.png){kind=icon}
.png){kind=icon}
Return
Inverse Design of Photonic Crystal Lasers Through Automatic Differentiation Optimization
.png){kind=icon}
PRE
AI
M
S
DOI：10.1109/jstqe.2026.3687110.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The behavior of (embedded) photonic crystal resonator lasers is inextricably linked to the unit cell profile. However, an overwhelming fraction of the design process unfortunately relies on the judicious manual choice of the permittivity profile. This is disadvantageous for the fact that this throws away information regarding the objective function (the physical response of the laser). We remedy this by employing gradient-based protocols that split the computational design task into two steps—between a 2D eigenvalue problem for the photonic eigenmodes (modeling) and the computational graph (gradient-propagation): (i) a forward passage through the simulation to predict the photonic-crystal surface-emitting laser (PCSEL) behavior (far-field and mode quality factors), and (ii) a backpropagation step where the gradient of a (user) defined cost function is computed to update the unit cell design. The performance of this method exceeds contemporary unit cell designs for large-scale PCSELs in both far-field and intracavity field performance metrics.
Keywords:
Photonic crystals
inverse design
surface-emitting lasers
bound states in continuum
Journal
I
IF:
5.1
Papers:
5.6K
Citations:
1.2W
Organization
Cited Papers