本文提出了一种基于多级微反射镜和栅格分束器的静态轻型傅立叶红外变换光谱仪, 通过两个多级微反射镜实现光程差的空间离散和干涉图的静态二维采样, 通过引入栅格分束器有效降低了系统的体积和重量。作为该光谱仪的核心光学器件, 多级微反射镜的阶梯高度一致性、面型平整度和结构精度是决定采样间隔、分辨率和噪声等仪器指标的主要因素。本文提出了基于MOEMS技术的厚度依次减半多层膜法, 制作了台阶高度为0625 μm, 阶梯数为32的低阶梯多级微反射镜。测得实际阶梯高度平均值为6269 nm, 表面粗糙度均方根值为172 nm。分析了阶梯高度误差对光谱复原的影响, 提出了两种阶梯高度误差校正方法, 分别为通过修正因子来减小膜厚监控误差, 和利用最小二乘余弦多项式算法对复原光谱进行校正。校正后的复原光谱误差(SCE)降低为234%, 满足系统对光谱复原的要求。最后, 将该低阶梯多级微反射镜置入光谱仪中, 得到乙腈样品的干涉图和复原光谱图。

In this study, a static and light Fourier transform infrared spectrometer based on stepped mirrors and a grid beam splitter was proposed. By introducing two stepped mirrors into the interference system, the optical path difference is discretized and the 2-dimensional sampling of the interferogram is obtained. Furthermore, by introducing the grid beam splitter into the interference system, the volume and weight are decreased. Stepped mirrors as the core optical devices of such a spectrometer, its step height consistency, face flatness and the structure′s precision determine the spectral sampling interval, resolution and noise of the system. We propose a method based on MOEMS technology involving multiple depositions accompanied by a 50% reduction in thickness at every iteration to fabricate a low-stepped mirror with 32 steps and 0625 μm in step height. The test results show that the root-mean-square of roughness is 172 nm and that the average height of the real steps is 6269 nm. The effect of the height error on the recovered spectrum is analyzed. In order to reduce the influence of this error, two methods are proposed: one is through using tooling factor to reduce the monitoring error of the film thickness, thus reducing the height error; the other is through using the least-squares approximation cosine polynomial algorithm to correct the recovered spectrum. The spectrum-constructing error(SCE) is reduced to 234%, which meets the requirements of spectral restoration. Finally, the experiment was carried out using low stepped mirrors and the interferograms were obtained before and after the addition of the sample. The absorption spectrum of the sample acetonitrile can be obtained using a Fourier transform.