Optical interconnects are of great interests as their electrical counterparts reach the physical limits. The two main research motivations are: (1) on-chip optical interconnects to sustain the growth of microprocessors and (2) the optical datalinks that bridge the communication between server racks within mega data centers and high-performance computers.

As ultra-high speed photodetectors are readily available for commercial use, optical interconnects often see the performance bottleneck at the laser end. In terms of energy per bit, direct modulation is always the most efficient method to operate a laser. However, the interaction between carriers and photons inside the laser active region gives rise to the relaxation resonance, which sets the limit for modulation speed. Increasing the bias current is the easiest way to increase the relaxation resonance frequency, until damping effect takes over. Moreover, high current density unfortunately leads to material degradation and shortens the device lifetime.

To achieve a better performance scaling, we investigated the concept of direct gain modulation, which from our study showed the addition of a zero frequency to the conventional two-pole transfer function of directly modulated lasers, and promised a higher modulation bandwidth. To realize direct gain modulation, we proposed a novel field-induced charge-separation laser (FICSL) structure in a three-terminal vertical-cavity surface-emitting laser (VCSELs) embodiment. In this presentation, the theory, device design, modeling, growth, fabrication, and experimental results of FICSLs will be discussed.