ni 10 10.24 102.4
The results were considered satisfac- Cu 10 10.68 106.8
tory (table 7). Zn 10 10.36 103.6
As 10 10.89 108.9
Sample analysis Ag 10 9.20 92.0
Cd 10 9.25 92.5
This study analyzed 13 toxic elements Ba 10 9.16 91.6
in seven TCMs purchased on the mar- Hg 5 4.87 97.3
ket. Each sample was analyzed eight Tl 10 9.58 95.8
times and the accuracy (RSD %) varied Pb 10 9.03 90.3
between 0.3 % - 6.8 % (table 8). Table 7
Results of the recovery experiments.

5989-5591EN 64
Name of sample Gegen Zhike Guifudi Huangli Jinsang Naodes Shugan
Soup San huang Pill anshangqing Pill sanjie Pill heng Pill Pill
Element
Be Found
value 0.020 0.009 0.032 0.034 0.060 0.037 0.013
RSD% 5.2 2.6 1.0 5.1 2.0 1.4 5.2

Cr Found
value 0.34 0.46 1.40 1.58 4.89 1.74 4.76
RSD% 2.0 2.0 1.9 2.6 2.3 2.5 0.8

Mn Found
value 34.77 17.80 54.78 30.37 87.23 23.96 124.4
RSD% 0.7 2.4 1.1 2.20 1.21 2.62 3.10

Ni Found
value 1.09 1.17 1.24 1.64 3.09 1.50 3.15
RSD% 0.6 0.9 1.4 0.8 2.2 1.8 1.8

Cu Found
value 1.49 1.68 4.08 8.85 15.23 3.56 5.38
RSD% 0.6 0.6 0.6 2.5 2.3 2.38 0.46

Zn Found
value 5.66 4.82 21.27 18.57 37.07 15.33 22.59
RSD% 0.6 1.1 2.4 1.3 1.5 1.7 1.7

As Found
value N.D. 0.26 0.46 0.79 0.89 0.56 29.14
RSD% 1.6 1.4 3.9 3.5 5.7 3.3 1.5

Ag Found
value 0.0009 0.0008 0.005 0.007 0.079 0.006 0.007
RSD% 5.3 6.3 2.1 5.8 3.5 5.6 4.3

Cd Found
value 0.022 0.042 0.10 0.11 0.15 0.077 0.082
RSD% 5.5 1.5 1.2 2.7 1.9 3.6 2.0

Ba Found
value 5.86 8.94 15.32 44.76 167.8 136.7 12.98
RSD% 1.1 1.7 1.4 2.1 2.8 0.3 0.9

Hg Found
value 0.003 0.002 0.85 0.088 0.027 0.076 5.05
RSD% 6.8 5.8 3.3 5.5 5.0 3.5 0.6

Tl Found
value 0.008 0.027 0.025 0.020 0.033 0.036 0.014
RSD% 1.3 2.4 2.7 3.6 3.2 1.1 4.1

Pb Found
value 0.15 0.16 0.93 1.54 2.14 0.68 1.69
RSD% 0.7 1.1 2.6 2.1 1.7 1.5 1.3
Table 8
Analytical results of TCMs (n = 8). Unit: µg gPT.

65 5989-5591EN
References

1.
Wang Xiao-ru, “Application of
Inductively Coupled Plasma Mass
Spectrometry.” Chemical Industry
Press. Bei Jing., 102, 2005.

2.
Dong Hong-bo, Han Li-qing, Dong
Shun-fu, “Instrumentation & Analysis
Monitoring”
(3): 29~30, 2002.

3.
Yu Zi-li, Guang Lei, “Analytical
Technique of Metal Ion”. Chemical
Industry Press. Bei Jing, 225~226,
2004.

4.
Determination of Lead (Pb), Cadmium
(Cd), Arsenic (As), Mercury (Hg) and
Copper (Cu), Ch. P 2005 (English ver-
sion), Appendix IX B, p. 54-56

Hua Zhang is postgraduate student in

the chemistry department of the
Capital Normal University, Yan-zhi Shi
and Ying-feng Wang are the professors
at the Analytical Center of the Capital
Normal University, Beijing, China. Yu-
hong Chen is Application Engineer at
Agilent Technologies Co. Ltd, China.

5989-5591EN 66
 Fast determination of five toxic elements
in Traditional Chinese Medicine (TCM)
by ICP-MS

Application Note Binfeng Xia

Zhuqing Lu
XinMei Wang
Ke Wang
Shen Ji

Abstract

Five toxic elements in 10 different types of Traditional Chinese Medicines (TCM)

Agilent Equipment were analyzed by Inductively Coupled Plasma Mass Spectrometry (ICP-MS).
Agilent 7500 ICP-MS
To do this, a sample digestion method and standard operation procedures were
Application Areas developed. Reproducibility and recovery were tested for method validation.
Pharmaceutical
The Agilent ICP-MS method is proven to be highly sensitive, and its fast acquisition
time makes it highly suitable to analzye trace toxic elements in TCMs.

67 5989-7590EN
Introduction Experimental curves
Traditional Chinese Medicine (TCM) Instrumentation and reagents The TCMs were dried at 60 °C for four
herbs and their manufactured prod- In this study a 7500c ICP-MS instru- hours, and ground into powder. 0.5 g
ucts have been used for thousands of ment was used,which uses an powder were weighed precisely into a
years for prevention and treatment of Octopole Reaction System. Sample microwave digestion tube (50 mL, PFA),
disease in China. TCM materials are digestion was done by microwave and then 10 mL 65 % HNO3 was added.
made up of plants, animals and miner- digestion. Concentrated HNO3 (68% The powdered TCM and HNO3 were
als with the majority being various w/v, GR, Merck) was used for sample placed in a microwave digestion sys-
parts of medical plants. The contents digestion and cleaning of the digestion tem and subjected to a standard pro-
of heavy metals in TCMs have been tubes. Deionized water was used gram of heating. Following digestion
commonly studied from the viewpoints throughout. the solubilized TCM samples were
of toxicity and bio-availability. The transferred into a PTFE volumetric flask
roots or leaves of the plant can absorb Operation parameters and the digestion tube was washed
heavy metals such as Pb, Cd, Hg and The ICP-MS operation parameters are three times using pure water.
As from the atmosphere, water and listed in table 2.
soil, and these can finally accumulate
at a particular position in the plant Sample digestion and calibration
through the cytosol. Different regula-
Regulations (µg/kg) Pb Cd Hg As
tions propose to limit the content of
Proposed regulation on “TCM Quality Standard”
heavy metals in herbs (table 1). Chinese Pharmacopeia 5000 500 200 200
The most stringent of these regula- Chinese regulations on imported herb supplements 5000 300 200 2000
tions is that from the Food and Drug French regulations on imported herb supplements 5000 200 100 5000
Administration (FDA). Therefore, it US FDA regulations on the medicines and functional food 1000 300 26 20
is important to determine these ele- South-East Asia regulations on imported herb supplements 20000 N/A 500 5000
ments as accurately and as unam Table 1
biguously as possible. Regulations on heavy metals in TCM or herb supplements.

Inductively coupled plasma mass

Working parameters
spectrometry (ICP-MS) has quickly
RF power 1350 W
become the technique of choice for
Nebulizer PFA 200 µL/min nebulizer
the determination of elements in a Spray Chamber Scott double pass 2±0.1 ºC
wide range of samples. It has been Data Acquisition Mode Spectrum Analysis Mode and Full Quant Mode
used widely for measurement of trace Sampling depth 7.0 mm
and ultra-trace elements in the envi- Carrier gas flow rate 1.18 L/min
Total Acquisition Time 90 s
ronmental and biological materials.
The advantages of choosing ICP-MS Table 2
Instrumental parameters of Agilent ICP-MS.
for measurement include rapid and
simultaneous multi-elemental
determinations, low spectral interfer-
ences, excellent detection limits and
STD As Pb Cd Cu Hg
the ability to analyze isotopic ratios.
Blank 0 0 0 0 0
1 1.0 1.0 0.5 50 0.2
In this study, the ICP-MS technique 2 5.0 5.0 2.5 100 0.5
was applied to measure five toxic ele- 3 10.0 10.0 5 200 1
ments in TCM samples from different 4 20.0 20.0 10 500 2
areas in China. 5 5
Table 3
Calibration curve range (µg/L).

5989-7590EN 68
200 µL of a 1 µg/mL Au solution was
also added into the digested sample to
stabilize Hg. The solution was finally
made up to 50 mL. The system blank
solution was made by the same proce-
dure except that no TCM was added.
The calibration curve standards were
also prepared from 10% HNO3.
The internal standard (1:10 diluted
Agilent Internal Standard Mix, part
number 5183-4680) was automatically
added on-line to samples during analy-
sis. Ge (72) was used as internal stan-
dard reference element of Cu and As.
In (115) was used as internal standard
reference element of Cd (114). Bi
(209) was used as internal standard
reference element of Pb and Hg.
The samples were run according to a
modified method based on USEPA
200.8
Results and discussion
Calibration linear range
All five element’s calibration curves
show good linearity, R2 values are
between 0.9990 and 0.9999.
The results prove that the ICP-MS Sample spike recoveries
method and the sampling and diges- 0.25 g powder was taken from each of
tion method are very good for trace the ten TCMs, six times parallel sam-
toxic elements analysis, even at low pling was done for each material.
concentration levels. Using the measured concentrations of
the metals in the TCM guidelines, six
Blank spike recoveries undigested samples were spiked using
Cd, As, Pb, Cu, Hg were spiked into the standard stock solutions. All spikes
digestion tube, followed by microwave were done in pairs and the levels cho-
digestion and then made up to a blank sen for spiking were 40 %, 50 % and 60
spike recovery solution with concen- % of the levels found in the originally
trations similar to TCM samples. The analysed TCMs. Then the samples
spike recoveries test results are were digested and measured by ICP-
shown in table 5. A total of eight blank MS. The results for the Xiyangshen
spike recoveries solutions were made drug are summarised in Table 6. Even
for statistical purposes. though the spike levels were low (with
The results again illustrate that respect to the elements in the sample
microwave digestion is a very good itself) the recoveries are excellent and
method for sample digestion without highlight that the digestion method is
loss of target elements effective, without loss of material.
during digestion.
No Cu As Cd Hg Pb
1 5962 33.8 94.9 127 68.5
2 6394 29.7 91.4 141 66
3 6577 30.7 93.6 157 68.8
4 6312 32.1 92.2 137 66.9
5 7000 28 100.9 169 76.8
AVG 6449 30.9 94.6 146 69.4
RSD(%) 5.9 7.2 4.0 11.4 6.2
Table 4
Reproducibilities for Xiyangshen samples (µg/kg).

Reproducibility of the sample

Element Cu As Cd Hg Pb
preparation
Avg(%) 98.8 106.0 102.2 110.0 100.7
Ten TCMs were investigated: RSD 1.8 1.0 1.1 1.3 1.9
Xiyangshen, Danshen, Huangqi,
Table 5
Ganchao, Huangbo, Baishao, Chenpi, Blank spike recoveries.
Lingzhi, Jinyinhua, and Huanxieye
drugs. All samples were ground into No Cu As Cd Hg Pb
powder and five repeated samples 1 112.3 111.0 97.3 88.3 130.6
(0.5 g powder for every sampling) were 2 120.8 90.6 110.9 101.2 122.7
taken from 10 batches. Powders were 3 89.6 94.7 111.0 78.4 120.7
digested by microwave digestion. 4 110.3 88.5 108.1 93.2 134.4
5 91.6 107.8 111.7 80.7 108.4
Reproducibility was good for all sam-
6 117.4 89.2 101.8 97.6 111.3
ples. The Xiyangshen medicine is Avg 107.0 97.0 106.8 89.9 121.4
taken here as an example. RSD 12.4 10.2 5.5 10.2 8.5
Table 6
Spike recoveries for Xiyangshen after microwave digestion and ICP-MS analysis.

69 5989-7590EN
Results and discussion TCMs Cu As Cd Hg Pb
Baishao 5403 376 72 1775 280
Average Huangqi 7692 313 27 4805 383
Ten types of TCMs were selected as Lingzhi 8162 224 65 247 154
Jinyinhua 11698 3105 113 127 1885
targets, and ten samples were collect-
Ganchao 8968 536 12 89 278
ed from different local areas for each Danshen 10404 594 69 22 948
type. The samples were analyzed by Xiyangshen 5281 82 104 25 131
ICP-MS, the average results are listed Huanxieye 5198 276 12 159 249
in table 7. While the data in table 7 Huangbo 2457 323 14 77 769
Chenpi 2842 328 32 166 1298
suggests that all measurements were
below the regulated levels for heavy Table 7
metals, please note that these are Average results (µg /kg).
averages; some samples were above
Elem Range (µg/kg) 2000 3000 5000 10000 20000
the regulated limits. Interestingly, even
Cu 1289~19038 97 83 68 20 0
for the same TCM type, the toxic ele-
ment concentrations varied from very Table 8
Cu in 100 TCMs (sample amounts).
low trace levels to highly contaminat-
ed, when the samples are taken from Elem Range (µg/kg) 100 500 1000 2000 5000
different sources. For example, Hg con- Pb 69~3909 92 34 24 6 0
centrations in 10 Baishao samples var-
Table 9
ied from 5 µg/kg to 14 µg/g, with Pb in 100 TCMs (sample amounts).
almost 3000 times difference.
Elem Range (µg/kg) 100 200 300 500 1000
Statistical results for 100 TCM sam- As 12~10348 81 62 38 23 14
ples, summarized in tables 8 to 10 Cd 2.4~212 21 2 0 0 0
These results show how many samples Hg 0.4~36009 34 18 17 11 5
out of 100 have been detected to show Table 10
high toxic elements content. As, Cd, Hg in 100 TCMs (sample amounts).

Conclusion digestion, this has reduced the matrix

All samples contained toxic elements
below proposed limits, but element
analysis by ICP-MS of ten different
Traditional Chinese Medicines (see
table 1) from different geographical
locations showed that variability for
the amount of each element was very
high. This might correlate with the dif-
fering contamination level and different
contamination types due to soil and
water contamination in different areas.
Taking good care when growing plants,
such as in Good Agriculture Practice
(GAP) of TCMs will be crucial in the
future to maintain consistent quality.
The Agilent ICP-MS has been proven to
be an ideal analytical method for deter-
mination of toxic elements in TCMs,
since this system can determine trace
level impurities very efficiently. In com-
bination with optimized microwave
interferences significantly. Additionally,
ICP-MS enables very fast measurement
of a full set of elements, saving time for
other laboratories tasks.
References
1.
“Determination of Toxic Elements in
Traditional Chinese Medicine Using
Inductively Coupled Plasma Mass
Spectrometry”; Agilent Publication Binfeng Xia, Zhuqing Lu, XinMei Wang,
Number 5989-5591EN, 2006. Ke Wang, Shen Ji,
Shanghai Institute for Food and Drug
2. Control (China).
This Paper is an extract of a Chinese
publication: Modern Instrumentation,
first edition, 2004.

5989-7590EN 70
Quality Control for
Residual Solvents

71
72
 Better precision, sensitivity, and higher
sample throughput for the analysis of
residual solvents in pharmaceuticals
Using the Agilent 7890A GC system with G1888 headspace
sampler in drug quality control

Application Note Albert E. Gudat

Roger L. Firor
Ute Bober

5 10 15 20 25

Abstract
Laboratories using headspace GC for the analysis of pharmaceutical impurities
face a number of instrument-related issues:
• The area precision in headspace analysis is impacted by atmospheric pressure
changes.
• The sensitivity is poor for some low-concentration analytes.
• The presence of high-boiling impurities noticeably extends the analysis time
per sample and may damage the analytical column.
The determination of residual solvents is the most common application for head-
space GC in pharmaceutical quality control. In this Application Note an estab-
lished Agilent 6890 GC method for residual solvents analysis is transferred to the
new Agilent 7890A GC without any major changes. The results on both systems
are compared. Overall, the Agilent 7890A GC system delivers at least the same or
Agilent Equipment better performance than the Agilent 6890N GC system. Further it is demonstrated
7890A GC system,
G1888 headspace sampler that the new technology implemented in the Agilent 7890A GC can significantly
improve area and retention time repeatability and sensitivity. It can drastically
Application Area
Pharmaceutical quality control reduce the overall analysis time, hence increasing sample throughput and thereby
Impurity analysis productivity.

73 5989-6023EN
Introduction
Because many solvents pose a major
risk to human health, national and
international regulatory bodies such
as the United States Food and Drug
Administration (U.S. FDA), the United
States Pharmacopoeia (USP), the
European Pharma-copoeia (EP), and
the International Conference on
Harmonization (ICH) require analysis
for residual solvents in pharmaceutical
drug substances, excipients and final
products. Solvents are divided into
three classes on the basis of possible
risk. Class 1 solvents should be
avoided. Class 2 solvents should be
limited. Class 3 solvents are consid-
ered to have low toxic
risk. The ongoing trend toward lower
contaminant levels designated as safe
requires more sensitive and accurate
methods of analysis. New USP <467>
regulations for residual solvents begin
in July 2007. The goal of this initiative
is the final alignment with the ICH
Q3C(R3) guideline, which has also
been adopted by the EP.
The analysis for residual solvents in
pharmaceutical products and for sol-
vents considered extracta-
bles/leachables in pharmaceutical
packaging materials is typically done
using headspace (HS) GC with a
flame-ionization detector (FID) or, for
identification and confirmation, with
mass-selective detection (MSD). This
has been covered in references 1-3.
Residual solvents is the most common
application for headspace GC in phar-
maceutical quality control.
Laboratories currently using HS GC
face a number of instrument-related
issues:
5989-6023EN
• The area precision in HS analysis can
be compromised primarily due to
atmospheric pressure variations
influencing the amount of analytes
injected from the sampling loop in
the HS gas sampling valve (GSV).
• The sensitivity is poor for some low-
concentration analytes e.g., benzene.
• Sample turn-around time is excessive
and caused by late-eluting impurities
and high-boiling solvents/diluents
e.g., 1,3-dimethyl-2-imidazolidinone
(DMI) with boiling point of 225 oC.
Further, when the need for new analyt-
ical equipment arises, the first ques-
tion is whether an established validat-
ed method can be easily transferred to
the next generation of instruments
without any additional method devel-
opment and resulting in no or minimal
revalidation effort. The purpose of this
study was to at the least demonstrate
equivalence of the HS/7890A GC/FID
and HS/6890N GC/FID systems when
both are operated without pressure
regulation of the sampling loop con-
tent of the HS GSV. But more impor-
tantly, to also show how new capillary
flow technology, fifth generation pneu-
matics, and state-of-art electronics
implemented in the 7890A GC have
effectively addressed the above issues
with significant improvements in area
and retention time precision, sensitivi-
ty and productivity with increased
sample throughput for residual solvent
analysis.
Note: A list of acronyms and short-
hand terms used in the text, figures
and descriptions of experiments and
calculation formulas are included in
the appendix.
74
Experimental
Both 6890N and 7890A GC were
equipped with an Agilent headspace
sampler, volatiles interface (VI) and FID.
Table 1 gives the experimental condi-
tions used with the HS/VI/6890N/FID
and HS/VI/7890A/FID systems. The
7890A system is operated with and
without backpressue regulation on the
HS sampling loop, whereas the 6890N
does not have backpressure regula-
tion. For the purpose of calculating the
repeatability expressed as %RSD val-
ues in area and retention time and
determining statistical Method
Detection Limit (MDL) for each ana-
lyte, 20 identical samples were pre-
pared. A standard solution in water
was first prepared in a 100 mL volu-
metric flask by adding Restek class 1
and class 2 standards with an
Eppendorf pipette. 5 mL of the aque-
ous standard was subsequently trans-
ferred quickly to 10 mL headspace
vials containing 3 g sodium sulfate and
immediately sealed with Teflon-seal
caps. Each vial was then vortex mixed
for half a minute. Ten of these samples
were subsequently used with the
6890N system and ten with the 7890A
system, both systems being operated
without pressure control on the HS
sampling loop. At the end of the HS
equilibration, the HS vials were pres-
surized to 14.000 psi by an auxiliary
(AUX) Electronic Pneumatic Control
(EPC) module and injected in either
the 6890N GC or the 7890A GC system.
The same DB 624 column was used in
the two GCs for the sequence of injec-
tions in order to eliminate the influ-
ence of batch-to-batch variations in
column quality.
Another set of 20 samples was pre-
pared in the same way as described
above for use with the 7890A system 6890N or 7890A GC G1888A Headspace Sampler
but now implementing backpressure
Injection port Volatiles interface Loop size 1 mL
regulation on the HS sampling loop. Temperature 160 °C Vial pressure 14.0 psig
Figures 1, 2 and 3 show diagrams of Split ratio 2:1 Headspace oven 85 °C
the 7890A system where new pneu- Carrier gas Helium Loop temp 100 °C
matic features of pressure regulation Carrier flow 9 mL/min Transfer line temp 120 °C
GC oven program Equilibration time 30 min, low shake
of the HS sampling loop, HS vial pres-
Initial temperature 35 °C GC cycle time 50 min
surization and backflush can be Initial time 20 min Pressurization 0.15 min
applied. At the end of the HS equilibra- Rate 25 °C/min Vent (loop fill) 0.5 min
tion, the HS vials were pressurized to Final temp 250 °C Inject 0.5 min
20.000 psi by an AUX EPC channel and Final time 15 min
Column: 30 m x 0.45 mm x 2.55 µm DB-624
the loop was regulated at 5.000 psi
Agilent part number 124-1334
with the backpressure regulator chan-
nel of the Pneumatic Control Module Standards
(PCM). ICH class 1 and 2 Restek #36228 (Class 1)
#36229 (class 2A)
#36230 (class 2B)
Table 1
Instrument conditions for residual solvents analysis.

Forward flow

H Electronic BP control Restrictor

S
Vial pressure
v
e Dual
n HS TR line AUX mode
t module PCM FID
VI 7890 GC
G1888
Headspace sampler Effluent
splitter

30 m x 0.45 mm id x 2.55 μm DB624

Figure 1
Block diagram of the 7890A GC configuration with backflush capability used in the backflush
experiments.

PS Split vent
Carrier AUX EPC
Vent trap
Vial pressure HS Vent
from Aux 25 psi
module 4 MFS He
3 Flow source
5
6 2 restrictor PS S/SL inlet FID
1
To volatiles interface
0.1 psi Splitter
HS vial DB624
Pressure difference: (HS vial – BPR) = 10 to 15 PSI Backflush of late eluting solvent ( such as DMA or DMI)

Figure 2 Figure 3
Headspace (HS) sampling scheme with backpressure regulation (BPR). Schematic diagram of the reversed column flow used for backflushing of
late eluting solvents.

75 5989-6023EN
 7890A GC with BPR* 7890A GC at Atm P (no BPR) 6890N GC at Atm P ( HS valve)
Excipient
limit Repeatability Excipient Repeatability Excipient Repeatability Excipient
ICH concentration [ %RSD] N=8 MDL# [ %RSD] N=8 MDL # [ %RSD] N=8 MDL #
Residual solvents Class [ppm] tR Area [ppm] tR Area [ppm]
[ppm] tR Area
benzene 1 2 0.014 2.43 0.1 0.012 5.62 0.2 0.010 9.52 0.2
1,2-dichloroethane 1 5 0.005 4.47 0.7 0.02 8.03 0.5 0.016 8.63 0.5
1,1-dichloroethene 1 8 0.013 3.24 0.8 0.011 16.63 3.3 0.022 9.82 1.1
methylene chloride 2 600 0.009 2.85 54.9 0.016 7.15 61.4 0.018 8.20 62.7
hexane 2 290 0.014 4.18 23.1 0.027 7.15 33.2 0.020 10.61 39.1
cyclohexane 2 3880 0.042 3.59 341.0 0.012 4.29 299.9 0.018 9.79 501.1
trichloroethylene 2 80 0.012 2.69 5.8 0.016 5.29 7.9 0.013 7.91 7.9
toluene 2 890 0.025 2.11 46.3 0.024 5.41 85.9 0.031 7.90 90.3
ethylbenzene 2 369 0.002 2.27 24.4 0.002 4.90 35.3 0.003 7.40 35.5
ortho xylene 2 195 0.001 1.86 9.8 0.001 5.12 19.0 0.002 7.00 18.4
Average for 29 solvents 0.013 2.83 0.017 8.77 0.021 9.34
* Backpressure regulation (HS-valve outlet pressure is regulated) # Method detection limit
Table 2
Retention time and area repeatability and calculated MDLs of representative residual solvents for the 7890A and 6890N HS/VI/GC/FID systems.

Results from the above experiments faced to the headspace transfer line. !os;+
§sCm?#C#)
s9;ss)
"?
are summarized in table 2. From the 29 The VI configuration is anticipated to
solvents some representative class 1 have a limited backflush flow rate with A typical chromatogram of residual
and 2 solvents were selected and the the 0.45 mm diameter column, a limi- solvents is shown in figure 4 and
results summarized in this table. tation not observed with the S/SL repeatability data for area and reten-
Further, the average values for all 29 inlet. The VI configuration was not tion time of an early- and late-eluting
solvents are shown. A series of experi- tested for backflush operation. peak are shown in figure 5. An
ments was also performed to demon- improvement in area precision by a
strate potential sensitivity gains real-
ized from pressurizing the HS loop. 22, 23, 24, 25, 26,27, 28
1 4 5 6 7,8 10, 11, 12 17 20
29
This time, however, instead of prepar-
ing 5 mL aqueous standards in HS vials
containing the 3 g of sodium sulfate, a
Restek class 2B standard was used as
is. A 5 µL capillary tube was filled by 21
capillary flow action with the standard;
the outside of the tube was carefully
wiped with tissue paper, quickly trans- 16
ferred to an empty 10 mL headspace 3
vial and immediately capped. This pro-
cedure ensured accurate and repro-
ducible sample preparation by eliminat- 2
ing user bias to the extent possible in
preparing identical samples. Results 9 13,14 19
15 18
from this series of experiments are
shown in Figure 7. The error bars in
the figure represent a 95 % confidence
5 10 15 20 25 30
level (±2 times standard deviation or
sigma). 1 Methanol 11 Cyclohexane (co-elute with 10 & 12)
2 1,1 Dichloroethylene 12 1,1,1 Trichloroethane (co-elute with 10 & 11)
3 Acetonitrile 13 Benzene
Finally, the system was reconfigured to 4 Methylene chloride 14 1,2 dimethoxyethane 21 2 hexanone
facilitate a column backflush to quickly 5 Trans 1,2 dichloroethene 15 1,2 dichloroethane 22 Chlorobenzene
remove late eluting impurities or high 6 Hexane 16 Trichloroethylene 23 Ethylbenzene
7 Cis 1,2 dichloroethene 17 Methyl cyclohexane 24 DMF
boiling solvent or diluents. This config-
8 Nitrobenzene (co-elute with 7) 18 1,4 dioxane 25 M-xylene
uration is shown in figure 3. Rather 9 Trichloromethane 19 Pyridine 26 P-xylene
than using the volatiles inlet (VI), a 10 Carbon tetrachloride 20 Toluene 27 O-xylene
split/splitless (S/SL) inlet was inter- 28 N,N dimethylacetamide
29 Tetralin
Figure 4
Gas chromatogram of class 1 and 2 residual solvents.

5989-6023EN 76
factor of 3 was typically observed, as
in this example for 1,1-dichloroethyl- 1,1-dichloroethylene (8 ppm) o-xylene (195 ppm)
pA
pA
ene. However, in some cases up to a 5000
26 N=8 N=8

1,1-dichloroethylene
factor of 4 was determined, for exam-

o-xylene
24 4000
ple, for o-xylene. Overall, the perfor- 22 3000
mance of the 7890A system is better 20
18 2000
than the 6890N system. A summary of
16 1000
performance characteristics for some 14
representative class 1 and 2 analytes 0
1.94 1.96 1.98 2 2.02 min 23.2 23.22 23.24 23.26 23.28 min
is given in table 2.
Repeatability 6890 GC 7890 GC Repeatability 6890 GC 7890 GC
(%RSD) (%RSD)
Improving peak area precision
From the results presented in table 2 Retention time ± 0.022 % ±0.013 % Retention time ±0.002 % ±0.001 %
and figure 6 the following conclusions Area 9.8 % 3.2 % Area 7.0 % 1.9 %
can be drawn:
Improvement by a factor of ~ 3 Improvement by a factor of ~ 4
• Overall, the 7890A and 6890N show
the same area repeatability when no
backpressure regulation of the HS Figure 5
Examples of improved area and retention time precision by applying backpressure regulation
sampling loop is applied. Under the (backpressure regulation: 5.000 PSI, headspace vial pressure: 20.000 PSI).
same conditions the results don’t
exhibit significant differences. The
average area precision [%RSD] for 29
Backpressure Headspace vial Averaged area
residual solvents (shown on the bot-
regulation pressurization Repeatability (%RSD)
tom line of table 2) for both systems
is 9 %. No No 15.8% (N=8)
• The 7890A GC with backpressure
10.000 PSI 25.000 PSI 4.4% (N=7)
regulation of the HS sampling loop,
is at least 3 times better than the 5.000 PSI 20.000 PSI 2.8% (N=8)
7890A GC (or 6890N GC) operated
without pressure regulation of the HS Improvement by a factor of 5
sampling loop. For individual analytes
Figure 6
an improvement by a factor of 4 was Backpressure regulation – Effect of reducing atmospheric pressure variation at vent. Variability in
observed in some cases as shown in loading the gas sampling valve can occur based on atmospheric pressure differences.
figure 5 for o-xylene.
• The observed differences are even ferences. This could happen when • With backpressure regulation of the
more apparent when considering a running the same method in different HS sampling loop the averaged reten-
very turbulent day with large varia- geographic locations at different tion time repeatability measured on
tions in atmospheric pressure. This is altitudes or like in this case during a the 7890A GC was best. It improved
when a series of measurement were turbulent stormy day with large varia- by a factor of two relative to the
performed. The data presented in fig- tions in atmospheric pressure. With 6890N GC without any pressure regu-
ure 6 was obtained on such a stormy backpressure regulation the gas sam- lation.
day. Under extreme weather condi- pling valve can operate under a con-
tions the averaged area repeatability stant set of conditions and precision Optimizing sensitivity with backpres-
can increase to 16 % when no back- and sensitivity improve. sure regulation
pressure regulation is applied. Overall sensitivity could be doubled by
However, a method with optimized Improving retention time precision applying backpressure regulation com-
backpressure regulation fully com- Similar to the results obtained for peak pared to the 6890N or the 7890A GC
pensates for the atmospheric pres- area precision a positive impact on the without backpressure regulation.
sure instabilities and the measured repeatability of retention time was Figure 7 shows the area changes for
averaged repeatability returns to the observed: 1,4-dioxane with HS-vial pressurization
same value of 3 % that was obtained • In comparison to the 6890N GC the and HS sampling loop pressurization. The
under stable weather conditions in averaged retention time repeatability more we pressurize the vial, the more
the previous experiments. The reason for all 29 residual solvents measured we dilute the HS sample. This is clearly
for this is that variability in loading on the 7890A GC was generally better, shown when the x-axis is zero (where
the gas sampling valve can occur no matter whether backpressure regu-
based on atmospheric pressure dif- lation was applied or not.

77 5989-6023EN
the backpressure regulator is not used
and the loop is exposed to atmospher- 200
ic pressure). Pressurizing the HS vial
to 14, 35 and 60 psi gives the highest
peak area at 14 psi. When regulating 150
the pressure in the loop with the BPR,

1,4-dioxane peak area

we see an increasing area count that
reaches a maximum and then decreas-
es and eventually would give zero area 100
counts. Once the top of the curve is
reached, the depressurization of the
HS vial through the HS loop (the vent- 50
ing cycle) is opposed by the excessive P(vial) = 14 PSI
high backpressure and the sample P(vial) = 35 PSI
flow through loop will diminish and P(vial) = 60 PSI
may even reverse. After we reach the 0
0 5 10 15 20 25 30
top of the curve, we no longer trap a
BPR (PSI) on headspace gas sampling valve loop
representative sample in the loop. The
pressure difference (PHS-Vial – BPR)
Figure 7
should be 10 to 15 psi in order to col- Improving sensitivity with the 7890 GC – variation in peak area with headspace vial and sampling
lect and inject a proper HS sample. loop pressure.

Increase efficiency with backflush 1. 1,1 dichloroethene

2. Carbon tetrachloride
The backflush capability of the 7890A A 2,3 3. 1,1,1 Trichloroethane No backflush 7
GC allows to remove late eluting com- 4. Benzene
pounds by reversing the flow. The ben- 5. 1,2 dichloroethane
6. DMSO 6
efits are: 7. DMI (1,3-Dimethy-2-imidazolidinone)
1 4 5
• Shorter analysis time
• Extended capillary column life time. 5 10 15 20 25
B Backflush from 10 to 16 min at 250 °C oven temperature
Because this system has EPC, as soon
as the last analyte of interest has elut- Net gain of time: 14 min.
ed from the column, the AUX module
can be pressure-programmed to a 5 10 15 20 25 min
higher pressure (25 psi in this exam- C Blank run after backflush
ple) at the same time that the
split/splitless inlet is programmed to a
5 10 15 20 25 min
lower pressure (0.1 psi in this exam-
ple). Now the flow in the column is Figure 8
reversed, backflushing the remaining Example of how backflushing helps to decrease analysis time and increase workload efficiency
(chromatogram of a sample containing all ICH class 1 residual solvents): A. Initial situation
eluents out through the split vent of (no backflush) B. Reduced analysis time using backflush C. Blank run after backflush.
the inlet. A schematic overview of
the functionality is given in figure 3.
For the backflush experiment samples run was 14 minutes (figure 8B). The Equal in linearity
containing only class 1 residual sol- high-boiling compounds were success- To measure linearity five dilutions
vents were prepared in DMSO and fully removed as can be seen from the were prepared ranging from 1/10th to
DMI, both high boiling diluents (figure chromatogram of the blank run that two times the limit concentration. Based
8). A typical chromatogram of such a was executed afterwards (figure 8C). on the USP <467> method where 100
sample lasting more than 30 minutes In this example backflush improves mg of the excipient/drug product is
is shown in figure 8A. Since all the efficiency by almost doubling sample dissolved in 5 mL of water with 3
class 1 solvents elute in 10 minutes throughput. The fast oven cool-down grams of Na2SO4, the solution concen-
at 35 °C isothermal, a backflush was of the 7890A GC further contributes to tration in 5 mL of water was converted
initiated from 10 to 16 minutes with those time savings. to the concentration of the residual
an elevated column temperature of 250 solvent in the 100 mg of excipient with
°C. As a result the net gain of time per

5989-6023EN 78
the formula: ce [ppm] = 50 · cv. In the Methylene Chloride Benzene 1,4-Dioxane Chloroform Trichloroethylene
following text this concentration is
Linearity 0.99945 0.99859 0.99606 0.99362 0.99967
described as excipient equivalent con-
Slope 6.0935 50.2950 0.4214 2.7160 14.3362
centration ce while cv is the vial solu- Intercept 228.3704 4.6045 11.4008 5.2606 106.9926
tion concentration. LOD 10.2 0.02 2.5 0.25 0.07
LOQ 10.4 0.06 8.4 0.82 0.23
The linearity results for some repre- Table 3a
sentative residual solvents for the Linearity, LOD and LOQ results for the 7890A Headspace GC/FID system.
7890A GC are compared to similar
experiments for the 6890N GC1 system Methylene Chloride Benzene 1,4-Dioxane Chloroform Trichloroethylene
in figures 9 and 10, respectively. Linearity 0.9988 0.9995 0.9996 0.9991 0.9991
Overall, the 7890A and 6890N GC sys- Slope 252.82 2106.2 15.268 192.41 535.39
tems compare well. All calibration Intercept 19.987 0.0015 0.3239 0.1851 3.7229
curves are linear over a range from Table 3b
1/10th to 2 times the limit concentra- Linearity results for the 6890N Headspace GC/FID system.
tion. The signal-to-noise (S/N) data,
Methylene chloride Chloroform
limit of detection (LOD) and limit of
6000 LOD = 10.2 300 LOD = 0.25
quantitation (LOQ) indicate that the 2 2
LOQ = 10.4 LOQ = 0.82 1
Area

Area
4000 1 200
systems are similar in performance 3 3
2000 100
(only data for the 7890A is shown here 45 5
0 0 4
in tables 3a and 3b). Results from the 0 600 Amount [ppm] 0 60 Amount [ppm]
MDL calculations (see table 2 and the y = 6.09347347x - 228.37038 y = 2.71598018x + 5.2606117
appendix for the MDL equation) and a r2= 0.99945 r2= 0.99362
comparison of S/N ratios calculated Benzene Trichloroethylene
for samples at the limit concentration 200 LOD = 0.02 LOD = 0.07
150 2 2000 2

Area
indicate that the 7890A GC system is LOQ = 0.06 1 LOQ = 0.23
Area

100 1
at least two times better in sensitivity 50 3 1000
3
45
than the 6890N GC system. 0 0 45
0 2 Amount [ppm] 0 100 Amount [ppm]
y = 50.2949812x - 4.6044753 y = 14.3361657x - 106.99264
r2 = 0.99859 r2= 0.99967
Conclusion
Overall, it was demonstrated, that the 1,4-Dioxane LOD (ppm) where S/N = 3
7890A GC delivers better results than LOD = 2.5 LOQ (ppm) where S/N = 10
300 2
the 6890 GC. In summary: LOQ = 8.4 1 * Concentrations shown are
Area

• It was possible to directly transfer an
established method from the 6890 to
the 7890A GC without any method
development or altering the perfor-
mance.
• Without backpressure regulation the
7890A GC shows the same or better
performance.
• The backpressure regulation from the
7890A GC eliminates the influence of
atmospheric pressure variations.
• With optimized backpressure regula-
tion of the headspace sampling loop
of the 7890A GC area precision
[%RSD] could be improved by a fac-
tor of 3 to 5
• Under the same conditions retention
time stability increased to ±0.001 min.
• Sensitivity was doubled by pressuriz-
ing the headspace sampling loop of
the 7890A GC.
200
3 excipient equivalent concentration ce [ppm]
100
45
0
0 500 Amount [ppm]
y = 0.4214021x + 11.400823
r2 = 0.99606
Figure 9
Linearity plots for some residual solvents determined for the 7890A Headspace GC/FID system*
• The backflush capability of the 7890A where you can achieve superior
GC significantly reduces overall results for both identification of
analysis time (in the example by 50 %). unknowns and quantitation of target
• Both systems are equal in perfor- compounds.
mance when evaluating linearity
data.
• The experimental setup in this appli-
cation is suitable for the routine
analysis of residual solvents.
However, it does not provide any fur-
ther information when unknowns are
present. The solution is to couple the
GC to the Agilent 5975B Series MSD,

79 5989-6023EN
References Methylene chloride Chloroform
5000 400
1.
4000
Roger L. Firor, “The Determination of 300

Area

Area
3000
Residual Solvents in Pharma-ceuticals 2000 y = 252.82x + 19.987
200
y = 192.41x + 0.1851
Using the Agilent G1888 Network 1000 R 2= 0.9988
100 2
R = 0.9991
Headspace Sampler,” Agilent 0 0
0 5 10 15 20 0 0.5 1.0 1.5 2.0
Application Note, Publication Number μg/mL μg/mL
5989-1263EN, 2004.

140 Benzene Trichloroethylene

2. 1400
120 1200
Roger L. Firor and Albert E. Gudat, “The 100 1000

Area

Area
80 800
Determination of Residual Solvents in 60 y = 2106.2x - 0.0015 600 y = 535.39x + 3.7229
Pharmaceuticals using the Agilent 40 2 400 2
20 R = 0.9995 200 R = 0.9991
G1888 HS/6890N GC/5975 inert MSD 0 0
System,” Agilent Application Note, 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0 0.5 1.0 1.5 2.0 2.5 3.0
μg/mL μg/mL
Publication Number 5989-3196EN,
2005. 200 1,4 dioxane
160
Area

3. 120
80 * x-axis = vial solution concentration cv
Albert E. Gudat and Roger L. Firor, “The y = 15.268x + 0.3239
40
Determination of Extractables and R 2= 0.9996
0
Leachables in Pharmaceutical 0 2 4 6 8 10 12
Packaging Materials using μg/mL
Headspace/GC/MS,” Agilent Figure 10
Application Note, Publication Number Linearity plots for some residual solvents determined for the 6890N Headspace GC/FID system.
5989-5494EN, 2006.

Appendix
List of acronyms
%RSD – percent relative standard deviation LOD – limit of detection (S/N = 3)
[limit] – limit concentration LOQ – limit of quantitation (S/N = 10)
Atm P – atmospheric pressure MDL – method detection limit (statistical)
AUX – auxiliary MFS – mass flow sensor
BP – back pressure min – minutes
BPR – backpressure regulation MSD – mass selective detector
Cv – vial solution concentration P(Vial) – headspace vial pressure
Ce – excipient equivalent concentration PCM – pneumatic control module
Cal – calibration PS – pressure sensor
DMA – dimethyl acetamide PSI – pounds per square inch
DMI – 1,2-dimethyl-2-imidazolidinone RT – retention time
DMSO – dimethyl sulfoxide S/N – signal-to-noise ratio
EP – European Pharmacopoeia S/SL – capillary split/splitless inlet
EPC – electronic pneumatic control TR – transfer
FID – flame ionization detector U.S. FDA – United States Food and Drug Administration
GC – gas chromatograph USP – United States Pharmacopoeia
GSV – gas sampling valve VI – volatiles inlet
HS – headspace X – proportional valve
ICH – International Conference on Harmonization

Statistical Method Detection Limit (MDL)

MDL = s · tK?PTOCTPm+'(m@&&L= s · 3.143 Roger L. Firor and Albert E. Gudat are
Application Chemists at Agilent
Where
t(n-1, 1-alpha) = Student’s t value for the 99% Technologies, Inc., USA.
confidence level with n-1 Ute Bober is Program Manager at
degrees of freedom Agilent Technologies, Waldbronn,
n = number of trials Germany.
s = standard deviation of the 7 trials
USEPA Method 524.2 (Revision 4, August 1992)

5989-6023EN 80
Purification and Profiling of
Bioactive Compounds

81
82
Isolation of formononetin and other
phytoestrogens from red clover with
the Agilent 1100 Series purification
system

Application Edgar Nägele

Udo Huber

HO O

O
OH

Abstract

Isolating active natural products from plant origin is frequent in the pharmaceu-
tical industry when searching for new drugs. In this Application Note we show
how this task can be accomplished using the Agilent 1100 Series purification
systems AS (analytical scale) and PS (preparative scale)1. The isolation of for-
mononetin and other phytoestrogens from red clover serves as an example to
demonstrate the excellent performance of the Agilent purification solution at
analytical and preparative scale flow rates.

83 5988-5747EN
Introduction
Estrogens are used to treat
menopause disorders and osteoporo-
sis, because these disorders are
caused by a low hormone level.
Unfortunately, the steroids used have
a high rate of undesired side effects,
for example, thrombosis. Comparative
studies between Asian and Western
populations showed that these disor-
ders and diseases are a lot less com-
mon in Asia. This is explained by the
Asian soya-based diet, which contains
high levels of phytoestrogens. Phyto-
estrogens2 are currently under investi-
gation especially because of their impor-
tance in hormone replacement therapy
and cancer prevention without side effects.
Equipment Preparative scale system:
• Two Agilent 1100 Series
The experiments were performed preparative pumps
using the following systems: • Agilent 1100 Series
Analytical scale system: preparative autosampler
• Agilent 1100 Series vacuum • Agilent 1100 Series column
degasser organizer
• Agilent 1100 Series • Agilent 1100 Series diode-
quaternary pump array detector
• Agilent 1100 Series well-plate • Agilent 1100 Series
autosampler preparative fraction
• Agilent 1100 Series thermo- collector PS
statted column compartment The systems were controlled using the
• Agilent 1100 Series diode array Agilent ChemStation (rev. A.09.01) and
detector the Purification/HighThruput software
• Agilent 1100 Series fraction (rev. A.01.01).
collector AS

The identification of phytoestrogenes Absorbance

with isoflavonoid-structure in red [mAU]
clover (Trifolium pratense, L., Legu- 140
* = other phytoestrogens
minosae) extract using fluorescence3
120
and UV-visible detection is described *
in another Application Note4. In the 100
Application Note here, we describe the
separation and isolation of formononetin 80 *
and other phytoestrogens from red
60
clover in analytical and preparative *
* Formononetin
*
scale using the Agilent 1100 Series 40
purification systems AS and PS. *
20

0 5 10 15 20 Time [min]

Figure 1
Chromatogram of crude red clover extract.

Columns ZORBAX SB-C18 3 x 150 mm, 5 µm

Mobile phases: 0.1 % HOAc in water, 0.1 % HOAc in acetonitrile
Gradient: 20 % B to 45 % B in 20 min, 45 % B to 100 % B in 1 min
100 % B for 4.5 min, 100 % B to 20 % B in 0.5 min
Stop time: 25 min
Post time: 5 min
Flow: 0.7 ml/min
Injection: 5 µl
Column temp.: 35 °C
UV detector: DAD 260 nm/16 (ref. 800 nm/100) standard flow cell (10 mm pathlength)

5988-5747EN 84
Results and Discussion
Absorbance
Extraction [mAU]
50 µl
25 g dried red clover was extracted
ultrasonically with 250 ml methanol 4000
containing 25 ml 0.1 m H2SO4 for three
hours. The mixture was removed
from the ultrasonic bath and stirred 3000 30 µl
overnight. After filtration the solution
was evaporated at 40 °C to about 30
2000
ml and filtered again.
5 µl
Analytical method development 1000
An analytical scale method was devel-
oped based on the method described 0
previously4 to replace the phosphoric
acid by acetic acid. Figure 1 shows 0 5 10 15 20 25 Time [min]
the resulting chromatogram. Formo-
nonetine and several other phytoestro- Figure 2
gens4 are marked in the chromatogram. Volume overloading experiment.

Volume overloading experiment

Since concentration overloading was
not possible due to the fixed concen-
tration of the extract, volume overload-
ing had to be done to isolate the com-
pounds. Injecting up to 50 µl crude
extract sample still lead to sufficient
separation for analytical scale purifica-
tion (figure 2).

85 5988-5747EN
Isolation of phytoestrogens in
analytical scale. Absorbance
A common method to isolate com- [mAU]
pounds from complex natural extracts
is fractionation by time slices. Because 2500
of the good separation achieved in the
overloading experiments, peak-based 2000
fractionation was used for the red
clover extract. The chromatogram is 1500
shown in figure 3 – vertical lines indi-
cate the collected fractions. The ana- 1000
lytical method described in figure 1
was used with an injection volume of
500
50 µl. Fractions were collected
between 6 and 20 minutes, based on
threshold only (500 mAU). 0

0 5 10 15 20 Time [min]
Isolation of higher amounts
To gain more phytoestrogen material,
Figure 3
pooling of fractions from several runs Analytical scale fractionation of red clover extract
was carried out. This means repetitive
injections were performed from one
sample vial and the resulting fractions
were collected in the same fraction Absorbance
vials. The pooling feature is described [mAU] * = other phytoestrogenes
in detail in the User’s Guide5. 450 µl of *
1000
sample was injected in nine 50-µl *
injections and the resulting fractions
were pooled automatically. Reanalysis 800 Formononetin
of the fractions showed good results * *
which demonstates the excellent per- 600
formance of the instrument and soft- *
ware (figure 4). 400

200
*

0
0 5 10 15 20 Time [min]

Figure 4
Re-analysis of fractions from pooling.

5988-5747EN 86
Another possibility to purify more
Absorbance
material is to scale-up to a larger col- [mAU]
umn. Based on the analytical scale col- 2500
umn overloading experiment (figure 2),
scale-up calculations were done to
2000
inject 450 µl in one single injection.
This was achieved on a 9.4 x 150 mm
column at a flow rate of 7 ml/min. 1500
Since the Agilent 1100 Series well-
plate autosampler AS can only be used 1000
up to a flow rate of 5 mL/min purifica-
tion was transferred to a purification
500
system PS. Figure 5 shows the chro-
matogram that was obtained. The
lower peak heights and areas result 0
from the shorter pathlength of the 0 5 10 15 20 Time [min]
preparative flow cell (3 mm) compared
to the standard flow cell (10 mm). Re-
analysis of fractions showed compara-
Figure 5
ble purities as for the pooling experi- Preparative scale fractionation of red clover extract.
ment. The injected sample volumes
were the same (450 µl) for both the Columns: ZORBAX SB-C18 9.4 x 150 mm, 5 µm
pooling experiments and when using Mobile phases: 0.1 % HOAc in water, 0.1 % HOAc in acetonitrile
Gradient: 20 % B to 45 % B in 20 min, 45 % B to 100 % B in 1 min
the 9.4 mm id column. The solvent 100 % B for 4.5 min, 100 % B to 20 % B in 0.5 min
amounts used were also comparable Stop time: 25 min
(approximately 200 ml). However, the Post time: 5 min
main advantage of purification on the Flow: 7 ml/min
9.4-mm id column is the time that is Injection: 450 µl
Column temp.: ambient
saved in gaining the same amount of UV detector: DAD 260 nm/16 (ref. 800 nm/100) preparative flow cell (3 mm pathlength)
purified material. Fraction collection: based on threshold only (100 mAU) between 5.5 and 21 minutes

Isolation of phytoestrogens in prepar-

ative scale
To purify even higher amounts of sam-
ple the method was further scaled up
to a 21.2 x 150 mm column. At this
scale it was possible to inject 2300 µl
of sample in a single injection. The

87 5988-5747EN
chromatogram and the method are Absorbance
shown in figure 6. Re-analysis of frac- [mAU]
tions showed that scale-up was possi- 2500
ble without losing any performance with
regard to the purity of the fractions. 2000

1500
Conclusion

In this Application Note we showed 1000

the development of an analytical scale
method to separate the compounds in 500
a complex crude plant extract from
red clover containing formononetin 0
and several other phytoestrogen using
the Agilent 1100 Series preparative 0 5 10 15 20 Time [min]
system AS. Based on this method an Figure 6
analytical scale preparative separation Fractionation on a 21.2-mm preparative column.
with peak-based fraction collection
was carried out. To obtain more phy- Columns ZORBAX SB-C18 21.2 x 150 mm, 5 µm
toestrogen material the pooling fea- Mobile phases: 0.1 % HOAc in water, 0.1 % HOAc in acetonitrile
ture of the Agilent 1100 Series purifi- Gradient: 20 % B to 45 % B in 20 min, 45 % B to 100 % B in 1 min
cation system AS was used. 100 % B for 4.5 min, 100 % B to 20 % B in 0.5 min
Stop time: 25 min
The purity of the compounds gained Post time: 5 min
was determined by the reanalysis of Flow: 35 ml/min
the fractions. Based on the analytical Injection: 2300 µl
scale results the method was scaled Column temp. ambient
up. The purification was repeated on UV detector: DAD 260 nm/16 (ref. 800 nm/100) Preparative flow cell (3 mm)
Fraction collection: based on threshold only (100 mAU) between 5.5 and 21 minutes
two different columns on the Agilent
1100 Series purification system PS to
gain higher amounts of the desired
compounds in single runs. The results References 4.
“Separation of phytoestrogenes in red
in preparative scale were comparable
1. clover by reverse phase HPLC with UV-
to the results achieved on the analyti-
“New perspectives in purification with visible and fluorescence detection”,
cal scale system.
HPLC and HPLC/MS” Agilent Agilent Technologies Application Note,
Technologies Brochure publication publication number 5988-2399EN 2001.
number 5988-3673EN, 2001.
5.
2. “Agilent 1100 Series Purification
Julia Barrett, “Phytoestrogen: Friends System”, Agilent Technologies User's
or Foes”, Environmental Health Guide, part number G2262-90001, 2001.
Perspectives, (104), 478, 1996.

3.
“Sensitive and Reliable Fluorescence Edgar Nägele and Udo Huber are
Detection for HPLC”, Agilent application chemists at
Technologies Brochure, publication Agilent Technologies GmbH,
number 5968-9105E, 1990. Waldbronn, Germany.

5988-5747EN 88
Analysis of a complex natural product extract
from ginseng – Part I: Structure elucidation
of ginsenosides by rapid resolution LC–ESI
TOF with accurate mass measurement

Application Note Edgar Nägele

Abstract

Since prehistoric times extracts from herbs have been used for medical treat-
ment of disease. Their activity and effects on humans were found by trial and
error over generations. A good example of achieved efficiency is traditional
Chinese medicine (TCM). In Western medicine drugs derived from natural origin
are gaining importance due to their potential. However, Western pharmaceutical
quality standards require a deep knowledge about the ingredients in medicine
based on natural products.

This Application Note will demonstrate the use of the rapid resolution LC with
rapid resolution high throughput (RRHT) columns for the separation of the ingre-
dients found in a complex ginseng root extract and the measurement of accu-
rate masses by an ESI orthogonal acceleration time-of-flight MS (oaTOF) for the
structure elucidation. The use of a rapid resolution LC system together with an
ion trap MS for structure elucidation will be discussed in part two of this work8.

89 5989-4506EN
Introduction
Crude extracts of herbal and animal
origin have been used for medical
treatment of disease in all ancient cul-
tures around the globe since prehis-
toric time. Their activity for treatment
of different diseases and other effects
on humans were found by trial and
error over hundreds of years and the
knowledge about this medicine was
inherited from generation to genera-
tion. A good example of the efficiency
achieved during this process of opti-
mization is the herbal based traditional
Chinese medicine (TCM). Since these
drugs are often complex mixtures con-
taining hundreds of chemically varied
substances with different effects or
synergisms a quality control and quali-
ty assurance of medical potency is
very difficult. A widely used and
accepted method by the U.S. Food and
Drug Administration (FDA) and World
Health Organization (WHO) is chro-
matographic fingerprinting1,2.
In addition to the importance of the
chromatographic fingerprint, which is
capable of identifying a particular herb
and distinguishing between closely
related species in a qualitative manner,
the quantitative analysis of medical
plant extracts is also gaining impor-
tance. For traditional medicines a
quantitative analysis is crucial to their
quality control and to determine the
absolute content of pharmaceutical
effective substances as well as poten-
tially toxic undesirable natural sub-
stances3. In Western medicine drugs
derived from natural origins are gain-
ing importance due to their potential.
However Western pharmaceutical
quality standards require a deep
knowledge about the ingredients in
5989-4506EN
medicine based on natural products.
Since traditional herbal medicines
often contain hundreds of substances
with only a few bioactive compounds
it is necessary to develop new strate-
gies to screen these plant extracts for
biologically active compounds and for
their pharmacological effectiveness
on animal or cellular models as well
as receptor and enzyme based tests4.
A famous Asian herb, which has been
used in herbal medicine for more than
5000 years is the ginseng (Panax
species) root. Pharmacological effects
of ginseng which have been reported
are, for example, stimulatory and
inhibitory effects on the central ner-
vous system (CNS), antistress, antihy-
perglycemic, antineoplastic and
immunomodulatory effects5. The main
active compounds of the ginsenosides
are triterpene saponins from which
more than 80 have been isolated and
characterized during the past years.
An enormous amount of work has
been done during the last 30 years to
develop analytical methods for the
analysis of ginseng extracts and med-
ical formulations. The method of
choice for the analysis of complex
natural product extracts, like those
derived from the ginseng root, is high
performance liquid chromatography
(HPLC)6. LC/MS equipment such as
LC/ESI oaTOF for accurate mass mea-
surement and LC/ion trap or LC/triple
quadrupole instruments for structure
elucidation by MS/MS and MSn are
currently being used to determine the
complex and similar structures of gin-
senosides7. This Application Note will
demonstrate the use of the Agilent
1200 Series Rapid Resolution LC sys-
tem with RRHT columns for separat-
ing the ingredients found in a complex
90
ginseng root extract and the measure-
ment of accurate masses by an ESI
oaTOF MS. For the structure elucida-
tion CID fragmentation was carried out
and the measured accurate masses
were used to calculate empirical for-
mulas of the fragments. The use of the
high resolution LC system together
with an ion trap MS for structure eluci-
dation by MSn will be discussed in
part two of this work8.
Experimental
Equipment
• Agilent 1200 Series binary pump SL
with degasser. This pump has the
capability to deliver a pressure of up to
600 bar, which is necessary to per-
form high resolving HPLC analysis on
a 1.8-µm particle size RRHT column
to get the best resolution perfor-
mance.
• Agilent 1200 Series high performance
autosampler SL with thermostat. This
autosampler
is especially designed to work
together with the 1200 Series binary
pump SL at the lowest delay volume.
• Agilent 1200 Series thermostatted
column compartment (TCC). This TCC
is ready for use together with the
high pressure binary pump with
optional separate heat exchangers
and post column cooling under opti-
mized delay volume conditions and
alternating column regeneration with
an optional 2-position/10-port valve.
• Agilent 1200 Series diode-array
detector SL (DAD). This DAD is capa-
ble of acquiring data at a sampling
rate up to 80 Hz. This device has a
built-in data storage capability.
• Agilent 6200 Series MSD TOF.
2 -μL UV flow cell
Orthogonal acceleration time-of- 1.8 -μm SB C18
flight mass spectrometer with dual column
DAD SL
sprayer interface for mass calibration Low delay
to acquire molecular masses with To MS
volume heat
highest accuracy. This time-of-flight exchanger Column compartment
mass spectrometer is capable of
acquiring data at 40 Hz and pos/neg High-performance
switching. Degasser autosampler SL
• Picard TOF software A02.00.
Software used for data acquisition Binary pump SL Thermostat
with the TOF LC/MS system.
• Analyst Software. Software for TOF 650 mm x 0.17 mm capillary 320 mm x 0.12 mm capillary
and UV data analysis.
• Column. ZORBAX SB C18, Figure 1
2.1 x 150 mm, 1.8 µm Agilent 1200 Series binary LC system for MS using a low delay volume configuration.

Sample
Powdered freeze-dried Asian ginseng achieve the highest possible resolu-
root (1g) (Panax ginseng) was treated tion, which is demonstrated by the UV
ultrasonically for 30 minutes in 10 mL analysis of a complex natural product
methanol, filtered and directly used for extract obtained from Asian ginseng
analysis. root (Panax ginseng) (see below). The
full performance of the Agilent 1200
The set-up of the Agilent 6200 Series Series binary system in the high reso-
MSD TOF system is shown in figure 1. lution configuration is demonstrated
In this set-up the Agilent 1200 Series in another publication9. It is also pos-
binary pump SL is connected to the sible to use this system in a high
Agilent 1200 Series high performance throughput environment by making
autosampler SL with a 0.17-mm i.d. minor changes10.
stainless steel capillary. To reduce
delay volume, the seat capillary in the Methods:
Agilent 1200 Series high performance • The Agilent 1200 Series binary pump
autosampler SL has an 0.12-mm i.d., SL was operated under the following
which is the same kind of capillary conditions:
that connects the low delay volume Solvent A: water + 0.1 % TFA
(1.4 µL) heat exchanger in the Agilent Solvent B: AcN + 0.1 % TFA
1200 Series thermostatted column Flow: 0.5 mL/min
compartment to the column. A 2-µL Gradient: 0 min 5 % B,
cell is built into the Agilent 1200 1 min 5 % B,
Series diode-array detector SL for UV 60 min 85 % B,
detection. The outgoing capillary is 61 min 95 % B,
connected directly to the sprayer of 70 min 95 % B
the electrospray source at the time-of- Stop time: 70 min
flight mass spectrometer, which is Post time: 15 min
capable of aquiring spectra at 40 Hz.
This instrument set-up is optimized to
• The Agilent 1200 Series high perfor-
mance autosampler SL was used to
make injections of 10 µL sample and
the samples were cooled to 10 ºC.
The sample loop was switched to
bypass after 1 minute to reduce
delay volume.
• The Agilent 1200 Series thermostat-
ed column compartment, equipped
with the 1.4-µL low delay volume
heat exchanger, was set at 50º C.
• The Agilent 1200 Series diode-array
detector SL was operated at 80 Hz
for data acquisition at a wavelength
of 220 nm/4, ref. 360/100 with a
2-µL flow cell, 30-mm path length.
• Agilent 6200 Series MSD TOF was
operated under the following condi-
tions:
Source: ESI in positive mode with
dual spray for reference mass
Dry gas: 12 L/min
Dry Temp.: 200 ºC
Nebulizer: 35 psi
Scan: 200-1300
Fragmentor: 150 V or 300 V for CID
Skimmer: 60 V
Capillary: 3000 V

91 5989-4506EN
Conditions for both experiments: Column:

15.065
Pumps mAU 2.1 x 150 mm ZORBAX SB C-18, 5 µm
Solvent A: H2O + 0.1 % TFA

15.555
350
Solvent B: ACN + 0.1 % TFA

15.967
Agilent 1100 Series

17.136
Gradient: 10 % to 95 % ACN in 40 min, 300
hold for 1 min 400 bar system
Flow rate: 0.4 mL/min 250

15.159
Autosamplers
200

15.411
Injection volume: 3 µL
Thermostatted column compartments 150
Temperature: 50° C
Detectors 100
DAD 2-µL cell and 20 Hz, 220 nm
50
13 14 15 16 17 18 min

15.079
Results and discussion mAU
Column:
250 2.1 x 150 mm ZORBAX SB C-18, 1.8 µm

15.472
15.790
To compare the resolution perfor- Agilent 1200 Series

mance of the Agilent 1200 Series
Rapid Resolution LC system to an
Agilent 1100 Series standard LC sys-
tem, the analysis of a complex natural
product extract was performed on an
Agilent 1100 Series system with its
maximum pressure at 400 bar
equipped with a 5-µm particle size col-
umn and in comparison on an Agilent
1200 Series system with its maximum
pressure at 600 bar equipped with a
RRHT column with a 1.8-µm particle
size. The system backpressure was
about 520 bar using the 2.1 x 150 mm,
1.8-µm column. The resulting UV chro-
matograms acquired at 220 nm clearly
demonstrate the better resolution of
the peaks on the Agilent 1200 Series
system (figure 2). The peak width
(FWHM) of the majority of the peaks
in the UV chromatogram is below
0.1 min with baseline separation. This
excellent resolution, which can be
achieved with the Agilent 1200 Series
pump in combination with the RRHT
column results in significantly more
MS information useful for compound
identification.
16.746
200 600 bar system
150
100 14.787
14.937
50
13 14 15 16 17 18 min
Figure 2
Ginseng extract using a 400 bar system with a 5-µm particle column and using a 600-bar system
with a 1.8-µm particle column.
After separation of the individual com- the species dependent ginsenosides
pounds, which are components of the Rf and F11 are discussed in another
crude complex extract from the gin- publication11.
seng root, by means of rapid resolution
HPLC on a 1.8-µm particle size RRHT The ginsenoside Rb1 eluting at 27.7
column, they are subjected to accurate min was identified by its protonated
mass measurement by means of an molecular ion at m/z 1109.6129 and
ESI oaTOF. The main constituents calculation of the corresponding
elute between 20 and 33 minutes and empirical formula with a mass accura-
are marked as ginsenosides Re, Rf, F11, cy of -1.90 ppm relative to 2.10 mDa. A
Rb1, Rc, Rb2 and Rd in the base peak loss of one molecule of water leads to
chromatogram (figure 3). The ESI the ion at m/z 1091.6012 with -0.91
oaTOF data of the ginsenosides Rb1, ppm mass accuracy. A CID experiment
Rc and Re were analyzed in more was performed for structure elucida-
detail for structure elucidation while tion at an elevated skimmer voltage,

5989-4506EN 92
which presented additional informa-
tion in the TOF spectrum (figure 4).
9.0e6
The loss of one of the glucose chains
Rd
resulted in the ions at m/z 785.5047 8.0e6 Re
and m/z 325.1136 with mass accura-
7.0e6 Rb 1
cies of 0.53 ppm and -0.39 ppm,
Rb2
respectively. A series of ions resulting 6.0e6 Rc
from a consecutive loss of water from

Intensity cps
5.0e6
the ion m/z 785.5047 was also identi- Rf
fied with high mass accuracy. Further 4.0e6
loss of the second glucose chain from
3.0e6
the molecule resulted in the ion at m/z
425.3784 with -0.14 ppm mass accura- 2.0e6
cy. This ion is derived from the triter-
1.0e6 F11
penoid core structure common to all
ginsenosides. A loss of one molecule 0.0
of water resulted in the ion at m/z 19.0 21.0 23.0 25.0 27.0 29.0 31.0 33.0 35.0
407.3679 with 0.10 mDa and -0.30 ppm Time, min
mass accuracy. This set of CID frag- Figure 3
ments together with the molecular ion RR-LC-TOF basepeak chromatogram of the area containing the main components of the Asian
and the set of calculated empirical for- ginseng root extract.
mulas confirm the structure of the gin-
senoside Rb1. The fragmentation pat- Measured mass Calculated mass Formula Mass accuracy Mass accuracy
tern is shown in figure 4 and the mass [mDa] [ppm]
accuracies and empirical formulas are 1109.6129 1109.6108 C54H93O23 2.10 -1.90
summarized in table 1. 1091.6012 1091.6002 C54H91O22 1.00 -0.91
785.5047 785.5051 C42H73O13 -0.40 0.53
767.4950 767.4946 C42H71O12 0.40 -0.58
749.4854 749.4840 C42H69O11 -1.40 1.88
425.3784 425.3783 C30H49O 0.10 -0.14
407.3679 407.3678 C30H47 0.10 -0.30
343.1248 343.1240 C12H23O11 0.80 2.23
325.1136 325.1135 C12H21O10 0.10 -0.39

Table 1
Empirical formulas and achieved mass accuracies for the structure elucidation of ginsenoside RbT
by ESI oaTOF.

HO O
425.3784 O
O 425.3784
4.8e5 OH
OH O
CH3
HO
4.4e5 407.3679 HO CH3
OH OH OH
4.0e5
Intensity, counts

CH 3
3.6e5 343.1248
CH 3 CH 3
C30H49O = 425.3784
3.2e5
325.1136 325.1136 CH 3 -H2O
2.8e5 HO O
O C30H47 = 407.3679
2.4e5 H3 C CH3
2.0e5 OH
749.4854 HO 785.5047 C42H73O13 = 785.5047
1.6e5 1109.6129
HO O -H2O
O
1.2e5 767.4950 C42H71O12 = 767.4950
OH
8.0e4 1091.6012 HO -H2O
4.0e4 343.1248 785.5047 OH C42H69O11 = 749.4850
0.0
100 200 300 400 500 600 700 800 900 1000 1100
m/z, amu
Figure 4
Accurate mass measurement of ginsenoside RbT and its CID fragments by ESI oaTOF.

93 5989-4506EN
The ginsenoside Rc with 29.3 min min retention time are structure iso- m/z 947.5585 with a high mass accura-
retention time was identified after mers with the same empirical formula cy of 0.60 mDa or -0.59 ppm (figure 6).
empirical formula calculation by its and molecular mass. In the ginseno- The CID fragmentation gave further
molecular ion at m/z 1079.5991 and a side Rc the sugar arabinose is in the evidence for the identity of this sub-
product obtained by the loss of a mole- furanose form and in the ginsenoside stance. After cleavage of the saccha-
cule of water at m/z 1061.5888 with Rb2 the arabinose is in its pyranose ride moieties there are two remaining
mass accuracies of 1.02 ppm and 0.78 form. main fragments. The fragment
ppm, respectively. The fragmentation obtained after a loss of a molecule
obtained at a higher skimmer voltage The last of the main ingredients, which of glucose with m/z 767.4957 was
starts with a loss of the sugar residues was classified in the examined gin- measured with a mass accuracy of
arabinose and glucose followed by a seng root extract is ginsenoide Re. -1.50 ppm. A consecutive loss of water
loss of water from the remaining mole- This compound, which elutes at yields the ion at m/z 749.4855. After a
cule fragment (figure 5). The fragmen- 20.8 min, was identified by empirical further loss of the glucose disaccha-
tation of the sugar moieties resulted in formula calculation from the mass at ride the fragment of the triterpenoide
the initial ions at m/z 785.5060 and at
m/z 755.4936 with mass accuracies of
-1.12 ppm and 1.26 ppm, respectively.
After complete loss of all saccharide
Measured mass Calculated mass Formula Mass accuracy Mass accuracy
moieties the triterpenoide core struc- [mDa] [ppm]
ture was detected at m/z 425.3785
with high mass accuracy -0.37 ppm for 1079.5991 1079.6002 C53H91O22 -1.10 -1.02
1061.5888 1061.5896 C53H89O21 -0.88 0.78
the calculated empirical formula. The 785.5060 785.50.51 C42H73O13 0.90 -1.12
cleaved glucose chain was detected at 767.4939 767.4946 C42H71O12 -0.70 0.85
m/z 325.1124 with 0.22 pmm mass 749.4846 749.4840 C42H69O11 0.60 -0.81
755.4936 755.4946 C41H71O12 -1.00 1.26
accuracy. The fragmentation pattern is 737.4830 737.4840 C41H69O11 -1.00 1.34
shown in figure 5 and the achieved 719.4723 719,4734 C41H67O10 -1.10 1.56
mass accuracies for all fragments are 425.3785 425.3783 C30H49O 0.20 -0.37
407.3679 407.3678 C30H47 0.10 -0.30
summarized in table 2. The ginseno- 325.1134 325.1135 C12H21O10 -0.10 0.22
sides Rc and Rb2, which elute at 28.7
Table 2
Empirical formulas and achieved mass accuracies for the structure elucidation of ginsenoside Rc
by ESI oaTOF.

1079.5991 785.5060
C42H73O13 = 785.5060
1.6e5 O O
425.3785
425.3776 O -H2O
OH 767.4939 C42H71O12 = 767.4939
1.4e5 HO OH O
CH3
OH H O CH3 -H2O
1061.5888

1.2e5 OH OH C42H69O11 = 749.4846

Intensity, counts

737.4830 CH 3
425.3785
1.0e5
407.3678

CH 3 CH3 C30H49O = 425.3785

755.4936

325.1134 -H2O
CH3
8.0e4
325.1134

767.4939

O C30H47 = 407.3679
HO
749.4846

O
737.4830

6.0e4 H 3C CH3
719.4723

OH C41H71O12 = 755.4936
HO 755.4936
4.0e4 -H2O
785.5060

HO O
O C41H69O11 = 737.4830
2.0e4 OH -H2O
HO C41H67O10 = 719.4723
0.0 OH
150 250 350 450 550 650 750 850 950 1050
m/z, amu

Figure 5
Accurate mass measurement of ginsenoside Rc and its CID fragments by ESI oaTOF.

5989-4506EN 94
core structure at m/z 441.3739 with a Measured mass Calculated mass Formula Mass accuracy Mass accuracy
mass accuracy of 0.35 ppm occurred. [mDa] [ppm]
After a further successive loss of 947.5585 947.5579 C48H83O18 0.60 -0.59
water, this fragment yields to the ions 767.4957 767.4946 C42H71O12 1.10 -1.50
at m/z 423.3625 and m/z 405.3519 749.4855 7494840 C42H49O11 1.50 -2.01
441.3731 441.3733 C30H49O2 -0.20 0.35
with 0.45 ppm and 0.55 pmm respec- 423.3625 423.3627 C30H47O -0.20 0.45
tively. The calculated empirical formu- 405.3519 405.3521 C30H45 -0.20 0.55
las and mass accuracies of the mea- Table 3
sured fragments are summarized in Empirical formulas and achieved mass accuracies for the structure elucidation of ginsenoside Re
table 3. by ESI oaTOF.

767.4946
HO
O 441.3731
423.3625 OH O
CH3
4.0e5 CH 3
HO
OH OH
3.6e5
CH 3
3.2e5 CH 3
CH 3
2.8e5
CH 3
2.4e5 441.3731
HO
H 3C CH 3
2.0e5 O
HO
O
1.6e5
OH C30H49O2 = 441.3733
1.2e5 441.3731
HO -H2O
405.3519
8.0e4 749.4855 O C30H470 = 423.36275
HO O
767.4957
CH 3 -H2O
4.0e4 947.5585 C30H470 = 405.35.19
0.0
100 200 300 400 500 600 700 800 900 OH OH
m/z, amu

Figure 6
Accurate mass measurement of ginsenoside Re and its CID fragments by ESI oaTOF.

Conclusion
The featured application with the
Agilent 1200 Series Rapid Resolution
LC system together with the Agilent
6200 Series MSD TOF proves its capa-
bility for use in structure elucidation of
natural products in highly complex
extracts from plant origins. In this
application a highly complex extract
from ginseng root was analyzed with
the Agilent 1200 Rapid Resolution
LC/TOF system. A comparison to the
predecessor Agilent 1100 Series sys-
tem clearly demonstrated the higher
resolution that is achieved by the new
system equipped with a 1,8-µm parti- possible to use the high resolution
cle size column. Complex structures of LC/TOF technology to monitor the
three ginsenosides, which are the content of an extract before usage in a
main compounds of the extract, could pharmaceutical formulation. This is
be elucidated by the interpretation of achieved in a fully automated manner
the obtained TOF spectra. The pro- and confirms the compound’s identity
posed structures were confirmed by by measuring its mass with highest
empirical formula calculation of the accuracy and empirical formula confir-
molecular ions as well as for the frag- mation.
ments of the molecules obtained by
CID. All measured masses have accu-
racies in the lower single digit range
and therefore confirm the structures
with highest confidence. With this
detailed knowledge about the ingredi-
ents of the natural plant extract it is

95 5989-4506EN
References
1.
U.S. Food and Drug Administration,
“Guidance for Industry botanical Drug
Products”, 2000.
2.
World Health Organization, “General
Guidelines for Methodologies on
Research and Evaluation of Traditional
Medicine”, 2000.
3.
Drasar P., Moravcova J. “Recent
advances in analysis of Chinese med-
ical plants and traditional medicines.”,
J. Chrom. B 1-2, 812, 3–21,
4.
Huang X., Kong L., Li X., Chen X., Guo
M., Zou H. „Strategy for analysis and
screening of bioactive compounds in
traditional
Chinese medicines.”, J. Chrom. B 1-2,
812, 71-84, 2004.
5.
Attele A.S., Wu J.A., Yuan C.S.
Biochem. Pharmacol. 58, 1685-1693,
1999.
6.
Fuzzati N. “Analysis methods of gin-
senosides.”, J. Chrom B1-2, 812, 114-
133, 2004.
5989-4506EN
7.
Wang X., Sakuma T., Asafu-Adjaye E.,
Shiu G. K. „Determination of ginseno-
sides in plant extracts from Panax gin-
seng and Panax quinquefolius L. by
LC/MS/MS.”, Anal. Chem. 71, 1579-
1584, 1999.
8.
Naegele, E., “Examination of a com-
plex natural product extract from gin-
seng - Part II: Structure elucidation of
ginsenosides by high resolution LC-ion
trap by MSn” Agilent Application
Note, publication number 5989-
4705EN, 2006.
9.
“Performance of Agilent 1200 SL LC
system for highest resolution.” Agilent
Application Note, publication number
5989-4489EN, 2006.
10.
“Performance of the Agilent 1200 SL
HPLC System for Ultra-Fast LC
Applications with 2.1-mm i.d.
columns.” Agilent Application Note,
publication number 5989-4502EN,
2006.
11.
Naegele, E, “Examination of a complex
natural product extract from ginseng -
Part III: Species differentiation of gin-
seng and authentification of ginseng
products by LC/MS”, Agilent
Application Note, publication number
Edgar Nägele is Application Chemist at
5989-4706EN, 2006.
Agilent Technologies, Waldbronn,
Germany.
96
Analysis of a complex natural product
extract from ginseng – Part II: Structure
elucidation of ginsenosides by high
resolution ion trap LC/MS

Application Note
Edgar Nägele

Abstract

Since prehistoric time extracts from herbs have been used for medical treatment
of disease. Their activity and effects on humans were found by trial and error
over generations. A good example of achieved efficiency is traditional Chinese
medicine (TCM). In Western medicine drugs derived from natural origins are
gaining importance due to their potential. However, Western pharmaceutical
quality standards require a deep knowledge about the ingredients in medication
based on natural products. This Application Note will demonstrate the use of
the Agilent 1200 Series Rapid Resolution LC system with Rapid Resolution High
Throughput (RRHT) columns for the separation of the ingredients found in a
complex ginseng root extract. The information obtained with an ion trap MSn
and MRM is used to determine the structure of the ingredients and for quality
control purposes.

97 5989-4705EN
Introduction
Crude extracts of herbal and animal
origin have been used for medical
treatment of disease in all ancient cul-
tures around the world since prehis-
toric time. Their activity for treatment
of different diseases and other effects
on humans were found by trial and
error over hundreds of years and the
know-ledge about this medicine was
passed on from generation to genera-
tion. A good example of the efficiency
achieved during this process of opti-
mization is the herbal based traditional
Chinese medicine (TCM). A famous
Asian herb, which has been used in
herbal medicine for more than 5000
years is the ginseng root (Panax gin-
seng). The main active compounds of
the ginsenosides are triterpene
saponins of which more than 80 have
been isolated and characterized during
the past years. A lot of work was done
during the last 30 years to develop
analytical methods for the analysis of
ginseng extracts and medical formula-
tions. The method of choice for the
analysis of complex natural product
extracts such as those derived from
the ginseng root is high performance
liquid chromatography (HPLC)1.
LC/MS equipment, e.g LC/ESI oaTOF
for accurate mass measurement and
ion trap LC/MS or triple quadrupole
LC/MS instruments for structure eluci-
dation by MS/MS and MSn are cur-
rently used to determine the complex
and similar structures of
ginsenosides2. In particular, it is possi-
ble to confirm the authenticity of the
pharmaceutical ginseng products and
differentiate between their active
ingredients using the ion trap MSn
fragmentation patterns3.
5989-4705EN
This Application Note will demon-
strate the Agilent 1200 Series Rapid
Resolutioin LC system with 1.8-µm
columns for the separation of the
ingredients found in a complex gin-
seng root extract. The information
obtained with an ion trap MSn and
MRM is used to determine the struc-
ture of the ingredients and for quality
control purposes. The use of a high
resolution LC system together with an
ESI oaTOF for accurate mass mea-
surement is described in Part I in this
series of Application Notes4.
Experimental
Equipment
• Agilent 1200 Series binary pump SL
with a degasser. This pump has the
capability to perform high resolution
HPLC analysis on a 1.8 -µm particle
size RRHT column and achieve the
best performance.
• Agilent 1200 Series high perfor-
mance autosampler SL with a ther-
mostat. This autosampler is espe-
cially designed to work with the
Agilent 1200 Series binary pump SL
to ensure lowest delay volumes.
• Agilent 1200 Series thermostatted
column compartment (TCC). The TCC
is ready for use with the Agilent
1200 Series binary pump SL and
optional separate heat exchangers, as
well as post column cooling, under
optimized delay volume conditions,
together with alternating column
regeneration with an optional 2-
position/10-port valve.
• Agilent 1200 Series diode array
detector SL (DAD). This DAD is
capable of acquiring data with a
sampling rate of up to 80 Hz. In case
98
of network problems, this device has
a built-in data storage capability.
• Agilent 6330 Ion Trap LC/MS.The ion
trap is operated with a standard ESI
source and able to scan with 26,000
m/z /sec. The MSn spectra are
acquired data dependent and fully
automated.
• The software used for instrument
control was ChemStation B01.03, ion
trap software 5.3, and for data analy-
sis the ion trap data analysis soft-
ware 3.3.
• Column: ZORBAX SB C18,
2.1 x 150 mm, 1.8 µm
Sample
Powdered freeze-dried Asian
ginseng root (1g) (Panax ginseng) was
treated ultrasonically for 30 minutes in
10 mL methanol, filtered and directly
used for analysis.
The set-up of the Agilent 6330 Ion Trap
LC/MS system is shown in figure 1.
The Agilent 1200 Series binary pump
SL is connected to the Agilent 1200
Series high performance autosampler
SL (ALS SL) with a 0.17-mm i.d. stain-
less steel capillary. To reduce delay
volume, the seat capillary in the ALS
SL has an i.d of 0.12 mm. The same
kind of capillary connects to the low
delay volume (1.6 µL) heat exchanger
in the Agilent 1200 Series thermostat-
ted column compartment, which is
connected to the column. A 2-µL cell is
built into the Agilent 1200 Series
diode-array detector SL for UV detec-
tion. The outgoing capillary is directly
connected to the sprayer of the elec-
trospray source of the 6330 Ion
Trap LC/MS, which is capable of
acquiring data with a scan speed of
26,000 m/z/sec. This instrument set-up
is optimized to achieve the highest
possible resolution, which is demon-
strated by the UV analysis of a com-
plex natural product extract obtained
from Asian ginseng root (Panax gin-
seng). To illustrate the performance, a
comparative analysis on an Agilent
1100 Series LC system (5-µm particle
size column) and on an Agilent 1200
Series Rapid Resolution LC system
(1.8-mm particle size column) is pre-
sented4. The resulting UV chro-
matograms acquired at 220 nm clearly
demonstrate the better resolution of
the peaks on the Agilent 1200 Series
Rapid Resolution LC system. The peak
width (FWHM) of the majority of the
peaks in the UV chromatogram is
below 0.1 min with baseline separa-
tion. The full performance of the
Agilent 1200 Series Rapid Resolution
LC system in the high resolution con-
figuration is documented in a separate
Application Note5. It is also possible to
use this system, with minor changes,
in a high throughput environment,
which is described in another
Application Note6.
Methods
• The Agilent 1200 Series binary pump
SL was operated under the following
conditions:
Solvent A: water + 0.1 % TFA
Solvent B: AcN + 0.1 % TFA Flow:
0.5 mL/min
Gradient: 0 min 5 % B,
1 min 5 % B,
60 min 85 % B,
61 min 95 % B,
70 min 95 % B
Stop time: 70 min
Post time: 15 min
2 µl UV flow cell
1.8 µm SB C18
column
DAD SL
Low delay
volume heat To MS
exchanger ColCom
degasser h-ALS-SL
binary pump SL cooler
650 mm x 0.17 mm capillary 320 mm x 0.12 mm capillary
Figure 1
Agilent 1200 Series Rapid Resolution LC system for MS in low delay volume configuration.
• The Agilent 1200 Series high perfor- Results and discussion
mance autosampler SL was used for
injections of 10 µL sample and the After separation of the individual com-
samples were cooled to 10 ºC. The pounds, which are components of the
sample loop was switched to bypass crude extract, from the ginseng root by
after 1 minute to reduce delay volume. means of high resolution HPLC on a
• The Agilent 1200 Series thermostat- 1.8-µm particle size column, they are
ted column compartment SL was subjected to fragmentation for MSn.
adjusted to 50 ºC equipped with the The major ingredients elute between
low delay volume heat exchanger. 20 and 33 minutes and are recorded as
• The Agilent 1200 Series diode-array ginsenosides Re, Rf, F11, Rb1, Rb2, Rc,
detector SL was operated at 80 Hz Rd in the base peak chromatogram
for data acquisition at a wavelength (figure 2). The high resolution of the
of 220 nm/4, ref. 360/100 with the column used resolves a large amount
2-µL flow cell, 30-mm path length. of minor compounds from the ginseng
• The 6330 Ion Trap LC/MS was oper- extract, which may also be analyzed
ated under the following conditions: because the high scan rate of 26,000
Source: ESI in positive mode. m/z/sec allow sufficient ion trap
Dry gas: 5.0 L/min MS/MS and MSn data to be acquired.
Dry temp.: 300 ºC The ion trap MS/MS and MS3 data of
Nebulizer: 15 psi the ginsenosides Re, Rb1 and Rc were
Target: 125,000 investigated in more detail for struc-
Max. accum. time: 100 ms ture elucidation while the species-
Scan: 200-1300 dependent ginsenosides Rf and F11 is
Averages: 2 discussed in another part of this
MSn: Automated MS/MS study7.
and MS3
99 5989-4705EN
The simple MS scan delivers the mass
of the molecular ion in its protonated Intens. Rb2
x107
Re Rb1 Rc Rd
and sodiated form (figure 3). The ratio 4
of protonated and sodiated ions Rf
3
depends on the electrospray source
F11
temperature because the sodiated 2
complexes are more stable at higher 1
temperatures than the protonated 0
ions, which decompose due to a loss 15 20 25 30 35 40
Time [min]
of water and other fragmentations.
The MS scan delivers the ions at Figure 2
m/z 928.9 [M+H-H2O]+, 946.9 [M+H]+, Base peak chromatogram of ginseng extract by high resolution LC ion trap on a 1.8 µm particle
969.1 [M+Na]+ for ginsenoside Re; column.
at m/z 1090.1 [M+H-H2O]+, 1108.5
[M+H]+, 1131.1 [M+Na]+ for gin-
senoide Rb1 and at m/z 1059.71
+MS, 19.8 min
[M+H-H2O]+, 1077.75 [M+H]+,
BPC 100-1300
1101.1 [M+Na]+ for ginsenoside Rc. m/z 946.9 [M+H] +
The conditions for the MS/MS and m/z 969.1 [M+Na] +
MS3 fragmentation in the 6330 Ion Intens.
m/z 928.9 [M+H-H2O] +
x10 7 946.9 Ginsenoside Re
Trap LC/MS were essentially chosen 5
to produce sodiated ions. The frag- 969.1
4
mentation of these ions gave much 3
clearer fragmentation and additional
2
structural information compared to the 928.9
1
protonated ions, whose fragments
0
were similar to the CID fragments dis- 200 400 600 800 1000 1200 1400 m/z
cussed in part 1 of this study4. Intens.
Only one fragment is produced in +MS, 27.3 min 1108.5
Ginsenoside Rb1
x10 7
MS/MS for the ginsenodide Re at m/z 1.0 BPC 100-1300
789.9 by a loss of a molecule of glu- m/z 1108.5 [M+H] +
0.8
m/z 1131.1 [M+Na] +
0.6 m/z 1090.1 [M+H-H O] +
2
0.4
1090.1 1131.1
0.2
0.0
200 400 600 800 1000 1200 1400 m/z
Intens.
x10 7 +MS, 28.5 min 1077.7 Ginsenoside Rc
1.5 BPC 100-1300
m/z 1077.7 [M+H] +
1.0 m/z 1101.1 [M+Na] +
m/z 1059.7 [M+H-H2O] +
0.5 1101.1
1059.7
0.0
200 400 600 800 1000 1200 1400 m/z

Figure 3
Mass spectra of ginsenosides Re, Rc and RbT.

5989-4705EN 100
cose from the sodiated molecular ion
(figure 4). In MS3 this sodiated frag- Intens.
HO
x10 7 789.5 +MS2 (969.1), 19.8 min O 789.5
ment is cleaved into two parts, where-
8 OH O
CH3
as the detected ion at m/z 349.2 6 HO
OH OH
CH3

comes from the cleaved disaccharide 4 CH3

CH3
moiety. In comparison, for the ginseno- 2 CH3

side Rc there are two fragments 0 CH3

200 600 1000 1400 1800 m/z
HO
obtained by MS/MS. There is also one Intens. H 3C CH3
O 349.2
at m/z 789.5 and another one at m/z x106 349.2 +MS3 (969.1->789.5), 19.9 min HO
O

335.1 for the arabinose saccharide (fig- 0.8 OH

HO
0.6
ure 5). The dissacharide at m/z 365.1 is 0.4 HO O
O

CH3
cleaved off from the ion at m/z 789.5 0.2
0.0
in the MS3 stage. The ginsenoside Rb1 200 600 1000 1400 1800 m/z OH OH

has another different fragmentation

pattern where the molecular ion Figure 4
is cleaved to the ions at m/z 789.5 and MS/MS and MSV fragmentation for structure elucidation of ginsenoside Re with ion trap LC/MS.
m/z 365.1 by a loss of the
glucose chain in the MS/MS stage. In
the MS3 stage the fragment obtained Intens.
at m/z 789.5 also loses the same dis- x107 789.5 +MS2 (1101.1), 28.5 min O O 335.1 789.5
O
accharide at m/z 356.1 (figure 6). 0.8 HO
OH
O
OH
0.6 O H HO
C H3
C H3
0.4 OH OH

It is possible to distinguish between 0.2 335.1 C H3

C H3 C H3
0.0 365.1
the different ginsenosides contained in 200 600 1000 1400 1800 m/z C H3
Intens. O
the ginseng root extract with these HO
O
x10 6 365.1 +MS3 (1101.1->789.7), 28.6 min H 3C C H3
MS/MS and MS3 fragmentation pat- 8 HO
OH

terns because the different saccharide 6 HO

O
O

molecules connected to the triter- 4 OH

HO
2
penoide core structure will be cleaved 0
OH

off in a different but characteristic 200 600 1000 1400 1800 m/z
way. If the specific fragmentation pat-
tern in MS/MS and MS3 mode of the Figure 5
various ginsenosides is known, it MS/MS and MSV fragmentation for structure elucidation of ginsenoside Rc with ion trap LC/MS.
helps to detect the presence of a spe-
cial compound in the plant extract in a
very specific manner with the ion trap
MRM. MS/MS-MRM on the sodiated Intens.
x107 789.5 365.1 789.5
+MS2 (1131.4), 27.3 min HO
O
O
6 O
OH O
OH CH 3
4 365.1 HO
HO CH 3
OH OH OH
2 CH 3

CH 3 CH 3
0 365.1
Intens. 200 400 600 800 1000 m/z O
CH 3
HO
O
x106 365.1
OH
H3 C CH3
5 +MS3 (1131.4->789.5), 27.4 min HO
4 HO
O
O
3
OH
2 HO
1 OH
0
200 400 600 800 1000 m/z

Figure 6
MS/MS and MSV fragmentation for structure elucidation of ginsenoside RbT with ion trap
LC/MS.

101 5989-4705EN
molecular ions of the ginsenosides Re,
Rb1 and Rc shows exactly their pres-
ence in the crude ginseng root exact
Intens.
(figure 7). The obtained MS/MS spec- UV chromatogram, 216-224 nm TIC +All MSn
tra are in accordance with the spectra mAU Intens
125 Re Rd .

obtained in the experiments described Rb1 Rb2 x10 6

100 3
above (figures 4-6). In addition, the Rc
75
2
isomeric ginsenosides Rb2 and Rd are 50
also detected. 25 1
0 0
Conclusion 18 20 22 24 26 28 30 32
Time [min]

The Agilent 1200 Series Rapid

Resolution LC system in combination
with the Agilent 6330 Ion Trap LC/MS
proves its capability in structure eluci-
dation of natural products in highly
complex extracts from plant origin. In
this study a highly complex extract
from ginseng root was analyzed using
the Agilent 1200 Series Rapid
Resolution LC system and the Agilent
6330 Ion Trap LC/MS system. Complex
structures of three ginsenosides,
which are the main compounds of the
extract, could be elucidated by the
interpretation of the MS/MS and MS3
ion trap data obtained. The detailed
knowledge of the different fragmenta-
tion data can be used to control the
quality of natural extracts or pharma-
ceutical products by ion trap MRM
when applied to special ingredients.
Figure 7
MS/MS-MRM of the sodiated molecular ions of ginsenosides Re, RbT and Rc together with the
UV chromatogram obtained from a crude ginseng root.
References
1.
Fuzzati N. “Analysis methods of gin-
senosides.”, J. Chrom B1-2, 812, 114-
133, 2004.
2.
Wang X., Sakuma T., Asafu-Adjaye E.,
Shiu G. K. „Determination of ginseno-
sides in plant extracts from Panax
ginseng and Panax quinquefolius L.
by LC/MS/MS.”, Anal. Chem. 71,
1579-1584, 1999.
3.
Chan T.W.D., But P.P.H., Cheng S.W.,
Kwok I.M.Y., Lau F.W., Xu H.X.
“Differentiation and authentication of
Panax ginseng, Panax quinquefolius,
and ginseng products by using
HPLC/MS.” Anal. Chem. 72, 1281-
1287, 2000.
4.
“Examination of a complex natural
product extract from Ginseng – Part I:
Structure elucidation of Ginsenosides
by high resolution LC – ESI oaTOF with
accurate mass measurement”, Agilent
Application Note, Publication number
5989-4506, 2006.
5.
“Performance of the Agilent 1200
SL HPLC System for Ultra-Fast LC
Applications with 2.1-mm i.d.
columns.” Agilent Application Note,
publication number 5989-4502EN,
2006.
6.
“Performance of Agilent 1200 SL
LC system for highest resolution.”
Agilent Application Note, publication
number 5989-4489EN, 2006.
7.
“Examination of a complex natural
product extract from Ginseng - Part III:
Species differentiation ginseng and
authentification of ginseng products
by LC/MS”, Agilent Application Note,
Publication Number 5989-4706EN,
2006.
Edgar Nägele is Application Chemist at
Agilent Technologies, Waldbronn,
Germany.

5989-4705EN 102
Analysis of a complex natural product
extract from ginseng – Part III: Species
differentiation of ginseng plants and
authentication of ginseng products by LC/MS

Application Note
Edgar Nägele

Abstract

Since prehistoric times herbal extracts have been used for the treatment and
prevention of disease. The same plant often shows different patterns of active
ingredients depending on the region, climate, growing conditions or sub-species.
Some plants used in traditional Chinese medicine (TCM) are a good example of this
behavior. A typical plant is the ginseng root, which is widespread in various sub-
species on the entire Asian and American continents. This Application Note will
demonstrate the use of the Agilent 1200 Series Rapid Resolution LC system with
Rapid Resolution High Throughput (RRHT) columns for the separation of the ingre-
dients found in a complex ginseng root extract. Detection and identification of the
different compounds by LC/ESI orthogonal acceleration time-of-flight (oaTOF) and
LC/ion trap mass spectrometry was implemented for the differentiation of the
regional or biological origin. The possibility to provide evidence for the authenticity
of a ginseng product is also demonstrated by determining the ingredient profile
based on known structural information of the ingredients.

103 5989-5493EN
Introduction
Traditional natural medicine from dif-
ferent and regionally separated cul-
tures often uses the same plant
species for the preparation of herbal-
based extracts and tinctures for the
treatment of disease. The plants used,
which are often cultivated as sub-
species or grown under different con-
ditions, show a distinctive pattern of
active ingredients depending on these
influences. A typical botanical exam-
ple for this behavior is the ginseng
root (Panax spec.), which has been
used in traditional Chinese medicine
(TCM) for thousands of years. This
herb is widespread in various sub-
species on entire Asian and American
continents1. The method of choice for
the analysis of complex natural prod-
uct extracts like the ginseng root is
high performance liquid chromatogra-
phy (HPLC)2. The main active com-
pounds are triterpene saponins called
ginsenosides of which more than 80
have been isolated and characterized
during the past several years. LC/ESI
oaTOF for accurate mass measure-
ment and LC/ion trap or LC/triple
quadrupole instruments for structure
elucidation by MS/MS and MSn are
currently used for the determination of
complex and similar structures of gin-
senosides3. It is possible to differenti-
ate between the origin and the gin-
seng sub-species and to confirm the
authenticity of pharmaceutical ginseng
products using fragmentation patterns
obtained by means of tandem mass
spectrometry of the pharmaceutical
active ingredients4. This Appli-cation
Note will demonstrate the use of the
Agilent 1200 Series Rapid Resolution
LC system with Rapid Resolution High
Through-put (RRHT) columns for the
separation of the ingredients found in
a complex ginseng root extract from
Asian ginseng (Panax ginseng) and
5989-5493EN
American ginseng (Panax quinque-
folius). Detection and identification of
various compounds by LC/ESIoaTOF
accurate mass measurement and
LC/ion trap MSn mass spectrometry
was implemented for the differentia-
tion of the regional or biological origin.
The possibility to provide evidence for
the authenticity of a ginseng product
is also demonstrated by determining
the ingredient profile based on known
structural information of the ingredi-
ents. The detailed structure elucida-
tion of ginsenoides by means of ESI
oaTOF for accurate mass measure-
ment including CID fragmentation and
ion trap MSn is described in detail in
parts I and II of this study5,6.
Experimental
Equipment
• Agilent 1200 Series binary pump SL
with degasser. This pump has the
capability to perform high resolving
HPLC analysis on a 1.8-µm particle
size column to get the best resolu-
tion performance.
• Agilent 1200 Series high performance
autosampler SL (h-ALS SL) with ther-
mostat. This autosampler is especial-
ly designed to work together with the
binary pump SL with lowest delay
volumes.
• Agilent 1200 Series thermostatted
column compartment (TCC). The TCC
is ready for use with the binary pump
SL. Optional extras include separate
heat exchangers for pre-column heat-
ing and post-column cooling under
optimized delay volume conditions
and a 2-position/10-port valve for
alternating column regeneration.
• Agilent 1200 Series diode array
detector SL (DAD). The DAD is capa-
ble of acquiring data with a sampling
rate of up to 80 Hz.
• Agilent 6210 MSD TOF. Orthogonal
acceleration time-of-flight mass
104
spectrometer with dual sprayer inter-
face for mass calibration to acquire
molecular masses with highest
accuracy. This time-of-flight mass
spectrometer is capable of acquiring
data at 40 Hz with pos/neg switching.
• Agilent 6330 Ion Trap. Ion trap mass
spectrometer for MSn tandem mass
spectrometric experiments with scan
rate up to 26,000 m/z per second and
automated data-dependent MSn
capabilities.
• The software used for LC/ion trap
instrument control was ChemStation
B01.03, ion trap software 5.3, and for
data analysis the ion trap data analy-
sis software 3.3.
• The software used for LC/TOF instru-
ment control was the Mass Hunter
Workstation A02.00 for data acquisi-
tion and AnalystQS for data analysis.
• Column: ZORBAX SB C18,
2.1 X 150 mm, 1.8-µm
Sample
1. Powdered freeze-dried Asian gin-
seng root (1 g) (Panax ginseng) and
powdered freeze-dried American gin-
seng root (1 g) (Panax quinquefolius)
were treated in an ultra sonic bath
for 30 minutes in 10 mL methanol,
filtered and directly used for analysis.
2. Syrup-like Korean ginseng extract
pharmaceutical product (ILHWA Co.,
LTD, Korea) was dissolved (1 g) in
100 mL water/MeOH (1/1, v/v) and
used directly after filtration.
System set-up
The set-up of the LC/MS system is
shown in figure 1. The Agilent 1200
Series binary pump SL is connected to
the Agilent 1200 Series h-ALS SL with
a 0.17-mm i.d. stainless steel capillary.
To reduce delay volume, the seat capil-
lary in the h-ALS SL has 0.12 mm i.d.
The same kind of capillary connects to
the low delay volume (1.6 µL) heat
exchanger in the TCC, which is con-
2 μL UV flow cell
nected to the column. For UV detection 1.8 μm SB C18
a 2-µL flow cell is built into the DAD column
SL. The outgoing capillary is directly DAD SL
Low delay
connected to the sprayer of the elec- volume heat
To MS
trospray source at the ESI oaTOF or to exchanger Column
the ion trap mass spectrometer. This Compartment
instrument set-up is optimized to
achieve the highest possible resolu- Degasser h-ALS-SL
tion, which is demonstrated by the
comparative UV analysis of a complex Binary pump SL Cooler
natural product extract obtained from
Asian ginseng root (Panax ginseng)5.
650 mm x 0.17 mm 320 mm x 0.12 mm capillary
Using the 2.1 x 150 mm, 1.8-µm col- capillary
umn, the system back pressure was
typically about 560 bar. The peak width Figure 1
Agilent 1200 binary LC system for MS in low delay volume configuration.
(FWHM) of the majority of the peaks in
the UV chromatogram was below 0.1
min with baseline separation. The full volume. tions:
performance of the Agilent 1200RR LC • The Agilent 1200 Series thermostat- Source: ESI in positive
system in the high resolution configu- ted column compartment was adjust- mode.
ration is outlined in a separate perfor- ed to 50 ºC, equipped with the 1.6-µL Dry gas: 5.0 L/min.
mance note7. It is also possible to use low delay volume heat exchanger. Dry temp.: 300 ºC
this system with minor changes in a • The Agilent 1200 Series diode array Nebulizer: 15 psi
high throughput environment8. detector SL was operated at 80 Hz for Target: 150,000
data acquisition at a wavelength of Max. accum.
Methods 220 nm/4, ref. 360/100 with the time: 100 ms
• The Agilent 1200 Series binary pump 2-µL flow cell, 3-mm path length. Scan: 200-1300
SL was operated under the following • The TOF mass spectrometer was Averages: 2. Automated
conditions: operated under the following MS/MS and MS3
Solvent A: Water + 0.1 % TFA conditions:
Solvent B: AcN + 0.1 % TFA Source: ESI in positive mode
Flow: 0.5 mL/min.
Results and discussion
with dual spray for The separation of the individual com-
Gradient: 0 min 5 % B reference mass. pounds in the samples of Asian gin-
1 min 5 % B Dry gas: 12 L/min. seng (Panax ginseng) and American
60 min 85 % B Dry temp.: 200 ºC ginseng (Panax quinquefolius) using
61 min 95 % B Nebulizer: 35 psi. high resolution HPLC on a 1.8-µm par-
70 min 95 % B Scan: 200-1300 ticle size column unravels the differ-
Stop time: 70 min. Fragmentor: 150 V or 300 V ences in the individual composition of
Post time: 15 min. for CID the crude plant extracts from the dif-
• The Agilent 1200 Series high perfor- Skimmer: 60 V ferent ginseng sub species. Obviously,
mance autosampler SL was used to Capillary: 3000 V the differences lie not only in the con-
make sample injections of 10 µL and • The ion trap mass spectrometer was centrations of the ginsenoides but also
the samples were cooled to 10 ºC. operated under the following condi- in the presence of different com-
The sample loop was switched to
pounds in the individual species. The
bypass after one minute to reduce delay

105 5989-5493EN
main components elute between 20
and 40 minutes and are evidently the Intens.
common ginsenosides Re, Rg1, Rb1, x107
6 Re/Rg1 F11
Rb2, Rc and Rd as well as the special Rb1 Rd
ginsenoides Rf and F11 in the ion trap 4 Rb 2
MS base
2 Rc
peak chromatogram (figures 2 and 3).
Despite the good resolution of the
0
1.8-µm particle column, their resolution 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0
can be improved depending on the Time [min]
temperature. For instance, at a column
temperature of 50 °C the geniseno- Figure 2
Base peak chromatogram of the separation of compounds contained in a crude extract from
sides Re at m/z 946.5 [M+H]+ and gin- American ginseng (Panax quinquefolius) on a RRHT column at 50 °C.
senoside Rg1 at m/z 823.5 [M+Na]+
are not resolved (figure 2).
Intens
Nevertheless, they are clearly separat-
ed at 80 °C and are contained in both x10 7 A F11
5
ginseng samples (figure 3). To distin-
4 Rb1 Rd
guish the subspecies of Asian and 3
Re Rb 2
American ginseng by LC/MS the iso- 2
Rg1 Rc
meric ginsenosides Rf and the pseudo- 1
ginsenoside F11 are very useful. The 0
analysis of Asian ginseng shows the 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0 Time [min]
Intens
ginsenoside Rf in protonated and sodi- x10 7 B Rb2
ated form at m/z 801.5 and m/z 823.5, 4 Rb1 Rd
Rc
respectively. The protonated form of 3
Re
the isomeric pseudoginsenoside F11 at Rf
2 Rg 1
m/z 801.5 is only detectable as a trace
1 F11
compound (figure 3). In contrast, the
American ginseng species only con- 0
20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0 Time [min]
tains the pseudoginsenoside F11 in a
larger amount and not the ginsenoside
Rf. Both isomeric compounds have the Figure 3
Base peak chromatograms from the separation of compounds in a crude extract from
same empirical formula (C42H72O14). A) American ginseng (Panax quinquefolius)
The compounds are constitutional iso- B) Asian ginseng (Panax ginseng) sub species on a RRHT column at 80 °C.
mers only differentiable by their struc-
tural formula. To distinguish between of pseudoginsenoside F11 obtained identity of the molecular ion with 0.41
both isomeric forms of these species- at a retention time of 28.5 min (figure ppm mass accuracy. Four of the empir-
typical compounds, it is necessary to 4). Ultimately the distinctive furan ring ical formulas for the fragments result-
perform MS/MS experiments. The ion fragment is released at m/z 143.1. To ing from the typical consecutive loss
trap MS/MS experiments of the confirm the structural identity of the of five molecules of water resulting in
American ginseng sample show the proposed fragments, the experiment the ions at m/z 475.3782, 457.3676,
typical sequence of a loss of five water was repeated with a high resolution 439.3569 and 421.3464 were confirmed
molecules at m/z 475.4, 457.4, 439.3, LC/MS oaTOF instrument for high with mass accuracies less than 2 ppm.
421.3 and 403.3 after the cleavage of mass accuracy measurement and The cleaved disaccharide fragment
the glucose-rhamnose disaccharide empirical formula confirmation (figure was confirmed by the accurate mass
moiety at m/z 309.1, which is charac- 5). The measurement of the accurate at m/z 309.1176 with 3.10 ppm accura-
teristic for the molecular constitution mass of m/z 801.4997 confirmed the cy and the small furan ring fragment at

5989-5493EN 106
m/z 143.1065 with 4.90 ppm. The table
OH
in figure 5 summarizes the measured H3 C
C H3
M(C42 H 72 O14 )=800.49
masses and the calculated accuracies
C 8 H15O2
[M+H] + = 801.50
for all obtained fragments derived from O
OH 143.10 [MH-(-Glc-Rha)-H2 O] + = 475.37
pseudoginsenoide F11.In comparison, H3 C
[MH-(-Glc-Rha)-2H2 O] + = 457.36
the compound obtained at a retention [MH-(-Glc-Rha)-3H2 O] + = 439.35
CH3 C H3
time of 27.7 minutes which is typical [MH-(-Glc-Rha)-4H2 O] + = 421.34
for Asian ginseng, is ginsenoside Rf. C H3 C30 H 51O4
[MH-(-Glc-Rha)-5H2 O] + = 403.33
475.37
This compound also contains a proto- HO
H3 C C H3
nated molecular ion of m/z 801.5 in O
C12 H21O9 Intens. 143.1
HO
the LC/MS analysis. The fragments of O 309.11 457.4 +MS2(801.5), 28.5min
x10 6
this molecule are typical and can be OH 3 439.3
HO 2
used to distinguish it from ginsenoside
O
1 125.1 421.3
F11. After the cleavage of the diglucose HO O
207.1 309.1 403.3 475.4 603.3
chain at m/z 325.0, a consecutive loss C H3 0
100 200 300 400 500 600 700 800 m/z
of four molecules of water is identified
OH OH
by MS/MS analysis (figure 6). The
mass of the resulting ions at m/z Figure 4
459.4, 441.4, 423.4 and 405.4 is differ- MS/MS of 24(R) pseudoginsenoside FTT typical for American ginseng (Panax quinquefolius).
ent from the pattern obtained for the
molecule F11. The fragment at m/z
143.1 was not observed. To confirm the ride was measured at m/z 325.1134 means of ESI oaTOF as indicated in
identity, a LC/MS oaTOF analysis was with 0.22 ppm mass accuracy. The table figure 3. The detection of the typical
performed. The TOF spectrum shows in figure 6 summarizes the measured ginsenosides Rf and F11 by LC/MS on
the molecular ion at m/z 801.4999 masses and the calculated accuracies either ion trap or ESI oaTOF can be
with a high mass accuracy of 0.16 ppm for all obtained fragments derived from used to determine the species of a gin-
(figure 7). The molecules obtained ginsenoside Rf. Other ginsenosides seng plant and therefore the origin of a
after the loss of water at m/z were detected by ion trap MS/MS pharmaceutical ginseng preparation.
459.3834, 441.3726, 423.3620 and and/or accurate mass This was performed with a black,
405.3515 were confirmed with accura- syrup-like ginseng extract purified for
cies between 0.9 and 1.6 ppm. The measurement with empirical formula pharmaceutical use, which was pur-
mass of the cleaved glucose disaccha- calculation of the molecular ions by chased from a Korean vendor. The

457.3676
3.2e5 +TOF MS: 28.530 min Mass Mass
Measured Calculated
Formula accuracy accuracy
801.4997 mass mass
2.8e5 [mDa] [ppm]

801.4997 801.5000 C42H73O14 -0.35 0.41

2.4e5
143.1065 783.4892 783.4895 C 42H71O13 -0.30 0.34
439.3569
Intensity counts

2.0e5
765.4791 765.4789 C42H69O12 0.20 -0.25

1.6e5 475.3782 475.3787 C30H51O4 -0.50 1.12

457.3676 457.3682 C30H49O3 -0.60 1.25

1.2e5
439.3569 439.3574 C30H47O2 -0.70 1.60

8.0e4 783.4892
421.3464 421.3470 C30H43O1 -0.60 1.52
421.3464
4.0e4 309.1176 309.1186 C12H21O9 -1.00 3.10
309.1176 475.3782 765.4791 143.1065 143.1072 C8H15O2 -0.70 4.90
0.0
150 250 350 450 550 650 750 850
m/z, amu

Figure 5
HR-LC ESI oaTOF from an extract of American ginseng (Panax quinquefolius) 24(R) pseudoginsenoside FTT.

107 5989-5493EN
analysis shows the pattern of gin-
HO
senoides Rb1, Rb2, Rc and Rd typical CH3
CH 3 M(C42H 72O14 )=800.49
for the Asian ginseng sub species OH [M+H]+ = 801.50
Panax ginseng (figure 8). The peaks CH 3 [MH-(-Glc-Glc)-H2O] + = 459.38
for the ginsenosides Re and Rg1 are CH 3 CH 3 [MH-(-Glc-Glc)-2H 2O] + = 441.37
clearly separated by RRHT column CH3 [MH-(-Glc-Glc)-3H 2O] + = 423.36
C30H51O 3
chromatography. In the end the [MH-(-Glc-Glc)-4H 2O] + = 405.35
HO 459.38
appearance of the characteristic gin- H 3 C CH 3
HO O Intens.
senoside Rf supplies the evidence for O C12H21O10 x104 441.4 +MS2(801.5), 27.7min
325.11 3
the Asian ginseng species. It is possi- OH
HO 2
ble to use the ion trap MRM on one HO 1
O
O 423.4 459.4
or more specific masses of typical 325.1 405.4 603.3 621.1 763.2
OH
0
ginsenoides to look only for character- HO 300 350 400 450 500 550 600 650 700 750 800 m/z
istic ginsenoides in a sample6. OH

Figure 6
MS/MS of ginsenoside Rf typical for Asian ginseng (Panax ginseng).

423.3620
4.0e5
3.6e5 Mass Mass
Measured Calculated
Formula accuracy accuracy
3.2e5 mass mass
[mDa] [ppm]
Intensity counts

2.8e5 801.4999 801.5000 -0.10 0.16

C42H73O14
2.4e5 783.4888 783.4895 -0.70 0.85
C 42H71O13
2.0e5 765.4780 765.4789 C42H69O12 -0.9 1.18

1.6e5 765.4780 459.3834 459.3838 C30H51O3 -0.40 0.91

1.2e5 441.3726 441.3726 441.3733 C30H49O2 -0.70 1.48

405.3515 423.3620 423.3627 -0.70 1.63

8.0e4 783.4888 C30H47O1
325.1134 801.4999
405.3515 405.3521 C30H45 -0.60 1.54
4.0e4 459.3834
325.1134 325.1135 C12H21O10 -0.10 0.22
0.0
150 250 350 450 550 650 750 850
m/z, amu

Figure 7
HR-LC ESI TOF of an extract of Asian ginseng (Panax ginseng) ginsenoside Rf .

Conclusion Intens.
x10 7 .. TIC +All MS Rb 1
The Agilent 1200 Series Rapid 1.25 Rb 2 Rd
Resolution LC System together with 1.00
Rc
the Agilent 6330 ion trap and the 0.75 Re
Agilent 6210 TOF are powerful analyti- 0.50 Rg 1 Rf
cal tools for the detection of struc- 0.25
turally complex compounds contained 0
22 24 26 28 30 32 34 36 38 Time [min]
in crude natural product extracts also
found in the purified pharmaceutical
Figure 8
products manufactured thereof. In this Ion trap MS analysis of a purified pharmaceutical ginseng extract from Asian ginseng
application a highly complex root (Panax ginseng) from Korea.

5989-5493EN 108
extract from different ginseng species
was analyzed using an Agilent 1200
Series Rapid Resolution LC/Trap sys-
temn and an Agilent 1200 Series Rapid
Resolution LC/TOF system. With the
resulting highly resolved compounds
of the extract, it was possible to identi-
fy compounds which are typical for the
regional species of ginseng as well as
the different compositions of the com-
mon ginsenosides in ginseng root
extracts and pharmaceutical products.
Complex structures of two species-typ-
ical ginsenosides could be elucidated
by the interpretation of the obtained
MS/MS ion trap data and could be
confirmed by accurate mass measure-
ment using the ESI oaTOF. The
detailed knowledge about the different
fragmentation was used to determine
the regional origin of a pharmaceutical
ginseng extract.
References
1.
Chuang W.-C., Wu H.-K., Sheu S.,
Chiou S.-H., Chang H.-C., Chen Y.-P.,
Planta Med. 61, 459pp., 1995.
2.
Fuzzati N. “Analysis methods of gin-
senosides.”, J. Chrom B1-2, 812, 114-
133, 2004.
3.
Wang X., Sakuma T., Asafu-Adjaye E.,
Shiu G. K. „Determination of ginseno-
sides in plant extracts from Panax gin-
seng and Panax quinquefolius L. by
LC/MS/MS.”, Anal. Chem. 71, 1579-
1584, 1999.
4.
Chan T.W.D., But P.P.H., Cheng S.W.,
Kwok I.M.Y., Lau F.W., Xu H.X.
“Differentiation and authentication of
Panax ginseng, Panax quinquefolius,
and ginseng products by using
HPLC/MS.” Anal. Chem. 72, 1281-
1287, 2000.
6.
“Analysis of a complex natural product
extract from Ginseng - Part II:
Structure elucidation of ginsenosides
by high resolution ion trap LC/MS”
Agilent Application Note, Publication
Number 5989-4705EN, 2006.
7.
“Performance of Agilent 1200 SL LC
system for highest resoution.” Agilent
Application Note Publi-cation number
5989-4489EN, 2005.
8.
“Performance of the Agilent 1200 SL
HPLC System for ultra-fast LC applica-
tions with 2.1-mm i.d. columns.”
Agilent Application Note, Publication
Number 5989-4502EN, 2005.

5.
“Analysis of a complex natural product
extract from Ginseng - Part I: Structure
elucidation of Ginsenosides by rapid
resolution LC – ESI TOF with accurate
mass measurement”, Agilent
Edgar Nägele is Application chemist at
Application Note, Publication number
Agilent Technologies, Waldbronn,
5989-4506EN, 2005.
Germany

109 5989-5493EN
5989-5493EN 110

Agilent Equipment:
1200 Series Rapid Resolution LC system
6210 Time-of-Flight MS
MassHunter Workstation software
Application Area:
Natural product analysis in drug
discovery
Agilent MassHunter – Fast computer aided
analysis of LC/ESI-TOF data from complex
natural product extracts
Part 1: Analysis of Agilent 6210 TOF data with the Molecular
Feature Extractor in MassHunter Workstation software
Application Note Edgar Naegele
Abstract
This Application Note describes:
• Accurate mass measurement with LC/ESI-TOF for the determination of complex
natural compound structures
• Fast computer aided identification and elucidation of chemical structures of new
natural product compounds
• The use of the Agilent 1200 Series Rapid Resolution LC (RRLC) system with rapid
resolution high throughput (RRHT) columns for separation of ingredients in a
complex ginseng root extract, and the use of the Agilent 6210 ESI-TOF mass
spectrometer for accurate molecular mass measurement.
• Processing of TOF data with the Molecular Feature Extractor in the MassHunter
Workstation software, to generate molecular features, which are discrete molec-
ular entities defined by the combination of retention time and mass.
111 5989-5928EN
Introduction
Since prehistoric times, herbal
extracts have been used for the treat-
ment of disease. Their effects on
humans were found by trial and error
over generations. A good example for
the efficiency achieved is traditional
Chinese medicine (TCM). In Western
medicine, drugs of natural origin are
gaining importance due to their poten-
tial. But Western pharmaceutical qual-
ity standards require a deep knowl-
edge about the ingredients in medi-
cine based on natural products. A
famous Asian herb, which has been
used in herbal medicine for more than
5000 years, is the ginseng root (Panax
species). Pharmacological effects of
ginseng that have been reported are,
for example, stimulatory and inhibitory
effects on the central nervous system
(CNS), antistress, antihyperglycemic,
antineoplastic and immunomodulatory
effects1. The main active compounds
of the ginsenosides are triterpene
saponins, of which more than 80 have
been isolated and characterized during
the past years.
A lot of work has been done during
the last 30 years to develop analytical
methods for the analysis of ginseng
extracts and medical formulations. The
method of choice for the analysis of
complex natural product extracts, like
the ginseng root, is high performance
liquid chromatography (HPLC)2. For
the determination of the complex and
similar structures of ginsenosides,
modern LC/MS equipment such as the
LC/ESI-TOF for accurate mass mea-
surement and LC/ ion trap or LC/triple
quadrupole instruments for structure
elucidation by MS/MS and MSn are
currently used3. However the time
consuming bottleneck is the examina-
tion of the acquired MS data to identi-
fy and elucidate chemical structures of
new natural product compounds.
This Application Note demonstrates
5989-5928EN
the use of the Agilent 1200 Series
Rapid Resolution LC (RRLC) system
with rapid resolution high throughput
(RRHT) columns for the separation of
ingredients in a complex ginseng root
extract, in combination with the
Agilent 6210 ESI TOF mass spectrome-
ter for accurate measurement of their
molecular masses. The TOF data
obtained were processed by the
Molecular Feature Extractor (MFE) of
the MassHunter Workstation software
to generate molecular features, which
are discrete molecular entities defined
by the combination of retention time
and mass. Natural product compounds
were identified from this simplified MS
data. The composition differences
between ginseng sub-species is eluci-
dated by software-aided data compari-
son4 and an example for an automated
approach to analyze complex natural
product extracts is given5.
Experimental
Equipment
• Agilent 1200 Series binary pump SL
with degasser. This pump has the
capability to perform high resolving
HPLC analysis on a 1.8 µm particle
size column for best resolution perfor-
mance.
• Agilent 1200 Series high performance
autosampler SL (h-ALS) with thermo-
stat. This autosampler is especially
designed to work together with the
Agilent 1200 Series binary pump SL.
• Agilent 1200 Series thermostatted
column compartment (TCC). This TCC
is ready for use with the binary pump
SL with optional separate heat
exchangers and post column cooling
under optimized delay volume condi-
tions, as well as alternating column
regeneration with an optional
2-position/10-port valve.
• Agilent 1200 Series diode array
detector SL (DAD SL) which can
acquire data at a sampling rate of up
to 80 Hz.
• Agilent 6210 MSD TOF. Orthogonal
112
acceleration time-of-flight (TOF)
mass spectrometer with dual sprayer
interface for mass calibration to
acquire molecular masses with high-
est accuracy. This TOF mass spec-
trometer is capable of acquiring data
at 40 Hz and with positive/negative
switching.
• Column: ZORBAX SB C18,
2.1 x 150 mm, 1.8 µm particle size.
• Software: TOF instrument control
software MassHunter Workstation
A02.00 for data acquisition, Analyst
software for data review and
Molecular Feature Extractor (MFE)
software for data processing.
In this LC/TOF instrument set-up, the
binary pump SL, which is in the low
delay volume configuration, is con-
nected to the h-ALS SL with a 0.12-mm
ID stainless steel capillary. To reduce
delay volume, the seat capillary in the
h-ALS SL has a 0.12 mm ID. The same
kind of capillary connects to the low
delay volume (1.6 µL) heat exchanger
in the TCC, which is connected to the
column. For UV detection, a 2-µL cell is
built into the DAD SL. The outgoing
capillary is directly connected to the
sprayer in the electrospray source of
the TOF mass spectrometer. This
instrument set-up is optimized to
achieve the highest possible resolution.
Sample preparation
Powdered freeze-dried Asian and
American ginseng root (1g) (Panax
ginseng and Panax quinquefolius) was
ultrasonically treated for 30 minutes in
10 mL methanol, filtrated and directly
used for analysis.
Method
• The Agilent 1200 Series binary pump
SL was operated under the following
conditions:
 Solvent A: Water + 0.1 % TFA;
Solvent B: AcN + 0.1 % TFA. 1.8e7
Flow: 0.5 mL/min. 1.6e7
Gradient: 0 min 5 % B, 1.4e7
1 min 5 % B, 60 min

Intensity, cps
85 % B, 61 min 95 % B, 1.2e7
70 min 95 % B. 1.0e7
,
Stop time: 70 min. 8.0e6
Post time: 15 min.
• The Agilent 1200 Series high perfor- 6.0e6
mance autosampler SL was used for 4.0e6
injections of 10 µL sample and the 2.0e6
samples were cooled to 10 ºC. The
0.0
sample loop was switched to bypass 22.0 24.0 26.0 28.0 30.0 32.0 34.0 36.0 38.0
after 1 minute to reduce delay volume. Time, min
• The Agilent 1200 Series thermostated Figure 1
column compartment was adjusted High resolution LC on a 1.8 µm particle column with an extract from Asian ginseng root (Panax
to 50 ºC and equipped with the low ginseng) for MS analysis by ESI-TOF.
delay volume heat exchanger.
• The Agilent 1200 Series DAD SL was A
operated at 80 Hz for data acquisition 2.5
Processed TIC
at a wavelength of 220 nm/4, ref.
Intensity (106)

2.0
360/100 with the 2-µL flow cell, 3-
mm path length. 1.5
• The 6210 TOF MS was operated 1.0
under the following conditions:
0.5
Source: ESI in positive mode
with dual spray for 0.0
reference mass. 15 20 25 30 35
Dry gas: 12 L/min B min
Dry Temp.: 200 ºC 800
Processed
Nebulizer: 35 psi 700
Intensity (103)

EIC:Group #7 (RT=26.312) --- 20 Features

Scan: 200-1300.
Fragmentor: 150 V or 300 V for
CID
Skimmer: 60 V
Capillary: 3000 V
Results and discussion
The ingredients of a ginseng root
extract from an Asian ginseng root
(Panax ginseng) were separated with
the Agilent 1200 Series RRLC system
on a RRHT column (1.8-µm particle
size) with subsequent ESI TOF mass
spectrometry (Agilent 6210 TOF). The
high resolution LC provided an excel-
lent separation of the major and minor
ingredients of the natural product
extract (figure 1). The acquired TOF
data were extracted by the MFE soft-
600
500
400
300
200
100
0
15 20 25 30 35
min
Figure 2
A) Software processed TIC of Asian ginseng extract and grouped molecular features.
B) Feature group # 7 at retention time 26.3 min groups 20 molecular features.
ware and the identified molecular ions processed data, the feature group #7
of the comprised compounds were at the retention time of 26.3 min will
separated from undefined background be examined in more detail to identify
ions (figure 2). In this process, the these compounds in the ginseng
identified ions were clustered to mole- extract. At this retention time, 20 dif-
cular features comprising isotope com- ferent features, molecular ions togeth-
pounds and adducts. From the er with their different adducts and iso-

113 5989-5928EN
 A C B D

Figure 3
2D contour plot of TOF spectrum of ginseng extract, retention time vs. mass. A) Unprocessed 2D contour plot at retention time 24.8 – 28.0 min and
mass 900 – 1250. B) Magnified contour plot around the unprocessed ginsenoside Rb1 ions. C and D) Contour plot of processed data of the same mass
and retention time area.

Original RT = (26.266, 26.358)

topes, are grouped together. The
processed EIC of feature group #7 at a 700
retention time of 26.3 min is shown in 600 A
figure 2B. The capability to extract 500
Intensity (103)

groups of ions as molecular features 400

can be optically displayed by a detailed 300
2D contour plot of the spectrum, which 200
shows retention time and the mass of 100
the ions (figure 3). The unprocessed 0
chromatograms of the protonated gin- m/z 200 400 600 800 1000 1200
senoside Rb1 ion at m/z 1109.6 and Processed Group #7 (RT=26.312) --- 20 Features
the sodiated ion at m/z 1131.4, both at 425.3775 1131.5926
700 B 767.4935
a retention time 26.3 min, are showing
600
a lot of unspecific chemical noise ions
Intensity (103)

500
(figures 3A and 3C). These unspecific
400 785.5045
ions can not be grouped to a molecular
feature because of an unclear isotopic 300
distribution and are eliminated during 200 1109.6104
data processing. The ions remaining at 100
the same retention time window are 0 200
grouped together to molecular features m/z 400 600 800 1000 1200
comprised of adducts and isotopes of
the basic ion (figures 3B and 3D). The Figure 4
A) Unprocessed mass spectrum at 26.3 min. B) Processed mass spectrum at 26.3 min showing
magnified window (figures 3C and 3D) molecular features, which belong to the same group.
clearly demonstrates the simplification
of the data. The data enhancement can The significant simplification of the fragments of the molecular ion are
also be demonstrated by the compari- mass spectra resulting from the gener- generated by collision induced decay
son of the original mass spectrum of ation of molecular features and elimi- (CID) for structure elucidation (figure
ginsenoside Rb1 (figure 4A) to the nation of chemical noise allows an 4B). These fragments, which elute at
processed mass spectrum (figure 4B). easier interpretation, especially when the same time, are included in the

5989-5928EN 114
same group of molecular features
together with the molecular ion. The HO O
O
interpretation of the spectrum from O 425.3775
ginsenoside Rb1 for structure elucida- OH O
OH CH3
tion is shown in figure 5. The protonat- HO
HO CH 3
ed molecular ion was found at m/z
OH OH OH
1109.6104 and the sodium adduct at
m/z 1131.5926. The fragment at m/z CH3
785.5045 is derived from the loss of
CH3 CH3
one glucose moiety. An additional loss
of a molecule of water results in the
CH 3
fragment at m/z 767.4935. The loss of O
the second glucose moiety yielded the HO
O
ion at m/z 425.3775 from the ter- H 3C CH3
penoide core structure. The identified OH
molecular features for ginsenoside Rb1 HO C 42 H 73 O 13 = 785.5045
and its CID fragments are summarized in HO
785.5045
O -H 2 O
table 1. The protonated, sodiated and O
potasiated adducts are found with high C 42 H 71 O 12 = 767.4935
OH
mass accuracies of -0.14, -0.41 and -
HO C 30 H 49 O = 425.3775
0.32 ppm, respectively. The calculated
mass accuracy of the molecule is -0.41 OH
ppm. In addition, the mass accuracies
for the obtained CID fragments and Figure 5
Structure of ginsenoside Rb1 obtained by interpretation of the processed mass spectrum with
their adducts are calculated (table 1). molecular features included.
This confirms the empirical formula
and the proposed molecular structure.
The content of a natural product
extract often depends on the sub-
Processed TIC
species, the growth region and growth 4
conditions. To decide if a natural plant
extract contains the right active ingre- 3
Intensity (106)

F 11
dients in proper concentration, a com-
2
parison to a standard extract is neces-
sary. For ginseng the different sub-
1
species contain common ginsenosides
as well as specific ginsenosides. A
0
typical example is Asian ginseng
20 25 30 35 min
(Panax ginseng) (figure 2) in compari-
son to American ginseng (Panax quin-
quefolius). The simplification achieved
Figure 6
by the generation of molecular fea- Software processed TIC of American ginseng extract and grouped molecular features.
tures unravels the differences (figure
6). Especially the high content of gin-
senoside F11 eluting at 24.5 minutes
and the absence of ginsenoside Rf
(eluting at 24 minutes in figure 2) is
typical for the American ginseng sub-
species5.

115 5989-5928EN
Species RT m/z Mass Accuracy RT m/z Mass Accuracy RT m/z Mass Accuracy RT m/z Mass Accuracy
M+H 26.31 1109.6104 26.31 785.5045 26.31 767.4935 26.31 425.3775
M+H+1 26.31 1110.6152 26.31 786.5081 26.31 768.4976 26.31 426.3813
M+H+2 26.31 1111.6176 1108.6031 - 0.14 ppm 26.31 787.5108 784.4972 0.10 ppm 26.31 769.5004 766.4862 0.68 ppm 26.31 427.3850 424.3702 0.75 ppm
M+H+3 26.31 1112.6219 26.31 788.5134 26.31 770.5027 26.31 428.3882
M+H+4 ----- ----- ----- ----- 26.31 771.4965 ----- -----

M+Na 26.31 1131.5926 26.31 807.4486 26.31 789.4756 ----- -----

M+Na+1 26.31 1132.5964 26.31 808.4902 26.31 790.4809 ----- -----

M+Na+2 26.31 1133.599 1108.6034 - 0.41 ppm 26.31 809.4950 784.4976 0.39 ppm 26.30 791.4883 766.4864 0.42 ppm ----- ----- ----- -----

M+Na+3 26.31 1134.6021 26.30 810.4938 ----- ----- ----- -----

M+Na+4 26.31 1135.5998 ----- ----- ----- ----- ----- -----

M+K 26.30 1147.5664 1108.6033 - 0.32 ppm ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- -----

M 26.31 1108.6034 - 0.41 ppm 26.31 784.4972 + 0.10 ppm 26.31 766.4862 + 0.7 ppm 26.31 424.3702 + 0.75 ppm

Table 1
Identified molecular features and calculated mass accuracies for empirical formula confirmation. (Ginsenoside Rb1, CXWH&UOUV , M 1108.6029;
Fragment CWUHZUOTV , M 784.4973; Fragment CWUHZSOTU, M 766.4867; Fragment CV SHW&O, M 424,3705).

Conclusion
The ingredients of highly complex nat-
ural product samples can be accurate-
ly separated by the Agilent 1200 Series
RRLC system with 1.8 µm particle size
columns. By connecting the ESI-TOF
MS, a large amount of accurate mass
data for empirical formula calculation
and compound structure elucidation
can be created. This data is processed
by the Molecular Feature Extractor of
the MassHunter Workstation soft-
ware, which generates molecular fea-
tures, including the molecular ion, its
isotopes and adducts, to simplify the
large amount of data and to help with
data interpretation. The highly accu-
rate measured masses of the different
molecular species of molecular ions
and CID fragments, included in the
molecular features, are used for empir-
ical formula calculation and structure
elucidation.
References
1.
Attele A.S., Wu J.A., Yuan C.S.,
Biochem. Pharmacol. 58, 1685-1693,
1999.
2.
Fuzzati N., “Analysis methods of gin-
senosides.” J. Chrom. B1-2, 812, 114-
133, 2004.
3.
Wang X., Sakuma T., Asafu-Adjaye E.,
Shiu G. K., „Determination of ginseno-
sides in plant extracts from Panax gin-
seng and Panax quinquefolius L. by
LC/MS/MS.” Anal. Chem. 71, 1579-
1584, 1999.
4.
Edgar Naegele: “Agilent Mass- Hunter
– Fast computer-aided analysis of
LC/ESI-TOF data from complex natural
product extracts
Part 2: Comparison of Agilent 6210
TOF data from different biological ori-
gin with Mass Profiler in MassHunter
Workstation software“, Agilent
Application Note, Publication number
5989-6076EN, 2007.
5.
Edgar Naegele: “Agilent Mass- Hunter
–Fast computer aided analysis of
LC/ESI-TOF data from complex natural
product extracts Part 3: Automated
analysis of Agilent 6210 TOF data from
complex natural product extracts”,
Agilent Application Note, Publication
number 5989-6077EN, 2007.
Edgar Naegele is Application Chemist
at Agilent Technologies, Waldbronn,
Germany.

5989-5928EN 116
 Agilent MassHunter – Fast, computer-aided
analysis of LC/ESI-TOF data from complex
natural product extracts
Part 2: Comparison of Agilent 6210 TOF data from different
biological origin with the Mass Profiler in MassHunter software

Application Note Edgar Naegele

Abstract

This Application Note describes:

• Fast computer-aided identification of differences in complex natural product
extracts
• Accurate mass measurement with LC/ESI-TOF for the determination of complex
natural compound structures
• The use of the Agilent 1200 Series Rapid Resolution LC (RRLC) system with
Rapid Resolution High Throughput (RRHT) columns for separation of ingredi-
Agilent Equipment:
1200 Series Rapid Resolution LC system ents in a complex ginseng root extract, and the use of the Agilent 6210 ESI-TOF
6210 Time-of-Flight MS mass spectrometer for accurate molecular mass measurement
MassHunter Workstation software
• Processing of TOF data with the Mass Profiler of the Agilent MassHunter
Application Area: Workstation software to identify statistically significant differences in a set of
Analysis of complex natural products in samples from two different ginseng subspecies
drug discovery
• Identification of the different compounds by empirical formula calculations
based on highly accurate mass measurement

117 5989-6076EN
Introduction
Herbal extracts have been used since
prehistoric times for the treatment of
disease. The effects of these extracts
on humans were determined by simple
trial and error over generations. A
good example for the efficiency
achieved is Traditional Chinese
Medicine (TCM). In Western medicine,
drugs of natural origin are gaining
importance due to their therapeutic
potential. However, Western pharma-
ceutical quality standards require a
deeper know-ledge of the ingredients
in medicines based on natural prod-
ucts. A famous Asian herb, which has
been used in herbal medicine for more
than 5000 years, is the Asian ginseng
root (Panax ginseng).
Apart from the Asian subspecies,
other subspecies of the ginseng plant
are available such as the American
subspecies (Panax quinquefolius). The
pharmaceutically active compounds of
each subspecies have different pat-
terns of occurrence and concentration1.
Therefore, it is important to know the
composition of the plant extract,
depending on its biological origin, to
anticipate its medical activity before
use. The main active compounds of
ginsenosides are triterpene saponins,
of which more than 80 have been iso-
lated and characterized during the
past years.
Much work has been done during the
last 30 years to develop analytical
methods for the analysis of ginseng
extracts and medical formulations. The
method of choice for the analysis of
complex natural product extracts such
as the ginseng root is high-perfor-
mance liquid chromatography (HPLC)2.
For determination of the complex and
similar structures of ginsenosides,
modern LC/MS equipment is currently
5989-6076EN
used such as LC/ESI-TOF for accurate
mass measurement and LC/ion trap or
LC/triple quadrupole systems for
structural elucidation by MS/MS and
MSn3. However, the time-consuming
bottleneck remains the examination of
the acquired MS data.
This Application Note describes the
analysis of TOF data with the Mass
Profiler of the Agilent MassHunter
Workstation software for fast and easy
identification of differences in the
extracts from different ginseng sub-
species. This note also describes the
use of the Agilent 1200 Series Rapid
Resolution LC (RRLC) system with
Rapid Resolution High Throughput
(RRHT) columns for the separation of
the ingredients in complex ginseng
root extracts together with the Agilent
6210 ESI-TOF mass spectrometer for
accurate measurement of their molec-
ular masses.
Processing of the TOF data with the
Molecular Feature Extractor (MFE) of
the Agilent MassHunter Workstation
software to generate molecular fea-
tures for the identification of natural
product compounds is described in
part 1 of this study4. An example for an
automated approach to analyze com-
plex natural product extracts is
described in part 3 of this study5.
Experimental
Equipment
• Agilent 1200 Series binary pump SL
with degasser. This pump is capable
of performing high resolution HPLC
analysis on 1.8 µm particle size
columns for best resolution perfor-
mance.
• Agilent 1200 Series high performance
autosampler SL with thermostat. This
autosampler is designed specifically
to be used with the binary pump SL.
118
• Agilent 1200 Series thermostatted
column compartment. This column
compartment is supplied ready for
use with the binary pump SL. Also
available are heat exchangers for
post-column cooling under optimized
delay volume conditions, as well as a
2-position/10-port valve for alternat-
ing column regeneration.
• Agilent 1200 Series diode array
detector SL. This detector is capable
of acquiring data at a sampling rate
up to 80 Hz.
• Agilent 6210 Time-of-Flight mass
spectrometer. This orthogonal accel-
eration TOF MS has a dual sprayer
interface for mass calibration and
acquisition of molecular masses with
highest accuracy, and is capable of
acquiring data at 40 Hz and with pos-
itive/negative switching
• Column: ZORBAX SB C18,
2.1 x 150 mm, 1.8 µm particles.
• Software: TOF instrument control
software Agilent MassHunter
Workstation revision A.02.00 for data
acquisition, Agilent Analyst Software
for data review, and Agilent Mass
Profiler software for data processing.
In this LC/TOF system setup, the bina-
ry pump SL is configured for low-delay
volumes and connected to the high
performance autosampler SL with a
0.17 mm ID stainless steel capillary. To
reduce delay volume, the seat capillary
in the high performance autosampler
SL has an ID of 0.12 mm. The same
type of capillary is used to connect the
low delay volume (1.6 µL) heat
exchanger, which is connected to the
column, in the thermostatted column
compartment. For UV detection, a 2 µL
cell is built into the diode array detec-
tor SL. The outgoing capillary is con-
nected directly to the sprayer in the
electrospray source of the mass spec-
trometer. This instrument setup is opti-
mized to achieve the highest possible
resolution.
Sample preparation
Powdered freeze-dried Asian and
American ginseng root (1 g each) was
treated ultrasonically for 30 minutes in
10 mL methanol, filtered and directly
used for analysis.
Method
• The Agilent 1200 Series binary pump
SL was operated under the following
conditions:
Solvent A: Water + 0.1 % TFA
Solvent B: ACN + 0.1 % TFA
Flow: 0.5 mL/min
Gradient: 0 min, 5 %B
1 min, 5 %B
60 min, 85 %B
61 min, 95 %B
70 min, 95 %B
Stop time: 70 min
Post time: 15 min
• The Agilent 1200 Series high
performance autosampler SL was
used to make sample injections of
10 µL sample and the samples were
cooled to 10 °C. The sample loop was
switched to bypass after one minute
Fragmentor: 150 V ural product extracts (figure 1). For the
Skimmer: 60 V data analysis with Mass Profiler, five
Capillary: 3000 V repeated injections of each extract
sample were measured by the same
Results and discussion LC/MS TOF method as described in
The ingredients of the Asian and the experimental section. The acquired
American ginseng root extracts were TOF data were extracted by the
separated with the Agilent 1200 Series Molecular Feature Extractor software
RRLC system using an RRHT column as described4. In this process, the
(1.8 µm particle size) with subsequent identified ions were clustered to mole-
ESI-TOF mass spectrometry (Agilent cular features comprising isotope
6210 TOF). The high resolution LC compounds and adducts. The files
facilitated excellent separation of the obtained were grouped according to
major and minor ingredients of the nat- the biological origin of the measured
(A) TIC of TOF-MS from Asian ginseng
Intensity [cps] Rb 1
Rb 2
3.5e6 Re Rc Rd
3.0e6
Rg 1 Rf
2.5e6
2.0e6
1.5e6
1.0e6
5.0e5

to reduce delay volume. 0.0

8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0 20.0 21.0

• The Agilent 1200 Series thermostated

Intensity [cps] (B) TIC of TOF-MS from American ginseng
column compartment was adjusted F11
11
4.0e6
4.0e6 (B) TIC of TOF
-
to 50 °C and equipped with the low Rb11
delay volume heat exchanger. Re Rd
3.5e6
3.5e6
Rb22
• The Agilent 1200 Series DAD SL was 3.0e6
3.0e6
operated at 80 Hz for data acquisition
2.5e6
2.5e6
at a wavelength of 220 nm/4 nm, ref- Rg1
1
erence 360 nm/100 nm with the 2 µL 2.0e6
2.0e6
flow cell, 3 mm path length flow cell. Rc
1.5e6
1.5e6

• The Agilent 6210 TOF MS was oper-

1.0e6
1.0e6
ated under the following conditions:
Source: ESI in positive 5.0e5
5.0e5

mode with dual

0.0
spray for 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0 20.0 21.0
reference mass Time [min]
Dry gas: 12 L/min
Dry temp.: 200 °C Figure 1
Nebulizer: 35 psi TIC of TOF-MS from an Asian (A) and an American (B) ginseng root extract between 8 and 21
minutes showing the main ginseosides.
Scan: 200-1300.

119 5989-6076EN
extract samples into two respective 1108.6029 and RT = 11.90) were displayed in a logarithmic (log2) plot
groups for Asian and American gin- enlarged. The standard deviations for showing the abundance ratio of Asian
seng and loaded into the Mass Profiler mass and retention time, the relative and American ginseng extracts (
software. All 671 identified molecular standard deviation of the abundance in figure 3). In the plot there are five lines
features were displayed in a plot of each group as well as the high relative for selected levels of abundance differ-
mass against the retention time plot in mass accuracy of 1.38 ppm for this ence in the two sample groups.
the Mass Profiler software to inspect particular compound demonstrate the Molecular features lying on the line in
the quality of the data (figure 2). The high quality of the data. the middle (1x) are equal in both
molecular features for the known gin- For the differential analysis of both groups, molecular features within the
senoside Rb1 (C54H92O23 at M = groups, the features of each group are 2x margins are up to twice the abun-

Mass vs. Retention Time

Mass [MU]

1200

HO O
Mass accuracy = 1.38 ppm O O

1000 HO
OH

OH
HO
OH

OH OH
O
CH3
CH3

CH3

CH3 CH3

CH3
HO O
O

800 Retention time SD = 0.006 HO

OH
H3C CH3

HO O
O

Mass SD = 0.0004 HO
OH
Mass

OH

Abundance RSD = 0.23

600

400

200

0
0 5 10 15 20 25 30
Retention Time (min) Retention time [min]

Figure 2
Plot of mass against retention time shows all 671 molecular features found in the Asian (red) and American (blue) ginseng extracts and molecular
features of ginsenoside RbT (CXWH&UOUV at M = 1108.6029 and RT = 11.90).

26
Log2 Abundance of American ginseng

24 Pseudoginsenoside F11
OH
CH3
H3C

22 O
OH
H3C

20 CH3 CH3

CH3

HO

18 H3C CH3
HO
O
O

Ginseoside Rc
OH

16
HO
O O
O O
HO O OH
CH 3 HO OH O
CH 3
OH HO CH 3
OH OH
14 OH OH +4X
CH3

+2X CH3 CH3

CH3

12 1X
HO
O
O

H3C CH3
-2X OH
HO

10 -4X HO
O
O

OH
HO
OH
8
10 12 14 16 18 20 22 24
Log2 Abundance of Asian ginseng
Figure 3
Differential analysis of Asian and American ginseng, showing pseudoginseoside FTT exclusively in American ginseng and higher concentration of
ginsenoside Rc in Asian ginseng.

5989-6076EN 120
dance in one group and within the 4x
margins up to four-fold. Beyond these
margins a feature is nearly unique or
exclusively present in one group. An
example for a compound which has a
higher concentration in the Asian gin-
seng sample group is the ginsenoside
Rc (figures 1 and 3). The comparison
of the abundances in each sample of
the two groups clearly shows a signifi-
cant eight-fold higher occurrence of
ginsenoside Rc in Asian ginseng (table
1). The calculated relative mass errors
are in the low single digit ppm range.
The average for the Asian ginseng
sample group is 1.71 ppm and the
average for the American ginseng
sample group is 0.90 ppm. The molecu-
lar feature, which is easily recogniz-
able as almost exclusively present in
the American ginseng samples, is the
special the compound pseudoginseno-
side F11 (figure 3). This compound has
more than 100-fold higher abundance
in the samples from the American gin-
seng root extract (table 2). For this
compound, the average relative mass
accuracy for the Asian ginseng sample
group is 1.71 ppm and the average for
the American ginseng sample group is
1.26 ppm.
Conclusion
The ingredients of highly complex nat-

ID Name RT Mass Abundance Mass error [mDa] Rel. mass error [ppm] Av. rel. mass error [ppm]
1 American Ginseng_1 12.550 1078.5909 143,961 -1.50 1.36
2 American Ginseng_2 12.505 1078.5912 243,837 -1.20 1.09
3 American Ginseng_3 12.557 1078.5921 240,121 -0.30 0.25 0.99
4 American Ginseng_4 12.491 1078.5919 266,422 -0.50 0.44
5 American Ginseng_5 12.493 1078.5904 252,872 -2.00 1.83
6 Asian Ginseng_1 12.538 1078.5905 2,092,954 -1.90 1.73
7 Asian Ginseng_2 12.534 1078.5912 2,023,553 -1.20 1.09
8 Asian Ginseng_3 12.545 1078.5898 2,058,617 -2.60 2.38 1.71
9 Asian Ginseng_4 12.545 1078.5906 2,052,396 -1.80 1.64
10 Asian Ginseng_5 12.534 1078.5905 2,042,638 -1.90 1.73

Table 1
Retention times, abundancies and mass accuracies of ginsenoside Rc (CXV H&SOUU at M = 1078.5924) in Asian and American ginseng.

ID Name RT Mass Abundance Mass error [mDa] Rel. mass error [ppm] Av. rel. mass error [ppm]
1 American Ginseng_1 11.328 800.4900 2,835,414 -2.20 2.70
2 American Ginseng_2 11.347 800.4910 2,836,847 -1.20 1.50
3 American Ginseng_3 11.337 800.4924 2,815,218 0.20 -0.24 1.53
4 American Ginseng_4 11.333 800.4909 2,797,743 -1.30 1.60
5 American Ginseng_5 11.335 800.4900 2,864,285 -2.20 1.60
6 Asian Ginseng_1 11.337 800.4914 23,132 -0.80 1.00
7 Asian Ginseng_2 11.337 800.4930 22,875 0.80 -1.00
8 Asian Ginseng_3 11.344 800.4927 20,865 0.50 -0.60 1.26
9 Asian Ginseng_4 11.346 800.4942 23,578 2.00 -2.48
10 Asian Ginseng_5 11.342 800.4912 22,556 -1.00 1.25

Table 2
Retention times, abundancies and mass accuracies for Pseudoginsenoside FTT (CWUHZUOTW at M = 800.4922), present nearly exclusively in American
ginseng samples.

121 5989-6076EN
ural products can be separated with
very low standard deviations of reten-
tion times using the Agilent 1200
Series RRLC system and 1.8 µm parti-
cle size columns. Connection to the
Agilent 6210 ESI-TOF MS facilitates
acquisition of highly accurate and
repeatable mass data. With these sys-
tem prerequisites, the data can be
processed by the Mass Profiler soft-
ware for statistical evaluation of the
differences in the abundance of the
molecular features in the different
sample groups. This facilitates identi-
fication of differences in concentra-
tion and abundance of single com-
pounds in very complex samples such
as natural product extracts.
5989-6076EN
References
1.
Chuang W.-C., Wu H.-K., Sheu S.,
Chiou S.-H., Chang H.-C., Chen Y.-P.,
Planta Med. 61, 459 pp, 1995.
2.
Fuzzati N., “Analysis methods of gin-
senosides.” J. Chrom. B1-2, 812, 114-
133, 2004.
3.
Wang X., Sakuma T., Asafu-Adjaye E.,
Shiu G. K., “Determination of ginseno-
sides in plant extracts from Panax gin-
seng and Panax quinquefolius L. by
LC/MS/MS.” Anal. Chem. 71, 1579-
1584, 1999.
4.
Edgar Naegele: “Agilent Mass- Hunter –
Fast computer-aided analysis of
LC/ESI-TOF data from complex natural
product extracts – Part 1:Analysis of
Agilent 6210 TOF data with the
Molecular Feature Extractor in
MassHunter Workstation software“,
Agilent Application Note, Publication
number 5989-5928EN, 2007.
5.
Edgar Naegele: “Agilent Mass- Hunter –
Fast computer aided analysis of
LC/ESI-TOF data from complex natural
product extracts – Part 3: Automated
analysis of Agilent 6210 TOF data from
complex natural product extracts”,
Edgar Naegele is Application Chemist
Agilent Application Note, Publication
at Agilent Technologies, Waldbronn,
number 5989-6077EN, 2007.
Germany.
122
Metabolic Profiling

123
124
124
 An interwoven, multi-algorithm approach
for computer-assisted identification of drug
metabolites
Rapid identification of drug metabolites from accurate QTOF MS
and MS/MS data by Agilent MassHunter metabolite identification
software

Application Note Edgar Nägele, Frank Wolf,

Uwe Nassal, Rainer Jäger,
Horst Lehmann, Frank Kuhlmann,
Karina Subramanian

Abstract

This Application Note is based on scientific poster MOP 253 presented at the

Agilent Equipment:
ASMS conference of mass spectrometry, June 4, 2007, in Indianapolis, Indiana,
1200 Series Rapid Resolution LC system USA. The note describes:
6510 quadrupole time-of-flight LC/MS
• Rapid, computer-assisted identification of drug metabolites using a multi-algo-
ZORBAX RRLC column
MassHunter metabolite identification rithm approach.
software
• High resolution chromatographic separation of drug metabolites from an in
Application Area: vitro experiment.
Metabolite identification in drug discovery
and development
• High mass accuracy measurement of drug metabolites by QTOF MS and
MS/MS.

125 5989-7375EN
Introduction
In modern pharmaceutical drug dis-
covery it is of crucial importance to
identify all possible metabolites of a
new chemical entity because of possi-
ble toxic effects on humans and to
evaluate its potential as new drug sub-
stance. Today, high resolution and high
mass accuracy QTOF MS and MS/MS
data1, which are acquired from in vitro
as well as in vivo metabolism experi-
ments, are used for metabolite identifi-
cation in the different stages of the
drug discovery and development
process. To make use of all potential
information contained in such data, it
is essential to use different and com-
plementary computer algorithms for
data analysis. Each of these algo-
rithms provides an individual result
based on the specific functionality and
the analyzed part of the data. The real
advantage of computer-assisted data
analysis comes into being when these
algorithms work together in an inter-
woven fashion and contribute to a
overall result of higher confidence. To
do so, all scores produced by the indi-
vidual algorithms are combined and
weighted with user-settable factors.
This Application Note describes an
example for the analysis of QTOF data
from a metabolite identification experi-
ment using the drug compound
Nefazodone2 by means of a the
Agilent MassHunter metabolite identi-
fication software which uses the
described approach.
5989-7375EN
Experimental
Equipment
• Agilent 1200 Series Rapid Resolution
LC system, including degasser, binary
pump SL, high performance autosam-
pler SL with thermostat, thermostat-
ted column compartment and diode
array detector SL
• Agilent ZORBAX SB-C18 column,
2.1 x 150 mm, 1.8 µm particle size
• Agilent 6510 quadrupole time-of-
flight LC/MS system
Sample preparation
• Stock solutions:
20 mg/mL S9 liver homogenate
preparation 0.1 mg/mL nefazodone in
water 1.6 mg NADP in 1.6 mL
0.1 M phosphate buffer at pH 7.4
50 mM isocitrate/MgCl2 (203 mg
MgCl2.6H2O + 258.1 mg isocitrate in
20 mL water) Isocitrate dehydrogenase
0.33 U/mL
Figure 1
System configuration for the metabolite identification experiment, comprising the Agilent 1200
Series Rapid Resolution LC with 1.8 µm particle size column and Agilent 6510 QTOF LC/MS.
126
• NADPH regeneration system:
1.6 mL NADP solution + 1.6 mL isoci-
trate solution + 100 µL isocitrate
dehydrogenase solution
• Incubation mixture:
3.85 µL substrate + 200 µL NADPH
regeneration system + 746.15 µL
phosphate buffer + and 50 µL S9 liver
homogenate
Incubation was carried out at 37 °C for
60 minutes, a 100 µL aliquot was taken
at 0 and 60 min. The reaction was
stopped by adding 6 µL perchloric acid
and 100 µL acetonitrile to the aliquots
followed by centrifugation for 15 min
at 14.000 g. The supernatant was
evaporated to dryness using a
SpeedVac concentrator and reconsti-
tuted with water containing 0.1 %
formic acid (FA) for LC/MS analysis as
described below. Incubations stopped
at 0 min were used as controls.
High resolution LC/MS method
• Agilent 1200 Series binary pump SL
Solvent A: Water + 0.1 % FA,
Solvent B: ACN + 0.1 % FA
Flow rate: 0.5 mL/min
Gradient: 0 min, 5 %B;
15 min, 75 %B;
15.1 min, 95 %B;
16 min, 95 %B
Stop time: 16 min
Post time: 10 min
• Agilent 1200 Series autosampler SL
Injection volumes:
1-10 µL with needle
wash
Sample temperature: 4 °C
Automated delay volume
Cl
reduction O
• Agilent 1200 Series diode array O
N
detector SL Detection wavelength: N
N
210 nm, (± 4 nm) N N
CH 3 Nefazodone
Reference wavelength:
360 nm (± 16 nm)
Flow cell: 2 µL volume, Figure 2
3 mm path length Setup of metabolite relevance score for each individual algorithm and weighted overall identification
relevance score threshold.
• Agilent 1200 Series thermostatted
column compartment Data analysis which are new or increased in the
Column temperature: 60 °C The first step in the analysis of the metabolized sample are considered
data comprises the introduction of the potential metabolites and are subject-
QTOF MS and MS/MS method compound structure formula and cal- ed to further analysis by different algo-
• Agilent 6510 QTOF LC/MS system culation of the molecular weight fol- rithms, which can be specified by the
Source: ESI, positive mode lowed by the comparison between the user (figure 2). The algorithms can
with dual spray for data file that contains the metabolite identify and qualify new metabolites or
reference mass compounds (sample, incubation time can just qualify metabolites found by
solution t>0) and the data file that contains another algorithm. The results of all
Dry gas: 12.0 L/min (m/z 121.005 only the parent drug (control, incuba- metabolite identification algorithms
and m/z 922.00) tion time t=0). In this comparison, all are weighted and combined into a final
Dry temperature: 300 °C detectable mass signals are extracted identification relevance score.
Nebulizer pressure: 60 psi from the MS level data using the Metabolites are qualified when their
Mass range: 100-1000 Molecular Feature Extraction (MFE) final score is above a defined rele-
Fragmentor voltage: 200 V algorithm. Related compound isotope vance threshold. The results from all
Skimmer: 60 V masses and adduct masses are algorithms are collected in a results
Capillary voltage: 4000 V grouped together into discrete molecu- table and can be inspected at-a-
Collision energy: 35 V lar features, and chemical noise is glance.
Data dependent MS/MS: 2 com- removed. The compound lists of the
pounds, 2 MS/MS spectra, metabolized sample and the control
exclusion for 0.03 min are then compared. All compounds

127 5989-7375EN
 Compound Isotopic Fragment EIC of RAD UV Built-in Molecular Mass Metabolite Molecular
Correlation Pattern Pattern expected Chrom. Chrom. Biotransf. Formula defect Prediction structure
Matching Matching masses assignment filter elucidation
CH3
18
13
H
O

MS MS/MS 401 O
O N
O

S
N
O
N S
OO
N 11 12 17
OO
CH39 H
–C2H6+O
N O
N S N
O OO
F O
N F O
N
1 19 14 16
N F O F
2 1018 8 15
H
F O m/z 597 F O
F
F
H H
F
m/z 584
m/z 570 13
- CH2 3 5 7
+ O2 11 412 17
+O2
O
O
6
O N O

MFE
N S

83 N S
N
N O OO N
CH39 H
MFE C12 H14 N9 Cl F
F O
O
N
m/z 611
OO

F
F O

F
N

m/z 568, major F O

O
N
N S
OO

2
1 19
10 8
14
15
16

C13 H23 N3 P S Cl
m/z 554, major
H H
F
F

+HF
F

- CH2
3 5 7
C14 H24 N2 S2 Cl
- CH2
O
O
O
- CH2
N S
N
O
N
O 4 6
C14 H22 N O3 S Cl N O OO N
N S
129 N S N
OO

–H2O
O OO N O
F O
CH3 CH3
…
N N

C14 H16 N6 O Cl
F O F F O

Cpd Cpd F
F
m/z 625, major F Parent, [M+H]+ m/z 639
+ O2
F
F
m/z 540 18
13
H
21
20
H H
N
C15 H25 O P S Cl
+ O2

O
11 12 17 22 28 26
List List +O6+S2 C16 H19 N3 P Cl
O
N S
N
O
N

…
O
OO
H
N N O
N S N
O
O OO N
9 14 16 23 25
256 C17 H20 N2 S Cl F O
N
F
F O
N
F
F O

F O 8 15 24 CH3
F
F F
H 27

… C18 H21 O P Cl
O
m/z 341, major
m/z 641 m/z 655, 2 in rat species

CH3
7

Metabolites

Figure 3
Metabolite identification software workflow, featuring different data analysis algorithms. Each time a new metabolite candidate is found, a new row
is added to the table. Columns are added to the table to confirm an existing metabolite candidate and to show the result of the individual algorithm.

Results and discussion

The metabolite identification workflow

(figure 3) used interwoven multiple
algorithms to populate a results table
with potential metabolites. A plurality
of different procedures was used to
identify the metabolites. Each time a
new metabolite candidate was found,
a new row was added to the table.
For each individual algorithm that
was used to find or qualify a potential
metabolite, an additional column in the
results table was used to display the
results. The following points describe
the individual algorithms and their
Figure 4
interaction to generate the final result Comparison table between control sample and metabolite sample with e.g. di hydroxyl and
table. dechlorination metabolites at mass 451.2578 and 501.2140 respectively.

1. Sample-control comparison (figure

4) – In a sample comparison table,
the compounds found in the
metabolite sample (time > 0, parent
drug was metabolized) and the con-
trol (time = 0) are compared and
aligned by mass and RT. This
allowed detection of both expected
and unexpected metabolites in the
metabolized sample.

5989-7375EN 128
 A
2. Isotopic pattern matching HO
O
(figure 5) – The isotopic pattern O
N
of a metabolite coming from an N
N
expected biotransformation was N N
CH3
compared to the theoretical
pattern of the biotransformed parent
drug, while the pattern of an unex-
pected metabolite was compared to
the theoretical isotope pattern of the
parent drug.

3. Fragment pattern matching or

MS/MS correlation (figure 6) – This
procedure correlated the MS/MS
spectrum of each potential metabo-
lite with the MS/MS spectrum of
the parent drug. Using this proce-
dure mass shifts in the fragment
ions due to biotransformations could B
Cl
be detected and visualized. O
O
N
N
N OH
N N
CH3

Figure 5
A) Isotopic pattern matching for dechlorinated metabolite by comparison with calculated isotopic
pattern (CIP) after application of biotransformation to parent drug formula.
B) Isotopic pattern matching of chlorinated hydroxymetabolite by comparison to the calculated
isotopic pattern of the chlorinated parent drug.

Cl
O
O
N
N OH
N
N
N CH 3

C15H20N3O3
290.1500
C15H20N3O2
274.1550

Figure 6
MS/MS fragment pattern matching between protonated parent drug (m/z 470.2323) and proto-
nated hydroxy metabolite (m/z 486.2265) with biotransformation mass shift assignment. The
MS/MS fragments at m/z 274.1550 from the parent drug and at m/z 290.1500 are related by a
shift of 15.9999 for the metabolic hydroxylation reaction.
129 5989-7375EN
4. Extraction of chromatograms (figure
7) – This included generation of
extracted ion chromatograms (EIC)
directly from the data, and genera-
tion of extracted compound chro-
matograms (ECC) from extracted
molecular features.

5. Compound search in RAD (radioac-

tivity detection) chromatograms (not
presented here).

6. Compound search in UV (ultraviolet)

chromatograms or other detection
methods (not presented here).

7. Biotransformation labeling
(figures 6 and 10) – Expected
metabolites were confirmed by com-
parison of parent ion mass shifts
with a table of known biotransfor-
mations and these compounds were
labeled with the name of biotrans-
formation reaction in the result table
(figure 10). Figure 7
The extracted ion chromatograms (EIC) were directly obtained from the measured data for the
control as well as metabolite sample, and the extracted compound chromatograms (ECC) were
8. Molecular formula assignment obtained from extracted molecular features – shown here shown for two different hydroxy
(figures 8 and 9) – Molecular formu- metabolites at m/z 486.1903 at RT 7.2 and 7.9 min. The metabolites were not present in the control
la assignment was based on the sample.
assumption that only one elemental
composition fits to the measured
accurate mass of the product and
that subsets of the same elemental
composition must explain the prod-
uct fragment masses and their neu-
tral losses in the MS/MS spectrum.

9. Mass defect filter – Potential

metabolites with a mass defect out-
side a defined mass defect window
around the parent drug were filtered
out.

Figure 8
Calculated formula for highest score hydroxyl metabolite and abundance for isotopic pattern and
masses. The relative mass error was calculated to be 2.42 ppm.

5989-7375EN 130
10. Metabolite prediction – Structures
from manual or computer-assisted
metabolite prediction were
assigned to the identified com-
pounds in the result table.

11. Molecular structure elucidation –

All data were consolidated for
structure elucidation and structure
formula assignment.
12. Population of the final metabolite
result table (Figure 10): The identi-
fied metabolites were collected in a
result table, which shows the
major information about the com-
pounds and the qualification from
each individual algorithm as well
as additional available information.
There is the possibility to produce
and to report more detailed result
tables.
Figure 9
Measured masses of MS/MS fragments and calculated fragment formulae for the the protonated
hydroxy.metabolite at m/z 486.2255. The loss masses as well as the calculated fragment formula
of the loss are displayed. The relative mass error of fragment m/z 290.1500 was calculated to
be -0.20 ppm.
[ "?9+
;s)
"?C
Metabolites from the drug compound mass accuracy MS and MS/MS data
Nefazodone were automatically identi- were acquired with low single digit rel-
fied by means of a computer-assisted ative mass error and used for molecu-
approach, which applied several inter- lar formula generation (MFG). In the
leaved algorithms to QTOF MS and final at-a-glance result table, an overall
MS/MS data. The comparison of con- relevance score was created for the
trol and metabolized sample was identified metabolites, which was cal-
based on molecular feature extraction culated from the weighted relevance
(MFE) to extract metabolites. High score from each algorithm.

Figure 10
Final result table for overall qualified metabolites, which are related to a known biochemical metabolic reaction. Qualified results form the algorithms
“sample control comparison”, “isotopic pattern matching”, “fragment pattern matching” and “mass defect filter” are marked in green. Additional
information such as “assigned biotransformation2”, “calculated formula” and “available MS/MS” spectra are marked in blue. Structures can be
assigned manually. Metabolites not related to a known biotransformation can also be extracted from the data by the same algorithms.

131 5989-7375EN
References

1.
Hopfgartner G., Chernushevich I. V.,
Covey T., Plomley J. B., Bonner R.,
“Exact mass measurement of product
ions for the structural elucidation of
drug metabolites with a tandem
quadrupole orthogonal acceleration
time-of-flight mass spectrometer”, J.
Am. Soc. Mass Spectrom., 10, 1305-
1314, 1999.

2.
Peterman S. M., Duczak J. N.,
Kalgutkar A. S., Lame M. E., Soglia J.
R., “Application of a linear ion
trap/orbitrap mass spectrometer in
metabolite characterization studies:
Examination of the human liver micro-
somal metabolism of the non-tricyclic
anti-depressant nefazodone using
data-dependent accurate mass mea-
surement”, J. Am. Soc. Mass Edgar Nägele is Application Chemist,
Spectrom., 17, 363-375, 2006. Frank Wolf, Uwe Nassal, Rainer Jäger
and Karina Subramanian are Software
R&D Scientists, Horst Lehmann and
Frank Kuhlmann are Product Managers,
all at Agilent Technologies in
Waldbronn, Germany.

5989-7375EN 132
133
Learn more
www.agilent.com/chem/tcm

Buy online
www.agilent.com/chem/store

Find an Agilent customer center in your country

www.agilent.com/chem/contactus

U.S. and Canada

1-800-227-9770, agilent_inquiries@agilent com
Europe
info_agilent@agilent.com

Asia Pacific
inquiry_lsca@agilent.com
www.agilent.com/chem/tcm
© Agilent Technologies, Inc., 2010
Published April 1, 2010
Publication Number 5989-9912EN