密集波导阵列实现无失真光传输

提高成像能力或分辨率是光学显微面临的重要挑战之一。在过去的三百多年中,科学家们一直在致力于制造更好的显微镜。在很长一段时间里,显微镜的性能极限由两个因素决定: 被观察物体的对比度和显微镜中光学元件的分辨率。而在过去的50年里,科学技术的飞速发展提高了物体的对比度和光学质量。超透镜就是近年来出现的一项新兴技术。超透镜利用了波的特性,能够解析原本隐藏在视线之外的细节。

近期,南京大学的研究人员发表了一个波导阵列的研究结果,该阵列在性能上具有超透镜诸多优点的同时,制造上也不存在超透镜制造过程中相关的技术困难。研究成果发表在Advanced Photonics上(Wange Song, Hanmeng Li, Shenglun Gao, et al. Subwavelength self-imaging in cascaded waveguide arrays[J]. Advanced Photonics, 2020, 2(3): 036001)。

级联波导阵列的亚波长自成像。(a)超透镜波导阵列补偿正负耦合。(b)级联波导、以及(c)模拟场演化的相应结果。(d)模拟“0” / “1”编码信号通过波导阵列传输的信号结果。级联波导的输出完美地再现了输入信号。(e)所制备的级联样品的扫描电镜图像。CCD记录光波从输入到输出通过(f)直和(g)级联波导阵列。

超透镜

了解图像是如何形成的有助于理解超级镜头。当光线照在物体上时,物体会将光向四面八方散射。图像的细节是由光线散射的强度和方向决定的。然而,有限的镜头尺寸限制了光线的捕捉量。通过镜头捕捉到的光线重建的图像将不会体现未进入镜头的光线所携带的细节。因此这样的成像并不完美。

倏逝波不会传播,因此没有供镜头捕捉光线的角度。并且,倏逝波会很快以指数形式消失,在几个波长内强度便会非常接近于零。因而一个工作距离与显微镜相当的透镜无法捕捉到这些所谓的倏逝波。

研究人员设计了超透镜来捕捉这些细微的倏逝波。要做到这一点,透镜必须由具有负折射率的超材料构成(普通材料具有正折射率)。然而,超材料并不容易制造,而且性能也不好。大多数光线会在超透镜表面由于巨大的阻抗失配而被反射,而超透镜内部的物质会吸收大量的光线。因此这样的透镜虽然能捕捉精细的细节,但成像效率和对比度很差。

这正是南京大学的研究人员希望解决的问题。他们的镜头由一系列彼此非常靠近的波导组成。每个波导从波导开口的正前方捕获光线。光被传输到波导阵列的另一端,然后完成图像重建。

波导耦合控制

光在靠近的波导中传播会发生相互耦合,因此密集排列的波导不能传输图像。如果在密集的波导阵列中传输图像,则图像将完全被破坏。

为了解决这个问题,研究人员研究了波导之间的耦合作用。在直平行波导中,阵列之间的耦合系数是一个正值。两个平行排列的弯曲波导间的耦合系数则可能是负数。由此,可以设计合适的波导阵列参数,使得在直波导阵列中由于耦合而发散光,在进入弯曲波导部分之后正好通过负耦合效应汇聚回来,最终实现所有的光回到初始的波导中输出。

研究人员用13个亚波长间距的波导组成的阵列演示了这种效应。实验证明,光线从单根波导输入后,尽管经历中间的发散过程,最终都可恢复到初始的波导中,完成了波导阵列在亚波长尺度下点对点成像的功能。在上述效应下,可以通过扫描波导阵列建立图像。此外,通过缩小波导的相关参数可以进一步提高分辨率。

该结构还可以应用在其他领域。与电子系统相比,用于计算和通信的集成光学电路体积很大,控制相邻波导之间的耦合是电路体积很大的原因。这项研究展示了获得高密度波导的新方法,最终将获得比高分辨率成像更广泛的应用。
 

Waveguide Array Transports Light Without Distortion

One of the challenges of optical microscopy is to continually increase the imaging power, or resolution. In the past three hundred odd years, scientists have been building ever-better microscopes. The limit, for a long time, was determined by only two factors: the contrast of the object being viewed, and the resolving power of the optics in the microscope. The last 50 years, in particular, have led to an explosion in techniques to improve both the contrast of object and the quality of the optics.

One such technology is called a superlens. The superlens makes use of some of the peculiarities of waves to be able to resolve details that would otherwise be hidden from view. Now, researchers from Nanjing University in China have published results on a waveguide array that provides many of the benefits of a superlens. Along with that, the waveguide array does not have the technological difficulties that are usually associated with superlens fabrication. 

That lens is super

To understand the superlens, it helps to understand how an image is formed. Let's begin with something like the head of a pin against a featureless background. When light is shone on the pin, it scatters in all directions. The details of the image are held in the intensity and directions that the light is scattered. However, lenses have a limited size, limiting the amount of light captured. The image that is reconstructed from the light captured by the lens will not have the details carried by the light that never reached the lens. Our image is imperfect.

For the finest of features, there is no angle at which a lens can capture the light, because the light doesn't travel. Instead, the wave dies rapidly (exponentially), and within a few wavelengths, the intensity is very close to zero. A lens, with a working distance typical of a microscope, will not capture these so-called evanescent waves.

A superlens is designed to capture these detail-holding evanescent waves. To enable that, the lens must be constructed from a metamaterial that has a negative refractive index (normal materials have a positive refractive index). However, metamaterials are not easy to make, and don't perform well. Most of the light that hits a superlens is reflected from it, while internally, the substances that are used to create the metamaterial absorb a lot of light. Hence, the lens captures fine details, but the image contrast is poor.

This is where the work of Song and coworkers comes into play. Their lens consists of an array of waveguides that are placed very close to each other. Each waveguide captures light from just in front of the waveguide opening. The light is transported to the other end of the waveguide array, where it is used to (in principle) recreate an image.

Waveguide flow control

Closely spaced waveguides don't transport images. When waveguides are close together, the light flows from one waveguide to another. An image will be completely randomized if it is transported in a dense array of waveguides.

To get around this problem, the researchers exploited how the coupling between the waveguides works. In straight parallel waveguides, the coupling between the arrays can be represented by a fixed positive number. This number gives the fraction of light that swaps waveguides as a function of distance. However, if the waveguides are parallel, but meander in a wave-like manner, then the coupling can be negative.

To be more concrete: imagine two waveguides that are close together and straight. Light enters one waveguide and spreads to the second at a rate given by the coupling constant. The light then enters the meander, which has a coupling coefficient that has equal size, but is negative. This section undoes the spreading exactly so all the light exits the same waveguide it entered. The researchers demonstrated this effect with an array of 13 waveguides. They showed that light would consistently exit the waveguide it was coupled into, despite severe mixing in the straight section.

Subwavelength self-imaging in cascaded waveguide arrays. (a) Compensated positive and negative coupling in waveguide array for superlens. (b) Cascaded waveguides, and corresponding result of the simulated field evolution in (c). (d) Simulated signal results of "0"/"1" coded signal transmission through cascaded waveguide arrays. The output in cascaded waveguides perfectly reproduces the input signal. (e) SEM figures of the fabricated cascaded samples. CCD recorded optical propagation from input to output through (f) straight and (g) cascaded waveguide arrays.

This is only the beginning of the story. Images can be built up by scanning the waveguide array. The resolution can be further increased by making the aperture of the waveguides smaller.

The demonstrated structure has other uses. Integrated optical circuits for computing and communications are, compared to electronic systems, large. The spacing is dictated by the need to control the coupling between neighboring waveguides. This research shows how to have high density waveguides without unwanted coupling. In the end, that could find applications more widespread than high resolution imaging.