Error due to thermal expansion of standards

[davidjr]

I am a chemistry hobbyist/citizen scientist. My lab temperature varies widely according to the weather (it is an unheated garage which is not especially well insulated/sealed). Range over the past year is probably about 4c to 29c, though usually on the lower end of that scale.

I have mostly been doing qualitative HPLC work, but now that I'm starting to do quantitative work, I'm concerned about errors being introduced due to density changes with lab temp.

In particular I am currently interested in formaldehyde analysis in ambient air, following EPA TO-11A. The method specifies that standard solutions should be prepared in acetonitrile, and also samples are to be eluted from cartridges using acetonitrile. Between 4c and 25c, a fixed mass of acetonitrile increases in volume by 3.4%. On the other hand, water increases in volume by only 0.29%.

If I prepare standards at one temperature and then use them later when the temperature has changed, obviously that is going to introduce an error. I'm less concerned about samples themselves since they would be prepared and then immediately analysed, so any temperature change between dilution to known volume and autosampler injection would be negligible.

What is the best approach to deal with this? Unfortunately, regulating the lab at a cozy 22c is not an option.

Is it a good idea to prepare the standards in ultrapure water rather than acetonitrile? I think DNPH-formaldehyde should be soluble at the concentrations needed.

Hydrolysis could be an issue but that would at least be detectable by the appearance of a DNPH peak where there was none before. I know DNPH derivatives are prone to hydrolysis in basic conditions, so maybe it would be worthwhile acidifying the water e.g. with 0.05% phosphoric acid. Apparently, DNPH-aldehyde derivitives are not prone to hydrolysis in acidic aqueous conditions, but DNPH-ketone derivitives are (see https://doi.org/10.1051/analusis:1999104)

[davidjr]

I tried preparing standards at 20ug/ml, 10ug/ml, 5ug/ml, 1ug/ml, and 0.5ug/ml using 10mM pH 5.0 acetate buffer as the diluent. Unfortunately, this appears to have resulted in partial hydrolysis.



So I will probably have to stick to using acetonitrile.

[James_Ball]

The EPA methods usually have QC pass fail in the 15-20% range just to allow for such problems. Even in multi million dollar buildings it can be difficult to hold the room temperature to close tolerances.

3.4% expansion will probably give less difference in the value returned by the instrument than just simple instrument variances will. If you want to be extra accurate you could store both standards and samples in a refrigerator at 6-10C and also chill the acetonitrile before making the final dilutions, then everything will be prepared in equivalent conditions, and any expansion that takes place while sitting in an autosampler would be accounted for since the standard and sample will undergo the same change in volume due to temperature. On the other hand I think you will observe that the results really don't change that much compared to just doing everything at ambient.

[davidjr]

I checked the TO-11A spec and it states:

• "13.2.3 Precision of response for replicate HPLC injections should be ±10% or less, day to day, for analyte calibration standards at 150 ng/mL or greater levels (as the carbonyl compound). At 75 ng/mL levels and below, precision of replicate analyses could vary up to 25%. Precision of retention times should be ±7% on a given day."
• "13.4.2 Precision for replicate HPLC injections should be ±10% or better, day to day, for calibration standards."

This does seem to allow for disregarding the error due to temperature change between standard preparation and use.

Thanks,
David