Disorder enhanced on-chip spectrometers.

准确测量光信号的功率和频率(或波长)分布的能力对于医学光谱学、确保食品或药品的安全、气体遥感和基础科学，如表征短激光脉冲或寻找太阳系外行星的大气等广泛应用至关重要。目前，这是使用光学光谱分析仪或光学单色仪来实现的，这有一个关键的限制。为了实现高分辨率，它们需要较大的光路长度，因此需要较大的占地面积(通常为0.5-1m量级的光路长度)。因此，这些设备既笨重又昂贵。虽然这不是以实验室为基础的低容量应用的问题，但这排除了它们的使用--因此也排除了高分辨率光谱学在大容量、占地面积和重量敏感的应用中的使用，例如集成到芯片实验室设备、移动电话和低质量卫星(例如Cube-Sat)中。这些应用只能由集成片上光谱仪提供服务。这里使用散斑光谱仪，使用光的随机散射来在超小的占地面积内实现高波长分辨率，如果不是因为产生散斑所需的典型的多次散射导致大多数光在可以被检测之前被散射出设备，那么将是非常有希望的。然而，在过去的十年里，几个小组(包括我自己)已经表明，散射点的统计分布可以用来控制光散射的量和方向(例如，在设备的平面内与平面外)。在这个项目中，我们将这些进步与散斑光谱仪结合在一起，即使用受控无序有效地产生散斑图案，同时几乎消除了面外散射和光学损失。在这一进步的基础上，我们将展示一种高分辨率、低占地面积的芯片上光谱仪，在不牺牲设备分辨率的情况下，其性能比最先进的水平高出几个数量级(在设备占地面积方面)。我们还将证明，这些设备适用于未来的大规模生产，使用现有的CMOS设施，适用于气体光谱分析和激光脉冲光谱分析，并与未来用于直接电子读出的光学探测器集成兼容。这将是集成光谱仪领域一项改变游戏规则的进步，并为未来集成散斑光谱仪的商业化奠定基础。

The ability to accurately measure the power and frequency (or wavelength) distribution of an optical signal is crucial to a vast range of applications, for spectroscopy in medicine, ensuring the safety of food or pharmaceuticals to remote sensing of gasses and fundamental science, e.g. characterising short laser pulses or finding the atmospheres of extrasolar planets. Currently, this is achieved using Optical Spectrum analyzers or optical monochromators, which have a key limitation. To achieve high-resolution they need a large optical path length and therefore large footprint (optical path length on the order of 0.5-1 m is common). Thus these devices are bulky and expensive. While not an issue for lab-based low-volume applications, this excludes their use - and thus the use of high-resolution spectroscopy - in large volume, or footprint and weight-sensitive applications, e.g. integration into lab-on-a-chip devices, mobile phones and low mass satellites (e.g. cube-sat). These applications can only be served by integrated on-chip spectrometers. Here the use of speckle spectrometers, using the random scattering of light to achieve a high wavelength resolution in an ultra-small footprint would be highly promising if it were not for the case that typical the multiple scattering needed to create the speckle results in most of the light being scattered out of the device before it can be detected. However, over the last decade, several groups (including myself) have shown that the statistical distribution of scattering sites can be used to control the amount and direction (e.g. within the plane of the device vs out-of-plane) of light scattering. In this project we merge these advances with speckle spectrometers, i.e. using controlled disorder to efficiently generate a speckle pattern, while virtually eliminating out-of-plane scattering and optical losses. Building on this advance we will demonstrate a high resolution, low footprint on-chip spectrometer that outperforms the state of the art by orders of magnitude (in device footprint) without sacrificing the device resolution. We will also demonstrate that these devices are suitable for future large scale manufacturing, using pre-existing CMOS facilities, are suitable for gas spectroscopy and laser pulse spectrum analysis and compatible with future integration with optical detectors for a direct electronic readout. This would present a game-changing advance in the field of integrated spectrometers and lay the foundation for future commercialization of integrated speckle spectrometers.