This page describes how to do a flame test for a range of metal ions, and briefly describes how the flame colour arises.
Flame tests are used to identify the presence of a relatively small number of metal ions in a compound. Not all metal ions give flame colours.
For Group 1 compounds, flame tests are usually by far the easiest way of identifying which metal you have got. For other metals, there are usually other easy methods which are more reliable - but the flame test can give a useful hint as to where to look.
Carrying out a flame test
Clean a platinum or nichrome (a nickel-chromium alloy) wire by dipping it into concentrated hydrochloric acid and then holding it in a hot Bunsen flame. Repeat this until the wire doesn't produce any colour in the flame.
Note: There will, in fact, always be a trace of orange in the flame if you use nichrome. You soon learn to ignore this. Platinum is much better to use, but is much, much more expensive. If you have a particularly dirty bit of nichrome wire, you can just chop the end off. You don't do that with platinum!
Dilute hydrochloric acid can be used instead of concentrated acid for safety reasons, but doesn't always give such intense flame colours.
When the wire is clean, moisten it again with some of the acid and then dip it into a small amount of the solid you are testing so that some sticks to the wire. Place the wire back in the flame again.
If the flame colour is weak, it is often worthwhile to dip the wire back in the acid again and put it back into the flame as if you were cleaning it. You often get a very short but intense flash of colour by doing that.
The colours in the table are just a guide. Almost everybody sees and describes colours differently. I have, for example, used the word "red" several times to describe colours which can be quite different from each other. Other people use words like "carmine" or "crimson" or "scarlet", but not everyone knows the differences between these words - particularly if their first language isn't English.
What do you do if you have a red flame colour for an unknown compound and don't know which of the various reds it is?
Get samples of known lithium, strontium (etc) compounds and repeat the flame test, comparing the colours produced by one of the known compounds and the unknown compound side by side until you have a good match.
Note: I don't have confidence in the caesium flame colour. There is total disagreement about this on the web and in the books I have looked at, and I have never seen this flame colour myself. However, I have received a helpful email from a student who says: "At my school we did some flame testing experiments, and . . . caesium is actually either blue or violet, depending on the way you look at it. I think it looks more violet than blue, but it sort of changes each time you do it." (Kara Gates, March 2006). If you thought chemistry was clear-cut, you are sadly mistaken!
The origin of flame colours
Flame colours are produced from the movement of the electrons in the metal ions present in the compounds.
For example, a sodium ion in an unexcited state has the structure 1s22s22p6.
When you heat it, the electrons gain energy and can jump into any of the empty orbitals at higher levels - for example, into the 7s or 6p or 4d or whatever, depending on how much energy a particular electron happens to absorb from the flame.
Because the electrons are now at a higher and more energetically unstable level, they tend to fall back down to where they were before - but not necessarily all in one go.
An electron which had been excited from the 2p level to an orbital in the 7 level, for example, might jump back to the 2p level in one go. That would release a certain amount of energy which would be seen as light of a particular colour.
However, it might jump back in two (or more) stages. For example, first to the 5 level and then back to the 2 level.
Each of these jumps involves a specific amount of energy being released as light energy, and each corresponds to a particular colour.
As a result of all these jumps, a spectrum of coloured lines will be produced. The colour you see will be a combination of all these individual colours.
The exact sizes of the possible jumps in energy terms vary from one metal ion to another. That means that each different ion will have a different pattern of spectral lines, and so a different flame colour.