Hello.

Thanks for providing detailed information about your issue. I haven't worked with triclosan but I went out to check the structure online.

Seems like you did a few good things trying different solvents and pH ranges. My two first two directions from here would be:

1) Are you filtering your samples through the nylon filter and collecting the pass-thru material for further analysis? If yes, I would speculate the triclosan is being bound to the filter, hence you lose your material there.

2) Is your analytical method verified at the expected concentration levels of triclosan? Is there a chance you are diluting the sample below the limit of detection? Is there a chance you have co-elution between analytes (did you verify by injection/analysis of calibration standards of each analyte on their own AND then mix all of them together to verify they can be analyzed together, i.e., to prove your method is specific and no co-elution, cross reaction or signal loss happens).

Hi Petroslav Chipinski

Thanks for getting back to me! I’m filtering my sample after reconstituting the extract, and I’m using nylon filters. I understand that nylon can sometimes reduce recovery, but I’m seeing almost zero recovery for triclosan, which seems unusual.

My list includes about 21 compounds, and this issue is only affecting triclosan and fenofibrate—where triclosan shows no peak at all, and fenofibrate has a low recovery of around 5-10%.

For my spiking, I’m adding 1 ppb to 200 mL of DI water and reconstituting it down to 1 mL, which should keep it well above the detection limits.

I’ve analyzed all the compounds together, and since I’ve established a calibration curve for each, co-elution doesn’t seem to be the problem.

Did you prove that the analyte is not leaking during the loading step? The method attached suggested to use HCl to acidify the sample during the loading step. This indicates that you need to have a much lowerer pH to retain it on the HLB SPE.

I have had good recoveries of triclosan using OASIS HLB cartridges in the past without the need to acidify the sample.

From the available information you have shared, your SPE method should work in theory. There should be no problem in using methanol as the elution solvent. Perhaps you could give us more information on your exact protocol (volume of solvent, flow speed, solvent used for reconstitution etc.).

As you have rightly said, to verify if indeed it is a sample preparation issue, try and isolate the problematic step by examining each SPE step one at a time:

- analyse the water sample that has passed through the cartridge during loading and check if you are indeed losing TCS. Still, if you were loading the sample too quickly unto the cartridge, I would expect poor recovery of the more hydrophilic compounds as well...

- If there is no problem in sample loading, pay special attention to the elution step (low enough flow-rate and sufficient volume).

- In what solvent are you reconstituting the sample after elution, methanol? Filter a standard and check how much TCS you are recovering.

Thank you all for your responses. Petroslav Chipinski you were correct - the issue was indeed with the nylon syringe filters. I confirmed this by analyzing a 200 ppb standard both directly and after filtration through nylon filters. The results showed almost no response for triclosan and only a very small peak for fenofibrate in the filtered standard. Switching to PTFE (hydrophobic) filters resolved the issue, and I was able to recover these compounds successfully.

However, I am still unsure about the reasons for loss observed with nylon filters, especially since I’ve come across studies in the literature where the same compounds were analyzed using nylon filters.

Dileep Singh Good to hear you found the problem.

When I develop methods for formulations where filtration is important, I usually screen at least two types of filter materials (such as nylon and PTFE) and two different manufactures for each (say, FischerSci and MilliporeSigma). Additionally, I would keep the diameter (for example, 33 mm for all, otherwise there will be difference in binding and filtration capacity due to the surface area) and the material of construction for the housing the same (such as polypropylene; this is to avoid difference due to degradation and leeching, especially with aggressive solvents or base/acids). Whenever possible, I also use centrifugation as the reference solution to evaluate differences due to filtration. Since centrifugation is more straightforward, if I can get a clear solution where the expected recovery matches the experimental, that's my go to.

It's not unusual to see difference in results between labs when filtration is used. Some suggested reasons are listed below.

It's common to see a difference of a few percents in recovery using the same filter material from different manufacturers (especially if the diameters are also different).

If standard syringe filtration is used, that's prone to user variations:

- One point of variation for example is: I often prefer to discard the first few drops of filtrate as it may contain leechables or particles from the filter and/or syringe. But sometimes there is insufficient starting solution to sacrifice this amount.

- Another variation may be the final step of the filtration: Some people push the syringe plunger all the way in, thus trying to collect every last drop from the filter. This allows collecting of all volume and material but also may rupture the filter or create bubbles (and with that aggregates when we work with polymeric substances); it may also force out particles that would otherwise be retained by the filter.

- The total sample load( concentration and volume) and turbidity of the solution is important. If you only had one analyte in a clear solution, then that would filter more efficiently that filtering a solution that contains 10 analytes + matrix additives that is also turbid. The filter would then quickly exceed its binding capacity and may leave no pores for the analytes that would normally pass to traverse through the membrane.

Thanks, Petroslav Chipinski, for the detailed response.