Large-scale characterisation of silicon T centres with tuning-free broadband cavities
This paper presents a large-scale characterization of silicon T centres using an array of 400 low-Q, tuning-free confocally-addressable cavities that achieve a 6.5-fold signal enhancement and a 42-fold measurement speedup while enabling the statistical inference of T centre density without requiring individual cavity tuning or collection efficiency estimates.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer
To build a future internet that is unhackable and incredibly fast, scientists are working to create networks that transmit information using the delicate states of individual particles of light. A critical piece of this puzzle is finding a way to link a stationary bit of information, stored in a solid object, with a flying bit of information, carried by a photon. Silicon, the material that powers our computers, has emerged as a promising host for these stationary bits because it is compatible with existing manufacturing technology and can be purified to create a quiet environment for quantum states. Within this silicon, a specific defect known as a T centre acts as a tiny light source that emits photons at a wavelength ideal for long-distance travel through fiber optic cables. However, these T centres are notoriously shy; they emit very few of the specific photons needed to build a network, making them difficult to study and slow to measure.
For years, researchers have tried to coax more light out of these silicon defects by placing them inside microscopic mirrors called cavities. These structures can trap light and force the defect to emit more photons, but there is a catch. Traditional high-performance mirrors are so precise that they only work if the defect's color matches the mirror's setting exactly. Because every T centre is slightly different and manufacturing is never perfect, scientists have had to individually tune each mirror to its specific defect. This process is so slow and labor-intensive that it has limited studies to just a handful of examples, preventing a broad understanding of how these quantum sources behave. A team of researchers at Delft University of Technology has now solved this bottleneck by designing a new kind of mirror that does not require individual tuning, allowing them to measure hundreds of T centres at once and revealing their true density in the material.
The researchers realized that instead of building a mirror that is perfectly tuned to a single color, they could build a mirror that is slightly less sharp but covers a wider range of colors. They designed a structure known as a bullseye cavity, which looks like a series of concentric rings surrounding a central disc. While a standard mirror might be like a narrow tunnel that only lets one specific color through, these new cavities use the periodic nature of the rings to create a weak mirror that confines the light to the central disc and diffracts the radiated light out of the plane. This sends far more emission into the collecting lens than would naturally escape from the silicon, while the confinement creates a Purcell enhancement that boosts the emission rate. By carefully calculating the size of the rings and the central disc, the team created a design where the mirror's natural range of colors overlaps with the emission of the T centre, even if the manufacturing process introduces small errors. This approach removes the need to adjust each mirror individually, as the design is robust enough to work across a large group of defects.
To prove this concept, the team fabricated a massive array of four hundred identical bullseye cavities on a single silicon chip. They first spent time calibrating their manufacturing process, running test patterns to understand exactly how the physical size of the rings shifted the color of the light they produced. Once they understood these shifts, they adjusted their design to ensure that the final array of four hundred cavities would all be resonant with the T centre emission, despite the tiny variations inherent in making them. The result was a collection of cavities with an average quality factor of 298, a measure of how well they trap light, and a spread in their tuning that was small enough to cover the natural variation of the defects without needing adjustment.
When they turned on the lasers to measure the light coming from these four hundred sites, the improvement was immediate and dramatic. The cavities boosted the signal from the T centres by a median factor of 6.5 times compared to unpatterned silicon. This increase meant that the researchers could gather the same amount of data in a fraction of the time. Where a standard measurement might take minutes to get a clear signal, the new setup allowed them to collect data in just 15 seconds per cavity. This speed-up of 42 times transformed the workflow, allowing the team to measure the entire array of four hundred sites in just 5.5 hours, with the actual time spent collecting light taking up less than a third of that duration. The rest of the time was spent on the mechanical tasks of moving the microscope and checking the equipment, which are now the limiting factors rather than the faintness of the signal.
Beyond simply speeding up the measurements, the sheer volume of data allowed the team to answer a question that had previously been difficult to solve: how many T centres are actually present in the silicon sample? Because the formation of these defects is a random process, some cavities in the array captured a T centre while others did not, and the strength of the signal varied depending on exactly where the defect sat inside the cavity. By comparing the statistical spread of the signals they measured against computer simulations, the researchers could infer the density of the defects without needing to know the exact efficiency of their light-collecting equipment. Their analysis suggested that there are approximately 4.2 T centres per square micrometer in the sample. This method of using large-scale statistics to determine density is a significant step forward, as it relies on the internal consistency of the data rather than external assumptions about the setup.
The work demonstrates that by accepting a moderate level of imperfection in the design, scientists can achieve a level of scalability that was previously impossible. The team showed that they do not need to chase perfect, individually tuned mirrors to study quantum emitters; instead, they can use a broad, robust design to survey large populations of defects. This approach shifts the focus from the painstaking effort of tuning single devices to the statistical understanding of how these quantum systems behave in bulk. The ability to measure hundreds of T centres quickly and accurately provides a new window into how sample processing affects the formation and properties of these defects. This understanding is essential for scaling up current experiments, moving from proof-of-concept demonstrations to the creation of large numbers of working quantum devices on a single chip, bringing the vision of a quantum network closer to reality.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.