III-nitride semiconductor lasers grown on Si

Abstract

III-nitride semiconductor laser directly grown on Si is a potential on-chip light source for Si photonics. Moreover, it may greatly lower the manufacture cost of laser diodes and further expand their applications. Therefore, III-nitride lasers grown on Si have been pursued for about two decades. Different from GaN homoepitaxy on free-standing GaN substrates, III-nitride semiconductors grown on Si substrates are usually rich with strain and threading dislocations due to the large mismatch in both lattice constant and coefficient of thermal expansion between GaN and Si substrates, which hindered the realization of electrically injected lasing. The key challenges in the direct growth of high-quality III-nitride semiconductor laser materials on Si substrates, as well as their corresponding solutions, are discussed in detail. Afterwards, a comprehensive review is presented on the recent progress of III-nitride semiconductor lasers grown on Si, including Fabry-Pérot cavity lasers, microdisk lasers, and the lasers with nanostructures, as well as the monolithic integration of lasers on Si. Finally, the further development of III-nitride semiconductor lasers grown on Si is also discussed, including the material quality improvement and novel device structures for enhancing optical confinement and reducing electrical resistance, with a great prospect for better performance and reliability.

Introduction

Silicon photonics compatible with large-scale and large-wafer-size manufacturing foundries are deemed as a promising way to revolute communication and computation technologies [[1], [2], [3]]. However, Si with an indirect band structure, usually used for complementary metal oxide semiconductor (CMOS), cannot emit light efficiently. Recently, Ge and SiGe alloys with a hexagonal crystal structure were reported to emit light efficiently [4]. Nevertheless, they are not fully compatible with CMOS-used Si with a cubic lattice structure. Therefore, the lack of an efficient on-chip laser source remains as the major roadblock of Si photonics.

III-V compound semiconductor lasers on Si may be a potential option for on-chip light source [[5], [6], [7], [8]]. Given the low throughput of the off-chip integration via chip bonding process, it is highly desirable to achieve lasers directly grown on large-size Si [5,8]. Liu’s group grew high-quality III-V semiconductor epilayer with a recorded threading dislocation density (TDD) on the order of 10⁵ ​cm⁻², by combining a nucleation layer and dislocation filter layers on Si, and then realized high performance III-V quantum dot lasers grown on Si [5]. However, the defect density in the laser material is still a little high for III-V semiconductor lasers. In contrast, III-nitride semiconductors are less sensitive to structural defects [9], and thus III-nitride lasers on Si may be a potential on-chip light source for Si photonics.

III-nitride semiconductor materials with a direct bandgap have a wide emission ranging from ultraviolet (UV) to near infrared (IR) region, and have been widely used for highly efficient light-emitting diodes (LEDs) and laser diodes (LDs) with a huge commercial success [[10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21]]. The 2014 Nobel Prize in physics was awarded to the inventors of GaN-based blue LEDs. The state-of-art threshold current density (J_th) of III-nitride LDs is less than 0.6 ​kA/cm², and the highest wall-plug efficiency is over 50% from Nichia [14]. And III-nitride LDs have been widely used in displays [14,15], lighting and optical storage [22], and also showed a great potential for wide applications in monolithic integration [23], visible light communications [24], optical clock [19], material processing [15], quantum technology [19], and medical instruments [17].

Today most of the GaN-based LDs are produced on 2-inch free-standing (FS) GaN substrates (~$4000/pc), which makes the LD chip cost 2-3 orders of magnitude higher than that of LEDs grown on Si or sapphire substrates. Additionally, FS GaN substrates often have issues of non-uniformity in offcut angle, defect density, and residual stress [25], which affect the device yield, further increasing the LD chip cost. GaN-based LDs grown on large-size low-cost Si substrates may revolutionize the LD industry by slashing the cost and further boost their applications [[26], [27], [28], [29]]. Therefore, III-nitride LDs grown on Si have received wide attention and been pursued for decades.

Similar with GaN-on-GaN LDs, GaN-on-Si LDs can be processed by standard micro/nanofabrication technologies, including lithography, dry etching, deposition, facet cleavage, etc. However, distinctly different from the homoepitaxy of GaN-on-GaN LDs, the heteroepitaxy of GaN-on-Si LDs encounters the issues of huge mismatch in both lattice constant and thermal expansion coefficient between GaN and Si. These issues usually not only induce a large tensile stress, wafer bowing, and even microcrack networks in the epilayer, but also cause a high TDD (10⁸-10⁹ ​cm⁻²), 2-3 orders of magnitude higher than that of GaN-on-GaN LDs. The large tensile stress poses a daunting challenge for the growth of GaN-on-Si LDs including about 2-μm-thick AlGaN cladding layers, while the high TDD affects the internal quantum efficiency and the optical gain by causing severe non-radiative recombination and formation of V-shape pit defects along with deteriorated interfaces of multiple quantum wells (MQWs).

This review article began with a discussion on the key challenges of the direct growth of III-nitride semiconductor lasers on Si, followed by the recent progress of GaN-on-Si lasers and their monolithic integration. Additionally, various methods of further improving the material quality and several novel device structures are suggested for boosting the device performance and reliability.