Could you try injecting a smaller sample? I think the detector/ software is being overloaded. I've seen this on a few HPLC systems. What I expect you to see is, starting with a small peak, as the peak gets larger, you will start to see this negative peak form.

I think that they use a signed integer in the software to store the peak height. if the peak goes above a certain level, the data overflows into the most significant bit, which is used by the microprocessor to indicate a negative number. Let's pretend that the data is stored as a single byte (it isn't, but a single byte can be used to demonstrate). A byte can hold an range of 0 -> 255 (in binary, 00000000 -> 11111111). If a signed number is required, left most bit is used to indicate the sign (1 = negative), so it can only hold a range of 0->127 for a positive or negative number (01111111 = 127). Now add one to 127; you now get 10000000, which is "negative zero" or 0 if the program is using a signed byte. Add one more, and we get 10000001 which is now -1 as a signed byte. As the peak get to 128 (now zero for an signed byte), the trace falls to 0, as the peak gets larger, the peak goes negative.

Looks like injection solution / sample prep effects at T0, but impossible to comment further without you first providing all of the method details. For chromatography questions, we must have the HPLC method conditions. "Help us to help you".

(1) What are the exact column dimensions (ID x L, mm)?

(2) What is the flow rate?

(3) What is the mobile phase composition?

(4) Isocratic or gradient? *If gradient, please provide the exact profile and composition.

(5) What is the injection volume?

(6) What is the injection solution?

(7) What are the exact detection settings?

Any other information about sample prep that you think might be useful would also be appreciated too.

Could you try injecting a smaller sample? I think the detector/ software is being overloaded. I've seen this on a few HPLC systems. What I expect you to see is, starting with a small peak, as the peak gets larger, you will start to see this negative peak form.

I think that they use a signed integer in the software to store the peak height. if the peak goes above a certain level, the data overflows into the most significant bit, which is used by the microprocessor to indicate a negative number. Let's pretend that the data is stored as a single byte (it isn't, but a single byte can be used to demonstrate). A byte can hold an range of 0 -> 255 (in binary, 00000000 -> 11111111). If a signed number is required, left most bit is used to indicate the sign (1 = negative), so it can only hold a range of 0->127 for a positive or negative number (01111111 = 127). Now add one to 127; you now get 10000000, which is "negative zero" or 0 if the program is using a signed byte. Add one more, and we get 10000001 which is now -1 as a signed byte. As the peak get to 128 (now zero for an signed byte), the trace falls to 0, as the peak gets larger, the peak goes negative.

Jack Silver if this is correct you are a genius. I just got results almost identical to those in this post and I am simply reacting PDAM with a small carboxylic acid. We started with 100 µM which was diluted to 25 µM and then injected as our highest concentration (10 µL injection). We are currently waiting for the lower concentrations to run through the system, as in right this minute it is running. Hopefully it will look better at lower concentrations. I love Research Gate!

Well it looks like those negative peaks were not the derivatized carboxylic acid, because they did not change in magnitude throughout the standard curve. However, I suspect we lost our compound in a precipitate. We tried derivatizing the compound with PDAM in a water/methanol mix. Specifically, the compound was dissolved in water, then was mixed with PDAM dissolved in pure methanol. It was 1:3 water to methanol, and I saw an orange precipitate form during the incubation, which we pelleted and saved. We injected the supernatant, which must have some reacted PDAM or degraded PDAM, because we saw several fluorescent peaks, some negative, some low but positive. For such a simple reaction (small carboxylic acid, PDAM, and solvents) we saw a lot of peaks?

I think it's a change of refractive index, maybe a compound or solvent coeluting with the main compound. But I agree with William Letter, we need more information

Jack Silver

Sphurti Adigal - I'm sorry, I don't have a good answer for you other than to do two injections at the differing concentrations. One injection, not diluted, allows measurement of the small peaks and the diluted injection is used to measure the large peak. The issue you have is related to "dynamic range", the ratio between the largest and smallest values a signal can assume. You have a very large amount of one compound. When it is diluted to keep the response within the limits that you can quantify, the other compounds are diluted below the limit of detection or limit of quantitation.

Depending on how you get your sample, perhaps you can change the sample preparation. If you are using solid phase extraction and the compounds have very different polarities, maybe you can capture the compounds as you do now. Instead of washing immediately with a strong solvent, maybe wash in steps, increasing the strong solvent concentration with each step. You can then analyze a sample from each wash step separately. As I don't know how you are preparing your sample for for analysis, this idea may not be useful, but I suggest it to let you think about your sample preparation.