Академический Документы
Профессиональный Документы
Культура Документы
Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma
a r t i c l e
i n f o
Article history:
Received 17 July 2009
Received in revised form
11 September 2009
Accepted 17 September 2009
Available online 24 September 2009
Keywords:
UHPLC
Method development
DryLab
Sub-2 m particles
Computer-assisted optimization
a b s t r a c t
Many different strategies of reversed phase high performance liquid chromatographic (RP-HPLC) method
development are used today. This paper describes a strategy for the systematic development of ultrahighpressure liquid chromatographic (UHPLC or UPLC) methods using 5 cm 2.1 mm columns packed with
sub-2 m particles and computer simulation (DryLab package). Data for the accuracy of computer modeling in the Design Space under ultrahigh-pressure conditions are reported. An acceptable accuracy for
these predictions of the computer models is presented. This work illustrates a method development
strategy, focusing on time reduction up to a factor 35, compared to the conventional HPLC method
development and exhibits parts of the Design Space elaboration as requested by the FDA and ICH Q8R1.
Furthermore this paper demonstrates the accuracy of retention time prediction at elevated pressure
(enhanced ow-rate) and shows that the computer-assisted simulation can be applied with sufcient
precision for UHPLC applications (p > 400 bar). Examples of fast and effective method development in
pharmaceutical analysis, both for gradient and isocratic separations are presented.
2009 Elsevier B.V. All rights reserved.
1. Introduction
The expression high performance liquid chromatography was
created by Horvth et al. in 1967 [1]. As experimenting in his lab at
Yale with supercially porous pellicular materials, the pressure
went up the rst time above 1000 psi, Horvth said, this is not
LC anymore, this is high-pressure LC (HPLC). The consequent and
visionary work of another Hungarian, Halsz with small particles
35 years ago laid down the fundaments for columns packed with
ne particles and he made separations of 15 compounds possible
in 60 s as early as in 1974 [2]. The understanding of the fundamentals of reversed phase chromatography (RPC) performed by Csaba
Horvth and his team at Yale, was strongly inuencing the future
developments in HPLC such as the concepts of the Design Space
and DryLab already in 1976 [3]. The work towards smaller and
smaller particles was continued and in the year of 2004 Waters
introduced the new technology called UPLC offering new possibilities to reduce analysis time by a factor of 34 using higher pressures
up to 1000 bar.
Corresponding author.
E-mail address: fekete.szabolcs1@chello.hu (S. Fekete).
0021-9673/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2009.09.043
7817
7818
Table 1
Experimental retention times and resolutions vs. predicted from the two-dimensional gradient timetemperature model.
Peaks
1
2
3
4
5
6
7
8
9
Resolution
Experimental
Predicted
Differencea
% errorb
Experimental
Predicted
Differencea
% errorb
0.65
0.72
0.94
1.14
1.49
1.61
1.73
1.86
2.30
0.64
0.73
0.97
1.13
1.48
1.59
1.69
1.81
2.29
0.01
0.01
0.03
0.01
0.01
0.02
0.04
0.05
0.01
1.56
1.37
3.09
0.88
0.68
1.26
2.37
2.76
0.44
2.61
8.41
5.56
9.9
2.75
2.98
2.05
11.52
3.09
8.62
5.38
9.68
2.80
2.51
1.94
11.83
0.48
0.21
0.18
0.22
0.05
0.47
0.11
0.31
15.53
2.44
3.35
2.27
1.79
18.73
5.67
2.62
0.02
1.60
0.25
6.55
Average
a
b
7819
separation. The predicted optimum condition was set and experimental chromatograms were recorded. Fig. 4 shows the predicted
and experimental chromatograms.
The precision of prediction in the case of isocratic %
Btemperature model under UHPLC conditions was evaluated
with the comparison of predicted and experimentally obtained
retention times and resolutions (Table 2).
The predicted and experimentally obtained chromatograms
(retention times and resolution) were compared. The predicted
retention times were in excellent agreement with the experimental
ones, the average of retention time errors was 0.74% (see Table 2).
The resolutions also were predicted with high accuracy (average of
errors was 3.76%). So we can state that computer-assisted simulation can be applied with high precision for UHPLC conditions in the
case of simultaneous optimization of isocratic % B and column temFig. 2. Predicted (A) and experimental (B) chromatograms were optimized by
7 and 21 min gradient basic runs at two different column temperature (35 and
65 C). Column: Restek Pinnacle C18 1.9 m (50 mm 2.1 mm), mobile phase
A: acetonitrilewater 595 V/V%, mobile phase B: acetonitrile, gradient elution (3570% B, in 2.3 min), ow: 0.5 mL/min (p = 299 bar), column temperature:
50 C, injection volume: 1 L, detection: 220 nm, analytes: a neutral polar API
(steroid) and its related impurities and degradation products: (1) 6-alpha-hydroxyethinylestradiol, (2) 6-beta-hydroxy-ethinylestradiol, (3) 6-keto-ethinylestradiol,
(4) unknown degradant, (5) estradiol, (6) 9,11-didehydro-ethinylestradiol, (7)
ethinylestradiol, (8) unknown degradant and (9) unknown impurity.
7820
Table 2
Experimental retention times and resolutions vs. predicted from the two-dimensional isocratic % Btemperature model.
Peaks
1
2
3
4
5
6
7
Resolution
Experimental
Predicted
Difference
% error
Experimental
Predicted
Differencea
% errorb
0.45
0.62
0.87
0.96
1.12
1.22
2.23
0.45
0.63
0.86
0.95
1.11
1.22
2.24
0.00
0.01
0.01
0.01
0.01
0.00
0.01
0.00
1.59
1.16
1.05
0.90
0.00
0.45
6.02
5.84
1.80
2.44
1.50
10.40
6.11
5.95
1.83
2.70
1.58
10.71
0.09
0.11
0.03
0.26
0.08
0.31
1.47
1.85
1.64
9.63
5.06
2.89
0.01
0.74
0.15
3.76
Average
a
b
Fig. 5. Two-dimensional resolution map of the gradient time (min) against mobile
phase pH for the separation of basic API and its related impurities and degradation
products.
Fig. 6. Predicted (A) and experimental (B) chromatograms. Column: Zorbax SB C18
50 mm 2.1 mm, 1.8 m, mobile phase A: methanolbuffer 595 V/V% (buffer:
10 mM phosphate + 0.1% triethylamine, pH 6.7), mobile phase B: methanolbuffer
8020 V/V% (buffer: 10 mM phosphate + 0.1% triethylamine, pH 6.7), gradient elution (initial 0% B, at 0.7 min 0% B, at 3.1 min 65% B and 100% B at 10 min), ow:
0.5 mL/min (p = 531 bar), column temperature: 30 C, injection volume: 3 L, detection: 230 nm, analytes: basic drug API and its related impurities and degradation
products: (1) peak of light stress origin (unknown) (2) 1-naphtol (3) duloxetine (4)
duloxetine-3-isomer impurity (5) dimethyl-duloxetine impurity and (6) duloxetine
impurity A.
7821
Table 3
Experimental retention times and resolutions vs. predicted from the two-dimensional gradient timemobile phase pH model.
Peaks
1
2
3
4
5
6
Resolution
Experimental
Predicted
Difference
% error
Experimental
Predicted
Differencea
% errorb
1.38
4.52
6.9
7.17
8.08
8.94
1.27
4.5
6.95
7.12
8.04
8.98
0.1
0.02
0.05
0.05
0.04
0.04
8.66
0.44
0.72
0.70
0.50
0.45
25.13
25.31
1.92
8.66
10.78
23.94
26.80
2.13
9.34
11.33
1.19
1.49
0.21
0.68
0.55
4.97
5.56
9.86
7.28
4.85
0.05
1.91
0.82
6.50
Average
a
b
Fig. 8. Experimental (AD) and predicted (EH) chromatograms. Column: Restek Pinnacle DB Biphenyl 1.9 m (50 mm 2.1 mm), mobile phase A: acetonitrilewater
595 V/V%, mobile phase B: acetonitrile, gradient elution (3068% B, in 6 min), ow: 0.4 mL/min (A and E), 0.5 mL/min (B and F), 0.6 mL/min (C and G) and 0.8 mL/min
(D and H), column temperature: 45 C, injection volume: 2 L, detection: 270 nm, analytes: a neutral polar API and its related impurities: (1) unknown impurity, (2)
bicalutamid-beta-anilin, (3) bicalutamid, and (4) bicalutamid-beta-sulphenyl. The prediction is based on basic runs performed with the ow-rate of 0.4 mL/min.
7822
Fig. 9. Experimental and predicted retention times plotted against pressure. Analytes: (A) bicalutamid-beta-anilin, (B) bicalutamid, and (C) bicalutamid-beta-sulphenyl.
4. Conclusion
The separation power of short columns packed with sub-2 m
particles are reported many times. But in this study it is also proved
that by using computer modeling, the time required for method
development can be considerably reduced.
It is possible to develop methods for pharmaceutical analysis
(assay, impurity proling, cleaning validation) within a day or even
in a few hours. If a 50 mm 2.1 mm sub-2 m column is applied
during the systematic method development, basic gradient runs
with 7 and 21 min (at a ow-rate of 0.40.5 mL/min) can provide reliable accuracy for the computer model simulation under
ultrahigh-pressure conditions if gradient separation is necessary.
The average of predicted retention time errors was lower than 2%,
which can be considered as a highly accurate prediction, so the suggested fast gradient initial basic runs can be applied in daily routine
work resulting in signicant time saving. Based on our experiments
we can state that DryLab separation modeling can be applied for
elevated pressure (not only in HPLC practice) with high accuracy.
The average of retention time errors did not exceed 5% when the
ow-rate (pressure) was duplicated (p 600 bar). When the owrate was enhanced with a factor of 1.25 and 1.50 compared to the
ow applied for basic runs the prediction error was approximately
3 and 4% (respectively).
The second point is that the column technology used for sub2 m particles is well developed to reduce the silanol activity of
the stationary phase and thus the computer simulation can be
applicable also for the separation of basic solutes with reliable
precision.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
Fig. 10. Retention time prediction error (average) against pressure. Column:
Restek Pinnacle DB Biphenyl 1.9 m (50 mm 2.1 mm), mobile phase A:
acetonitrilewater 595 V/V%, mobile phase B: acetonitrile, gradient elution
(3068% B, in 6 min), ow: 0.4 mL/min (p = 290 bar), 0.5 mL/min (p = 362 bar),
0.6 mL/min (p = 435 bar) and 0.8 mL/min (p = 581 bar), column temperature: 45 C,
analytes: a neutral polar API and its related impurities.
[12]
[13]
[14]
[15]
[16]
[17]
[18]
7823