I'm not aware of any race that makes smite evil more powerful, though Aasimar might be an obvious choice because it has higher charisma and because its favored class is paladin.
Paladin smites themselves, since a paladin gets only one per day (extra smiting makes it five, I think) are not going to be that impressive. However, holy martial strike, an epic level feat which requires smiting and which does 2d6 divine damage to evil through any melee weapon--now that's a feat to be sure you get.
If the paladin will encounter a lot of demons, Knight of the Chalice is a good prestige class (adds 4d6 divine damage against demons to any melee weapon and holy aura vs. demons on command).
If smiting really is your focus, consider Fist of Raziel. The Fist gains more smites and acquires the ability to use the smite to do holy damage to up to five targets (the one he hits and others nearby). A neat twist if you want to use it is that the Fist of Raziel extends spellcasting for the highest divine class. If you take a few paladin levels, then switch to cleric and make sure you have more cleric than paladin, the Fist increases cleric caster level and spells known at level up. Then you have a holy warrior who can cast 9th level spells and even epic spells if you want. Think of it as a cleric who fights better than the norm or a warrior who casts spells far beyond what a typical paladin type could do.
If you are campaigning against undead, there is no paladin prestige class that specifically addresses that. You might almost be better off with a fighter/soldier of light/champion of torm build. Soldier of Light can convert turn undead attempts to positive energy bursts that harm undead. The soldier also gets an energon campanion who becomes more powerful based on total character levels, not just soldier of light levels. The COT levels you use to get smite evil; the soldier's smite undead doesn't qualify you for holy martial strike later. You can adjust the balance to more fighter or more soldier of light depending upon the needs of different modules.