[2lrigbal]

I am still working on finding the concentrations of Arsenic and Selenium. Here's what I am working with: AAnalyst400 atomic absorption spectrometer and MHS 15 hydride generating system. 

My question is this:  Does anyone know of a way to find get consistent results to find the concentration of As(up to 5ppm) and Se(up to 1pmm)?  My samples are TCLP liquid extracts.

The problem I have with the hydride generating system is that it measures in concentrations for As up to 50ppb.  I need to reach concentrations up to at least 5ppm.  For Se, it measures up to 100ppb, I need to find up to 1ppm.

I thought that I could perhaps use flame AA with a nitrous oxide-acetylene flame.  I am having trouble getting consistent results.  The characteristic concentration/sensitivity check using is 25ppm for As.  That concentration is given if I use the 10cm burner head with an air-acetylene flame (very poor results).  That means the with the NOX flame, the absorbance would be half that, or 0.1.  That said, the absorbance of a 5ppm solution sould be 1/5 that of the 25ppm solution, or 0.02.  Additionally, using the 5cm burner head requires installing the spacer, which reduces sensitivity, but minimizes chemical interferences.  The spacer reduces sensitivity by 2-3x.  So that brings our absorbance down even lower.   I am using the background corrector to improve the signal to noise ration, but I am still not getting consistent results. 

Any suggestions? Is there another alternative?  I don't think they'll jump at getting me an ICP-MS, so something relatively inexpensive as well...  (I can't get a graphite furnace either...)

[enahs]

That said, the absorbance of a 5ppm solution sould be 1/5 that of the 25ppm solution, or 0.02.

No. You can not assume your detector has a linear response like that over such a large range unless it is specified to, or you have tested it. That could be some of your problem as well.

Use an internal standard to get more consistent data. Just googeling internal standard gives plenty of results to explain how; but I am sure you will hit your head and say "duh".

[2lrigbal]

Thanks enahs.  I discovered that the background noise is not due to complex sample matrices, but the spectrometer.  I can create a calibration curve with standards and run one of the standards again and get a slightly different result.  The recommended conditions that says the sensitivity check for As is 25ppm says that there is 1ppm relative noise.  Can using an internal standard still help in finding 5ppm?  If so, could I just spike the samples and use MSA instead?

[enahs]

If you have a low signal-to-noise ratio, an internal standard can help some, but will not solve it completley. However,  you can do multiple standard curves, as well as analysis, and with a little statistics account for a lot of the noise.

Though, a question. You say for example the equipment is designed for up to 50ppb for As but want to read 50ppm. And for Se it is designed up to 100ppb for Se but you want to read a 1ppm solution.

The solutions you want to read are too high of concentrations (ppm is more concentrated then ppb). Why can you not just dilute down into the optimal range of the equipment?

[2lrigbal]

I thought that I would probably have to do that for Se since my samples wouldn't have to be diluted as much and also because I wouldn't be able to run FAA due to the background noise being the same as the concnetration.

But wouldn't that be too much of a dilution factor for As concentrations to be accurate?  Would I be able to trust my results if I diluted it that much?

[enahs]

Ohh yes, a thousand fold dilution is no problem at all. It all comes down to the quality of your instruments and your technique. A thousand fold dilution does not require super special equipment, just good glassware.