Multidimensional Pattern Recognition in High-Resolution 2D and 3D Spectra of Gas-Phase Molecules
Multidimensional Pattern Recognition in High-Resolution 2D and 3D Spectra of Gas-Phase Molecules
复制标题
气相分子高分辨率 2D 和 3D 光谱中的多维模式识别
DOI：
10.1021/acs.accounts.2c00637
复制
发表时间：
2023
影响因子：
18.3
通讯作者：
Chen, Peter C.
中科院分区：
文献类型：
--
作者：
Chen, Peter C.
当分子从凝聚相转变到气相时，它们的光谱也发生了戏剧性的变化；凝聚相光谱中的每个峰都可以在气相中产生数千个峰，因为分子可以自由旋转，这些旋转运动是量子化的。这些气相光谱包含大量关于分子结构和行为的详细信息，但峰密度往往如此之高，以至于拥堵掩盖了指定峰和提取分子常数所需的模式。本文描述了连贯的多维技术如何不仅减少了气相光谱中的峰密度和拥塞，而且还创建了易于识别和解释的多维模式。首先，所有具有相同振动量子数的峰都形成了旋转图案，如X、双抛物线和星号。这些旋转模式由基本单元组成，可以提供有关分子结构、行为和旋转选择规则的即时信息。其次，这些旋转图案的组可以排列成振动图案，从而形成矩形或平行四边形的阵列。这些振动模式可以用来确定混波过程和测量振动常数。因此，相干多维光谱学自动分离振动和旋转信息，然后根据振动和旋转量子数对峰进行排序。此外，如果样品是由混合物组成的，那么这些模式还可以按物种对峰进行排序，更高维的技术甚至可以提供在混合物中选择物种的能力。这些技术已经成功地为样品产生了高度图案化的2D和3D光谱，否则会产生无图案的光谱，如同位素混合物和振动扰动的分子，如NO2。高密度的态可能会导致拥堵和扰动，从而难以使用传统上从一维光谱获得的信息准确分配峰：峰的强度及其频率。相干2D和3D技术非常适合于处理扰动并从中学习，因为多维空间中每个峰值的坐标包括多个频率值。当2D或3D频谱中沿一个频率轴扰动的峰值沿正交频率轴未扰动时，准确的指认是可能的。此外，图案经常在相邻的行或列中重复，因此不太拥挤的区域可以用来解析或识别严重拥堵的区域中的关键峰值或图案。微扰可以使多维旋转和振动图案中的间距稍微不规则，但这些自动生成的图案仍然很容易识别和分析。本文描述了三种高分辨率相干多维光谱技术，它们可以产生的图案类型，以及如何从这些图案中提取信息。这项工作是在斯佩尔曼学院进行的，这是一所历史上的黑人女子学院，所有学生都是本科生。由此产生的技术不仅对于处理一些最拥挤、最受干扰和最具挑战性的光谱系统非常有效，而且它们相对容易使用，设置价格适中，运行迅速。
ConspectusWhen molecules transition from the condensed phase to the gas phase, their spectra undergo a dramatic transformation as well; each peak in a condensed-phase spectrum can yield thousands of peaks in the gas phase because the molecules are free to rotate and those rotational motions are quantized. These gas-phase spectra contain a wealth of detailed information about molecular structure and behavior, but peak densities are often so high that congestion obscures the patterns needed to assign peaks and extract molecular constants. This Account describes how coherent multidimensional techniques not only reduce peak densities and congestion in gas-phase spectra but also create multidimensional patterns that are easy to recognize and interpret. First, all peaks with the same vibrational quantum numbers form rotational patterns such as X’s, double parabolas, and asterisks. These rotational patterns are composed of basic units and can provide immediate information about the molecule’s structure, behavior, and rotational selection rules. Second, groups of these rotational patterns can be arranged into vibrational patterns that form arrays of rectangles or parallelograms. These vibrational patterns can be used to determine wave-mixing processes and measure vibrational constants. Coherent multidimensional spectroscopy therefore automatically separates vibrational and rotational information and then sorts peaks by vibrational and rotational quantum number. Furthermore, if the sample is composed of a mixture, then these patterns can also sort peaks by species, and higher-dimensional techniques can even provide the ability to select a species in the mixture. These techniques have successfully produced highly patterned 2D and 3D spectra for samples that otherwise generate patternless spectra such as isotopologue mixtures and vibronically perturbed molecules such as NO2.High densities of states can lead to congestion and perturbations that make it difficult to accurately assign peaks using the information that is traditionally available from 1D spectra: a peak’s intensity and its frequency. Coherent 2D and 3D techniques are well-suited for dealing with and learning from perturbations because the coordinate of each peak in multidimensional space includes multiple frequency values. Accurate assignments are possible when peaks in 2D or 3D spectra that are perturbed along one frequency axis are unperturbed along an orthogonal frequency axis. Furthermore, patterns often repeat in adjacent rows or columns, so regions that are less congested can be used to resolve or identify key peaks or patterns in regions that are severely congested. Perturbations can make the spacings within multidimensional rotational and vibrational patterns slightly irregular, but these automatically generated patterns remain easy to recognize and analyze.This Account describes three high-resolution coherent multidimensional spectroscopy techniques, the types of patterns they can produce, and how information can be extracted from these patterns. This work is being conducted at Spelman College, a historically Black college for women where all of the students are undergraduates. The resulting techniques are not only highly effective for dealing with some of the most congested, perturbed, and challenging spectroscopic systems, but they are relatively easy to use, moderate in price to set up, and quick to run.