Software or Strategy for calibration curves, MS Excel, LC S.

[dresdentl]

What is the best software or strategy for calibration curves?

- I use MS Excel

- I use the software which is on the computer of the HPLC instrument.

According to my experience, the use of the automatic integration software which is on the HPLC instrument delivers the best results and it is the most convenient. In my case, I use LC Solution from Shimadzu.
After batch running you can automatically access the results (concentration). Important is that the method file is good adjusted for correct integration of the peaks.

MS Excel is only necessary for the validation procedure (accuracy, precision).

Is there any other software or strategy?

[tom jupille]

Once you have integrated the peaks, generating a calibration equation is a straightforward mathematical algorithm. In principle, it should not make any difference whether this is implemented in your data system or if you use Excel.

The catch with Excel is that you have to know *something* about statistics to choose (and interpret) the appropriate function, so in that sens the data system is "safer".

[dresdentl]

If I use Ms Excel, with linear regression the higher concentrations will be overweighted. With the HPLC instrument automatic integration software all points get equally weighted and I get a different linear equation.
Further I suppose that the HPLC software don't use one linear equation for calculating the concentrations, although a calibration curve is indicated.

So, I think that the HPLC software is the better choice.

[PeterS]

We use LC solutions too. In fact there no difference between the results in excel vs labsolutions. However if you want to to use a weighed line then LC solutions is more easy since excel does not support that as standard.

[dresdentl]

With LC solution all calibration points get equally weighted, if you choose none weighting. If I look at results, after automatic integration, then the indicated concentrations of all calibration points are near the theoretic value. If I use Excel (without any weighting formula) by using the indicated Areas from LC Solution, then the middle and high calibration points are the same as with LC Solution, but the 1-2 lowest calibration points are much higher then the theoretic value. Excel overweights the high values and the low points are neglected.
I think the concentration range you use can also play a role. I use 9 calibration points, from 0,039 to 10 µg/ml. Maybe, this is a too large concentration range to get good results with one calibration curve. If I use 2 calibrations curves one from 0,039 to 0,625 µg/ml and the other over the whole range, then I also get good results (same as with automatic LC Solution program).
On the other hand, if I use manuell integration with LC Solution, then LC Solution can also indicate wrong high values at low concentrations.

[carls]

What fitting function/style/type is used in the LC solution software? As others have indicated the only way the LC solution software will give a different result is if its using a different method to calculate concentration.

Point to point fitting will give good fit results across the entire range but will not tell you anything about the linearity.

Response factors are another common option.

Which are you using?

[tom jupille]

On the other hand, if I use manuell integration with LC Solution, then LC Solution can also indicate wrong high values at low concentrations.

This raises a huge "red flag". Integration and calibration plot generation are two different activities. If you integrate incorrectly, then you will *never* get a good calibration (garbage in = garbage out).

[dresdentl]

After batchrunning a calibration curve and the linear equation is indicated. The peaks are integrated automatically according to the method file. If the method file is good adjusted then all peaks get correctly integrated.

If you integrate manually, I think it is not incorporated in the indicated calibration curve (LC Solution) and therefore it would make no sense.

On the other hand, I can use areas determined by manuall integration for MS Excel calibration curve.


Both methods give very similiar results.