Non-standard pawn sacrifices

We all know about the common types of pawn sacrifices, e.g. pawn sacs to open lines for an attack, usually conducted against the king; sacrifices to create an important outpost for a piece (these are the positional sacrifices, so to speak); sacrifices to disrupt the enemy pawn structure and sacrifices to gain time to build up an attack. It's instructive therefore to consider other positions where, at first glance, a pawn sacrifice may not look obvious or sound, but upon deeper analysis turns out to be the only correct route.

One such kind is the pawn sacrifice whose object is merely to gain time to improve the positioning of the pieces, with no immediate plans for attack. These can occur in situations where one side is up material but is behind in development or has inferiorly placed pieces; then often the best course of action for the side is to choose a suitable time to give back the pawn and using the time gained improve the piece positioning. It is really tempting in such positions to try to hold on to the extra pawn. This is even more so because often this can be the most effective course of action, but sometimes returning the material can be the only way to gain a decent position.

Here is a position from the recent Topalov-Svidler game in Linares a few days ago which illustrates the above ideas.

Svidler has just played 19. ... Qa5, attacking the a3 pawn. Here the obvious move is 20. Qc3, protecting the pawn and offering an exchange of queens while ahead in material. Qc3 looks like a very good, solid move. I know that a lot of players here would play Qc3 without hesitation; after all, what can be more natural than to protect the attacked pawn and forcing the opposing queen to move in order to avoid the exchange. However, after 20. ... Qa4!, it turns out that matters are hardly as simple as that. White's pieces are passively placed; his d3 bishop and e2 knight are particularly badly placed. The constricted position of the d3 bishop means that the white rooks and queen are tied down to its defence. The c4 and a3 pawns are static weaknesses; and the knight on e2 has no good squares except c3 and d5, but the queen occupies c3 now! Meanwhile, Black's pieces are actively placed and he has threats like Rb8 or Rad8, and Nd4 is an annoying threat. White may not be actually worse here; but it will take lengthy manoevres for him to untangle his pieces, and it seems unlikely he will be able to exploit his extra pawn.

So Topalov, after 19. ... Qa5, played the stronger 20. Nc3!, and after 20. ... Qxa3 21. Nd5 (see position below), the transformation in the position is striking.

Material is now even, but it is obvious now that White is better. The knight on d5 is very powerful (and removing it with Bxd5 leaves White with a strong passed pawn on d5). The c-pawn is no longer a weakness but a strong passed pawn. The bishop on d3 is no longer under attack by the d7 rook, which now bites on granite. Meanwhile, Black's queen is vulnerable on a3, where it can now be attacked by a rook coming to a1, and the a7 pawn is a big weakness. White here has a sizeable advantage, and in fact Topalov went on to win a very nice game. An excellent example of giving up extra material for piece activity.

Such decisions come naturally to strong GMs, but for us ordinary mortals, it can often be difficult to correctly evaluate positions like this where the side with the more material may not always be better off. That is why it is important to study games played by masters, in order to gain a 'feel' for such positions so that when they arise in our own games we are comfortable playing them.