High-Performance Liquid
Chromatography Optimization
Introduction
In many pharmaceutical applications of high-performance liquid chromatography (HPLC),
it is of outmost importance to find optimal separation conditions for different products.
This application note shows how this can be achieved with the MultiSimplex software.
The application note is based on a study performed in a pharmaceutical industry
laboratory.
Problem
To find an optimal analytical method for a pharmaceutical product.
Control variables
Three control (experimental) variables were used. The settings are specified in the
table below.
| |
Ref. value |
Step size |
| pH |
5 |
0.5 |
| ACN, % |
16 |
1 |
| Temperature, °C |
45 |
5 |
Response variables
Three different response variables were used to characterize the outcome. The settings
are specified in the table below.
| |
Objective |
Low limit |
High limit |
Importance |
| Resolution 12 |
Maximization |
0 |
10 |
High |
| Resolution 23 |
Maximization |
0 |
10 |
High |
| Resolution 34 |
Maximization |
0 |
10 |
High |
The shape of the membership functions was proportional for all three
responses.
Results
The results from the 11 optimization trials are shown in the table below.
| pH |
ACN, % |
Temp., °C |
Rs 12 |
Rs 23 |
Rs 34 |
| 4.75 |
16.5 |
42.5 |
0.5583 |
1.343 |
1.884 |
| 4.75 |
15.5 |
47.5 |
0.8970 |
1.508 |
2.007 |
| 5.25 |
15.5 |
42.5 |
0.7992 |
2.132 |
0 |
| 5.25 |
16.5 |
47.5 |
0.5351 |
1.420 |
0 |
| 4.58 |
15.2 |
40.8 |
0.6888 |
2.937 |
1.879 |
| 4.25 |
14.5 |
37.5 |
0.6492 |
0 |
0 |
| 4.14 |
15.9 |
44.7 |
0.5692 |
0 |
0 |
| 4.97 |
15.6 |
43.1 |
0.7596 |
1.179 |
2.299 |
| 4.79 |
14.4 |
45.1 |
1.108 |
1.300 |
2.657 |
| 4.44 |
14.4 |
45.9 |
1.009 |
3.408 |
2.048 |
| 4.18 |
13.8 |
47.3 |
1.270 |
4.530 |
2.017 |
In this study there was a significant improvement in resolution, see figure below.
Click here to view resolution vs. trial number.
Conclusion
The HPLC-method for this pharmaceutical product was improved significantly. Some
additional trials are needed to ensure that the optimal conditions are found. The problem
is solved in an quick and easy way by using MultiSimplex for experimental optimization.
|
