Chromatography Forum

Is there a rule of thumb when setting the equilibrium temperature for headspace analysis? For instance, if I am trying to detect dimethylformamide (bp = 153C) should my equilibrium temperature be around 153?

Thank you.

What solvent are you using?

The oven temperature will effect the concentration that goes into the headspace gas. As the oven temperature increases, the amount of gas entering the headspace from the sample increases, increasing the pressure in the vial and delivering more analyte to the GC. This increases the sensitivity.
When you pick an oven temperature, consider the following:
• Over-pressurization of the headspace vial could be dangerous.
• Unstable compounds could degrade at elevated temperatures.
Do not set the oven temperature within 10°C of the boiling point of any solvents in the sample, except in special cases.
Loop and transfer line temperatures should be set higher than the highest temperature the oven will reach in a programmed sequence.

I would never exceed a temperature 10C below the boiling point of my solvent when seeking to equilibrate DMF from solution.

I would prefer to keep the vapor pressure of the solvent low, even though the vapor pressure of the DMF would also be low.

If I were doing a total vaporization method I would use a temperature above the boiling point of the dissolution solvent or higher.

best wishes,

Rod

An analyte's boiling point is of no special significance in headspace analysis - the analyte's concentration in the headspace shows a smooth increase with temperature.

NB that Rod's guidance about keeping the equilibrium temperature (i.e. the headspace oven temperature) at least 10C below the boiling point of the solvent refers to the bulk solvent that is used to dissolve the sample - not to the individual analytes that might happen to be residual solvents.

Peter

Thank you for the replies. I am a beginner trying to piece together something that is complicated with limited resources so I really appreciate any guidance I receive.

I am working on a series of methods for detecting Toluene, DMF, Xylene and NMP at <350ppm with the solvents being majorly composed of MEK and MIBK.

For instance I had tested an equilibrium temperature of 80C, wasn't seeing any NMP (bp = 202C) and turned up the equilibrium temperature to 120C on advice from agilent (oops?) (Loop = 125, Transfer line = 130) and seen some NMP come off.

In regards to sample size, what would you guys recommend for a paint type sample consisting of around 30% solids? I had been told 2mL (yes 2 millilitres!) which I knew was too much. I have microliter positive displacement pipettes available to me capable of delivering microliter increments from 1-10ul and 20-50ul.

Initially a method with 25ul sample was working great, now that I have moved to difference solvent blends I am having trouble due to interfering peaks, shouldering etc.

Thank you!

Just so I'm clear - you are trying to detect ul/l levels of analyte in two volatile ketones, by headspace ?

The problem is that you will always have very high concentrations of your volatile bulk solvents in the headspace, and this will give huge peaks that you are not interested in, resolution problems and analyte peak distortions due to solvent effects.

The attraction of headspace is that the solids and dissolved resins do not interfere. You might be able to retain that advantage if you go to complete evaporation which prevents the high volatility of the bulk solvents from making their peaks even bigger, in which case you need a small sample (Rod will probably advise), which you should weigh rather than relying on volumetrics. I suspect htough that the lower limit of detection might be too high.

Peter

Hello Peter, thank you for your reply.

In summary yes. The solvent combinations actually vary by the product with some containing MEK, MIBK plus alcohols such as Isopropanol, butanol etc.

In the mean time I have had success to an extent with one solvent combination but as you said the solvent peaks can be quite large and since there can be up to 5 "solvents" one is bound to overlap with one of the four contaminants we are trying to detect and quantify at ul/L concentrations.

At the moment I am trying to figure this all out based on my undergraduate experience which is obviously limited. I believe I have gotten it to the point where I am testing the limitations of a poorly maintained piece of equipment that requires servicing.

I have to say that quantifying ul/l concentrations of solvent impurities in mixtures of other solvents seems both unrealistic and uneccessary. There are ppm impurities in the highest grades of laboratory solvent and I struggle to believe that the performance of industrial chemicals can be adversely affected by impurities at these levels, which are more appropriate to residual solvents in drugs. Is the concern that the trace impurities will have adverse effects on people exposed to drying paint ?

Did your predecessors in the company achieve these detection limits with headspace ? - if they did just copy their method.

Peter

Peter, you are correct that the concern is trace exposure to the select contaminants and you are preaching to the choir in terms of the arbitrary specification set for this product.

There is no predecessor to follow. When I came it was routine to know that when the GC report said THF it actually meant ethyl acetate . There was never any quantitative work done, only approximate normalized percentage reports were required.

Your first step needs to be to sort out what can be separated from what - there is no point working out a headspace method unless you can see the peaks.

Go back to ordinary liquid injections - disconnect the headspacer and make up % level solutions of each target analyte in each bulk solvent. Inject these into the split splitless inlet, using liner with a wisp of glass wool it it, with a split ratio of 50:1 and see which of the analytes you can separate from each of the solvents. For those that do not separate you will need to try a column with a different stationary phase.

Once you have separations of analytes from solvents and solvent mixtures, then dilute the solutions by 100 times and re-inject, dilute the 1/100 dilution by 10 and re-inject, dilute by 10 again etc etc (since by this stage you know which analytes come out when you can make mixed standard solutions to save time) to get an idea of the lowest concentration that you can detect with an injection of a liquid standard. If your detector can give a decent peak with 1 ng, and you inject 1 ul you should get down to about 50 ul/l with a split of 50:1.

Now re-attach the headspacer and run the % standards again, if necessary swapping columns to do the various solvent - analyte combinations, and work back down in concentration.

Peter