Internal Standards for Solid Samples

[HILP]

Dear all,

for the last 15 years I have analysed with GC exclusively liquid samples with a variety of sample prep methods.

For the first time I am now confronted with solid samples. It appears that there are problems related to adding internal standards to complex, heterogeneous solid samples. The sample techniques being used are headspace SPME and various dynamic headspace methods. The samples in question are mostly meat and meat product samples, but also some extruded biscuit-type samples.

The analyses are presently being done without any internal standards during the extraction process, which concerns me (in addition to the usual extraction variance I am sure that the matrix will have a large effect on the partition coefficients). However, I am told that this is normal with solid samples.

My possibly naive view is that the samples could be homogenised, a liquid internal standard added, the vial closed and then the headspace extraction carried out. If necessary, a nitrogen atmosphere could be used to avoid oxidation of the sample material. Does this make sense?

If anybody has experience in adding internal standards to this kind of sample matrix, I would welcome any help you could give me.

All the best,

HILP

[Consumer Products Guy]

Yes, the samples NEED to be homogenized first, that's what my company always did.

[Peter Apps]

The principal problem is to get the internal standard mixed with the solid sample and in equilibrium with it. You need to homogenize the sample after the standard is added, and then give it as long as possible to get into equilibrium - the main constraint is that diffusion in solids is very slow.

The slow diffusion also causes problems during headspace and purge sampling, so it is way better to turn the sample into a runny slurry by homogenising it with water (which is the major part of a meat matrix in any case) before you do anything else with it.

Alternatively you can avoid having to use internal standards by doing multipe injections from one sample, equilibriating between each injection, but this is so time consuming that it really a method of last resort.

Peter