P/60 is good for looking at efficiency, but can't be assumed to hold in more minutes. Also, currently Matthews' P/60 is 2.92, while Barkov's is 2.43, but this is with Matthews' on-ice shooting percentage at an unsustainably high and league-leading 12.98%. Barkov meanwhile is at a fairly reasonable 9.04%. If Matthews' on-ice shooting percentage goes down to an even 10% and he gets a point on the same percentage of goals, this would put his P/60 at 2.26. Considering last year their P/60s were 2.18 for Matthews and 2.2 for Barkov while both were on more reasonable 7.69% for Matthews and 8.52% for Barkov, the idea that there's a great disparity in their efficiency seems flawed.
Now, maybe you want to argue that Matthews' game leads to more high end chances and thus naturally a higher on-ice shooting percentage. But Matthew' xGF isn't nearly as impressive as you suggest. It's currently at 42.43, which is below the 44 goals a 10% on-ice shooting percentage would give him that I calculated above, and far below the 57 he's currently at. And his xGF% is roughly the same as Barkov's (53.02 to 52.15), but lower relative (3.16 to 5.21).
And I just saw that Corsica has quite different numbers for Barkov at 5v5, putting him at 2.23 P/60. But again, that's basically the same as Matthews is his underlying numbers weren't highly inflated.