Chromatography Forum

Hello,
I have been tasked with conducting performance qualification (PQ) testing on our Agilent HPLC systems. We use butyl benzoate standards for this. I am curious as to why I get different responses on different systems using the exact same standards, same gradient, same mobile phases, same UV wavelength. For example, with a 10 mg/L standard, on one system I get peak area of 1654, peak height 493. On another system, I get peak area of 3718, peak height 860. The linearity and correlation for both systems is excellent over the range of standards. I've scoured the methods printouts for differences that would account for this, but I can't find any. What am I missing? I should add that am not a chemist by training. I'm a biologist who has been drafted for this task. Thanks.

I'm not sure you really have an issue. Let's just say that in 4 decades of cGMP pharmaceutical method development and validation that we ever even looked into such variation.

Small differences in tubing inner diameter, length, and flow cell can all make differences. I'll guess that your retention times are also not exactly the same on two systems.

To: Consumer Guy
Yeah, I don't think I really have an issue either, and I recall from my QA days reviewing HPLC data and seeing similar things, but I never really got an in depth explanation of it, just that it was "instrument variability". Now that I'm doing this stuff myself, I was wondering if there was more to it than that. Thanks for the reply!

Not "instrument" variability per se. That assumes it is the instruments fault, which it is not. The user is mostly responsible for excessive variability. In your case it could simply be detector lamp energy differences, but since you are asking... it sounds like you may be unaware of many other common things that can change the output from system-to-system.

This is due to hundreds of other HPLC system variables that most people with poor training in chromatography fail to understand and document in their methods. First, you are not comparing identical systems. No such thing exists. They are all different. Some more than others, and "close" is possible.Plumbing, as noted, is certainly one key difference that will change results as changed in tubing ID, length and flow path change results. Of the many other differences, you would need to compare lamp energy, slit width, wavelength and BANDWIDTH used, flow cell volume and path length (all of which can result in huge changes to the signal output. There are many more. Let us not forget the use of wrong method / instrument settings too, such as turning on the "reference wavelength" software feature (and all of the related settings it uses). That one can really screw up the comparisons too.We use many of these configuration related settings to BOOST signal for hard to detect compounds. *This is one of many reasons why formal instrument Performance Verification (and Qual) tests must be written for every single version/model of detector AND include variations for every option too (flow cells, lamps, slits etc) too. It is a lot of paperwork for just the detector.

To "Multidimensional": Hmmmm....yes, some of the many reasons I prefer biology over chemistry. If you necropsy 100 rats, the heart, brain and lungs are going to be in the same places 100 times! The "plumbing" is going to be (relatively) the same 100 times. No changing of the rules depending on temperature, atmospheric pressure, pH, etc. etc. Sigh....

Hi JHC,

While I find your comment funny, I think your comparison is a little off, like comparing apples to oranges. Rats (used in in-vivo studies) are cloned SPECIFICALLY for the repeatability and purpose. The whole point of the rats/mice is to be virtually identical in order to limit those outside variables and mute their affects on your assays. Meanwhile, LCs come in all different forms, purposes, brands, etc. Perhaps think of different LCs as different types of rats. All of the rats from one lot would have tiny intramolecular differences, making them suitable for an assay. Ideally, the same should be true for your LC (when it comes to simple things such as area, height, etc.). Now if you got a new lot of rats/mice you may notice small difference (such as the mice maybe being a little heavier overall), but your assay is still able to run perfectly fine because it fundamentally doesn't rely on this information. On another LC system this may also be the case, but fundamentally the experiment should still be fine if all of the parameters are followed correctly (column type, temperature, method settings, etc.) and minute differences such as peak height and area can typically be ignored interinstrumentally. For and should be documented. But you are still example, maybe one lab is using a PDA detector that is close to it's expiration and slightly saturated, while another lab has an LC with a fresh PDA detector. They will likely see different peak heights and areas, but, as Multi has stated, this is expected trying to keep the important analysis parameters (the retention time of peaks in your example perhaps) the exact same in the assay as a whole. The rules are never supposedly "changing" as you describe, but are intrinsic parts of the assay itself and are important to it's repeatability! I've thawed enough cells to know that expecting the same results every time is ridiculous, but for the many years of LC work that I have done and the only unexpected results I tend to see arise from the experimental work done during the chemistry, and never during the analysis on the system (although I haven't had anything break catastrophically or clog yet).

The peak area ratio of the two systems was 2.25 and the peak height ratio was 1.74. The discrepancy between the ratios suggests the peak width in the two systems was different too (the system with higher response likely showed broader peaks). That being said, I suspect the most likely cause of response difference between the two systems is the flow cell length.

Absorbance units are an absolutely-defined thing, not a wobbly number that's allowed to vary between instruments. If something has an OD of 1.0, then 90% of the light that went into the flow-cell got absorbed. This should be absolutely true, on all instruments that give a readout of OD. It's true on all spectrophotometers, so why should a PDA be any different?

But of course flow-cell geometry can vary. If you've got a different path-length, the same solution will give a different OD.

Variations from instrument to instrument (or in one instrument over time) that can help account for this sort of variation:
Lamp intensity
Lamp/other optical bench alignments
hazing of mirrors, windows and other optical components
pump flow rate variations
quality of connections in plumbing, lengths and diameters of tubing used between injector and detector

The peak area with UV absorbance detection depends on:
- the mass of the analyte injected to the column (i.e., the analyte concentration in the sample multiplied by the injection volume); the higher the mass, the larger the peak area unless the detector is overloaded or unless the mass is below the injected mass detection limit;
- the flow cell optical length;
- the wavelength (both its nominal value and the spectral bandwidth);
- the units (AU*min, mAU*min, mAU*s, V*min, mV*min, mV*s, "counts"*min, ...; here the x axis of the chromatogram is in time units);
- the flow rate (if the x axis of the chromatogram is in time units); since the time needed for the analyte zone to pass the detector cell is the reciprocal of the flow rate, the peak area is strictly inversely proportional to the flow rate as well; if the x axis of the chromatogram is in eluent volume units, the peak area (e.g., in mAU*ml) is independent of the flow rate.

The peak area with UV absorbance detection is independent of:
- the analyte band broadening (if the peak is high enough in comparison with the signal detection limit);
- the lamp intensity since the absorbance is the log of the reciprocal of transmittance (log(I0/I)).

The peak height with UV absorbance detection depends on:
- the mass of the analyte injected to the column;
- the flow cell optical length;
- the wavelength;
- the units (AU, mAU, V, mV, "counts");
- the analyte band broadening (and all the factors affecting it, namely the column, the extra-column path in the instrument, the mobile phase, the retention factor, the temperature, the flow rate,...); in contrast to its effect on the peak area, the flow rate influences the peak height only via its effect on the band broadening (see the van Deemter equation).

The peak height with UV absorbance detection is independent of the lamp intensity for the same reason as the peak area.

2 things...
" if the x axis of the chromatogram is in eluent volume units, the peak area (e.g., in mAU*ml) is independent of the flow rate."
- I agree with this 100%, but it is predicated on a *really big* "if".

If your lamp is failing, your peaks are going to be inconsistent at best. As someone who has seen significant changes in peak areas as a function of replacing an old lamp, I assure you that lamp intensity (whether a function of output at a given wavelength or alignment accuracy) matters!

2 things...
" if the x axis of the chromatogram is in eluent volume units, the peak area (e.g., in mAU*ml) is independent of the flow rate."
- I agree with this 100%, but it is predicated on a *really big* "if".

Yes, it is common to use time on the x axis in HPLC rather than the eluent volume pumped through the column. However, the volume is the intrinsic parameter describing the retention. Classical column liquid chromatography (e.g., in such its modern implementation as "solid-phase extraction" sample clean-up), without in-line detectors, chart recorders, and computers, deals with volume rather than time. As long as V = t*F, it is easy to convert the usual time-based x axis to the volumetric x axis in HPLC.

If your lamp is failing, your peaks are going to be inconsistent at best. As someone who has seen significant changes in peak areas as a function of replacing an old lamp, I assure you that lamp intensity (whether a function of output at a given wavelength or alignment accuracy) matters!

In theory, the lamp intensity (the incident light intensity I0) is not a parameter influencing the absorbance (in contrast to fluorescence detection where the intensity of fluorescence is directly proportional to I0). However, in practice some technical aspects may cause the lamp intensity to influence the peak area and the peak height with UV absorbance detection, especially at concentrations near the detection limit.

" lamp intensity " certainly does change measured Area amounts as the optical bench design used in an HPLC detector (UV/VIS) has adjustable slit-widths to reduce or increase the amount of light that is focused on the cell and detector. These features were developed to account for this change. Manufacturers offer bulbs with different energy outputs to extend the useful time of the bulb in the detector (can be seen by viewing their spectral energy levels across all wavelengths). Lamp energy varies during the lifetime of the bulb so engineers designed the detectors to have an optimized optical path and idealized slit of light shine through to maintain an even field of light on the detector. As the lamp ages, this slit needs to be opened up to try and maintain the same amount of output as you had before. Changing the bulb changes the amount of light focused on the cell/detector so we have adjustments for this built-in. These are different and not related to cell path-length (also very important). The SETTINGS used between detectors AND the differences in design for each optical bench result in differences in signal output.