I really don´t know if Sadler made a desperate attempt (´tho choosing a game against Morozevich to do so is very bad for the health generally) or he could just calculate or theoretize that giving up two central pawns (one of them being a passer) will translate into winning chances for him.

Sadler's quote regarding his pawn sacrifice 21..d5:
"...Black's idea is first to free d6 so that he can point his bishop against White's king in combination with the queen on c7, and then to break with ..e4 to open the b8-h2 diagonal. White's pieces are very awkwardly placed, which adds to the strength of the idea. In my game, this idea is even more dangerous as the pawn on h4 cramps the White kingside."

31 c3!? was an interesting idea that did not work out well; after 31 c4 White would have had good chances. Sadler's 33..Rf4! was the strongest way to continue the attack; Sadler showed that the obvious 33..Be5 could have been answered with 34 Bxe5..Rxe5 35 Bb3..Ng4 36 d6..Qxb3 37 d7..Qe6 38 Re8..Qb6+ 39 Rd4 with an unclear position. In fact, Dvoretsky later showed that this line is winning for White after 39..Qc6 40 Rxe5..Rxe5 41 Qd2..Qe6 42 Rxg4. Morozovich continued mistakenly to try and trap the queen with 34 Bb3?; better was 34 Be2..Rhf5 with unclear play. Note if 36 Ra2..then 36..Nxf2 37 Qb1..Ng4 38 Rxa3..Rf1+ 39 Qxf1..Bxh2+ 40 Kh1..Rxf1#.