Imec announced a breakthrough in silicon photonics – they learned to grow lasers on conventional 300 mm silicon wafers

The Belgian research center Imec has announced a breakthrough in the production of laser diodes using the classical CMOS process. Traditionally, substrates made of rare elements and compounds have been used for this, while inexpensive silicon has always been left behind. This hampered the development of silicon photonics, since growing laser elements in conventional chips was impossible. Imec’s experience changes the situation: lasers can now be grown on silicon.

Image source: AI generation Kandinsky 3.1/3DNews

To grow lasers using elements of groups III-V of the periodic table, substrates from compounds of the same groups, for example, gallium arsenide, indium phosphide and others, were traditionally used. Typically, after growing semiconductor laser structures, the substrates were disposed of, and the lasers themselves had to be somehow mounted on silicon in the case of silicon photonics applications. It was difficult, expensive and not conducive to sustainable green production. Growing lasers directly on silicon structures could greatly simplify the development of optical computing and other fields.

One of the main difficulties in the production of silicon lasers was the different thermal expansion coefficients of the materials. Imec managed to solve this problem by creating unique buffer zones – grooves around the sites on which gallium arsenide laser structures were grown, as well as by using the original ribbed nanostructure of diodes. As a result of the development, it was possible to produce GaAs lasers on a pilot lithography line on a conventional 300 mm silicon wafer using the CMOS process.

Image source: imec

Semiconductor lasers created in this way with a wavelength of 1020 nm have demonstrated the ability to emit up to 1.75 mW of optical power at a minimum threshold current of 5 mA. Lasers have shown stable operation at room temperature, which opens the way for them in areas such as intensive computing, computer vision and other promising areas.