We are using an LC method that involves resolving closely eluting peaks by using two UV wavelengths. In other words, "analyte a" has significant absorbance at two selected wavelengths, and "analyte b" has significant absorbance only at one of the wavelengths. By measuring the peak area of both wavelengths, we can determine the amount of both analytes by solving for the areas. Most of the time, this works well, but if we inject the same sample 5-10 times, we will get 1-2 chromatograms where the ratio of of the two wavelengths changes. As a general example, we may get a peak area of 10 on one wavelength and 5 on the other on a particular sample eight of ten injections, but we may get an area of 8 on one wavelength and 6 on the other for the other two- far enough off that something is different beyond typical detector noise.

My assumption is that even if the injection accuracy was poor, the ratio of the peaks at different wavelengths should never change as long as the analyte doesn't change.

This is on a reverse phase column using an isocratic ~80/20 methanol/water solution with phosphoric acid, and the sample is dissolved in the same solution as the mobile phase

Are there ways that we could see varying ratios of absorbance for the same compound? Short of the actual spectrum of the analyte changing, I'm struggling to understand how this could be the case.

Thank you for any help, my apologies for the generalizations and long winded description!