We are analysing a compound and found one impurity. The limit we explained for this is 0.10% max. When we analysed this product on Dionex RSLC having PDA detector we got this impurity with in the specification limit ( 0.075%). When we analysed same solution in the same condition with Agilent 1260 having PDA we got this impurity out of specification (0.117%).
Can anybody help me to understand why this phenomenon is happening?
First question is: what is S/N value obtained for impurity peak in both cases? If the imp. peak is small S/N= 10 - 20 or below), so the uncertainty of the result is high, I would suppose that statistically you obtained the same result. If not see question 2.
Second question: what was your reference peak? If you calculated imp. as percentage to main peak. the difference in your result could come from different sensitivity of the detectors. Every single detector has individual characteristics and sensitivity, thus it requires appropriate reference for testing. The best way is to dilute your test solution 1000x (to obtain final conc. of main component at a level 0,1% comparing to undilluted solution) and the area of the peak should be used as reference for calculation your imp in undilluted sample. The rule says: compare similar areas.
Third question: is impurity response factor the same (similar) to main compound response factor? If not you should be aware in comparison results from two detectors. The response factor for impurities will be found different depending on detector used. I would expect that PDA are prone to exhibit higher viariability in response factor between one another comparing to UV detectors.
And last question: is your impurity sufficiently separated from main peak in both HPLC systems? Resolution has impact on peak area especially in case of impurities eluting on main compound's tail. Apart from column efficiency, resolution depends also on detection cell geometry. The lower resolution the higher assay error because the signal from impurity can be affected by residual signal of main compound.
In case of gradient elution you should consider also dwel volume that affects retention time vs % of organics at the moment and thus local response factor of your compound.
Thans for replying Małgorzata
I can suspect on the PMT(Photonmultiplies tubes) which is used in detector. As you also aware that all the PDA detectors used on the same principle but the differences in PMT design or no. of PMTs used in the detectors are also an important factor.
Also I want to clear that the Agilent HPLC is normal HPLC while the Dionex HPLC is RSLC where the dwell volume is reduced as well as due to less ID of tubing we may get the change in the response(As the peak symmetry is changed and due to this the main peak response is vary.
again thanks
*Neither detector uses PMT's. All use photodiode arrays for detection.
What you are describing is completely normal. This is why method transfer can be complex and requires a great deal of skill and experience with the two systems to develop a method where the exact parameters are matched as close as possible.
The real answer to tor question is that the two systems being compared are completely different. HPLC systems are not like balances. Two analytical balances can be and should be calibrated to provide the same weight for the same sample under test. HPLC systems are far too complex with so many more variables and permutations that it is not reasonable to expect that two different HPLC systems will provide the same output (even two similar systems with identical modules will not give exactly the same results).
Let me just share with you a few examples and/or reasons for this variation.
(1) The two detectors used are different and have different responses. Response is not as important as wavelength accuracy (which is calibrated). The design of the two detectors and the settings used, esp on a DAD/PDA are almost certainly different (to name a few parameters: light output, reference wavelength on or off, slit width, bandwidth selected, sampling rate, flow cell volume and path length all effect the signal). Again, this results in variation.
(2) Mobile phase composition accuracy can and does vary between systems.
(3) The plumbing is often different. Different connection types, internal ID's, lengths between connections all result in changes, including dwell volume.
(4) Injector accuracy, design, configuration and RSD. = More variation.
(5) Software settings, esp integration algorithms used and exact integration settings for each chromatogram. Any change in the baseline may result in changes to the resulting integration.
(6) The actual chromatography itself on the specific column. Any changes in peak width, noise may result in variation.
(7) Examples of variabilty in technique used. For example, if the column was not properly washed and equilibrated before the injection. Maybe the column was given 5 minutes to equilibrate for the first run and two hours for the second run? *This is one area where I like to see methods programmed into a computer so the computer insures that every method is run in exactly the same way (ie. Same equil time, same wait time, same run time).
You can indeed set up two different HPLC systems to provide the same result (This is why we are able to transfer methods and validate them in different labs), but this should never be something that is expected to happen without accounting for some adjustments for how the two systems differ.
- Badges
- Science method