automatic liq-liq of pahs, bad recovery of last QC standards

Discussions about sample preparation: extraction, cleanup, derivatization, etc.

17 posts Page 1 of 2
Hello everybody,

I’m trying to validate an automated liq-liq extraction of PAHs in water and I’m having some trouble with the extraction efficiency.
First of all: to avoid PAHs to stick to the glass wall of sample-bottles, vials etc we add 5% MeOH to the water.

Setup is as follows:

- 8ml water is transferred into an amber 10ml HS vial
- Sequence setup:

# Spike all QC and calibration standards with calibration standard only.
# extract QC-sample 1
* Add internal standard
* Add extraction solvent (dichloromethane)
* Vortex
* Centrifuge
* Extract solvent from the vial and transfer to a 250µl vial (I can’t inject with the long needle I need to extract the solvent from the bottom of the vial)
* Inject 75µl
# Extract a batch of unknowns (n15)
# Extract calibration curve
# Extract a batch of unknowns (n15)
# Extract QC_2

Mechanically this all works fine, but when checking the results I find that QC_2 only has 50% recovery for the heavy PAHs. Naphtalene, acenaphtylene etc are ok, but starting from Chrysene they are not good at all.

Time between spiking and extraction of QC_2 is approx. 25h.

My first thought is that they stick to the wall of the vial, but I’m extracting with dichloromethane directly in that same vial, so how can they stick??

Photodecomposition is another option, but they are in amber vials. What I couldn’t find is which PAHs are more susceptive to photodecomposition, the smaller ones of the heavy ones?

I can rule out a drift in the sensitivity of the MS as being the cause, since my internal standard response is constant throughout the sequence.

Any advice would be greatly appreciated
Bart
What do you see if you re-inject QC1 at the end of the sequence ?

Peter
Peter Apps
I don't know. the problem is that I only transfer 150µl (to avoid water) and inject 75µl. the remaining solvent evaporates trought the pierced septum within those 24h, so I only have one shot.

I can try to manually extract the last bit of solvent from the 10ml vial and see what result I'll get from that.

Since my internal standards are ok, I conclude that the sensitivity of the ms is fine.
Whatever is occuring has to happen in the 10ml vial in the time between spiking and extraction.
BMU_VMW wrote:
I don't know. the problem is that I only transfer 150µl (to avoid water) and inject 75µl. the remaining solvent evaporates trought the pierced septum within those 24h, so I only have one shot.

I can try to manually extract the last bit of solvent from the 10ml vial and see what result I'll get from that.

Since my internal standards are ok, I conclude that the sensitivity of the ms is fine.
Whatever is occuring has to happen in the 10ml vial in the time between spiking and extraction.


what internal standards are you using?
Davide Balbo from Italy
How much dichloromethane do you use? I guess about 1mL since you don't have a lot of space left in the 10mL vial after adding 8mL water. It needs some space to extract well. Does the QC2 vortexes and then stands for 20 hours?

It's slightly offtopic but why don't you use hexane? It will be easier to separate since it's on top, it's cheaper and safer. The only good thing about dichloromethane for this extraction is that it's on the bottom which is a good thing for a classical separatory funnel extraction...

I have some trouble imagining the whole automated process.. Why is there so much time between the extraction of QC 1 and 2?

Why do you spike the calibration curve in water just to extract it again? This makes sense for standard addition but if you do have extraction losses you're masking them (calibration standards have the same losses). In the case that your extraction is not 100%, i would add salt. We always add salt for our PAH extraction just to be sure.

Sorry, more questions than answers!! Don't feel obliged answering them. For your original question i think the remarks of Peter and Hornet make sense. But i guess you do use at least a couple of the heavier deuterated PAHs as internal standard. Are they spiked at the same time as the analytes in the QC?
First off all thanks for the replies, I’ll try to answer as much off the questions as possible. If you still lack information please ask.
Each PAH has its one labeled internal standard
I use a Thermo Triplus RSH autosampler to do the sample prep for me. It is able to do a prep-ahead, but only when the instrument method of the sample it is working on is the same as the sample it has to do next. This creates a problem with the calibration curve and drift standards. To make them the autosampler has to spike different amounts of calibration standard an thus has to use a different method each time.
To avoid waiting a long time, I spike my standards and QC’s at the beginning of the sequence. Method is as follows:
- add calibration standard
- vortex for 15 seconds to mix the standard in the water
- do a fake injection with a dummy GC- and MS-run of +/- 0.2 min to make it easy controlling handshaking and ready-signals
- spike the next calibration level
After all calibration and drift standards are spiked I simply process all vials the same as the unknowns. This is with the same instrument method en thus the prep ahead works fine. Since the spiking only took 6 min/sample, I gain approx. 5h doing it like this.
We have 2 types of QC’s in our sequence:
- Drift standards: spiked with calibration-standard. Analyzed as first injection, after every 15 samples and at the end of the sequence. As in the name, they only make sure the sensitivity of the instrument is the same throughout the sequence.
- Control standards: They are used to check the calibration curve. It is a standard from a different vendor. It is spiked manually to check for errors in spiking by the robot.
The extraction process is:
- Add internal standard
- Add 2x 500µl dichloromethane (syringe takes 2x550µl but only adds 500µl to avoid air bubbles)
- Vortex 2 min
- Centrifuge 2min
- Extract 250µl dichloromethane and transfer 150µl to a 200µl insert vial. If water is transferred along with the DCM, it will stay in the 100µl that is not transferred to the vial. I cannot add 200µl to the vial due to pressure buildup when filling the vial.
- Inject 75µl

It is definitely true that using hexane, pentane, or a mixture of those and dichloromethane makes life a lot easier when doing automated extractions. But we are starting off with the PAH-analysis to get some experience with the autosampler and the gc-msms (trace 1310, TSQ8000 evo). If this works we would like to automated our pesticide analysis (approx. 100 compounds) in the same way. For some pesticides using dichloromethane as extraction solvent is the only way to get a good recovery. Since I have to make a dichloromethane method for pesticides, I do the same thing with PAHs, if not I would have to optimize 2 methodes.
For now we don’t use salt. It takes a lot of time to add exactly the same amount to each 10ml vial. Since we are using dichloromethane and my needle has to get it from the bottom of the vial I don’t want to much solids on the bottom of the vial so over-saturating is not an option.
You say that your internal standards are OK, does this mean that labelled molecules are stable and unlabelled ones not ?

Peter
Peter Apps
Interesting method. You're right, hexane can't be used for everything (organophosporus pesticides for instance). Since we have way more samples for PAHs, we've chosen to miniaturize that using hexane extractions, and still do the classic extraction for organophosporus (just once a month or something). We plan to move these to LCMSMS once we get a new system and just straight up inject our water samples.

I guess you're worried about water entering the system especially with 75µL injections! And no one there to visually check if both phases are separated well... I've had 25µL of water, saturated with NaCl, injected a couple of times in our MS/MS system, it's a nightmare.. It's mostly the salt which you don't have though :lol: i advice to stay clear of salts for now!

For your original question: you say you ruled out drift issues by looking at the internal standard response of the QC 2. But what about the drift standard itself at the end of the sequence?
Peter:
You say that your internal standards are OK, does this mean that labelled molecules are stable and unlabelled ones not ?

The kalibration-standard is spiked at the beginning of the sequence. The internal standard is spiked directly prior to extraction. That's why the IS is ok and the calibration (used as drift in this case) isn't.

Rndirk:
You're right, hexane can't be used for everything (organophosporus pesticides for instance).

That is exactly the problem

Since we have way more samples for PAHs, we've chosen to miniaturize that using hexane extractions, and still do the classic extraction for organophosporus (just once a month or something).

PAH: 20-30/week Pesticides up to 60/week

I've had 25µL of water, saturated with NaCl, injected a couple of times in our MS/MS system, it's a nightmare.. It's mostly the salt which you don't have though

we started of with salt: so been there, done that, broken column, salt up into the ms, ...... tell me about it ;-)

For your original question: you say you ruled out drift issues by looking at the internal standard response of the QC 2. But what about the drift standard itself at the end of the sequence?

The first drift (1st extraction+injection of the sequence) is fine, the last one, after approx 24-25h is ok for the early eluters, but starting from +/- chrysene I only get 50% recovery. That is my biggest problem.


I've did some extra test in the meantime:
I've emptied the 10ml vial in which the last drift-std was spiked. allowed it to dry a little and than added 1ml of DCM to extract all PAHs that might be left on the glass-wall. -> not a trace of PAH left so it appears that sticking to the wall is not the cause.

I am wondering if breakdown of PAH by e.g. photodecomposition is a possibility. What I don't know, and can't find, is if this occurs with the lighter or the heavy PAHs (or both).

next week I will try and put up an experiment with spiked water that is kept in transparant bottles in de sun, in amber bottles in the sun and in amber bottles in the fridge. I hope this might give me an answer.
I will do this in double:
- 1: extraction in the glass bottle as it is.
- 2: empty the bottle and extract only the PAHs that has got stuck onto the glass.
By comparing both I should get a pretty good idea of what is going on.

Bart
I understand why you use drift standards like you say:

- Drift standards: spiked with calibration-standard. Analyzed as first injection, after every 15 samples and at the end of the sequence. As in the name, they only make sure the sensitivity of the instrument is the same throughout the sequence.


But why spike them in water? If they're off you don't know if its the sensitivity of the instrument or the extraction efficiency....

For the same reason i'm sceptical about spiking the calibration standards in water...
Rndirk wrote:
I understand why you use drift standards like you say:

- Drift standards: spiked with calibration-standard. Analyzed as first injection, after every 15 samples and at the end of the sequence. As in the name, they only make sure the sensitivity of the instrument is the same throughout the sequence.


But why spike them in water? If they're off you don't know if its the sensitivity of the instrument or the extraction efficiency....

For the same reason i'm sceptical about spiking the calibration standards in water...


Calibration standard spiked into the water would be equal to the Procedural Standards used in some EPA Drinking Water methods, which compensates for extraction loss directly with the processing of the calibration. It can be great or a nightmare, depending on the analyte.

From the original post it seems the internal standards are consistent but the higher level standards are not. This makes me believe that with increasing concentration you are not extracting all of the analytes from the water, but at low concentrations you can, unless your internal standards are spiked at the level of the high standard. It might be the ratio of solvent to water is too low to get complete extraction at the higher levels. I know when doing manual separatory funnel extractions you extract three times with solvent to increase extraction efficiency. I don't think you could automate that but maybe if you can increase the solvent volume slightly it might work. Also you are spiking the standards of the calibration with methanol, is it possible the increased volume of methanol is shifting the solubility of the analytes to the water/methanol phase instead of the DCM phase. If so you may want to try using a higher concentration of analytes in the methanol and spiking less into the water.
The past is there to guide us into the future, not to dwell in.
sorry for a late respons due to a bank holiday on thuesday we had a long weekend ;-)

we are a drinking water lab and do spike our standards in water to compensate for extraction recovery.

From the original post it seems the internal standards are consistent but the higher level standards are not.

the internal std is ok, the high calibrations standards are fine aswell, it is the drift at the end of the sequence that is to low. it has been spiked at aprox the same time as the one at the beginning of the sequence and that one was ok.

It seems like there is a loss of analyte during the +/- 24h between the spiking and the extraction. I've already emptied the 10ml vial in which the drift was spiked, and extracted the empty vial with DCM. There were no PAHs present, so I think they don't stick to the wall. I still have to try to do a second extraction to see how much is left in the water.

It might be the ratio of solvent to water is too low to get complete extraction at the higher levels.

I use 1ml DCM for 8ml of sample (when extracting manually we use 6ml DCM for 100ml of sample) so there should be plenty of DCM

Because our samples arrive as 100ml and we only use 8ml we have to add 5% MeOH in order to keep the PAHs from sticking to the glass/suspended particles, ...... To keep the matrix as constant as possible I alsow add 5% MeOH to the water that is used to spike the standards in. I don't think the MeOH is the cause since the extraction of the standards and the first drift is just fine.
That would leave photo degradation as a possibility. Have you tried setting it up to run over a weekend and making sure all of the lights are off in that area to see if the recovery increases?

Oh just curious, with that extraction, can you get down to 0.02ppb detection limit for the Benzo[a]pyrene? That is what we are required to report here and so far I am stuck at 1L volume to 1ml final extract and SIM to reach it. I have started looking into large volume injection to possible make up for a lower sample volume.
The past is there to guide us into the future, not to dwell in.
I've set up a rather large experiment to test what the cause might be.
I want to check the effect of:
* light/temperatur (room temp in light or dark in the fridge)
* effect clear bottle vs amber bottle
* effect of the amount of MeOH present.

I've spiked drinking water with 0-5-10% MeOH, and all with 100ng/l PAH
it is divided into 100ml samples:

* 0% MeOH is in clear and amber in the light at room temp and in the fridge (only in clear since there is no light in the fridge)
* 5 and 10% MeOH are in the fridge only.

Every day I take 2 bottles of each type. One is extracted with 6ml DCM to see what the recovery is. the other is poured empty and than extracted with 6ml DCM to see how much of the original 100ng/l PAH stick to the glass.

(note: the internal standard is added in the DCM because it can not be spiked correctly in an empty bottle)

I hope this might give some insight in what is happening.



We have to report 2.5ng/l (0.0025ppb) for Benzo(a)pyrene and that is about the limit when extracting drinking water. We use 8ml sample (containing 5% MeOH so 7.6ml 'real' sample), extract with 1ml DCM, inject 75µl, and detect in SRM mode on a Thermo TSQ8000_evo.
That is pushing it quite low. I haven't tried with such a large volume yet, but I will have to try that on the 7000C with MultiMode Injector in solvent vent LVI.
The past is there to guide us into the future, not to dwell in.
17 posts Page 1 of 2

Who is online

In total there are 3 users online :: 0 registered, 0 hidden and 3 guests (based on users active over the past 5 minutes)
Most users ever online was 234 on Tue Feb 06, 2018 7:33 am

Users browsing this forum: No registered users and 3 guests