Small peaks on HPLC-RID Agilent infinity 1260

[felixbb]

Dear everybody,

This is my very first message on this board as my frustration goes bigger. I have tried to find answers on google and on this forum prior to posting, but couldn't find "that" perfect answer. If the answer to this question is very straightforward and i should have found it earlier, then i apologise for it.

I am trying to see my products on an HPLC-RID Agilent Infinity 1260. My column is a NH2 zorbax. I'm using a mixture of CH3CN:H2O, and operate the column at RT, and the detector at 40°C. The flow rate is 1ml/min and injection volume is 10microliters.

The products i have in my mixture are polyols, with a concentration going from 1 to 20 g/L (or mg/mL...). I see very small peaks, that i was able to identify with some standards, but here is the thing:

The peaks are so very small that i am not very confident in the precision(accuracy) of my areas. Plus, some peaks can quite easily be confused with noise. Is there anything i can do, except concentrating my sample, to react a higher intensity of peaks? I'm not sure what the impact of flow rate and/or column temperature would do on this setup. I read that the best temperature for the detector was 5°C above RT, but that is about it.

I am a total novice to HPLC, and even worse for HPLC-RID, and this is why i ask for your help.

I thank you for reading me, have a good day.

[shaun78]

I am not very familiar with Agilent's RID specifically.

However, most RIDs have adjustable gain settings. These settings can be used to increase/decrease the signal coming from the RID. Note that this increase/decrease applies to both your peaks and baseline noise.

The temperature of an RID is also very important in terms of noise. As a suggestion, I would check +/-5C from where you are currently at if you are able to do so. See if your signal gets any better with temperature changes. Again, note that an RID will take several hours to thermally equilibrate.

Lastly, one must always look towards the HPLC's pump as a source of noise. If you are running a mobile phase of 50:50 water:organic DO NOT put water on line A and organic on line B and have the instrument mix the solvent for you. Instead, mix the solvent yourself and run only from one line. The reason for this is because, though HPLC pumps are very accurate a RID is more sensitive to mobile phase changes than your pump is accurate; the RID will pickup on the small differences in mixing and give you a more noisy baseline.

As another rule of thumb, RID's typically use LEDs as their light source anymore (or at least Waters' 2414 does). If the Agilent RID uses LEDs as the light source, it is not to your advantage to turn the RID off if you anticipate use anytime soon. LEDs will last for about 10,000 hours in an RI and the time spent waiting for the instrument to warm back up does not justify shutting it down. If you must shut down, then get into the practice of turning on the RID before you leave work for the day. Set the HPLC to run at low flow (0.1 mL/min) over night. When you get back to work the following day, the RID will be ready for use.

I hope this helps you out... RID can be an art at times, especially on the development side of things.

[felixbb]

Thank you shaun78 for this answer. I'm not quite sure what those gain settings are, i will look into it. I suppose if the system is very well equilibrated, then this gain settings can be quite helpful.

Concerning temperature and solvent, this is what i have tried doing. I will put the temperature at +5 and see how it works.

If i were to change a small setting after a run to try to optimise the process, will i have to wait a few hours after each run? If so, it might take forever to find the right parameters!! If i leave the HPLC running overnight at very low flow, the system will still be fine when i turn the flow back up or i will also have to wait for equilibration?

[tom jupille]

If the peaks are so small as to "easily be confused with noise", the increasing the gain won't help because that will increase both the peak and the noise by the same proportion (as shaun78 indicated).

Refractive index detection is typically insensitive compared to UV detection with a good chromophore, but you should be able to to better than an LOQ of 10 micrograms.

If shaun78's suggestions don't do the trick, you might want to get hold of the service manual for the detector and re-run the performance qualification test to confirm that it is working to spec. A dirty flow cell, for example, can result in poor sensitivity.

[sepscientologist]

There is also an issue with acetonitrile/water mixtures with RID and amino columns. Agilent actually published a very nice
paper on this in LCGC Europe quite a few years ago. The issue was very small changes in temp in the column selector causing
small fluctuations in the ACN content and much larger chemical noise in the RID. I measured the noise to be about 100x higher
than using just water. To check this you can run the same assay but circumvent the column heater and see if your noise
is much lower.

[felixbb]

Thank you for both your answers.

I believe the mixture ACN:H2o on my column does not provoke too much of noise (i scanned the paper very quickly, and my chromatogram does not look like that). However, i will keep this in mind!

From this discussion my first troubleshot is that the system might (probably is) not well equilibrated enough. I'm going to try to wait a longer period of time to get rid of all the noise. Then i will be able to discern the peaks from the noise a bit better.

I read that the LOD was about 10ng. If i'm not mistaken, it means that a concentration of 1mg/ml is already more than enough, correct? (This is with a very new detector i think). This is what you mean with ", but you should be able to to better than an LOQ of 10 micrograms.", right tom jupile?

Anyways, your answers are very precious, thanks for sharing !

[sepscientologist]

Hmmm... at this point I have two suggestions. First, if you run a blank for maybe 30 minutes take a look at the total RI fluctuation. This is roughly the 5sigma noise for the detector if you ignore drift. I think it should be about 100 nanoRIU for your instrument.

Second, if your noise is reasonably close to spec then maybe your injector is just off. Try injecting 1 mg/mL glucose and share what you get with the chromforum community. I don't recall off the top of my head what kind of signal you should get but plenty of folks here will

[felixbb]

Dear sepscientologist,

The fluctuation of the RI is more in the range of +/- 300 (600 fluctuation) after more or less 2h of calibration in my case. It seems this is due to the pump, that has fluctuation of +/- 0.2 bar (up and down, so 0.4 in total). This might be a bit too much, isn't it ? The pump should be more stable than this i think.

If i were to run a 1mg/ml solution, what i get with 10microliter injection is hardly anything. I have done this for Ethylene glycol. That being said i'm not sure the response is the same, as i'm not completely sure how the RI "sees" those two molecules.

My answer to my problem was waiting for a longer calibration, using other settings for the detector and injecting bigger volumes, as well as using the mobile phase as solvent to avoid a too perturbation of the system because of solvent peak. With this new conditions, my chromatogram becomes decent.

In conclusion: My chromatogram is now decent, but there might be something slightly off with the instrument anyways (detector and/or pump).

[uzman]

Please try this :

Remove the column and connect a ZDV union to connect the tubings.
Run the pump at 2ml / min with mobile phase , and wait 5 minutes.
Fill the reference cell ( wait 2 minutes )
Close the reference cell (wait 2 minutes).
Zero detector.
Stop the pump.
Connect the column and continue.

Good luck !

[felixbb]

Dear Uzman,

I did not perform your troubleshooting, as i have never really touched an HPLC i'd rather not do this. However, i told my colleagues about this.


I have another question, which might or might not be related. My mixture should contain some N-containing molecules. The amine function should be charged at pH = +/- 7. I believe this is a trouble for my conditions CH3CN:H2O, as there might be an equilibrium between charged and uncharged form. Would the addition of some kind of buffer in the water phase be useful? Something to "stick" the pH at around 6-7, and make the column more "homogeneous" (if this makes sence)?

With Zorbax NH2, CH3CN:H2O, i should not be able to see the charged compounds as they should be disgusting tailing peaks that take forever to go out? Might this be a reason why my baseline is not really stable? I've been trying to search the litterature for a few days, and read that either a "C18 column" (which i do not possess) or a "ion exchange HILIC" (this would be NH2) could do the trick. However, i don't really understand how this is achieved.

EDIT:
One of the articles i read on this forum is this one: http://pubs.acs.org/doi/abs/10.1021/ac070997p ... I'm not entirely sure if this is in my topic, but i thought it was interesting.

I apologise if my writing seems confusing, the reason for this is that i am terribly confused by all this HPLC!
I look forward to reading your answers.