Singaporean researchers develop world’s smallest LED for high-resolution mobile phone microscopes and on-chip emitters

Singaporean researchers have made a groundbreaking discovery in the field of photonics by developing the world’s tiniest LED, which can convert mobile phone cameras into high-resolution microscopes. This innovative LED, smaller than the wavelength of light, has been used to create the world’s smallest holographic microscope. The team’s breakthrough technology has made it possible to modify the silicon chip and software in everyday devices, such as mobile phones, to turn them into microscopes. This development is significant for sustainable agriculture, as it allows for the miniaturization of diagnostics for indoor farmers.

The researchers also developed an advanced neural networking algorithm that can reconstruct objects captured by the holographic microscope. This allows for a more detailed examination of microscopic objects, like bacteria and cells, without the need for bulky microscopes or additional optics. This groundbreaking research has paved the way for an on-chip emitter smaller than a micrometer, which is a significant advancement in photonics.

The majority of photonic chips use off-chip sources for their light, leading to low overall energy efficiency and limiting the scalability of these chips. To overcome this challenge, researchers have developed on-chip emitters using rare-earth-doped glass, Ge-on-Si, and heterogeneously integrated III–V materials. Although these emitters have shown promising device performance, integrating their fabrication processes into standard complementary metal-oxide-semiconductor (CMOS) platforms remains a challenge.

SMART researchers have made significant progress in the development of a sub-wavelength silicon (Si) light-emitting diode (LED) that can be individually controlled at the nanoscale. While Si has shown promise as a material for small, controllable emitters, its indirect bandgap has limited its quantum efficiency, and available materials and fabrication tools have hindered the realization of small native Si emitters in complementary metal-oxide-semiconductor (CMOS) platforms.

In a recent Nature Communications paper, SMART researchers detailed their creation of the smallest Si emitter reported to date, with light intensity comparable to larger state-of-the-art Si emitters. This novel CMOS-integrated LED is room temperature-compatible, has a high spatial intensity of 102 ± 48 mW/cm2, and the smallest emission area of 0.09 ± 0.04 μm2 among all known Si emitters in scientific literature. To demonstrate its practical application, the team integrated this LED into a lensless holographic microscope, which requires no lens or pinhole, in a centimeter-scale all-silicon platform.

In a related breakthrough, SMART researchers also developed a new deep neural network architecture capable of reconstructing images from the holographic microscope. The paper, titled “Simultaneous spectral recovery and CMOS micro-LED holography with an untrained deep neural network,” was published in the journal Optica. These exciting developments have the potential to pave the way for smaller and more efficient photonics devices.

A major challenge in lensless holography is the computational reconstruction of the imaged object. Traditional reconstruction methods require detailed knowledge of the experimental setup, which can be sensitive to variables such as optical aberrations, noise, and the twin image problem, making accurate reconstruction difficult.

To address this, the research team developed an untrained deep neural network architecture that improves image reconstruction quality. This novel neural network incorporates total variation regularization to increase contrast and accounts for the wide spectral bandwidth of the source.

Unlike traditional methods that require training data, this neural network embeds a physics model within the algorithm, eliminating the need for training. Furthermore, the neural network can recover the source spectrum from a single diffracted intensity pattern, offering a significant departure from previous supervised learning techniques.

The untrained neural network developed in this study enables researchers to use new light sources without prior knowledge of the source spectrum or beam profile, such as the novel and smallest known Si LED fabricated through a fully commercial, unmodified bulk CMOS microelectronics.

The combination of CMOS micro-LEDs and the untrained neural network has the potential for various computational imaging applications. This breakthrough can lead to the development of a compact microscope for live-cell tracking or spectroscopic imaging of biological tissues such as living plants. The use of in-line holography microscopes has already been employed for many applications, including biological sample imaging, particle tracking, environmental monitoring, and metrology. The researchers also envision arraying these LEDs in CMOS to create programmable coherent illumination for more complex systems in the future.

Lead author Iksung Kang said that the breakthrough can be hugely impactful for various applications requiring the use of micro-LEDs. The LEDs could be combined into an array for higher levels of illumination needed for larger-scale applications, and this can be done without increasing the system’s complexity, cost, or form factor due to the low cost and scalability of microelectronics CMOS processes. The LED could even be integrated into a mobile phone camera to create a holographic microscope. The all-in-one micro-LED could be transformative for the field as the control electronics and imager could also be integrated into the same chip.

Principal investigator Rajeev Ram added that the LED has a wide range of other possible applications, including bio-imaging, bio-sensing applications, near-field microscopy, and implantable CMOS devices. The LED could be integrated with on-chip photodetectors, leading to further applications in on-chip communication, NIR proximity sensing, and on-wafer testing of photonics.

Leave a Comment