It really does depend more on your instrument than anything else, and your instrument manufacturer will be ale to help. For example, on the PE-FTIR that I use, 5 mg of a liquid would easily cover the ATR crystal. But doing that with a powder might not be enough to insure a clean signal. And I don't know if your instrument will let you use an IR card, but you may likewise get minimal signal if you all you can do is "stain" a KBr plate, ith a dilute evaporated solution.