The University of California Santa Barbara (UCSB) has grown 1.3 μm-wavelength indium arsenide (InAs) quantum dot (QD) lasers directly on silicon, with record device performance. The breakthrough approach promises an alternative to wafer bonding lasers grown on native substrates, which, in return, could help advance silicon photonics by way of cheap large-scale growth and high-performing QD optoelectronics.
The new QD lasers consist of billions of nanometer-sized indium arsenide (InAs) quantum dots, cladded on either side by gallium-arsenide-(GaAs)-related compounds. The structure is grown using a technique called molecular beam epitaxy (MBE) on germanium/silicon substrates.
“The excitement from this research comes from the fact that the laser material was epitaxially grown on silicon substrates and still performs so well,” says Alan Liu, graduate researcher in the Materials Department at UCSB. For decades, direct growth of lasers on silicon has been unsuccessful. The technology maturing, though, researchers have recently been focusing on growing QD lasers on silicon with wavelengths around 1300 nm, “an important wavelength for datacom that will be used heavily in data centers,” Liu says. The performance still wasn’t there, though. Until the UCSB team demonstrated that QD lasers grown on silicon can have minimal performance constraints imposed by the dislocations generated from growing on silicon.
Liu cites three benchmarks for the record success of his team’s groundbreaking QD lasers, compared with previous devices: a lasing threshold current as low as 16 mA (10 mA feasible), the UCSB QD lasers can output nearly 180 mW of power (previous record for lasers on silicon is around 45 mW for a quantum well laser), and the UCSB lasers can sustain lasing up to 120 degrees Celsius (previous record for lasers on silicon is 105 degrees Celsius for a quantum well laser).
The record achievements could have significant implications for the future of silicon photonics, as silicon wafers only cost about one 10th of InP wafers, and CMOS is also cheap and easily scalable to the high volumes needed for data centers. “This result is a key step towards realizing photonic integrated circuits on silicon,” Liu highlights.
“Quantum dot lasers on silicon would enable low cost silicon photonics by virtue of epitaxial growth,” Liu says, adding that his team is focused on extending photonics, used in virtually all long-distance communication technologies, to the short distances within data centers, and within commercial electronics, such as computers, memory and displays, replacing the copper interconnects that is dominant today but will become less feasible as data rates increase. “Low-cost silicon photonics can help facilitate and expedite this transition,” Liu explains.
The “most prevalent” application for such QD lasers grown on silicon, according to Liu, would be being integrated with other photonic devices as part of a data transmission system, especially once costs are brought down by making photonics compatible with the existing silicon microelectronics industry, which is what silicon photonics is all about, as Liu points out.
Aside from data transmission packages, other niche applications may include optical coherence tomography (a medical imaging technique), high power laser arrays, and the material itself can be used in other device geometries such as amplifiers or solar cells.
The researcher still sees “significant room for improvement.” Going forward, his team wants to achieve “lower thresholds, higher powers and higher temperature operation through both material and device improvements,” he says. Then he plans to demonstrate advanced photonic integrated circuits using this technology.
Written by Sandra Henderson, Research Editor Novus Light Technologies Today