Is there an easier method? 

Some preliminary information: 

Okay, so I was asked to develop a method to detect, not purify, just quantify the concentration of N-Nitrosodimethylamine (NDMA) in blood plasma collected from whole blood of patients eating a nitrate rich diet ...

which should lead to the production of NDMA by N2O3 formation in vivo releasing NO[+] nitrosonium ion which reacts with amines in a not surprisingly nitrosation process. This produces nitrosamines of to my knowledge an unknown concentration. So I would like to be able to detect NDMA in plasma at low concentrations, even if there is no more than 5 nanograms / L present.

This is a pretty difficult task given that it's more easy to detect micrograms / liter concentration than nanograms / liter concentration, even with MS-MS methods in the literature it seems like the lower limit of quantification is too high about 10 times to 1000 times too high which is not acceptable. Most of the research papers out there analyze techniques to detect NDMA in water samples because NDMA is suggested to have carcinogenic behavior, so it would be a good idea to reduce public exposure to this chemical. (There are some studies of NDMA in blood already out there, but for some reason the researchers have closed off scientific access to their papers if you don't have an expensive subscription, so I can't depend on that.)   

However, I would like to detect it in blood plasma which is a more complicated matrix than the water sources studied in the literature, though plasma is largely constituted in water. Furthermore, I was able to find a scientific paper that was able to quantify 500 femtomolar concentrations of some aromatic compounds (Yuye Shi, Wenfang Liu, and Chuanpin Chen). This blew my mind because the limit of detection for this technique is more than 1,000 times lower than my goal! It essentially involves centrifuging a mixture of colloidal silver particles to get the particles with a larger surface area. This process is beneficial because it increases the concentration of the nanoparticles which correlates with an increased chance in analyte absorption. Then the analyte is mixed in with the particles and centrifuged to bind with the aggregated particles increasing the surface area. Then the concentration of the analyte is determined with surface enhanced Raman Spectroscopy. 

This is all good. But the two chemicals (phenformin hydrochloride and risperidone) they detected were highly aromatic nitrogenous ring structures of higher molecular weight than NDMA (74.08 daltons). So I don't if it would work with NDMA since its not that big or aromatic ... perhaps! I think that maybe if I bound NDMA to a phenformin and used phenformin as an internal standard then I might be able to quantify NDMA by subtracting the areas under the corresponding curves to determine the concentration of NDMA so that I know that I'm not mistakenly quantifying phenformin as NDMA supposing the binding is unsuccessful.

What I like about their detection technique is that it is a Raman spectroscopy technique which suggests that it is rather desensitized to detecting the water which makes up a great deal of the plasma matrix that I intend to purify. 

But the problems that I expect to face is that I only want to bind my analyte to the colloidal silver beads and not anything else in the plasma matrix which might crowd out the space for my analyte (NDMA) to bind. So I would like to get most of the interferents out of solution, though maybe if there is nothing else like my NDMA tagged phenoform-like molecule, then hopefully the interference at the ~ 1000 wavenumber Raman shift shouldn't be too overwhelming. (How could I actually give NDMA a tag like that?) 

-----------------------------

So here is my method: 

1) Collect the whole blood in heparin tubes to prevent coagulation.

2) Centrifuge at 5,000 g for a couple minutes to separate Red Blood Cells from Plasma.

3) Collect the plasma in the supernatant.

4) Pass the plasma through chitosan - PEO20k membrane by dialysis to remove most proteins and separate the small molecules like NDMA though the pores. (Note: the pores of this membrane are about 50 - 80 nm in size.) 

5)  By a liquid extraction, transfer the solution that passed the membrane into a solution of dicholomethane (or some similar solvent) to allow NDMA to pass into the aqueous phase since it is a very water soluble molecule.

6) Extract the aqueous phase

7) Prepare with centrifugation the concentrated colloidal silver particles separately. 

8?) Affinity chromatography? The receptors that bind NDMA appear to be pretty expensive. So if it didn't work or if the elution process (involving urea?) is ineffective, then I would be a couple hundred dollars wrong. Yeah, I don't know.

9) Combine the concentrated colloidal silver particles with the solution containing NDMA (NDMA with an aromatic tag?) and centrifuge to form the aggregated analyte complexes.

10) Extract the precipitate, redissolve the precipitate in supernatant.

11) Detect NDMA with Raman spectroscopy.

Note: Because of all the prep work I don't expect to get the femtomolar concentrations they reported, but if I could detect NDMA to the picomolar range, then that would undoubtedly be wonderful!   

------------------------

That's it! That's my method! Is there anything wrong with it? Are there any ways in which I could make it better?  

Is there an easier way!? 

It's no easy task detecting such trace quantities of NDMA from blood samples! 

Here is that research paper that I was so excited about: http://pubs.acs.org/doi/abs/10.1021/acs.analchem.6b01194