Most of us take advantage of the autozero function on our HPLC data systems to make the chromatogram look more pleasing to the eye. The autozero function is quite simple – it just takes whatever signal the data system is receiving and sets it at the zero point on the y-axis of the output. This means that the baseline is at the bottom of the screen when we view it on the computer monitor, making it easier to visually follow the chromatogram as it elutes. Thus our chromatograms look like the blue tracing in the figure, with the baseline at zero and the peak rising from there. This all works quite well when there is little detector response for the mobile phase and the peak height is less than the maximum detector response.
There is a potential problem, however, when there is significant background signal and the peaks are large. Let’s consider the case of an ultraviolet (UV) detector. All UV detectors that I know of are linear to 1 absorbance unit (AU). When signals exceed 1 AU, there may be problems. Some detectors specify larger ranges, such as 0 – 2 AU, but it isn’t always clear if this extended range is a result of optical linearity or some mathematical transformation of the data to make the detector appear to be linear at > 1 AU. As a result, I don’t like to operate methods used for quantification with signals > 1 AU, and preferably < ≈0.8 AU for a little safety margin.
So what is the problem with autozero? The detector has a fixed range, let’s say it is a maximum signal of 1 AU for discussion purposes. This range includes both the signal that has been autozeroed out as well as the peak response. I’ve shown an extreme case for the red chromatogram in the figure, which duplicates the blue peak. In this case, there is nearly 0.5 AU of background in the un-zeroed signal, as shown. This means that the top half of the peak exceeds the 1.0 AU maximum for the detector. We really don’t know how the detector responds at > 1 AU without some testing. It could truncate the signal and flat-top the peak, but this is unlikely. More likely is that the portion of the peak above the 1 AU maximum will have a non-linear response. It would not be difficult to determine what happens by injecting a series of calibration standards of known concentration to observe where the signal became non-linear; you might even be able to correct for this.
When autozero is applied to the red chromatogram, the apparent baseline would be at 0 AU, as is the blue one, but we would have no visual clues to know that part of the peak was in the non-linear range of the detector. In this case, the peak would have 1 AU of absorbance and the baseline would have 0.5 AU, making the effective signal 1.5 AU, which is beyond the range of the detector. But the peak would appear to be on-scale because of autozero. We might be able to compensate for this non-ideal condition using the calibration technique mentioned above. But what happens if the baseline is gradually rising over the course of a batch of samples? For example, it might start off at 0.1 AU with a new column, but over a batch of 100 samples, the background might increase significantly if strongly UV-absorbing materials were eluted so late that the peaks were sufficiently broad to be undistinguishable from the baseline. Now you’d have a moving target, with different portions of the peak appearing in the non-linear region as a result of a combination of peak height and position in the run.
The bottom line here is to pay attention to the background signal in your chromatograms. Autozero doesn’t make the problem go away – it just hides it. Most of the time, backgrounds are small, so we don’t have problems, but don’t be lulled into complacency or you might run into problems.
This blog article series is produced in collaboration with John Dolan, best known as one of the world’s foremost HPLC troubleshooting authorities. He is also known for his research with Lloyd Snyder, which resulted in more than 100 technical publications and three books. If you have any questions about this article send them to TechTips@sepscience.com
