Generation of photon pairs from quantum dots with near-unity entanglement fidelity has been a long-standing scientific challenge. It is generally thought that the nuclear spins limit the entanglement fidelity through spin flip dephasing processes. However, this assumption lacks experimental support. Here, we show two-photon entanglement with negligible dephasing from an indium rich single quantum dot comprising a nuclear spin of 9/2 when excited quasi-resonantly. This finding is based on a significantly close match between our entanglement measurements and our model that assumes no dephasing and takes into account the detection system's timing jitter and dark counts. We suggest that neglecting the detection system is responsible for the degradation of the measured entanglement fidelity in the past and not the nuclear spins. Therefore, the key to unity entanglement from quantum dots comprises a resonant excitation scheme and a detection system with ultralow timing jitter and dark counts.

We analyze the degree of entanglement measurable from a quantum dot via the biexciton-exciton cascade as a function of the exciton fine-structure splitting and the detection time resolution. We show that the time-energy uncertainty relation provides means to measure a high entanglement even in presence of a finite fine-structure splitting when a detection system with high temporal resolution is employed. Still, in many applications it would be beneficial if the fine-structure splitting could be compensated to zero. To solve this problem, we propose an all-optical approach with rotating waveplates to erase this fine-structure splitting completely which should allow obtaining a high degree of entanglement with near-unity efficiency. Our optical approach is possible with current technology and is also compatible with any quantum dot showing fine-structure splitting. This bears the advantage that for example the fine-structure splitting of quantum dots in nanowires and micropillars can be directly compensated without the need for further sample processing.