A good gambit idea for Black in French Defence?

Here's one of the 'sharpest' games I've ever played with my seemingly (and usually) 'quiet' pet opening, the French Defence.

On move 7 I 'invented' a gambit idea, because the book moves which I knew about this position seemed to favor White.

Of course I've analyzed the game with a strong program, but it couldn't tell me for sure whether my gambit is sound, and the program seems to see some compensation for Black. Both my opponent and I were rated at ~1800 national Elo, so until White's blunder in time trouble 25.Qxc5?  he seemed to play reasonably well (at least tactically).

However, with the Black pieces I was still able to generate some strong attacking chances on White's king, so I'm wondering what kind of positional blunders by White were the reason for  White's problems, and what should I expect from stronger players if I try this gambit on them in future games? (Of course stronger players don't play the C01-French as often as we amateurs).

I've added many comments to most moves after Black's 7th move, which was my 'novelty'/gambit, but whether it's sound or maybe even good - I'm not sure, the way things worked out for Black in this game looks like the gambit is maybe even dangerous for White. A way for Black to fight for an advantage?

Well, I have never in my "chess career" (lasting for about three years :D) seen this opening. As White I would undoubtely play 2.d4, then after d5 I like Nd2, or something different what is not Advance variation. But ok, let's see.

I can say I am a gambiteer, or at least a huge fan of gambits. It is said that when you don't sacrifice a pawn you cannot win :) But with your gambit I see a big problem - 9.d5. Where will the knight go? after 9..Na5 10.a3 and black has to waste some tempi. The only alternative is Nb8 which kills the whole gambit-wanna-be thing.

So until you provide some response to 9.d5 my opinion is that this sacrifice of a pawn is actually a loss without any compensation.

EDIT: Oh well, now I see 9..Nb4... Nice example of tunnel vision. But well, after 10.Bc4 you can't take that pawn and must consider the threat a3. for example 10..Bf5 11.Bb3.

9.d5 Nb4 10.Bc4 Qc7 11.Bb3 Bf5 12.0-0 Rad8 is rather equal, and certainly easy for Black to play.

 The opening transposes to a French Exchange Variation, where it's actually White who wants to avoid the drawish old lines when they play an early c4 - accepting an IQP in hope of getting the initiative. So here as Black I tried to take the initiative away from them (otherwise Black can also try to play a somewhat stereotypical setup against the IQP, of course that's generally more solid, but it also gives White a lot of initiative and a good game.

So the d5 line was actually what I was hoping for, to see whether White can defend their IQP on d5 at all.

Generally the side with the IQP is supposed to advance the IQP when they can. But here the pawn seems to be weaker on d5 than on d4. In any case if Black manages to regain the pawn the position is totally open and if anyone then Black should have a slight advantage due to a lead in development (but a draw is very likely).

Looks fine to me.

Saccing c3/c6 or f3/f6 is a fairly common idea throughout many openings because it's a simple gain of a tempo if recaptured with a knight.
 
 

I often encounter this position in the Nf6 Scandinavian.  When I first starting experimenting with the opening, I thought c4! was a clear refutation. I was dumbfounded to later find only a handful of master games recorded in this line.  "How does black gain compensation for the pawn?", I wondered.  In this position, there is no clear plan for black to regain his pawn.  He is simply a tempo or two up (depending on if you consider c4 as valuable as a knight) and is going to force white to defend a backward pawn on d3 all game.  Houdini says black has a slight advantage, and has at least equalizing in the above diagram.

To give you another example, take a glance at the Smith-Morra- another opening that I am fond of.

 Whether or not the Morra is sound is very much debated, but pfren has told me he considers it solid and he knows his openings.  It seems like the literature on the Morra is ever-changing as black and white find new ideas on how to combat each other in it.  Many would scoff upon seeing it, "Blacks simply takes his time in a passive position and holds the pawn until the endgame!"  But white gets much more than two tempi.  Black is almost forced to play a6 in the first couple of moves to avoid tactical tricks, he has trouble developing his knight to f6 because of the dangerous e5 push, and his queen doesn't have any good places to go because of the open c file that white is sure to control with his rook (where black would usually like to put her queen on c7).  Couple this with the fact that black's bishop have a hard time leaving the first two ranks and the weak d pawn that will be attacked and I think most would give white at least a small advantage.
 
 

@Daeru

The idea 12.b3!? leads to really crazy variations where only the computer seems to know what it's doing.

On the above lines (9...Nb4 10.Bc4 Qc7 11.Bb3 Bf5 12.0-0 Rad8) 13.Be3 fails to do something due to 13...Nfxd5 14.Bxd5 Be6 15.Nd4 Bxd5. However, 13.Bg5! h6 14.Be3 is an improvement, as now on 14...Nfxd5 15.Bxd5 Be6 16.Nd4 Bxd5 white has the strong 17.Qg4.

So, Black has to take on move 16 with knight, although again white has "something" after 17.Nxe6 fe6 18.Nxd5 Rxd5 19.Qg4.