What Is an IR Spectrum Database?

Infrared spectrum is a molecule that can selectively absorb infrared at certain wavelengths, which causes the transition of vibrational and rotational energy levels in the molecule. The infrared absorption spectrum of a substance can be obtained by detecting the absorption of infrared rays. Spectrum ^[1] .

In organic molecules, the atoms that make up chemical bonds or functional groups are constantly vibrating, and the frequency of vibration is equivalent to that of infrared light. So, use

When a beam of infrared light with a continuous wavelength passes through a substance,

Division of infrared spectrum

The infrared spectrum is usually divided into three regions: the near-infrared region (0.75 to 2.5 m), the mid-infrared region (2.5 to 25 m), and the far-infrared region (25 to 300 m). Generally speaking,

Prism and

Infrared spectroscopy has a wide range of applicability to samples. Solid, liquid or gaseous samples can be applied. Inorganic, organic, and polymer compounds can be detected. In addition, infrared spectroscopy also has the characteristics of rapid testing, convenient operation, good repeatability, high sensitivity, small amount of sample, and simple instrument structure. Therefore, it has become the most commonly used and indispensable tool for modern structural chemistry and analytical chemistry. Infrared spectroscopy has also been widely used in the study of polymer configuration, conformation, and mechanical properties, as well as in the fields of physics, astronomy, meteorology, remote sensing, biology, and medicine ^[3] .

The position and intensity of the infrared absorption peak reflect the characteristics of the molecular structure and can be used to identify unknowns

Infrared spectrum of liquid water

The structural composition of the substance or its chemical group is determined; and the absorption intensity of the absorption band is related to the content of the chemical group, which can be used for quantitative analysis and purity identification. In addition, in the study of the mechanism of chemical reactions, infrared spectroscopy also played a certain role. But its most widely used is the structural identification of unknown compounds.

Infrared spectroscopy can be used not only to study the structure and chemical bonds of molecules, such as determination of force constants and criteria for molecular symmetry, but also as a method for characterizing and identifying chemical species. For example, gaseous water molecules are non-linear three-atom molecules. Its v 1 = 3652 cm, v 3 = 3756 cm, and v 2 = 1596 cm. In the infrared spectrum of liquid water molecules, due to the hydrogen bonding between water molecules, The telescopic vibration bands of v 1 and v 3 are superimposed together, and a wide band appears at 3402 cm, and its variable-angle vibration v 2 is located at 1647 cm. In heavy water, due to the larger atomic mass of deuterium than hydrogen, the overlapping bands of v 1 and v 3 of heavy water are shifted to 2502 cm, and v 2 is 1210 cm. The above phenomenon shows that although the structures of water and heavy water are very similar, the difference in infrared spectrum is very large.

Infrared spectra are highly characteristic, so it is common to use methods that compare with the infrared spectra of standard compounds for analysis and identification. Several standard infrared spectra have been compiled and published, such as the "Satler Standard Infrared Grating Spectrum The infrared spectra of more than 100,000 compounds were collected. In recent years, these maps have been stored in computers for comparison and retrieval.

The number of bands of certain groups or chemical bonds in the molecule in different compounds is basically fixed or changes only in a small wave range. For example,

It often appears at 1600 to 1750 cm and is called the characteristic wave number of carbonyl group. Many chemical bonds have characteristic wave numbers, which can be used to identify the type of compound and can also be used for quantitative determination. Due to the interaction of adjacent groups in the molecule (such as the formation of hydrogen bonds, coordination, conjugation, etc.), the chemical environment of the same group in different molecules will be different, so that their characteristic wave numbers will change to some extent Range ( see table below ).

Qualitative analysis of infrared spectrum

Infrared spectroscopy is one of the important methods for material identification. Its analysis can provide a lot of information about functional groups, and can help determine some and all molecular types and structures. The qualitative analysis has high characteristics, short analysis time, and required tests.

IR

The advantages of small sample volume, non-destructive sample and convenient measurement.

The traditional method of identifying substances by infrared spectroscopy usually uses the comparison method , that is, the method of comparing with a standard substance and consulting the standard spectrum, but this method requires higher samples and depends on the size of the spectral library. If a consistent spectrum cannot be retrieved in the spectral library, it can be analyzed by manual spectral analysis, which requires a lot of infrared knowledge and experience accumulation. The infrared spectra of most compounds are complex. Even experienced experts cannot guarantee that all the molecular structure information is obtained from an isolated infrared spectrum. If you need to determine the molecular structure information, you need to use other analytical tests. Means, such as NMR, mass spectrometry, UV spectroscopy, etc. Nevertheless, infrared spectroscopy is the most convenient and quick way to provide functional group information.

In recent years, computer-based methods have been used to analyze infrared spectra, and extensive research has been conducted at home and abroad. New results are constantly emerging, which not only improves the speed of understanding the spectrum, but also has a high success rate. With the continuous advancement of computer technology and the continuous improvement of spectrum interpretation ideas, computer-assisted infrared spectrum resolution will definitely have a more positive impact on the efficiency of teaching and scientific research.

Quantitative infrared spectrum analysis

The quantitative analysis method of infrared spectrum is based on Lambert-Beer law. Compared with other quantitative analysis methods, infrared spectroscopy quantitative analysis method has some disadvantages, so it is only used in special cases. It requires that the selected quantitative analysis peak should be

Prism

It has sufficient intensity, that is, the peak with a large molar absorption coefficient, and does not overlap with other peaks. Quantitative methods of infrared spectrum mainly include direct calculation method, working curve method, absorbance ratio method and internal standard method, which are often used for the analysis of isomers.

With the development of chemometrics and computer technology, quantitative analysis of infrared spectra using various methods has also achieved good results, such as least squares regression, correlation analysis, factor analysis, genetic algorithms, artificial neural networks, etc. The introduction makes it possible for the quantitative analysis of complex multicomponent systems by infrared spectroscopy.

Quantum mechanics research shows that the energy of molecular vibration and rotation is not continuous, but quantized,

IR

That is, it is limited to some discrete, specific energy states or energy levels. Taking the simplest diatomic as an example, if the interatomic vibration is considered to comply with the law of simple harmonic motion, its vibration energy E v can be approximately expressed as:

Where h is the Planck's constant; v is the vibrational quantum number (a positive integer); v 0 is the simple resonance frequency . When v = 0, the energy of the molecule is the lowest, which is called the ground state. When a molecule in the ground state is irradiated with infrared rays with a frequency of v 0, the molecule absorbs a photon with an energy of h v 0 and transitions to a first excited state to obtain an infrared absorption band with a frequency of v 0. Conversely, the molecules in this excited state can also return to the ground state by emitting infrared rays with a frequency of v 0. The value of v 0 depends on the reduced mass of the molecule and the force constant k. k depends on the distance between the nuclei of the atoms, the position of the atoms in the periodic table, and the bond order of the chemical bonds.

The larger the molecule, the more infrared bands. For example, a molecule containing 12 atoms should have 30 kinds of normal vibrations, and its fundamental frequency should also have 30 bands. It may also have weaker octave bands. The perturbation effects of frequency bands, combined frequencies, differential frequency bands, and vibration energy levels make the corresponding infrared spectrum more complicated. If the molecule is assumed to be a rigid rotor, its rotational energy Er is:

Where j is the rotational quantum number (a positive integer); i is the rotational inertia of a rigid rotor. Transitions can also occur between certain rotational energy levels, producing rotational spectra. In the process of molecular vibrational transitions, it is often accompanied by rotational transitions, which makes the vibrational spectrum appear band-shaped.

Auxiliary parsing

The structural identification of organic compounds has become more and more important in many fields such as organic chemistry, biochemistry, pharmacology, and environmental science. In various identification methods, infrared spectroscopy has become an organic structure identification because of its convenient and sensitive properties. Important means, in addition to its sensitive response to structural features, the direct connection of the infrared spectrometer to the computer has also created conditions for the introduction of some intelligent means related to computer science.

The appearance and wide application of various modern analytical instruments make it possible to obtain a large amount of information about the material system in a short time, which provides opportunities for data mining research in chemometrics. Spectra package recorded by spectrometer

infrared spectrometer

Contains a large amount of structural information, but it is not yet possible to directly calculate the complex molecular spectral spectrum. The analysis is mainly based on experience. For a person who is not engaged in structural identification for a long time, it is difficult to analyze a spectral spectrum. work. In fact, even for less complex molecules, it is difficult to specify the assignment of functional groups and peaks where all heteroatoms are located, and relying on various computer search systems will be subject to various restrictions, such as limited data in the spectral library, or measurement conditions (Instrument type, specific experimental conditions, etc.) Different from the conditions used in the standard map, the position of each absorption peak is changed. In addition, the infrared spectrum is very complicated, the masses of the atoms that make up the compound are different, the nature of the chemical bonds are different, and the order of the atoms and the spatial position of the atoms will cause differences in the infrared spectrum. These all complicate the analysis of the infrared spectrum. If the knowledge of infrared spectroscopy can be learned and stored by a computer, and computer-assisted analysis of the spectrum is completed, it is naturally a very meaningful thing.

For decades, people have been exploring the intelligent analysis of infrared spectra. With the computerization of commercial infrared spectrometers, many computer-assisted infrared spectrum recognition methods have appeared. These methods can be roughly divided into three categories: spectrum retrieval systems, expert systems, and pattern recognition methods.

IR Spectrum Search

The main advantage of spectrum search is that it can collect a large number of spectra. As long as the compound can be identified based on the spectrum of the unknown without other data (such as molecular formula), its procedure is relatively simple. But it also has some insurmountable disadvantages:

First, the capacity of the retrieval system is proportional to the number of compounds stored in the spectral library. It is impossible for us to collect all compounds in nature. The development of the spectral library always lags behind the development of organic chemistry. Secondly, with the development of technology, the spectroscopic instrument is continuously improved: the spectrum range is continuously expanded, the resolution is continuously improved, low-temperature technology is applied, and the emergence of some new instruments, which requires the original spectrum library to be continuously modified, and a large spectrum library It can't be done in a short time. Due to these characteristics of the retrieval method, it cannot be regarded as a complete means of structural identification.

expert system

Another method of computer-aided structural analysis is expert system. Its research areas include: mathematical proofs, programming, behavioral science and psychology, life sciences and medicine.

The general method for analyzing the spectra of the expert systems currently designed is: storing some rules of the chemical structure to form the spectrum in advance in the computer; inferring some imaginary structural formulas of the unknown from the spectral characteristics of the unknown spectrum; The theoretical spectrum of the imaginary structural formula, and then compare the theoretical spectrum with the experimental spectrum, and continue to modify the imaginary structural formula, and finally get the correct structural formula. However, the current absorption rules of various groups in molecules are mainly obtained through experience or artificial. The infrared spectra of a large number of known compounds were artificially compared, from which the absorption rules of various groups were summarized. Although the results truly reflected the correspondence between the infrared spectrum and the molecular structure, they were not accurate enough, especially these empirical formulas Knowledge is difficult to process with computers, making it difficult for computer expert analysis systems to be practical.

Pattern recognition

The development of pattern recognition began in the 1950s, and machines were used to replace people to classify and describe patterns, thereby achieving the recognition of things. With the widespread application of computer technology, the conditions for processing large amounts of information have been met. Pattern recognition has flourished in the 1960s, and the theoretical foundation was laid in the early 1970s, thus establishing its own unique discipline system. Pattern recognition has been applied to relevant aspects in the field of analytical chemistry, the most involved of which is the analysis of molecular spectra, which has been successful in some classification problems.

Quantum mechanics

Munk is equivalent to the first application of linear neural networks to the sub-structure analysis of infrared spectrum in 1990, which brought the analysis of infrared spectrum into a completely new field, and since then it has caused a boom in computer analysis of infrared spectrum. Subsequently, various methods, such as various artificial neural networks, partial least squares, and signal processing methods such as wavelet transform, were gradually introduced into the computer analysis of the infrared spectrum, so that pattern recognition has been well developed in the application of infrared spectrum.

Cabrol-Bass et al. Used a hierarchical neural network system to identify substructures of the infrared spectrum. First, the spectrum of 10,000 compounds was divided into five categories: benzene ring, hydroxyl group, carbonyl group, C-NH group, and C = C group. These groups were further classified into 33 substructures. Each lower-level network uses the output from the upper-level network. Take the 3596 ~ 500 cm-1 band every 12 cm-1 as the input of the neural network, and the output is "1" and "0", which represent the existence and non-existence of the substructure, respectively. A back-propagation neural network with a hidden layer of 30 nodes was used to identify each substructure, and a comprehensive but rough classification of compounds was made, involving some common compounds in the database.

Most of these studies use neural networks to identify substructures, but have not done in-depth research on specific classes of compounds, and have not discussed in depth the characteristic absorption peaks of compounds. In addition, the artificial neural network which is most commonly used in identifying sub-structures is not very accurate in predicting structural fragments, and the neural network has problems such as instability, easy to fall into local minima, and slow convergence.

Therefore, in recent years, people have been looking for a better pattern recognition method to analyze the structure of the infrared spectrum. Vapnik et al. Proposed a support vector machine (SVM) on the basis of statistical learning theory (SLT) in 1995, which seeks between the complexity of the model and the learning ability based on limited sample information The best compromise to get the best generalization ability. SVM has been successfully used in chemistry at present. SVM can better identify the substructure of infrared spectrum. Compared with ANN, SVM also has the advantages of stability and fast training speed, which is a good assistant. Infrared spectral analysis tools.