flip-chip integration

Flip-chip integration is a way to build a quantum processor out of two chips stacked face-to-face instead of one crowded single layer. You make the qubits on one chip and all the wiring, readout, and control routing on a second chip, then flip one over and join the two with tiny metal bumps so their surfaces nearly touch. The name is literal: one wafer gets flipped upside down and bonded onto the other, like pressing two slices of bread together with a grid of solder dots in between.

The reason to bother is space. On a single flat chip every control line, resonator, and bond pad has to share the same surface as the qubits, and they quickly run out of room to route around each other — wires start crossing, frequencies get crowded, and the design jams up. Splitting the job across two tiers gives the wiring its own floor. The bumps, often made of indium because it stays soft and conducts even at millikelvin temperatures, carry signals and ground between the tiers and also set the gap between them. That small vertical gap matters: it keeps the lossy bonding metal far enough from the qubits that it does not spoil their coherence.

It is one of the leading paths past single-plane limits, but it is genuinely hard. The two chips must be aligned to within microns and pressed flat across the whole area, or some bumps never connect and the chip is dead — so yield is a real fight. The indium and the new interfaces can add their own loss and trap stray two-level defects, and getting thousands of bumps to bond reliably and cold-cycle without cracking is still being worked out. Several groups have shown working flip-chip qubit devices, but it is an active engineering frontier, not a solved, off-the-shelf step.