Anything BP-like and fairly high in sulfur will do that. The initial reaction produces liquid polysulfide dross which then slowly reacts with atmospheric oxygen, producing a sulfate. Once the temperature rises high enough (around the sulfate melting temperature) and there is sufficient accumulated sulfate it reacts with whatever reducing agents are left unreacted in the polysulfide melt, typically excess carbon or metal powders which have been basically inert throughout the smolder-phase.
It is a source of very dangerous hold-fires at times. Consider this early experiment of mine:http://www.vk2zay.net/device.php?id=42
Although I didn't understand it at the time, the same basic process was involved here. The composition was essentially high-sulfur nitrate flash - Potassium Nitrate, Sulfur and Aluminium. The primer dross warmed up the 'nitrate flash' and started a slow 'dark' reaction which converted a large percentage of it to an unstable moltern oxidiser/polysulfide/Aluminium mixture which then reacted basically all at once rather than burning from one end. The resulting reaction was extremely violent despite the granular nature of Aluminium!
Try this, it is a "delayless" glitter composition, which is basically a flitter:
6 Potassium Nitrate
1 Charcoal (airfloat)
1 Aluminium (atomised)
It burns a large percentage of the Aluminium silver if used in a gerb or other kind of driver. It is fairly low dross and probably works OK as a star too, but I haven't tried that yet. It is *not* a very good composition, the numbers are all rounded and it is lacking some basic improvements, but it is a good base composition to demonstrate glitter reactions, test Aluminium batches and delay agents by making a quantity up and dividing it into test lots with different variables.
If you subst some of the Potassium Nitrate for Barium Nitrate you will start to get delayed flashes. Adding a half- part or so of Iron or Antimony Sulfide should start it glittering quite well, as should an Oxalate or Bicarbonate. Adding metal oxides brightens the silver sparks or flashes. It is an instructive exercise, but probably best done by engineering the compositions carefully using balanced equations for what you want as a product of each phase. One of these days I'll write up a worked example on my site, or maybe code it up in an online calculator.