RSD% failure - PAHs

Hi jerivera,

If you are still having problems, a couple points to consider:-

Your instrument, when it was new, would have been capable of detecting 10pg on column (in scan mode for 8270) for many PAHs (generally the late eluters would be worse). The top range should be in the order of >10ng on-column giving you a calibration range of ~1000:1 - Assuming you are using a 1-2µL injection and your bottom calibration of 0.5 corresponds to ppm concentration, that implies that you are injecting 0.5-1ng on column - The top range is 140-280ng? - This at the top capacity of your 1µm column and is close to saturating the system, suggesting that you inject less. Your low level calibration responses are <20,000, so if you do inject less, you might want to put the gain up. Allowing for that, I think that most of your problems are at the MS end of the analysis. As you know, 8270 extracts are often very dirty and contamination can quickly build up inside the injector, column and MS. If you ignore your lowest calibration point, the higher points can be plotted as reasonably straight line (These lines would cross the origin at ~+0.3 for 1-methylnaphthalene and benzo[k]fluoranthene and ~+0.1 for chrysene). This can often be caused by background chemical noise ("dirt").

Are you only analysing PAHs in these runs? They are very stable and the source temperature can (and should be) higher - As trifector posted, 350C is normal for the source temperature for this work, and you could put the Quad temp up to at least 180C which will help with background contamination build up. If you can put the EI Energy above 70eV it should also help as it will give normal spectra for the stable PAHs, but more fragmentation with background ions.  Like trifector, I would avoid hydrogen carrier gas on an older instrument  - If you do try it, it tends to remove existing contamination from inside the system and it might take days before you get a clean background. If you are analysing a lot of dirty samples, to help keep the source clean, you might consider using a post-column backflush fitting and turning the backflush on a minute or so after benzo(g,h,i)perylene elutes.

I suggest: 1) Cleaning the source; 2) Inject less sample - This will also put less contamination into the system (Hot split injection as suggested by Ron Tackett, but >=10:1) and turning the detector gain up - If the curve got worse, it indicates possible background contamination.

If you still can't get a good linear calibration, as you say, try quadratic: The new 8270E method, recently promulgated by the EPA, states in Section 7.7.1 "ICAL standards must be prepared at a minimum of five different concentrations from the secondary dilution of stock standards or from a premixed certified solution. Include a minimum of five different concentrations in the calibration for average response factor or linear (first-order) calibration models and six different concentrations for a quadratic (second-order) model with the low standard at or below the lower limit of quantitation (LLOQ)". Visual inspection of your curves indicates that you should get reasonable quadratic calibrations...

Regards, Tim

Hi jerivera,

If you are still having problems, a couple points to consider:-

Your instrument, when it was new, would have been capable of detecting 10pg on column (in scan mode for 8270) for many PAHs (generally the late eluters would be worse). The top range should be in the order of >10ng on-column giving you a calibration range of ~1000:1 - Assuming you are using a 1-2µL injection and your bottom calibration of 0.5 corresponds to ppm concentration, that implies that you are injecting 0.5-1ng on column - The top range is 140-280ng? - This at the top capacity of your 1µm column and is close to saturating the system, suggesting that you inject less. Your low level calibration responses are <20,000, so if you do inject less, you might want to put the gain up. Allowing for that, I think that most of your problems are at the MS end of the analysis. As you know, 8270 extracts are often very dirty and contamination can quickly build up inside the injector, column and MS. If you ignore your lowest calibration point, the higher points can be plotted as reasonably straight line (These lines would cross the origin at ~+0.3 for 1-methylnaphthalene and benzo[k]fluoranthene and ~+0.1 for chrysene). This can often be caused by background chemical noise ("dirt").

Are you only analysing PAHs in these runs? They are very stable and the source temperature can (and should be) higher - As trifector posted, 350C is normal for the source temperature for this work, and you could put the Quad temp up to at least 180C which will help with background contamination build up. If you can put the EI Energy above 70eV it should also help as it will give normal spectra for the stable PAHs, but more fragmentation with background ions.  Like trifector, I would avoid hydrogen carrier gas on an older instrument  - If you do try it, it tends to remove existing contamination from inside the system and it might take days before you get a clean background. If you are analysing a lot of dirty samples, to help keep the source clean, you might consider using a post-column backflush fitting and turning the backflush on a minute or so after benzo(g,h,i)perylene elutes.

I suggest: 1) Cleaning the source; 2) Inject less sample - This will also put less contamination into the system (Hot split injection as suggested by Ron Tackett, but >=10:1) and turning the detector gain up - If the curve got worse, it indicates possible background contamination.

If you still can't get a good linear calibration, as you say, try quadratic: The new 8270E method, recently promulgated by the EPA, states in Section 7.7.1 "ICAL standards must be prepared at a minimum of five different concentrations from the secondary dilution of stock standards or from a premixed certified solution. Include a minimum of five different concentrations in the calibration for average response factor or linear (first-order) calibration models and six different concentrations for a quadratic (second-order) model with the low standard at or below the lower limit of quantitation (LLOQ)". Visual inspection of your curves indicates that you should get reasonable quadratic calibrations...

Regards, Tim