基于超构表面的光谱成像及应用研究进展
1 引言
光不仅是人类观察客观世界的重要媒介,还具有传播信息和能量的作用。光束携带的信息包括在时域、频域以及空间的分布特征,人眼能直观感知的颜色和强度只是其中的一部分,其他隐含信息往往需要借助仪器来进行观测。光谱是物质的固有特征,通过分析光谱可以揭示物质的化学组成成分,在材料分析、食品安全、医学诊断和生物成像等领域有广泛的应用[1-5]。传统的光谱仪通常由棱镜、光栅等[6-7]分光器件通过复杂的光路实现光谱成像,受衍射效应限制,其光谱分辨率与光程存在反比关系,因此普遍存在体积大、集成度低、成本昂贵的缺点,严重限制了在内窥镜、水下探测等特殊场景的应用。随着市场上对集成化器件的需求增多,紧凑型光谱仪的研究成为了一个重要的课题。近几十年来,傅里叶变换光谱仪[8-9]、微环谐振腔[10-11]和光波导耦合[12]已被用于缩小光谱器件的体积。傅里叶变换光谱仪具有高光谱分辨率和高信噪比,但是无法处理非常不规则的光谱信号,并且处理数据的速度较慢,难以应对随时间强烈变化的动态光谱信号。微环谐振腔和光波导耦合光谱仪因其较小的尺寸可以集成到微型光学系统中,但是光谱分辨率受到制造工艺的限制,对微纳技术的要求较高。
21世纪初出现了关于超构表面的研究[13],作为由亚波长小单元组成的大面积纳米结构,超构表面具有可塑性强、灵活度高、易集成的特点。2011年,Capasso小组提出了广义斯涅尔定律[14],由此拉开了超构表面研究热潮的序幕。通过设计优化共振相位、传输相位和几何相位,超构表面可以有效调制入射光的光学参数,如振幅、相位和偏振[15-16]。由于超构表面在光场调控方面表现出的优异性质,因此可以实现传统折射或衍射光学难以实现的复杂功能,如全息显示[17]、消色差透镜[18-19]、光加密通信技术[20]和隐形斗篷[21]。超构表面作为一种二维材料,与三维超构材料相比的优势是可以最大限度地减小损耗并增加集成度,此外,还可以通过如光刻或电子束刻蚀的纳米材料制造方法相对容易地获得[22]。光学材料中普遍存在色散的现象,得益于超构表面对相位的有效调控,利用超构表面可以进行更高效率的分光,这对实现紧凑型光谱仪和光谱成像具有重要的意义。滤波是另外一种达到光谱成像这一目标的思路,通过设计后的超构表面具有频率选择的功能,可以用于轻薄高效的集成式光谱滤光器。本文首先基于不同原理从超色散、窄带滤波和宽带滤波这三个方面重点介绍了多种机理的超构表面光谱成像,然后回顾了基于超构表面光谱成像的应用。最后,对目前超构表面光谱成像工作进行了总结并对未来发展方向进行了展望。
2 多种机理的超构表面光谱成像
2.1 基于超色散机理的超构表面光谱成像
对于一般的具有固有折射率色散的光学介质材料,不同波长的光对应有不同的光学响应。在超色散材料中,光的传播速度和折射率会随着频率发生更加急剧的变化,这种特性使光的散射得到增强,一方面这会极大降低如通信、检测、成像等全彩光学应用的性能,对此有很多科研人员进行了消色差的研究[18, 23];另一方面利用这种超色散机制并加以调控,可以使超构表面在彩色成像或光谱层析分析等领域展现出优良的应用潜力[24-25]。
传统的彩色图像传感器通常用彩色滤光片获得不同的颜色,然而随着图像传感器的尺寸变得更小,每个像素接收到的光也变得更少,导致光的利用率降低。Nishiwaki等[26]提出采用一种折射率比周围材料更高的透明介质微型板状结构来诱导光的偏转,以达到分光的目的。如
图 1. (a)对称偏转器和非对称偏转器的结构示意图[26];(b)异常反射的实验表征设置和实际拍摄图片[27]
Fig. 1. (a) Structural schematic diagram of symmetric and asymmetric deflectors[26]; (b) Experimental characterization setup and actual photography of anomalous reflection[27]
除了上述基于异常反射的分光外,Capasso小组提出的离轴超构透镜也被证明可以在高光谱分辨率的情况下同时聚焦或分散不同波长的光。如
图 2. 离轴超构透镜。(a)显示坐标的超构透镜示意图,聚焦线沿x′轴(垂直于聚焦轴)的位移作为波长的函数[28];(b)由超构透镜和CMOS相机组成的紧凑光谱仪[29];(c)像差校正超构透镜和贝里相位透镜的实验表征[30];(d)光谱范围和光谱分辨率与结构参数之间的关系示意图[31]
Fig. 2. Off-axis meta-lens. (a) Schematic diagram showing the coordinates of the meta-lens, and the displacement of the focal line along the x′-axis (normal to the focal axis) as a function of wavelength[28]; (b) Compact spectrometer composed of meta-lens and CMOS camera[29]; (c) Experimental characterization of the aberration corrected meta-lens and the Berry phase lens[30]; (d) Schematic diagram of the relationship between spectral range, Spectral resolution and structural parameters[31]
其中,
根据上述原理他们先设计了一个可以将光以80°角聚焦的超构透镜[28],由于大角度聚焦,超构透镜具有超色散特性(0.27 nm/mrad),在电信区域的波长差异分辨率高至200 pm,
然而,对这种离轴超构透镜还没有进行更为详细的参数分析。周毅等[31]研究了基于超构透镜的光谱仪的结构参数对有效光谱范围和光谱分辨率的影响。如
值得注意的是,离轴超构透镜虽然具有较高的光利用率和较强的光谱分辨能力,但是却难以在采集光谱信息的同时兼顾三维空间分辨率,因此距离4D成像[32-33]的目标还有一段距离。2018年,Faraji-Dana等[34-35]介绍了折叠超构表面的概念并制作了一个总体积仅有8.5 mm3的超紧凑线扫描微型高光谱成像仪(hyperspectral imager,简写为HSI)。如
图 3. (a)折叠超构表面示意图[35];(b) SLIM系统中光谱重建算法的数值模拟结果[36] ;(c)高光谱成像系统的光学架构示意图[37];(d)色散实验的装置和实验结果的拼接图[38]
Fig. 3. (a) Schematic diagram of folded metasurface[35]; (b) Numerical simulation results of spectral reconstruction algorithm in SLIM system[36] ; (c) Schematic of an optical architecture for hyperspectral imaging system[37]; (d) Set up of the dispersion experiment and the splice diagram of experimental results[38]
2022年华夏等[36]利用横向色散的超构透镜阵列和一个单色成像传感器演示了超紧凑光谱光场成像(spectral light-field imaging,简写为SLIM),仅需使用一次快照就可以同时获得光谱信息和空间信息。在SLIM系统中,为了同时获得高光通量、高空间分辨率和高光谱分辨率,不可避免地会捕获带有光谱和空间混叠的图像。光谱信息和空间信息耦合在一起,故将探测问题转化为解决一个欠定的优化问题,通过引入光谱重建算法可以对其进行求解从而获取场景中每个位置的光谱信息。
区别于上述的几何相位原理,由于超构透镜的功能是实现在给定光谱带宽∆λ上的连续波长光的色散和沿焦点线的聚焦,并且所有入射角θ由仪器入口狭缝的有效高度定义,因此Billuart等[37]提出可以用一个函数F(x,y,λ,θ)来描述超构透镜的性质。如
其中,M是由坐标x和y定义的超构表面上的点,
当光正入射到一个平面反光镜上时,反射光线与入射光线重合,如果把平面反光镜换成垂直截面为抛物线的反光镜,反射光总是会聚焦到特定的聚焦线上。根据这个光学性质,Chen等[38]设计了一个具有抛物线相位轮廓的光谱调制超构表面。抛物线的公式可以写为
其中,
除了光谱信息外,对偏振信息的检测也是受到广泛关注的研究方向,而基于相位调控的超构表面往往对光的偏振也有所响应[39]。利用这一特性,可以设计出能同时检测光谱和偏振的超构透镜[5],然而在只需要探测光谱信息的情况时,超构表面对偏振的响应会极大地影响光谱信息的采集,因此需要设计偏振不灵敏的超构表面。由两个正交方向的多个子周期单元组成超构表面的方案可以使超构表面具有偏振独立衍射的性质[40]。因为每个子周期在两个正交方向上的占空比是独立的,因此可以同时操纵TE和TM偏振的有效指数,模拟结果表明,对于正常入射光,TE和TM在0.7 μm参考波长下的衍射效率分别为79.2%和79.3%。
综上所述,相比起传统光栅光谱仪,基于超色散的超构表面光谱成像可以在一定程度上减小光学元件和光学系统的体积,并且成像系统相比起传统基于多个透镜的成像系统具有更轻巧、更便捷的优点,通过不断改进设计方案,光谱分辨率可以得到进一步提升。需要注意的是,基于超色散的光谱仪都有一个共同的特点,即入射光需要保持正入射才能实现有效分光,因此在使用光谱仪之前还需要在光路中加入额外的准直光路。
2.2 基于窄带滤波机理的超构表面光谱成像
通过滤波进行光谱探测是实现光谱成像的另一种方案[41],然而传统滤波器一般需要经过多个介电层的沉积,制造过程相对复杂,阻碍了滤波光谱成像的发展道路。近十几年来,随着制作滤波器的材料范围扩大和工艺技术提升,基于窄带滤波的光谱仪展现出了巨大的光谱成像潜力。根据滤波的方式,窄带滤波器可分为透射型滤波、吸收型滤波和反射型滤波;根据滤波波长的可调谐性质,可以分为可调谐型滤波和阵列性滤波。下面分三种滤波方式介绍了不同类型的窄带滤波超构表面光谱成像的研究进展。
2.2.1 透射型超构表面滤波器
透射型窄带滤波器对透射光具有宽带吸收或宽带反射的作用,只允许特定波长窄带光透过,虽然高分辨率的微型集成滤光片是高分辨光谱仪的重要组成部分,但是传统的工艺水平很难制造出高透光率的集成滤波片。王少伟等[42-44]开发了两种高效的制造集成滤波器阵列的方法,一种是组合蚀刻技术,另一种是没有任何蚀刻工艺的组合沉积技术。利用后一种技术,他们制作并演示了基于128个通道的集成光栅滤波器阵列的高分辨率微型光谱仪,如
图 4. 基于透射型超构表面的光谱成像。(a)集成滤波阵列组成的紧凑型光谱仪[44];(b)具有不同纳米柱宽度的一组滤波器模拟透射光谱[46];(c)超构表面快照光谱成像仪的示意图[47];(d)多光谱拼接滤波器生成过程示意图[4];(e)未知源入射功率的目标检测策略框图[48]
Fig. 4. Spectral imaging based on transmission-type metasurface. (a) Compact spectrometer composed of integrated filter array[44]; (b) Simulated transmission spectra of a group of filters with different nanopillar widths[46]; (c) Schematic diagram of a hypersurface snapshot spectral imager[47]; (d) Schematic diagram of generating process of the multispectral filter mosaic[4]; (e) Block diagram of target detection strategy for unknown source incident power[48]
为了获得高透射率的尖锐透射峰,科研人员研究了不同结构和材料的滤波性质。等离子体纳米结构能够在亚波长范围内实现对光场的调制,其中金属-绝缘体-金属(metal – insulator – metal,简称为MIM)波导结构已被证实可以在可见波段将白光转换成特定颜色的光[45],并且波导的上下两层金属结构决定了其在电光系统中能很容易被集成,有利于压缩器件尺寸。
为了增强倏逝光场以增加透射,等离子体滤波片需要结合金属-介电界面中的表面等离子体激元。已有研究表明,由于表面等离子体激元的存在,在可见光和近红外波段区域中,光与金属和介电材料之间的界面具有电子振荡相互作用,从而导致了纳米孔阵列中特殊的光传输或共振[49-50],这使其表现出滤波片的特征,孔间距和孔尺寸对透射共振的峰值位置和带宽有调制作用。然而,传统纳米孔阵列的透射共振由于低带外阻塞和宽共振光带宽而不能有效地分离颜色。受益于表面等离激元-能量匹配特性,具有空腔的纳米孔阵列可以提供更高的共振传输效率和更窄的共振带宽。甘雪涛等[51]演示了一种基于高品质因子(Q)半导体平面光子晶体纳米腔的紧凑光谱仪,通过平面二维波导的耦合,在840 nm波长处分辨率高达0.3 nm。Najiminaini等[52]开发了一种二维快照多光谱成像仪,在透射光谱中,观察到与(1,0)和(1,1)表面等离激元激发相关的两个主要共振峰,其中(1,0)共振传输效率在55%到62%之间。
除了上述的滤色方法外,设计高分辨率滤光片最常见的方法是使用一对宽带高反射率反射镜形成一个法布里-珀罗(FP)谐振腔,超构表面的引入会进一步增强对光的调制。Horie等[46]将介电超构表面层放在具有较高Q的垂直FP谐振腔中,腔内的往返相位通过独立地调整纳米柱的宽度得到了极大的改变,最终在波长范围为1550±250 nm内获得了7个尖锐的透射峰,如
此外,利用光栅的色散性质,Kaplan等[53]提出了一种基于Ag光栅的金属谐振纳米结构滤色器,选择合适的光栅周期即可将入射光纳入特定共振波长的波导模式。利用这种原理获得的透射光透射率相对较高(约75%),然而滤波片的尺寸较大,约为1.25 cm
上述研究的滤波片工作波长都没有超过可见到近红外范围,Lee等[54]证明在太赫兹波段,也能通过网格滤波器阵列进行光谱编码,中心频率与滤波器在超构表面中的位置成正比,滤波范围为0.2 THz~2.0 THz。同年,科研人员还报道了一种具有更宽波长范围的多光谱材料[55],通过将可见和红外等离子体滤波器与太赫兹超材料吸收器杂交,它可以在RGB三原色波长、单个近红外波长、单个短波红外波长和两个中红外波长处实现滤波,此外它还可以吸收单个太赫兹频率。
基于窄带滤波器阵列的高光谱成像具有易与其他光电探测器集成的优点,2022年,Lee等[56]在CMOS图像传感器上制造了电介质多层滤波器,每个光谱通道的传输波长通过在相应的像素上嵌入相应尺寸的硅纳米孔阵列进行选择,最终在700~950 nm区域内获得了2 nm的高光谱分辨率,所有工作波长的透光率均超过60%。
直接把窄带滤波器镶嵌覆盖在CMOS/CCD上,是实现集成多光谱成像的一个想法,但同时也存在滤波片与像素之间易错位的问题[4, 57],难以从低分辨输入中重建高分辨率光谱图像。使用智能算法可以纠正这种错位问题。缪立丹等[4, 58-59]基于二进制树形网络解码估计了错位造成的缺失光谱分量,他们先生成了多光谱拼接滤波器,如
2.2.2 吸收型超构表面滤波器
吸收型滤波器对透射光具有特定窄带波长光的吸收,根据这一特性,基于吸收型窄带滤波的光谱成像系统通常将光电探测器集成在滤波片的后面,探测到的光强在吸收峰值处会骤然下降,另外一种方法是直接把滤波片作为电极,通过产生的光电效应的强度判断波长的位置[61]。
液晶分子会随着施加场的大小变化而改变转向,这会导致液晶折射率发生改变,因此液晶与亚波长光栅结合的器件可以看做一个可调谐滤波器[62],随着有效折射率的减小,滤波片的共振波长将发生小范围的蓝移。为了拓宽滤波范围,Grant等[61]将一个红外超构材料吸收层嵌入到一个标准的太赫兹超构材料吸收层中,在除了109 μm (2.75 THz)共振吸收峰外获得了中心波长为4.3 μm的窄带共振。如
图 5. 基于吸收型超构表面的光谱成像。(a)多光谱超构材料的三维原理图[61];(b)混合等离子体-焦电装置示意图[63];(c)在共振峰处激发的电场分布[63];(d)多层超构表面吸收器示意图[65]
Fig. 5. Spectral imaging based on absorption-type metasurface. (a) 3D schematic of the multispectral metamaterial absorber[61]; (b) Schematic diagram of the hybrid plasmonic− pyroelectric detectors[63]; (c) Electric field distributions excited at resonant peaks[63]; (d) Schematic illustration of the multilayer metasurface absorber[65]
除了利用光电响应检测吸收光外,结合了纳米光子学和热电学的谐振热电等离子体吸收器也可以用于光谱探测。2016年,Dao等[63]展示了一种具有窄带光谱选择性的混合等离子体-焦电装置,如
高分辨率光谱成像决定了吸收波的高吸收幅值和窄FWHM,然而前面提到的吸收型滤波器虽然能在特定波长范围内进行吸收滤波,但是还需要在获得尖锐的吸收峰方面做出进一步研究。2022年,一种惠更斯超构表面因为其对特定光的良好吸收属性而被用于多光谱成像[65],与只能支持如电偶极子(electric dipole,简称为ED)或磁偶极子(magnetic dipole,简称为MD)单一共振的普通介电超构表面不同的是,它可以同时激发统一波长的两种共振模式,这使得单层吸收器获得了超过70%的吸收,如
2.2.3 反射型超构表面滤波器
反射型窄带滤波一般通过测量透射波的突变来观察反射峰值。平面介电波导光栅中的共振现象已经得到了证明[67],特定的入射光波长和角度会耦合成一种被引导进入自由空间的模式,因此对应的模式不能经过结构传输到下一层。根据这种特性,Lin等[7]制造了一个波长范围为506~915 nm的基于梯度光栅周期的导模共振滤波器,如
图 6. 基于反射型超构表面的光谱成像。(a)导模谐振滤波器的原理图和透射光谱[7];(b)像素化介电超构表面的分子指纹检测[68]
Fig. 6. Spectral imaging based on reflection-type metasurface. (a) Schematic of the guided-mode resonance filter with gradient grating periods[7]; (b) Molecular fingerprint detection with pixelated dielectric metasurfaces[68]
本小节主要分析总结了基于透射型、吸收型和反射型超构表面的窄带滤波,与前两者相比,基于反射型超构表面的窄带滤波需要一定的空间收集光,因此体积相对较大。由于大部分的窄带滤波是通过共振而非衍射来选择波长的,所以对入射光的入射角度要求不高。需要注意的是,尽管窄带滤波器的峰值透射率可以通过一些方案提高,但是整个波段的平均透射率仍然很低,导致整个器件的效率较低,而且对于阵列型窄带滤波光谱仪,由于滤波片之间的距离较小,容易发生串扰,因此对滤波光的FWHM要求较高。
2.3 基于宽带滤波机理的超构表面光谱成像
综合利用超色散和窄带滤波的超构表面光谱成像来看,较强的光谱分辨能力和较高的光利用率很难兼顾,这是因为前者的分辨率与光程成反比,在原理上不利于小型化,而后者会过滤掉大部分光。近年来,得益于计算光谱[69-71]和算法[72-73]的发展,基于宽带滤波的超构表面[74-75]尽管不能像窄带滤波那样直接分辨光谱,但是可以通过复杂的后端算法重建光谱信息,高分辨率和易于集成化的微型成像系统使其在光谱成像领域中占领着重要的位置。
图 7. (a)基于光子晶体(PC)板的微型光谱仪[76];(b)光子晶体滤波器和新型滤波器的重建光谱对比图[88];(c)纳米柱的设计和制造图[91];(d)基于超构表面的多光谱和偏振检测原理图[92]
Fig. 7. (a) Micro-spectrometer based on photonic-crystal (PC) slabs[76]; (b) Comparison of reconstructed spectra of photonic crystal filters and novel filters[88]; (c) Design and fabrication diagram of nanocolumn[91]; (d) Schematic drawing of the metasurface-based multispectral and polarimetric detection[92]
其中,ei表示噪声信号。对F(λ)、Xi(λ)和A(λ)分别进行离散采样,得到f(λ)、xi(λ)和a(λ),则Yi的离散形式为
令
其中:
通过宽带滤波实现光谱成像有两个重要的步骤,第一步是获得随机分布的宽光谱曲线,第二步是利用算法重建光谱,由于欠定方程求解的特殊性,一般来说光谱响应曲线特征越明显,光谱重建能力越高。下面分这两个步骤来总结近年来出现的宽带滤波超构表面光谱成像。
2.3.1 随机分布的光谱响应曲线
由等离子体滤波器[84-85]为代表的共振滤波器是一种常见的滤波方法,但是这些滤波曲线的形状比较单一,不是理想的宽带滤波器。将超构表面嵌入一个光学腔,可以在多个波段支持共振[86],结合超构表面的传播相位和法布里-珀罗干涉原理,理论上该方法能将纳米腔厚度降低到传统的最小值λ/(2n)以下,极大地缩小滤波器的尺寸,然而在实践中制作过程比较复杂。近年来,由多个孔阵列构成的光子晶体板被用来获得宽带光谱响应[76, 81, 87-88],由于入射光束中的每一个波长分量在每个孔径处都有位移的衍射角,因此光束中的所有波长分量的总衍射信号是唯一的,即不同的孔径阵列对应了特定的光谱响应。Liu等[88]通过在光子晶体板上添加金(Au)纳米柱,引入了表面等离子体共振效应,使光子晶体板的透射曲线表现出窄带响应,从而进一步增强了滤光器的光谱重建能力,
此外,通过合理设计激发表面等离子体波的元原子,可以使超构表面具备同时检测入射光的偏振和光谱信息的能力[92-93]。
目前超构表面的随机光谱响应曲线绝大部分都采用正向设计的方法,这是一种较为直观简单的方法,但是操作效率较低,而且极有可能错失最优结果。以结果为导向的拓扑优化、遗传算法、基于伴随的梯度下降法和深度学习等逆向设计方法[94-96]可以较好地规避这一缺陷,能从众多不同排列和不同形状的结构中筛选得到目标光谱曲线,并且在优化的过程中还可以设置条件以获得角度不敏感、鲁棒性更好的光谱曲线。理论上来说,光谱响应曲线的特征越明显,曲线之间的差异性越大,越有利于进行光谱重建。Redding等[84]在演示一个基于硅芯片的多重光散射光谱仪时,提出通过光谱相关函数来量化在探测器上产生不相关强度分布所需的波长变化,然而对于利用宽带滤波机理的超构表面光谱成像,往往关注的是重建光谱的精度,对随机宽光谱曲线的优异程度并无量化的标准。
虽然已经有大量的结构材料可以获得高透射率的光谱响应,但是他们无一例外的都是属于被动超构表面,即每一种结构对应着固定的光谱曲线。因此研究能调谐光谱响应的主动超构表面对光谱光场调控具有重要的意义。目前,液晶[62, 97]、石墨烯[98]、多量子阱结构[99]和相变材料[100-103]是实现主动超构表面的方法。其中,考虑到高透射和稳定性等因素,基于过渡金属硫族化合物如GeSbTe(GST)的相变材料成为了光谱滤波主动超构表面的理想候选材料。与其他材料不同的是,GST可以在电或热的作用下从非晶相(a-GST)转换为晶相(c-GST),并在结晶时表现出显著的可逆折射率调制,如
图 8. 调谐型超构表面。(a)基于GSST的相变超构表面光谱调制器[103];(b) MIM结构的横切面图[101];(c) a-GST和c-GST作为介质层时的反射光谱和电场分布[101];(d)电可调谐滤色器示意图[105];(e)石墨烯超构表面调制器原理图[106]
Fig. 8. Tuned metasurface. (a) Phase-change metasurface spectral modulator based on GSST[103]; (b) Cross-sectional view of the MIM structure[101]; (c) Reflection spectrum and electric field distribution of a-GST and c-GST as dielectric layers[101]; (d) Schematic representation of the electrically tunable color filter[105]; (e) Schematic of graphene metasurface modulator[106]
除了在材料上选用可调谐的相变材料实现主动超构表面这种方法外,通过电来调谐液晶[62, 97]、石墨烯[106]超构表面也是一种实现主动超构表面的有效手段,如
2.3.2 光谱重建算法
基于宽带滤波的超构表面经过快照捕获的目标光谱是一种重叠的光谱,因此有必要通过光谱重建算法提取原始光谱信息。目前使用较多的算法有最小二乘法[90, 107]、Tikhonov正则化[82, 87]、压缩感知[78, 80-81]以及基于神经网络的深度学习[71, 108]方法。
对于如
图 9. 光谱重建算法。(a)光谱重建系统示意图[82];(b)宽带光谱的重建结果[82]; (c)基于CS理论的窄带光谱重建结果[81]; (d)参数约束光谱编码器和解码器的设计框架[83]; (e)基于深度学习的重建结果[113]
Fig. 9. Spectral reconstruction algorithm. (a) Schematic diagram of spectral reconstruction system[82]; (b) Reconstruction results of broadband spectra[82]; (c) Narrow band spectral reconstruction results based on CS theory[81]; (d) Design framework for parametric constrained spectral encoders and decoders[83]; (e) Reconstruction results based on deep learning[113]
其中,α是一个正则化参数,考虑到频谱不能是负数,因此需要对求解施加非负的约束,则上式可以写为
通过L-曲线法[109]或者广义交叉验证(GCV)法[110]可以自适应地选择正则化参数α,matlab中的正则化工具能具体地实现上述求解过程,宽带重建光谱如
基于稀疏采样的压缩感知(compressive sensing,简称为CS)也是一种被广泛应用的技术[111],与最小二乘法和正则化技术不同的是,CS理论只需要较少的随机稀疏采样信号即可反演出原始光谱图像,在有效地减小算法复杂程度的同时还具备较强的光谱重建能力。2022年,Xiong等[81]提出了一种基于可重构超构表面的硅实时超光谱成像芯片,他们利用CS理论重建了光谱,如
近年来,随着深度学习在各个方面的应用越来越广泛,利用深度学习进行光谱反演也成为了高光谱图像的一种重建方法。此外,传统的滤波器设计方式大多是启发式的,可能没有充分发挥滤波器的编码能力,而利用深度学习可以设计出最佳目标频谱对应的结构,
3 光谱成像的应用
与传统光谱仪相比,基于超构表面的光谱成像系统因为超高光谱分辨率、空间占用体积小和易与CMOS传感器直接集成等优点,在生物传感、遥感、医学诊断和人脸识别等领域都有着广泛的应用前景。
在分子层面上,探测蛋白质、DNA等大分子结构的状态可以使用化学荧光团对目标结构进行标记、然后通过探测器探测荧光团发出的光进行分析的研究方法,基于超构表面的光谱成像因为具有快照的优势而可以被用于快速荧光测量[114]。然而对于一些特殊情况,荧光标记可能会损坏分子或细胞的结构,因此需要采用无标记的技术。中红外光谱是一种强大的无损和无标记的技术,被广泛用于识别生物化学组成部分,但是由于中红外波长与分子尺寸不适配,光谱灵敏度往往会受到限制。Tittl等[68]提出利用亚波长谐振器的强近场增强可以克服这一限制,当由亚波长谐振器组成的滤波片的共振与吸收分子的光谱重叠时,增强的分子-谐振器的耦合会导致共振频率或强度的变化,如
图 10. (a)分子指纹检索和空间吸收绘图[68];(b)利用介电超构表面对石墨烯进行光学表征[66];(c)用于小鼠脑血流动力学成像的超构表面装置图[81]
Fig. 10. (a) Molecular fingerprint retrieval and spatial absorption mapping[68]; (b) Optical characterization of graphene using dielectric metasurface[66]; (c) Metasurface device diagram for mouse cerebral hemodynamic imaging[81]
与分子层面上的静态光谱探测相比,基于实时光谱成像的动态光谱探测在生物医学研究中有着更为巨大的潜力。一种基于压缩感知算法的超光谱成像芯片可以用于脑血流动力学成像[81],如
人脸识别是一种生物识别的身份认证方法,与传统密码不同的是,人脸这一生物特征特性不易被盗窃或改变,因此提供了更好的安全性能,在安防系统、日常生活中有着广泛的应用场景。然而在一些场景中可以通过打印的照片或面具来攻击人脸识别系统,尽管利用三维人脸识别或视频分析能检测出假脸,但是随着3D打印技术的兴起和仿生硅胶技术的发展,面具做得越来越逼真,难以识别。Kim等[116]基于结构光提出了一种全空间衍射超构表面,这可以为人脸识别和汽车机器人视觉应用提供超紧凑的深度感知平台。此外,光谱分析一直是鉴别不同材料的有效工具,Rao等[79]开发了一种能获得面部高精度高光谱信息的快照图像传感器,只需要50 ms即可高精度地测量面部的反射光谱并得到血红蛋白的吸收峰。
图 11. (a)对真脸和其他面具的光谱测量结果[79];(b)密码显示的工作原理图[117];(c)通过可见光波段不透明物体的近红外成像演示[119]
Fig. 11. (a) Spectral measurement results of a real face and other masks[79]; (b) Operation schematic of the crypto-display[117]; (c) Demonstration of NIR imaging through the object that is opaque at visible wavelengths[119]
除了通过光谱来研究探测物外,利用超构表面的光谱响应还能实现一种加密显示,Yoon等[117]提出了一种双模超构表面的概念,它可以同时控制透射和反射两种操作模式的相位和光谱响应。在透射模式下,超构表面通过调整入射光的相位分布,使其显示出“3.141592…”,在反射模式下,通过白光照明会产生一幅反射的“π”彩色图像,如
4 总结与展望
本文首先分析总结了多种机理的超构表面光谱成像,工作原理主要分为超色散、窄带滤波和宽带滤波三种,其中窄带滤波包含透射型、吸收型和反射型三种滤波方式,宽带滤波包含获得随机分布的光谱曲线和利用光谱重建算法重建光谱这两个关键的步骤。然后回顾了近年来超构表面光谱成像在实际中的应用研究。由于超构表面可以很容易地集成到二维平面中,因此基于超构表面的光谱成像对实现紧凑型光谱仪有重要的意义。
通过我们的分析还可以客观地看到,基于超构表面的光谱成像目前还存在一些瓶颈,主要包括:1)超色散的原理限制了集成化。利用超构表面本身存在的色散,并通过相位调控使波长在空间中依次排列实现的光谱成像具有较高的光利用率和较强的光谱分辨能力,然而对于集成度和分辨率难以兼顾。2)窄带滤波对工艺技术的要求较高。窄带滤波利用频率的主动选择进行光谱成像,通过牺牲光利用率达到了非常高的集成度,但是普通的微纳制造技术难以达到对精确度的要求,光谱分辨率受限于制造工艺的发展。3)宽带滤波需要复杂的重建算法。宽带滤波将对硬件的重心转移到后端光谱重建算法上,在具有较高的光利用率的同时保持了高集成度,分辨率与算法的精确度息息相关,因此需要大量的数据集和时间训练算法。虽然基于超构表面的光谱成像还存在上述的一些不足,但随着技术的发展和研究的进一步深入,光谱重建算法的效率将得到进一步优化,制造工艺的精度也将得到提升,我们相信未来能突破现存的瓶颈,实现真正意义上的高分辨紧凑型光谱仪。
Overview: As the inherent characteristics of substances, spectra can be well used to identify the chemical composition of substances. Spectral imaging is a technology that combines spectral information with spatial information, having a wide range of applications in the fields of material analysis, food safety, medical diagnosis and biological imaging. However, traditional spectrometers are usually composed of prisms, gratings and other splitter devices. Limited by the diffraction effect, their spectral resolution is inversely proportional to the optical path. Therefore, they generally have the disadvantages of large size, high cost and complex optical path, and their application in compact devices is limited. Although there have been Fourier transform spectrometer, micro ring resonator and other research related to reducing spectrometer volume, they still have some problems, such as not being able to deal with very irregular spectral signals, spectral resolution is limited by manufacturing technology, , which cannot solve the problem that the spectrometer is difficult to compact.
The metasurface is a kind of large area nano-structured surface composed of subwavelength small units, characterized by strong plasticity, high flexibility, and easy integration. The optical properties of the metasurface are determined by its micro - nano structure. By designing and optimizing the resonance phase, transmission phase, and geometric phase, metasurfaces can be used to effectively modulate the optical parameters of light on the plane, such as amplitude, phase, and polarization. Due to the excellent electromagnetic properties exhibited by metasurfaces, they can achieve complex functions that are difficult to achieve in conventional refraction and diffraction optics. The spectral imaging technology based on metasurfaces is an emerging optical imaging technology, which can perform high resolution and high sensitivity spectral imaging within micro imaging systems, providing an opportunity for achieving compact spectrometers. In this paper, we firstly discuss the spectral imaging of metasurfaces based on superdispersion, narrowband filtering and broadband filtering. Narrowband filtering includes three filtering methods: transmission type, absorption type and reflection type filtering, while broadband filtering includes two key steps: obtaining randomly distributed spectral curves and using spectral reconstruction algorithm to reconstruct spectrum. Compared with traditional spectrometers, the spectral imaging of metasurfaces based on superdispersion can reduce the volume of optical components to a certain extent, but it is difficult to balance integration and resolution. Narrowband filtering can be used for snapshot spectral cameras, but it has low light utilization and high technological requirements. Broadband filtering has high light utilization and strong spectral resolution, but it relies on spectral reconstruction algorithms, so it requires high algorithm requirements.Then, the recent applications of the spectral imaging based on metasurfaces are introduced, such as biosensing, medical diagnostics, and face recognition. Finally, the development direction and application prospects of the spectral imaging based on metasurfaces are prospected.
[1] Gowen A A, O'Donnell C P, Cullen P J, et al. Hyperspectral imaging - an emerging process analytical tool for food quality and safety control[J]. Trends Food Sci Technol, 2007, 18(12): 590-598.
[2] Delalieux S, Auwerkerken A, Verstraeten W W, et al. Hyperspectral reflectance and fluorescence imaging to detect scab induced stress in apple leaves[J]. Remote Sens, 2009, 1(4): 858-874.
[3] Kong L H, Yi D R, Sprigle S H, et al. Single sensor that outputs narrowband multispectral images[J]. J Biomed Opt, 2010, 15(1): 010502.
[4] Qi H R, Kong L H, Wang C, et al. A hand-held mosaicked multispectral imaging device for early stage pressure ulcer detection[J]. J Med Syst, 2011, 35(5): 895-904.
[5] Khorasaninejad M, Chen W T, Zhu A Y, et al. Multispectral chiral imaging with a metalens[J]. Nano Lett, 2016, 16(7): 4595-4600.
[6] Avrutsky I, Chaganti K, Salakhutdinov I, et al. Concept of a miniature optical spectrometer using integrated optical and micro-optical components[J]. Appl Opt, 2006, 45(30): 7811-7817.
[7] Lin H A, Huang C S. Linear variable filter based on a gradient grating period guided-mode resonance filter[J]. IEEE Photonics Technol Lett, 2016, 28(9): 1042-1045.
[8] Le Coarer E, Blaize S, Benech P, et al. Wavelength-scale stationary-wave integrated Fourier-transform spectrometry[J]. Nat Photonics, 2007, 1(8): 473-478.
[9] Kita D M, Miranda B, Favela D, et al. High-performance and scalable on-chip digital Fourier transform spectroscopy[J]. Nat Commun, 2018, 9(1): 4405.
[10] Vasiliev A, Malik A, Muneeb M, et al. On-chip mid-infrared photothermal spectroscopy using suspended silicon-on-insulator microring resonators[J]. ACS Sens, 2016, 1(11): 1301-1307.
[11] Nitkowski A, Chen L, Lipson M. Cavity-enhanced on-chip absorption spectroscopy using microring resonators[J]. Opt Express, 2008, 16(16): 11930-11936.
[12] Hartmann W, Varytis P, Gehring H, et al. Waveguide-integrated broadband spectrometer based on tailored disorder[J]. Adv Opt Mater, 2020, 8(6): 1901602.
[13] Falcone F, Lopetegi T, Laso M A G, et al. Babinet principle applied to the design of metasurfaces and metamaterials[J]. Phys Rev Lett, 2004, 93(19): 197401.
[14] Yu N F, Genevet P, Kats M A, et al. Light propagation with phase discontinuities: generalized laws of reflection and refraction[J]. Science, 2011, 334(6054): 333-337.
[15] Dai J Y, Zhao J, Cheng Q, et al. Independent control of harmonic amplitudes and phases via a time-domain digital coding metasurface[J]. Light Sci Appl, 2018, 7: 90.
[16] Yuan Q, Ge Q, Chen L S, et al. Recent advanced applications of metasurfaces in multi-dimensions[J]. Nanophotonics, 2023, 12(13): 2295-2315.
[17] 杨睿, 于千茜, 潘一苇, 等. 基于片上超表面的多路方向复用全息术[J]. 光电工程, 2022, 49(10): 220177.
Yang R, Yu Q Q, Pan Y W, et al. Directional-multiplexing holography by on-chip metasurface[J]. Opto-Electron Eng, 2022, 49(10): 220177.
[18] Wang S M, Wu P C, Su V C, et al. Broadband achromatic optical metasurface devices[J]. Nat Commun, 2017, 8(1): 187.
[19] Wang S M, Wu P C, Su V C, et al. A broadband achromatic metalens in the visible[J]. Nat Nanotechnol, 2018, 13(3): 227-232.
[20] Liu H C, Yang B, Guo Q H, et al. Single-pixel computational ghost imaging with helicity-dependent metasurface hologram[J]. Sci Adv, 2017, 3(9): e1701477.
[21] Yang Y H, Jing L Q, Zheng B, et al. Full-polarization 3D metasurface cloak with preserved amplitude and phase[J]. Adv Mater, 2016, 28(32): 6866-6871.
[22] SuV C, ChuC H, SunG, et al. Advances in optical metasurfaces: fabrication and applications [invited][J].Opt Express, 2018,26(10): 13148-13182.
[23] Chen W T, Zhu A Y, Sanjeev V, et al. A broadband achromatic metalens for focusing and imaging in the visible[J]. Nat Nanotechnol, 2018, 13(3): 220-226.
[24] Li K, Guo Y H, Pu M B, et al. Dispersion controlling meta-lens at visible frequency[J]. Opt Express, 2017, 25(18): 21419-21427.
[25] Xie T, Zhang F, Pu M B, et al. Integrated multispectral real-time imaging system based on metasurfaces[J]. Opt Express, 2020, 28(24): 36445-36454.
[26] Nishiwaki S, Nakamura T, Hiramoto M, et al. Efficient colour splitters for high-pixel-density image sensors[J]. Nat Photonics, 2013, 7(3): 240-246.
[27] Li Z Y, Palacios E, Butun S, et al. Visible-frequency metasurfaces for broadband anomalous reflection and high-efficiency spectrum splitting[J]. Nano Lett, 2015, 15(3): 1615-1621.
[28] Khorasaninejad M, Chen W T, Oh J, et al. Super-dispersive off-axis meta-lenses for compact high resolution spectroscopy[J]. Nano Lett, 2016, 16(6): 3732-3737.
[29] Zhu A Y, Chen W T, Khorasaninejad M, et al. Ultra-compact visible chiral spectrometer with meta-lenses[J]. APL Photonics, 2017, 2(3): 036103.
[30] Zhu A Y, Chen W T, Sisler J, et al. Compact aberration‐corrected spectrometers in the visible using dispersion‐tailored metasurfaces[J]. Adv Opt Mater, 2019, 7(14): 1801144.
[31] Zhou Y, Chen R, Ma Y G. Characteristic analysis of compact spectrometer based on off-axis meta-lens[J]. Appl Sci, 2018, 8(3): 321.
[32] Mohan A, Oberoi D. 4D data cubes from radio-interferometric spectroscopic snapshot imaging[J]. Sol Phys, 2017, 292(11): 168.
[33] Gómez-Sánchez A, Marro M, Marsal M, et al. 3D and 4D image fusion: coping with differences in spectroscopic modes among hyperspectral images[J]. Anal Chem, 2020, 92(14): 9591-9602.
[34] Faraji-Dana M, Arbabi E, Arbabi A, et al. Compact folded metasurface spectrometer[J]. Nat Commun, 2018, 9(1): 4196.
[35] Faraji-Dana M, Arbabi E, Kwon H, et al. Hyperspectral imager with folded metasurface optics[J]. ACS Photonics, 2019, 6(8): 2161-2167.
[36] Hua X, Wang Y J, Wang S M, et al. Ultra-compact snapshot spectral light-field imaging[J]. Nat Commun, 2022, 13(1): 2732.
[37] Billuart J, Héron S, Loiseaux B, et al. Towards a metasurface adapted to hyperspectral imaging applications: from subwavelength design to definition of optical properties[J]. Opt Express, 2021, 29(21): 32764-32777.
[38] Chen Y F, Zhao R Z, He H Y, et al. Spectrum dispersion element based on the metasurface with parabolic phase[J]. Opt Express, 2022, 30(18): 32670-32679.
[39] Gao H, Fan X H, Wang Y X, et al. Multi-foci metalens for spectra and polarization ellipticity recognition and reconstruction[J]. Opto-Electron Sci, 2023, 2(3): 220026.
[40] Zhu J C, Zhou J K, Shen W M. Polarisation‐independent diffraction grating based on dielectric metasurface[J]. Electron Lett, 2019, 55(13): 756-759.
[41] Eichenholz J M, Barnett N, Juang Y, et al. Real-time megapixel multispectral bioimaging[J]. Proc SPIE, 2010, 7568: 75681L.
[42] Wang S W, Chen X S, Lu W, et al. Integrated optical filter arrays fabricated by using the combinatorial etching technique[J]. Opt Lett, 2006, 31(3): 332-334.
[43] Wang S W, Li M, Xia C S, et al. 128 channels of integrated filter array rapidly fabricated by using the combinatorial deposition technique[J]. Appl Phys B, 2007, 88(2): 281-284.
[44] Wang S W, Xia C S, Chen X S, et al. Concept of a high-resolution miniature spectrometer using an integrated filter array[J]. Opt Lett, 2007, 32(6): 632-634.
[45] Xu T, Wu Y K, Luo X G, et al. Plasmonic nanoresonators for high-resolution colour filtering and spectral imaging[J]. Nat Commun, 2010, 1: 59.
[46] Horie Y, Arbabi A, Arbabi E, et al. Wide bandwidth and high resolution planar filter array based on DBR-metasurface-DBR structures[J]. Opt Express, 2016, 24(11): 11677-11682.
[47] McClung A, Samudrala S, Torfeh M, et al. Snapshot spectral imaging with parallel metasystems[J]. Sci Adv, 2020, 6(38): eabc7646.
[48] Jang W Y, Ku Z, Jeon J, et al. Experimental demonstration of adaptive infrared multispectral imaging using plasmonic filter array[J]. Sci Rep, 2016, 6: 34876.
[49] Ebbesen T W, Lezec H J, Ghaemi H F, et al. Extraordinary optical transmission through sub-wavelength hole arrays[J]. Nature, 1998, 391(6668): 667-669.
[50] Najiminaini M, Vasefi F, Kaminska B, et al. Experimental and numerical analysis on the optical resonance transmission properties of nano-hole arrays[J]. Opt Express, 2010, 18(21): 22255-22270.
[51] Gan X T, Pervez N, Kymissis I, et al. A high-resolution spectrometer based on a compact planar two dimensional photonic crystal cavity array[J]. Appl Phys Lett, 2012, 100(23): 231104.
[52] Najiminaini M, Vasefi F, Kaminska B, et al. Nanohole-array-based device for 2D snapshot multispectral imaging[J]. Sci Rep, 2013, 3: 2589.
[53] Kaplan A F, Xu T, Jay Guo L. High efficiency resonance-based spectrum filters with tunable transmission bandwidth fabricated using nanoimprint lithography[J]. Appl Phys Lett, 2011, 99(14): 143111.
[54] Lee K, Choi H J, Son J, et al. THz near-field spectral encoding imaging using a rainbow metasurface[J]. Sci Rep, 2015, 5: 14403.
[55] McCrindle I J H, Grant J P, Gouveia L C P, et al. Infrared plasmonic filters integrated with an optical and terahertz multi-spectral material[J]. Phys Status Solidi (A), 2015, 212(8): 1625-1633.
[56] Lee J, Park Y, Kim H, et al. Compact meta-spectral image sensor for mobile applications[J]. Nanophotonics, 2022, 11(11): 2563-2569.
[57] Chen Q, Das D, Chitnis D, et al. A CMOS image sensor integrated with plasmonic colour filters[J]. Plasmonics, 2012, 7(4): 695-699.
[58] Miao L D, Qi H R. The design and evaluation of a generic method for generating mosaicked multispectral filter arrays[J]. IEEE Trans Image Process, 2006, 15(9): 2780-2791.
[59] Miao L D, Qi H R, Ramanath R, et al. Binary tree-based generic demosaicking algorithm for multispectral filter arrays[J]. IEEE Trans Image Process, 2006, 15(11): 3550-3558.
[60] Jang W Y, Hayat M M, Tyo J S, et al. Demonstration of bias-controlled algorithmic tuning of quantum dots in a well (DWELL) MidIR detectors[J]. IEEE J Quantum Electron, 2009, 45(6): 674-683.
[61] Grant J, McCrindle I J H, Li C, et al. Multispectral metamaterial absorber[J]. Opt Lett, 2014, 39(5): 1227-1230.
[62] Chang A S P, Morton K J, Tan H, et al. Tunable liquid crystal-resonant grating filter fabricated by nanoimprint lithography[J]. IEEE Photonics Technol Lett, 2007, 19(19): 1457-1459.
[63] Dao T D, Ishii S, Yokoyama T, et al. Hole array perfect absorbers for spectrally selective midwavelength infrared pyroelectric detectors[J]. ACS Photonics, 2016, 3(7): 1271-1278.
[64] Mauser K W, Kim S, Mitrovic S, et al. Resonant thermoelectric nanophotonics[J]. Nat Nanotechnol, 2017, 12(8): 770-775.
[65] Wen S, Jin C Q, Yang Y M. Multilayer Huygens’ metasurface absorber toward snapshot multispectral imaging[J]. J Opt, 2021, 23(4): 044001.
[66] Yesilkoy F, Arvelo E R, Jahani Y, et al. Ultrasensitive hyperspectral imaging and biodetection enabled by dielectric metasurfaces[J]. Nat Photonics, 2019, 13(6): 390-396.
[67] Wang S S, Magnusson R. Theory and applications of guided-mode resonance filters[J]. Appl Opt, 1993, 32(14): 2606-2613.
[68] Tittl A, Leitis A, Liu M K, et al. Imaging-based molecular barcoding with pixelated dielectric metasurfaces[J]. Science, 2018, 360(6393): 1105-1109.
[69] Yao M D, Xiong Z W, Wang L Z, et al. Spectral- depth imaging with deep learning based reconstruction[J]. Opt Express, 2019, 27(26): 38312-38325.
[70] Wang Z J, Chen B, Lu R Y, et al. FusionNet: an unsupervised convolutional variational network for hyperspectral and multispectral image fusion[J]. IEEE Trans Image Process, 2020, 29: 7565-7577.
[71] Chen Y Y, Zhu Y L, Britton W A, et al. Inverse design of ultracompact multi-focal optical devices by diffractive neural networks[J]. Opt Lett, 2022, 47(11): 2842-2845.
[72] Chang C C, Lee H N. On the estimation of target spectrum for filter-array based spectrometers[J]. Opt Express, 2008, 16(2): 1056-1061.
[73] Morawski R Z, Miekina A. Improving absorbance spectrum reconstruction via spectral data decomposition and pseudo-baseline optimization[J]. IEEE Trans Instrum Meas, 2009, 58(3): 691-697.
[74] Chen Q, Chitnis D, Walls K, et al. CMOS photodetectors integrated with plasmonic color filters[J]. IEEE Photonics Technol Lett, 2012, 24(3): 197-199.
[75] Yoon Y T, Lee S S, Lee B S. Nano-patterned visible wavelength filter integrated with an image sensor exploiting a 90-nm CMOS process[J]. Photonics Nanostruct - Fundam Appl, 2012, 10(1): 54-59.
[76] Wang Z, Yi S, Chen A, et al. Single-shot on-chip spectral sensors based on photonic crystal slabs[J]. Nat Commun, 2019, 10(1): 1020.
[77] Yang J W, Cui K Y, Cai X S, et al. Ultraspectral imaging based on metasurfaces with freeform shaped meta‐atoms[J]. Laser Photonics Rev, 2022, 16(7): 2100663.
[78] Wu X L, Gao D H, Chen Q, et al. Multispectral imaging via nanostructured random broadband filtering[J]. Opt Express, 2020, 28(4): 4859-4875.
[79] Rao S J, Huang Y D, Cui K Y, et al. Anti-spoofing face recognition using a metasurface-based snapshot hyperspectral image sensor[J]. Optica, 2022, 9(11): 1253-1259.
[80] Wu Z P, Zhang Z Q, Xu Y J, et al. Random color filters based on an all-dielectric metasurface for compact hyperspectral imaging[J]. Opt Lett, 2022, 47(17): 4548-4551.
[81] Xiong J, Cai X S, Cui K Y, et al. Dynamic brain spectrum acquired by a real-time ultraspectral imaging chip with reconfigurable metasurfaces[J]. Optica, 2022, 9(5): 461-468.
[82] Kurokawa U, Choi B I, Chang C C. Filter-based miniature spectrometers: spectrum reconstruction using adaptive regularization[J]. IEEE Sens J, 2011, 11(7): 1556-1563.
[83] Song H Y, Ma Y G, Han Y B, et al. Deep‐learned broadband encoding stochastic filters for computational spectroscopic instruments[J]. Adv Theory Simul, 2021, 4(3): 2000299.
[84] Redding B, Liew S F, Sarma R, et al. Compact spectrometer based on a disordered photonic chip[J]. Nat Photonics, 2013, 7(9): 746-751.
[85] Wang Z, Yu Z F. Spectral analysis based on compressive sensing in nanophotonic structures[J]. Opt Express, 2014, 22(21): 25608-25614.
[86] Shaltout A M, Kim J, Boltasseva A, et al. Ultrathin and multicolour optical cavities with embedded metasurfaces[J]. Nat Commun, 2018, 9(1): 2673.
[87] Yang T, Xu C, Ho H P, et al. Miniature spectrometer based on diffraction in a dispersive hole array[J]. Opt Lett, 2015, 40(13): 3217-3220.
[88] Liu D L, Li Z H. New nano-structure spectrometer by introducing gold nano-pillars for spectral reconstruction ability improvement[J]. Opt Commun, 2022, 502: 127419.
[89] Stewart J W, Akselrod G M, Smith D R, et al. Toward multispectral imaging with colloidal metasurface pixels[J]. Adv Mater, 2017, 29(6): 1602971.
[90] Craig B, Shrestha V R, Meng J J, et al. Experimental demonstration of infrared spectral reconstruction using plasmonic metasurfaces[J]. Opt Lett, 2018, 43(18): 4481-4484.
[91] Xiao Y, Wei S, Xu J J, et al. Superconducting single-photon spectrometer with 3D-printed photonic-crystal filters[J]. ACS Photonics, 2022, 9(10): 3450-3456.
[92] Pelzman C, Cho S Y. Multispectral and polarimetric photodetection using a plasmonic metasurface[J]. J Appl Phys, 2018, 123(4): 043107.
[93] Pelzman C, Cho S Y. Plasmonic metasurface for simultaneous detection of polarization and spectrum[J]. Opt Lett, 2016, 41(6): 1213-1216.
[94] YangJ W, CuiK Y, HuangY D, et al. Angle-insensitive spectral imaging based on topology-optimized plasmonic metasurfaces[Z]. arXiv: 2212.07813, 2022. https://arxiv.org/abs/2212.07813.
[95] Liu T R, Fiore A. Designing open channels in random scattering media for on-chip spectrometers[J]. Optica, 2020, 7(8): 934-939.
[96] Ma T G, Tobah M, Wang H Z, et al. Benchmarking deep learning-based models on nanophotonic inverse design problems[J]. Opto-Electron Sci, 2022, 1(1): 210012.
[97] Zhao Q, Kang L, Du B, et al. Electrically tunable negative permeability metamaterials based on nematic liquid crystals[J]. Appl Phys Lett, 2007, 90(1): 011112.
[98] Song S C, Chen Q, Jin L, et al. Great light absorption enhancement in a graphene photodetector integrated with a metamaterial perfect absorber[J]. Nanoscale, 2013, 5(20): 9615-9619.
[99] Ebermann M, Neumann N, Hiller K, et al. Tunable MEMS Fabry-Pérot filters for infrared microspectrometers: a review[J]. Proc SPIE, 2016, 9760: 97600H.
[100] Cao T, Wei C W, Simpson R E, et al. Broadband polarization-independent perfect absorber using a phase-change metamaterial at visible frequencies[J]. Sci Rep, 2014, 4: 3955.
[101] Guo Z Y, Yang X, Shen F, et al. Active-tuning and polarization-independent absorber and sensor in the infrared region based on the phase change material of Ge2Sb2Te5 (GST)[J]. Sci Rep, 2018, 8(1): 12433.
[102] Julian M N, Williams C, Borg S, et al. Reversible optical tuning of GeSbTe phase-change metasurface spectral filters for mid-wave infrared imaging[J]. Optica, 2020, 7(7): 746-754.
[103] Tao C N, Zhu H Z, Zhang Y S, et al. Shortwave infrared single-pixel spectral imaging based on a GSST phase-change metasurface[J]. Opt Express, 2022, 30(19): 33697-33707.
[104] Aspnes D E. Local-field effects and effective-medium theory: a microscopic perspective[J]. Am J Phys, 1982, 50(8): 704-709.
[105] Lee Y, Park M K, Kim S, et al. Electrical broad tuning of plasmonic color filter employing an asymmetric-lattice nanohole array of metasurface controlled by polarization rotator[J]. ACS Photonics, 2017, 4(8): 1954-1966.
[106] Shrestha V R, Craig B, Meng J J, et al. Mid- to long-wave infrared computational spectroscopy with a graphene metasurface modulator[J]. Sci Rep, 2020, 10(1): 5377.
[107] Duempelmann L, Gallinet B, Novotny L. Multispectral imaging with tunable plasmonic filters[J]. ACS Photonics, 2017, 4(2): 236-241.
[108] Huang L Q, Luo R C, Liu X, et al. Spectral imaging with deep learning[J]. Light Sci Appl, 2022, 11(1): 61.
[109] Hansen P C. Analysis of discrete ill-posed problems by means of the L-curve[J]. SIAM Rev, 1992, 34(4): 561-580.
[110] Golub G H, Heath M, Wahba G. Generalized cross-validation as a method for choosing a good ridge parameter[J]. Technometrics, 1979, 21(2): 215-223.
[111] August Y, Stern A. Compressive sensing spectrometry based on liquid crystal devices[J]. Opt Lett, 2013, 38(23): 4996-4999.
[112] Jiang Y Y, Li G M, Ge H Y, et al. Adaptive compressed sensing algorithm for terahertz spectral image reconstruction based on residual learning[J]. Spectrochim Acta Part A Mol Biomol Spectrosc, 2022, 281: 121586.
[113] Zhang W Y, Song H Y, He X, et al. Deeply learned broadband encoding stochastic hyperspectral imaging[J]. Light Sci Appl, 2021, 10(1): 108.
[114] Connolly P W R, Valli J, Shah Y D, et al. Simultaneous multi-spectral, single-photon fluorescence imaging using a plasmonic colour filter array[J]. J Biophotonics, 2021, 14(7): e202000505.
[115] Meng J J, Weston L, Balendhran S, et al. Compact chemical identifier based on plasmonic metasurface integrated with microbolometer array[J]. Laser Photonics Rev, 2022, 16(4): 2100436.
[116] Kim G, Kim Y, Yun J, et al. Metasurface-driven full-space structured light for three-dimensional imaging[J]. Nat Commun, 2022, 13(1): 5920.
[117] Yoon G, Lee D, Nam K T, et al. "Crypto-display" in dual-mode metasurfaces by simultaneous control of phase and spectral responses[J]. ACS Nano, 2018, 12(7): 6421-6428.
[118] Xu Z J, Li N X, Dong Y, et al. Metasurface-based subtractive color filter fabricated on a 12-inch glass wafer using a CMOS platform[J]. Photonics Res, 2021, 9(1): 13-20.
[119] Park H, Crozier K B. Multispectral imaging with vertical silicon nanowires[J]. Sci Rep, 2013, 3: 2460.
Article Outline
万源庆, 刘威骏, 林若雨, 余浩祥, 王漱明. 基于超构表面的光谱成像及应用研究进展[J]. 光电工程, 2023, 50(8): 230139. Yuanqing Wan, Weijun Liu, Ruoyu Lin, Haoxiang Yu, Shuming Wang. Research progress and applications of spectral imaging based on metasurfaces[J]. Opto-Electronic Engineering, 2023, 50(8): 230139.