While I hate to say it, the movie "Armageddon" had the right idea. No, no, hear me out.
Let's assume that Poten (mass = 5 x 10^18 kg) is set to make bullseye impact - that is, its' trajectory passes through the center of the earth. Let's say that, 2 years before impact, a rather large nuke is set off at Poten's core. If large enough, this will produce an approximately spherical shell of debris. If the nominal diameter of the shell is 10 times the earth's diameter, then only about (2 x pi x R^2) / 4 x pi x (10 R)^2 of the total mass will impact the earth, where R is the earth's radius, about 6400 km. For these numbers, the expected impact mass is decreased by a factor of 200, for a total of about 2.5 x 10^16 kg.
The velocity required to get the shell to this speed is (6.4 x 10^7) / (6.2 x 10^7) m/sec, which just happens to be 1 m/sec, since a year is 3.1 x 10^7 sec.
The energy required to produce this velocity is, of course 1/2 mv^2, which in this case is about 2.5 x 10 ^ 18 J. Since 1 MT is equivalent to 4 x 10 ^ 15 J, we'll need about 600 MT. The largest nuke ever set off was the Tsar Bomba, and at 50 MT it was crippled by being configured to 50% of theoretical yield. So we'd only need 6 full-up Tsar Bombas to do the job.
For the next 3 months, all efforts are concentrated on tracking the fragments, identifying the ones still on a collision course, and prioritizing them by mass. Then a second wave of Bruce Willises arrives and sets more nukes, and the cycle repeats. There's some uncertainty as to how big the actual debris chunks are - the absolute best case is that only a few pieces are produced, and those will all miss. But let's say that the largest pieces are 1% of the original mass. Then you only need 6 MT per chunk for a successful deflection.
The process continues for the next year, and assuming a 99% deflection rate per 6 months, the mass impacting on earth will be on the order of 2.5 x 10^10 kg. This will, of course, be pretty nasty, but it's a million times less than Chicxulub. Just as important, the arriving swarm will consist of multiple objects rather than one big one, and the total effects will be much less, since atmospheric effects will be much more pronounced, and a lot of the smaller pieces won't make it to the ground. At a minimum, the impacts will be divided into two separate waves, corresponding to the leading and trailing surfaces of the debris shell.
All of this, of course, depends on getting the nukes to the right place at the right time, and I'm not real optimistic about this. But the priniciple seems sound.
And, as note, the ability to plant the nukes at the core of Poten and its brood is problematic but not necessarily fatal. It's not actually necessary to get to the exact center. An off-center blast is actually better, as long as the asymmetry axis is perpendicular to the line of motion. The result is a small, fast ejection in one direction with a larger piece moving more slowly in the opposite direction. If you can get deep enough to produce something like a 10-1 imbalance, you can essentially solve the whole problem with one shot, although the bomb has to be a bit bigger. As the pieces get smaller, you can get away with bigger asymmetries as long as you use bigger bombs, and with an upper limit of 100 MT per nuke that's not a problem.