Towards Gated Quantum Emitters from Undoped Nano-LEDs

Abstract

Quantum light emitters have the potential to transform emerging quantum technologies and their applications, such as secure quantum communication, metrology, and quantum computing. Ideally, these light sources emit on-demand at high rates and efficiencies with high degrees of single-photon indistinguishability. Additionally, these emitters can emit entangled photon pairs, are position-controllable, and wavelength-tunable. Current state-of-the-art single- and entangled-photon sources based on spontaneous parametric down-conversion (SPDC) and on-demand (or deterministic) implementations suffer from various drawbacks that make them deviate from ideality. SPDC sources emit probabilistically, and increasing their brightness degrades their single photon purity, indistinguishability, and entanglement fidelity. Of the various deterministic sources, optically-driven semiconductor quantum dots have very high single-photon efficiencies (~ 71%), purities (> 99%) and indistinguishabilities (> 99%), are position-controllable and wavelength-tunable. However, complex synchronized optical routing between the pump laser, sources and detectors is required to scale their usage. This would occupy a large footprint, restricting them to a laboratory setting. Quantum dots may be current-injected instead, but while gigahertz (GHz) emission frequencies are possible, the electron injection number is not controllable. This inability to control the electron injection is akin to non-resonant optical excitation in which there are many charges in the environment around the quantum dot, thus making current-injected quantum dots inferior to optically-driven quantum dots.

This thesis proposes a novel design for a high-rate, deterministic, electrically-driven quantum emitter that combines a gate-defined lateral planar p-n junction (or nano-light-emitting-diode or nano-LED) with a quantized charge pump along a quasi-one-dimensional channel in dopant-free GaAs/AlGaAs heterostructures. In contrast to other electrically-driven sources, our implementation allows for a precise control of the injected electron number via the quantized charge pump. In addition, by using gates to define the p-type and n-type regions of the junction instead of intentional dopants (as in conventional vertical p-n junctions), the charge carrier mobility in these heterostructures is much higher. The lack of dopants also allows p-type and n-type regions to exist simultaneously on both sides of the device (such devices are termed `ambipolar'), in turn allowing flexible operation. By operating the charge pump at GHz frequencies, this source could emit a billion photons per second. Integrating a cavity at the site of emission would boost the rate of emission and the efficiency, and could also increase the single-photon indistinguishability.

The following research obstacles were identified over the course of developing our nano-LED (the prerequisite for our quantum emitter): - quenching of device electroluminescence (EL) and time-instability of emissions due to parasitic charge accumulation, necessitating thermal cycling to reset the device; - alternate current pathways (both radiative and non-radiative) through the device mesa that reduce both internal and external quantum efficiency; - delocalized emission at mesa edges due to minority currents under the topgate edges, affecting extraction efficiency and position-controllability; and - multimode emission and slow rate of spontaneous emission that reduce extraction and collection efficiencies. Descriptions of our nano-LEDs and their emissions as well as solutions to the above obstacles obtained through experiment are summarized below.

The nano-LEDs discussed in this thesis are gate-induced either in GaAs rectangular quantum wells or at GaAs/AlGaAs single heterojunction interfaces. All nano-LEDs reported in literature are induced using the former and not the latter. In fact, a recent theoretical study concluded that radiative electron-hole recombination was impossible in nano-LEDs induced at single heterojunction interfaces. Our demonstration of EL from nano-LEDs induced at GaAs/AlGaAs single heterojunction interfaces is the first of its kind. Since the fabrication yield using single heterojunction wafers is higher than when using rectangular quantum wells, they offer an alternative for easier fabrication of the nano-LEDs.

To understand how the EL quenches in our nano-LEDs, we propose a scenario of localized parasitic charging that results in enhanced non-radiative recombination and causes a gating of the p-n channel that suppresses the diode current. To address this issue, we have devised a gate voltage sequence that we call the `Set-Reset' protocol. This protocol clears away accumulated parasitic charge, extending the lifetime of device operation without the need for thermal cycling.

Our nano-LEDs can be operated in four distinct circuit measurement configurations, depending on whether the left side is p-type or n-type (with the right side being n-type or p-type, respectively), and whether the left side is grounded or floating (with the right side being floated or grounded, respectively). EL from our nano-LEDs (induced at both quantum wells and single heterojunctions) is observed not only around the p-n junction interface, but also as far as the edges of the etched mesa, indicating the presence of unwanted radiative recombination pathways. The p-side is consistently brighter in the single heterojunction samples while the n-side was brighter for the quantum well devices.

A neutral and a negatively charge exciton peak was observed in the spectra from the n-side of the nano-LEDs. Spectra from the p-side were measured only for the single heterojunction devices, and showed the neutral exciton peak as well as a lower energy peak. The narrowest neutral exciton emission linewidths (0.70 meV) from lateral p-n junctions to date were recorded from the quantum well nano-LEDs. Our nano-LEDs were also shown to be compatible with radio frequency operation, necessary for quantized charge pump integration to create a quantum emitter.

To address the issue of delocalized emission and time-instability of EL, we fabricated and tested a nano-LED with a novel gate architecture that included two wide surface gates placed adjacent and perpendicular to the p-n channel. The extra gates add a degree of freedom that along with standard DC operation and the Set-Reset protocol opens up many measurement configurations. A downside is that these surface gates are prone to current leakage. Several measurement configurations were explored, with two standing out---one yielded localized emission at the junction interface while using the Set-Reset protocol; another yielded time-stability of emission in DC operation. A conceptual model has been laid out that is compatible with the results from these various operating configurations. From the time-stable measurements of EL intensity and p-n current, the internal and external quantum efficiencies were estimated to be ~ 1.19x10^(-3) and ~1.95x10^(-5), respectively. These values may be boosted in future designs by incorporating insulator-separated side gates, blocking gates, and a cavity around the emission region. The side gates and blocking gates will respectively time-stabilize and localize the EL emission during DC operation, and the cavity will increase the rate of spontaneous emission and shape the mode.

A long-standing problem in the field of deterministic quantum emitters is the fact that they emit light omnidirectionally and into multiple modes. Various confining structures such as tapered nanowires, micropillar cavities, photonic crystals, solid immersion lenses and circular Bragg gratings have been proposed and implemented in literature. We identified the circular Bragg grating cavity etched into a heterostructure with a Bragg mirror grown below the rectangular quantum well as the optimum solution for our nano-LED. Through simulation, both the Bragg mirror and circular Bragg grating designs were tuned to match the quantum well emission wavelength (~ 807.5 nm). The circular Bragg cavity etched into the Bragg mirror wafer around the emission region enhances the rate of spontaneous emission via the Purcell effect, and simultaneously funnels the emission into a single elliptical Gaussian mode for efficient collection. A split was included in the circular Bragg grating to make it compatible with our proposed emitter design. Theoretically, for in-plane exciton dipoles oriented parallel to this split, the cavity enhances the spontaneous emission rate by a factor of 5.3 at a center wavelength of 807.4 nm and a bandwidth of ~ 3.7 nm or ~ 7.0 meV. The split in the cavity causes emission to be linearly polarized. This linear polarization is unfortunately incompatible with the emission of polarization entangled photon pairs. The effective collection efficiency (from simulation) is ~ 30%, which is ~ 52 times greater than that of a device without a cavity. The inclusion of our cavity also boosts the internal and external quantum efficiencies by factors of 4.5 and 89, yielding values of ~ 5.32x10^(-3) and ~ 1.74x10^(-3), respectively. Design validation of the Bragg mirror using reflection measurements yielded a Bragg stopband frequency and bandwidth that closely match simulation. Simulated and measured reflection spectra from the circular Bragg gratings indicated a linear relationship between the ring width of the grating and the cavity resonance wavelength, with a consistent wavelength offset between simulation and measurement of ~ 16.3 nm. From these results, a cavity with a ring width of ~ 94.8 nm would most closely match the emission wavelength of ~ 807.5 nm.

Through our proposed and implemented solutions for the obstacles facing our nano-LEDs, we pave the way for the realization of a high-rate, electrically-driven quantum emitter.