They do NOT combine to produce nothing, they annihilate each other producing either a different particle-antiparticle pair (based on what is permissible relative to various conserved quantities) or electromagnetic annihilation quanta (photons) whose combined mass-energy is equal to the mass-energy of the original particle-antiparticle pair. Mass-energy over and above the combined rest masses of the pair is generally expressed as kinetic energy. You are conceptually confusing antimatter with negative mass.
That is because of a derivative of Heisenberg's Uncertainty Principle:
ΔE · Δt ≥ ħ/2
where:
ΔE = Uncertainty in energy magnitude
Δt = Uncertainty in time
ħ = h/2π, and h is Planck's constant
If the uncertainty in a particle's total mass-energy is large (i.e. it can potentially have a value over a very large range of possibilities), then the time-range during which it can exist as a virtual particle (before the pair annihilates) must be very small.
The reverse is also true:
If the uncertainty in a particle's timeframe of virtual existence is large (i.e. it can potentially have a value over a very large range of possibilities), then the range of mass-energy values that the pair can possibly have as virtual particles (before the pair annihilates) must be very small.
A virtual particle cannot be observed if it is created and subsequently annihilates within the Δt timeframe given by Δt = 2ΔE/ħ. In order for it to exist beyond the Δt timeframe (and be observable), it must become a real particle-antiparticle pair thru the absorption of energy.
This also goes to the heart of the understanding of how forces manifest in quantum field theory. Virtual photons (for example) can only have large energy values relative to the zero-point if they exist for a very short period of time. Likewise, they can only exist for a long period of time if they have very small energies relative to the zero point. (But note that it is also possible for a low energy photon to exist for only a short period of time, as this still satisfies the equation above).
Therefore, if an electric charge is a "source" (for lack of a better term) of its own associated intermediary particle (the photon), a high-energy virtual photon can only go so far at light speed and interact with another electric charge before it vanishes back into the virtual particle froth (transferring its energy/momentum to the absorbing electrically charged particle, which it acquired from the source particle). A lower energy photon can get farther and interact with electric charges father away, and very low energy photons can react at even greater distances. When this is analyzed statistically, it reproduces the inverse square law, because a photon always has a constant and known speed.