What Is Near Infrared Spectroscopy?

Near Infrared Spectrum Instrument (NIRS) is an electromagnetic radiation wave between visible light (Vis) and mid-infrared (MIR). The American Society for Materials Testing (ASTM) defines the near-infrared spectral region as 780-2526nm. The region is the first non-visible light region found in the absorption spectrum. The near-infrared spectral region is consistent with the combined frequency of the vibration of the hydrogen-containing groups (OH, NH, CH) in the organic molecules and the absorption regions of the frequency doubling at various levels. By scanning the near-infrared spectrum of the sample, the hydrogen-containing groups of the organic molecules in the sample can be obtained The characteristics information of the cluster, and the use of near-infrared spectroscopy to analyze samples have the advantages of convenience, fastness, efficiency, accuracy and low cost, without damaging the sample, consuming chemical reagents, and not polluting the environment. Favor.

Chinese name: Near infrared spectroscopy
Foreign name: Near Infrared Spectrum Instrument

Short name: NIRS
Meaning: Electromagnetic radiation wave

History of near infrared spectroscopy

The near-infrared spectral region was discovered by Herschel during the measurement of the infrared part of the solar spectrum in 1800. In order to commemorate the historical discovery of Herschel, people called the band between 780 and 1100nm in the near-infrared spectral region as the Herschel spectral region.

Infrared spectral analysis technology was recognized as an effective analytical method in the 1930s. At that time, infrared instruments were mainly used for the study of molecular structure theory. The spectral absorption band in the near-infrared region is a superposition of the frequency doubling, combination, and difference frequency absorption bands of the fundamental energy absorption of the chemical bonds (mainly CH, OH, NH) in the organic infrared region in the mid-infrared spectral region. Due to the serious overlap and discontinuity of the spectra in the near-infrared region, it is difficult to directly extract the information related to the component content in the near-infrared spectrum of the substance and give a reasonable spectral analysis. Organic matter has more absorption bands in the mid-infrared region, narrow bands, large absorption intensity, and significant characteristic absorption. Traditional spectrographers and chemical analysts are used to performing spectral analysis in the mid-infrared fundamental frequency absorption band. Therefore, the near-infrared spectral region is a spectral region that has been ignored and forgotten for a long time.

With the development of infrared instrument technology, the development of more stable power supplies, signal amplifiers, more sensitive photon detectors, and microcomputers has made the near-infrared spectral region an independent and unique information characteristic spectral region. . Karl Norris, the founder of the development of near-infrared spectroscopy technology, started the near-infrared spectroscopy technology in the 1950s with the support of the United States Department of Agriculture for the rapid composition of agricultural products (including cereals, feed, fruits, vegetables, etc.). Research on quantitative detection. Norris's early work was mainly to explore a reasonable method of near-infrared spectroscopy to study the spectral absorption characteristics and scattering characteristics of substances reflected by near-infrared light. He first proposed a multiple linear regression (MLR) algorithm The advantages of infrared spectrum calibration model establishment and spectral information extraction and analysis have played an important role in the formation of the later theoretical system of near-infrared spectroscopy technology. In the 1960s, the research group led by Norris conducted a large number of spectroscopic method demonstrations, including comparisons of visible, near-infrared band transmission, reflection, and transflective measurement methods. The reflection absorption spectra of plant leaves and grains are too obtained, which provides greater advantages and convenience for the development of near-infrared spectroscopy technology. At the same time, Norris developed the world's first near-infrared scanning spectrometer. This spectrometer was improved on the basis of the Cary 14 monochromator. It has the function of data transmission with a microcomputer. On the spectrometer, the advantage of the multiple linear regression analysis method in extracting the spectral information related to the components was demonstrated, and this instrument became the prototype of the development of the near-infrared spectral analysis instrument later.

Near infrared spectrum

The successful development of the cereal moisture near-infrared analyzer and its widespread use are a milestone in the development of near-infrared analysis technology. Water exists in any organism and has a large specific gravity, and the near-infrared absorption spectrum of water has strong characteristics, high absorption intensity, its frequency doubling and combined frequency absorption bands are separated from each other, and the spectral resolution is high. Therefore, the analysis performance of the near-infrared moisture analyzer is relatively stable and highly accurate, and it was first recognized by the agricultural and industrial circles in the near-infrared spectrum analysis instrument family. However, things always have two sides. The strong absorption of OH in water has a strong interference with the spectral analysis and content determination of other components in the material. How to eliminate the interference between water absorption on each component and other components It became a key problem in the near-infrared spectroscopy analysis technology, and the proposed analytical method of correlation spectrum calibration analysis effectively solved this problem. Shenk, Hoove, McClure, and Hamid, under the leadership of Norris, designed and completed near-infrared spectroscopic analysis instruments for quantitative analysis of forage and tobacco components in the 1970s.

Based on the accumulated experience of near-infrared spectroscopy analysis technology and the maturity of instrument development technology, many companies (such as Dickie-John, Bran Leubbe, and Technicon) have joined the commercialization team of near-infrared analytical instruments, of which Dickie-John Company Produced the world's first commercial filter-type near-infrared spectrometer. Bran Leubbe produced the world's first commercial raster-scanned near-infrared spectrometer. In 1971, the world's first commercial near-infrared Neotec company Grain Quality Analyzer entered the market. The promotion and use of near-infrared analysis technology in various applications in the entire agricultural field has made the technology into a mature stage in agricultural applications. 1975 Canadian Grain Commission The Canadian Grain Commission accepted the near-infrared method as the official method for the determination of wheat protein. 1984 American Association of Public Analysts (AOAC) # 989.03: NIR became the standard method for the analysis of protein, acidic detergent fibers, and neutral detergent fibers in feed. Near During the development of infrared instrument technology and calibration technology, many difficult problems were solved one by one, including the working stability of the instrument itself, the impact of the physical and chemical characteristics of the test sample on the calibration model, the impact of sample preparation, and the environment. Factors (such as temperature, humidity, ambient light, vibration, etc.), these problems have been satisfactorily explained through a lot of experiments and application discussions. New method of calibration in 1994: artificial neural network technology-solving nonlinearity. 1995 NIRSystems introduced digital signal control based holographic grating DDS system.

Before the 1980s, although the near-infrared analytical instrument used multivariate linear regression technology to establish a calibration model and obtained satisfactory results in the field of agricultural applications, how can multiple regression variables complete the composition to be measured under specific combinations? The related calculations between the infrared spectral absorbance data and the reference chemical data, what is the characteristic relationship between each spectral variable and the component to be measured, and the instability caused by the particle size and scattering effects of the sample still need to be reasonably explained. . For a long time, although the analytical performance of near-infrared analysis instruments has been recognized in the agricultural field, both researchers and users have regarded near-infrared analysis technology as a relatively formed "black box" technology. Until the method of multivariate statistical variables (chemometrics) was developed in the 1980s and the method was introduced into the near-infrared spectrum analysis and calibration technology, the near-infrared analysis technology truly reached the unity of calibration theory and practice. Promoted the development of this technology and chemometrics side by side, so the 1980s were called "the era of chemometrics".

During this period, a climax of using chemometrics for data pre-processing to achieve near-infrared spectral analysis and calibration model optimization was raised, which mainly aimed at the problem of scattering caused by factors such as sample particle size and packing density. Ian Cowe and Jim McNicol first applied the principal component regression analysis method to the data reduction and compression processing of near-infrared spectroscopy to achieve the stability of the calibration model. By optimizing the regression main factors, non-measurement factors such as particle size and distribution were excluded. ) And the effects of non-linear factors have achieved very good results. At the same time, they were surprised that the main factor used by the stable calibration model has a strong correspondence with the main near-infrared spectral absorption of the component to be measured, and can give a satisfactory explanation for the rationality of the calibration model.

Development of near infrared spectrometer

Kawalski and his graduate students first applied partial least squares regression technology to spectroscopy, but it has not been applied and popularized in near-infrared analysis technology until recent years. After the parallel development of theory and practice, chemometrics has formed a relatively complete system, which is mainly divided into two modules: qualitative and quantitative analysis. H. Mark and others discussed it in detail. The application and discussion of chemometrics in the near infrared field during this period can be mainly focused on the following aspects:

1) Non-linear regression calibration methods such as artificial neural networks, local weight regression, etc. are used for the discussion of multi-variable non-linear calibration models.

2) Selection method of optimal calibration sample set.

3) Discussion on large calibration sample set based on wavelet transform data compression technology.

4) The comparison of partial least square regression and other factor regression methods in selecting the best factor.

5) Optimization of the calibration wavelength channel.

6) The calibration model transfer problem of the same type of instrument.

With the continuous development of chemometrics in the field of near-infrared spectroscopy, researchers can more accurately grasp the linear correlation between the near-infrared spectral absorbance information and the chemical composition information of substances. There is not much improvement in the accuracy of the quantitative analysis results.

The performance of near-infrared spectral analysis instruments has been greatly improved with the continuous advancement of optical technology, electronic technology, hardware technology, and computer and software technology. High-signal-to-noise Fourier transform and raster scan type spectrometers have been successfully developed and started. Entering the instrument market, the development of filter-type near-infrared analyzers has entered the mature stage and has become a mainstream product in near-infrared instruments. At the same time, near-infrared spectroscopy analysis technology has also entered the practical application stage in fields other than agriculture (such as textile industry, chemical industry, pharmaceutical industry, paper industry, etc.), especially in industrial field analysis, online quality monitoring, etc. This technology shows its unique advantages. In the 1990s, many new near-infrared analysis instruments based on different spectroscopic principles, such as diode array types, acousto-optic modulation types, and imaging spectral types, appeared. These instruments have good development potential in fast field real-time measurement and are contemporary Typical representative of the development of near-infrared spectroscopy instruments.

After nearly half a century of development, the near-infrared spectrum analysis technology has become one of the most promising analysis technologies in the new century. Many countries have now established special scientific research capabilities to conduct research and development in related fields of equipment and equipment. Reducing instrument cost and maintaining sufficient analytical performance has become the dominant direction in the development of near-infrared instruments today. Many developed countries in Europe have adopted this technology as the standard technology for industry product quality assessment in many fields, almost completely replacing the previously widely used chemical analysis methods, and have achieved good results in terms of production efficiency and product quality.

China's research on NIR spectroscopy technology started late. In the late 1980s, the Changchun Institute of Optomechanics and Physics undertook the "Eighth Five-Year" scientific and technological project issued by the State Grain Bureau, and successfully developed a filter-type feed near-infrared analyzer. In the following ten years, filter-type near-infrared analysis instruments have been developed that can analyze corn, wheat, soybean and other food crops. At present, they are engaged in research and instruments in ginseng, human blood glucose, coal, honey, tea, etc. Development work.

At the same time, China has developed a raster-scanning near-infrared analyzer for the rapid quantitative detection of petroleum components in the petrochemical industry, and has achieved promising results. The research on near-infrared spectroscopy analysis technology in China has also been relatively mature, and it is estimated that the application and promotion of near-infrared analysis instruments in various fields can be completed in the next few years.

Near infrared spectroscopy physical mechanism

1 Simple resonance model

In order to reasonably explain the physical mechanism of the spectrum generated by the interaction of light and matter, physicists have established a variety of theoretical models such as rigid rotors, simple harmonic oscillators (linear harmonic oscillators), non-rigid rotors, non-harmonic oscillators, rotational models, and polyatoms Molecular vibration and rotation models. Among them, the molecular vibration frequency given by the diatomic molecular linear harmonic vibration model is located in the mid-infrared region. The rigid rotor model and rotation model are generally used to study the interaction mechanism of gaseous molecules and light. The following gives the theoretical derivation process of the diatomic molecule linear harmonic oscillator vibration model under the expressions of classical mechanics and quantum mechanics.

In classical mechanics, this two-body problem is generally reduced to a single-body problem with a reduced mass and a relative displacement x in the centroid coordinate system. Its equation of motion can be written as:

Near infrared spectroscopy

This is a classic electromagnetic wave expression, and its vibration frequency is:

Near infrared spectroscopy

Using this simpler vibrational frequency formula can explain many spectral absorption characteristics that appear in the infrared region. According to Newton's second law, the energy conversion expression of this diatomic elastic system is:

Near infrared spectroscopy

According to the law of conservation of energy, the kinetic energy and potential energy of the atoms in the system are converted into each other. When the diatomic motion reaches the maximum amplitude x, the kinetic energy of the system is zero. The expression of the potential energy can be obtained by integrating the two sides of the above formula:

Near infrared spectroscopy

Figure 1 shows the potential energy diagram of an elastic system. It can be seen from the figure that the change in potential energy is also continuous with the continuous change in the amplitude of the diatomic molecule.

Figure 1 Potential energy diagram of the diatomic molecular elastic system

But experimental observations at the end of the nineteenth century show that energy is not continuously absorbed,

Near infrared spectroscopy This is inconsistent with the above classic conclusions. In 1900, Planck proposed the quantum theory of light to reasonably explain this experimental phenomenon. He pointed out that the interaction between energy and the atomic system is discontinuous, and changes in energy can only be accomplished by the transitional absorption or radiation of particles between two discrete energy states. These energy states are quantized, and each energy level Energy is deterministic, and the transition between energy levels can only occur in the absorption and emission of electromagnetic waves under the effect of particles and external light energy or spontaneous radiation. Planck further gives the correlation between the energy change and frequency of the electromagnetic waves radiated and absorbed by the system:

Near infrared spectroscopy

Where h is the Planck constant. It can be seen from the formula that if the frequency range of the incident light is wide, since the energy between the energy levels is determined, only the optical radiation of a specific frequency can change the existing energy state and cause a transition, but this is only a kind of The ideal experimental model is that the true molecular energy levels are infinite. Without the constraints of the transition selection rule, even the absorption spectrum formed by such a simple model is extremely complicated. Molecules in space generate many energy levels due to the spin of their own center of mass and vibrations that periodically deviate from the equilibrium position. The absorption in the infrared band is mainly due to the mutual transitions between different vibration and rotational energy levels in the molecule. However, the energy required for the transition between the rotational energy levels is much smaller than the vibrational energy level. Due to the sensitivity limitation of the spectroscopic instrument detection, the rotational absorption band can only be observed in the absorption spectrum of gas, so for the study of liquid and solid As far as infrared absorption bands are concerned, only vibrational absorption modes of molecules are considered.

Near infrared spectroscopy

Figure 2 Absorption spectrum of the ideal single vibration mode model

The energy levels of vibration absorption modes and other forms of molecular energy are quantized. The vibration mode energy levels allowed by any particular system can be obtained by solving the energy eigenvalues of the Schrödinger equation in the representation of quantum mechanical states. The selection rule of the energy level transition is obtained by the orthogonality peculiar to the eigenfunction corresponding to the eigenvalue. The transition between the energy levels of molecular chemical bonds is selective to the frequency of incident light, so the purpose of determining molecular characteristics and structure can be achieved by detecting the light energy that interacts with the molecule. This is the infrared absorption spectrum used for substances The mechanism of qualitative analysis of ingredients.

The selection rules that allow transitions between the various energy levels of the molecule are given below:

1) In quantum mechanics, energy level transitions can only occur if the dipole transition matrix element of the eigenfunction corresponding to each energy eigenvalue in the energy representation is not zero when the quantum numbers differ by one. The energy levels of the simple resonance subsystem determined by this rule are equally spaced, and theoretically there is only one vibration absorption band.

2) The spectral absorption band can only occur when the molecule interacts with the incident light energy, that is, the light energy can be coupled to the vibration mode of the molecule. The coupling of energy is through the change of the electric dipole moment between unequal nuclear charges. Finished. Therefore, even for diatomic molecules of equal charge, that is, homonuclei, they cannot complete energy transfer and form transitions even if they have vibrational energy levels.

2 Non-harmonic resonance model

Although the quantum model of the linear harmonic oscillator can explain the characteristic absorption bands due to the fundamental vibrational mode of the molecules observed in the infrared region, experiments have found that the fundamental absorption frequency corresponding to the characteristic absorption band is doubled, tripled, etc. There is still a strong absorption band at the position, which indicates that using the molecule as a simple harmonic model is only a rough approximation, and it is difficult to explain the frequency doubling phenomenon that occurs in the near infrared region. This is completely contrary to the transition allowed by the fundamental frequency absorption band. Due to the influence of other molecules and their own spins in the space, the model does not fully meet the requirements of the linear harmonic oscillator. The elastic vibration between atoms does not follow Hooke's law. Frequency position, which is also the spectral mechanism on which the near-infrared spectroscopy analysis technology relies. When two atoms in the molecule approach or move away from each other and deviate from the equilibrium position, the potential energy of the molecular system will increase at a rapid rate due to the Coulomb force between the nucleus. At lower frequencies, the potential energy given by the nonlinear harmonic oscillator The curve is similar to the linear harmonic oscillator, and at higher frequencies, the potential energy curve of the nonlinear harmonic oscillator starts to become gentle when the potential energy increases to a certain degree. According to the law of conservation of energy, the decrease in potential energy is due to the atomicity in the molecular system. Increased internal energy caused by changes in vibration energy levels.

Near Infrared Spectroscopy Books

3 Physical properties of light and matter

In nature, light interacts with matter at all times and follows specific rules to transfer photon energy of specific frequencies to matter. When light radiation is incident on the surface of matter, there are usually three forms of energy transfer: reflection, absorption ,transmission. Among them, reflection can be divided into diffuse reflection and specular reflection. Diffuse reflection appears in two forms: body reflectance and surface diffuse reflection. Surface diffuse reflection and specular reflection follow the same law-the law of reflection, but surface diffuse reflection is also called specular reflection on an irregular reflection plane. Specular reflection and surface diffuse reflection are physical phenomena in which light is directly reflected when it passes through the surface of matter. Light does not interact with matter, so it does not carry any information related to the composition of the matter. Stray light has a large impact on the signal-to-noise ratio and accuracy of the instrument. In the design of the instrument and the preparation of the sample, it is required to consider how to eliminate the energy interference from the specular reflection to the greatest extent. Bulk diffuse reflection is a phenomenon in which light energy passes through the surface of a substance and interacts with its microstructure, and then exits and enters into other particles to interact with each other. The microstructure selectively couples and absorbs light vibrations with different frequencies according to different modes of chemical bond motion. The light energy that has not undergone coupling and absorption is folded out by the atomic nucleus after multiple reflections. The ratio between the light signal reflected by the body and the incident original light signal reflects the selective absorption characteristics of the material for light at different frequencies. That is, the absorption spectrum of the measured substance is formed, which reflects the rich microstructure information of the substance. The absorption spectrum data is the relative value corresponding to each frequency within the frequency range of the spectrum measurement. The intensity and position of these relative values can be used to derive the structure of the molecule through spectral theory.

The absorbance data is the ratio of the material to the near-infrared light radiant energy before and after the incident (dimensionally unit). It is obtained by the energy acquisition system (mainly the detector) of the near-infrared spectrum analysis instrument. The concentration of the components has a direct linear relationship, and the near-infrared light radiation can be used as an information carrier to measure the energy change in the near-infrared measurement band after passing through the material to measure the concentration of the material component.

It is difficult to fully meet the requirements of the above-mentioned conditions that meet the Lambert-Beer law in the near-infrared technology. This has led to the introduction of many interference factors in the measurement of the near-infrared spectrum, making the near-infrared absorbance data and chemical composition The direct linear correlation between the concentration data is reduced. By analyzing a large number of near-infrared spectra of substances, it is found that the main factors that seriously affect the linear relationship are the baseline translation and nonlinear shift caused by the physical characteristics of the substance to be measured (such as granularity, packing density, uniformity, etc.).

The absorption characteristics of near-infrared light radiation after interacting with matter are generally reflected in two forms: transmission and diffuse reflection. When the near-infrared light energy is detected by the detector after passing through the sample, the linear relationship between the attenuation of the energy and the concentration of the components in the material is a linear relationship, which fully considers the average caused by the scattering effect of light radiation between the particles of the material. Optical path increasing effect.

Near infrared spectrochemical characterization

1 Molecular vibration mode

Six vibration modes of methylene

To calculate the many possible vibration modes of polyatomic molecules,

Six vibration modes of methylene It is necessary to introduce the concept of degrees of freedom to determine the number of vibration modes of a molecular system. Defining a point in space requires three degrees of freedom, and n points requires 3n degrees of freedom. Among them, determining the planar and rotational motion of the entire molecule requires 3 degrees of freedom, so that 3n- 6 degrees of freedom. The movement of a chemical bond between two atoms in a molecule is a stretching vibration, which can be divided into symmetrical vibration and asymmetrical vibration, and the movement between atoms at a certain angle relative to the chemical bond is called bending vibration, which can be divided into shear Each of the vibration, motion, swing, symmetric twist and asymmetric twist motions in the near-infrared spectral region will produce doubled or combined frequency absorption, and the absorption intensity depends on the degree of non-harmonicity of the vibration. The chemical bond of the hydrogen atom with the smallest nucleus mass has the largest amplitude when vibrating, so all the vibration modes of the chemical bond deviate greatly from the vibration modes of the linear harmonic oscillator model. Many of the absorption bands observed in the near infrared region are hydrogen atoms The frequency doubling caused by the stretching vibration and the combined frequency absorption caused by the interaction between the stretching vibration and the bending vibration.

2 Spectral absorption band analysis

Since the advent of the first commercial near-infrared spectrometer in 1954, scientific researchers have actively promoted the use of this spectroscopic technology in various fields. During this period, a large number of experiments have been performed to analyze the composition of various substances in the near-infrared. The absorption band in the region, combined with the results of the mid-infrared material composition characteristic absorption spectrum analysis work, can compare the experimental observation results with the theoretical calculation results, and can more accurately analyze the near-infrared frequency doubling and combined frequency absorption spectra of many substances.

NIR spectral performance index

When evaluating a near-infrared spectroscopy instrument, you must understand the main performance indicators of the instrument, which will be briefly introduced below.

Wavelength range of the instrument

Fourier transform near infrared spectrometer

For any particular near-infrared spectroscopy instrument, it has its effective spectral range. The spectral range mainly depends on the optical path design of the instrument, the type of detector, and the light source. The wavelength range of near-infrared spectroscopy instruments is usually divided into two sections, a short-wave near-infrared spectral region of 700 to 1100 nm and a long-wave near-infrared spectral region of 1100 to 2500 nm.

Spectral resolution

The resolution of the spectrum mainly depends on the spectroscopic system of the spectroscopic instrument. For instruments using multi-channel detectors, it is also related to the pixels of the instrument. The narrower the spectral bandwidth of a spectroscopic system, the higher its resolution. For grating spectroscopic instruments, the resolution is also related to the design of the slit. Whether the resolution of the instrument can meet the requirements depends on the analysis object of the instrument, that is, whether the resolution can meet the requirements for extracting sample information. The structural characteristics of some compounds are relatively close. To obtain accurate analysis results, it is necessary to put forward higher requirements on the resolution of the instrument. For example, the analysis of xylene isomers generally requires the resolution of the instrument to be better than 1 nm.

3. Wavelength accuracy

Spectral instrument wavelength accuracy refers to the difference between the wavelength of a certain peak of a standard substance measured by the instrument and the calibrated wavelength of that peak. Wavelength accuracy is important to ensure model transfer between NIR spectroscopy instruments. In order to ensure the effective transmission of the calibration model between instruments, the accuracy of the wavelength is required to be better than 0.5nm in the short-wave near-infrared range and better than 1.5nm in the long-wave near-infrared range.

4. Wavelength reproducibility

Wavelength reproducibility refers to the multiple scans of a sample. The difference between the peak positions is usually expressed by the standard deviation of the wavelength or wave number obtained by measuring a certain peak position multiple times. -1). Wavelength reproducibility is an important indicator of the stability of the instrument. It has a great impact on the establishment of the calibration model and the transfer of the model, and it also affects the accuracy of the final analysis result. The reproducibility of general instrument wavelength should be better than 0.1nm.

5. Absorbance accuracy

Accuracy of absorbance refers to the difference between the measured absorbance value and the calibration value of the substance when the instrument performs transmission or diffuse reflection measurement on a standard substance. For those near-infrared methods that directly quantify the absorbance value, the accuracy of the absorbance directly affects the accuracy of the measurement results.

6. Repeatability of absorbance

Absorbance reproducibility refers to the difference between the absorbance of different measurements at each scan point when the same sample is scanned multiple times under the same background. It is usually expressed by the standard deviation of the absorbance obtained by measuring the position of a peak multiple times. The reproducibility of absorbance is an important index for near-infrared detection, which directly affects the effect of model establishment and measurement accuracy. Generally, the reproducibility of absorbance should be between 0.001 and 0.0004A.

7. Absorbance noise

The absorbance noise is also called the stability of the spectrum, which means that the sample is scanned multiple times within a certain wavelength range to obtain the mean square error of the spectrum. Absorbance noise is an important indicator of instrument stability. The signal-to-noise ratio can be calculated by comparing the sample signal intensity with the absorbance noise.

8. Absorbance range

The absorbance range is also called the dynamic range of the spectrometer, and it refers to the ratio of the highest absorbance and the lowest detectable absorbance determined by the instrument. The larger the absorbance range, the larger the linear range that can be used to detect the sample.

9. Baseline stability

Baseline stability refers to the flatness of the instrument relative to the baseline obtained from the reference scan. The flatness can be measured by the magnitude of the baseline drift. The stability of the baseline has a direct impact on our ability to obtain a stable spectrum.

10. Stray light

Stray light is defined as the sum of the amount of light reaching the sample and the detector in addition to the required analysis light. It is the main reason for the non-linearity of the instrument measurement. Especially for the design of grating instruments, the control of stray light is very important. Stray light affects the instrument's noise, baseline, and spectral stability. Stray light is generally required to be less than 0.1% of transmittance.

11. Scanning speed

Scanning speed refers to the time required to complete one scan in a certain wavelength range. The time required for an instrument to complete a scan varies greatly between design approaches. For example, the charge-coupled device multi-channel near-infrared spectroscopy instrument needs only 20 ms to complete a scan, and the speed is fast; generally, the scanning speed of the Fourier transform instrument is about 1 / s; the scanning speed of the traditional raster scanning instrument is relatively slow. The faster scanning speed is only about 2 times / s.

12. Data sampling interval

Near infrared spectrum

The sampling interval is the wavelength difference between two spectral signals that are recorded continuously. Obviously, the smaller the interval, the richer the sample information, but the larger the spectrum storage space. If the interval is too large, the sample information may be lost. A more appropriate data sampling interval design should be smaller than the resolution of the instrument.

13. Testing method

The sampling method refers to the sample spectrum collection form that the instrument can provide. Some instruments can provide multiple forms of spectral acquisition such as transmission, diffuse reflection, and fiber measurement.

14. Software Features

Software is an important part of modern near-infrared spectroscopy instruments. The software generally consists of two parts: spectrum acquisition software and spectral chemometrics processing software. The former is not very different from the instruments of different manufacturers, while the latter is very different in software function design and content. Spectrum chemometrics processing software generally consists of three parts: preprocessing of spectra, establishment of qualitative or quantitative correction models, and prediction of unknown samples. The evaluation of software functions depends on whether the content of the software can meet the needs of actual work.

Near infrared spectral reflection

When near-infrared light is irradiated, resonance occurs between the light and the group with the same frequency, and the energy of the light is transferred to the molecule through the change of the molecular dipole moment. The frequency of near-infrared light is not the same as the vibration frequency of the sample, so light at this frequency will not be absorbed.

Therefore, when a near-infrared light with a continuously changing frequency is used to irradiate a sample, the near-infrared light passing through the sample is weakened in some wavelength ranges due to the selective absorption of near-infrared light of different frequencies by the sample, and in other wavelength ranges. Inside, the transmitted infrared light carries information about the composition and structure of organic matter.

The optical density of transmitted or reflected light is analyzed by a detector to determine the content of this component. The detection performed by measuring the information carried by transmitted light is called near-infrared transmission technology. The measurement performed by measuring the information carried by reflected light is called near-infrared reflection technology .

Near infrared spectroscopy

The advantages of NIR spectroscopy are:

1) Fast analysis speed. Once the near-infrared spectrum analyzer is calibrated, it can complete the simultaneous measurement of multiple components of the sample to be measured in less than one minute. If a diode array detector combined with an acousto-optic modulation spectrometer is used, The measurement results can be given in a few seconds, which can completely realize the online quantitative analysis of the process.

2) No chemical pollution to the sample. Depending on the particle size of the sample to be measured, it may require a simple physical preparation process (such as grinding, mixing, drying, etc.), and the measurement process can be completed without any chemical intervention. It is called a green analysis technology.

NIR spectroscopy model

3) The operation of the instrument is simple, and the quality level of the operator is relatively low. The software design can realize extremely simple operation requirements, and the human error introduced during the entire measurement process is small.

4) High measurement accuracy. Although the accuracy of this technology is slightly inferior to that of traditional physical and chemical analysis methods, the measurement accuracy given is sufficient to meet the actual requirements of quality monitoring in the production process, so it is very practical.

5) Low analysis cost. Since no chemical reagents are required during the entire measurement process, measurement after the calibration of the instrument is a very simple task, so there is almost no loss.

Near-infrared spectroscopy instrument

Near-infrared spectroscopy instruments can be classified into four types: fixed wavelength filters, grating dispersion, fast Fourier transform, and acousto-optic tunable filters.

Grating dispersion

The filter type is mainly used as a special analysis instrument, such as a grain moisture analyzer. Due to the limited number of filters, it is difficult to analyze samples of complex systems.

Raster scan type has higher signal-to-noise ratio and resolution. Because the moving parts (such as the grating axis) in the instrument may have wear problems during continuous high-intensity operation, which affects the reliability of the spectral acquisition, it is not suitable for online analysis.

Fourier transform near-infrared spectrometers have higher resolution and scanning speed. The weak point of this type of instrument is also that there are moving parts in the interferometer and it requires a stricter working environment.

The acousto-optic tunable filter is a birefringent crystal. The scanning wavelength is adjusted by changing the RF frequency. The entire instrument system has no moving parts and has a fast scanning speed. However, the resolution of such instruments is relatively low and the price is high.

With the maturity of array detection device production technology, NIR instruments using fixed optical paths, grating beam splitting, and array detectors are characterized by their stable performance, fast scanning speed, high resolution, high signal-to-noise ratio, and good performance-price ratio. People are paying more and more attention. Among the array detectors that match the fixed optical path, there are two types of charge-coupled device (CCD) and diode array (PDA). CCD is mostly used in the near-infrared short-wavelength spectrometer, and PDA detector is used in the long-wavelength near-wavelength. Infrared area.

Qualitative analysis of near-infrared spectroscopy

In the near-infrared spectrum, the peak position, number of peaks, and peak strength of the near-infrared spectrum are different according to the different chemical components contained in different types of substances, and the frequency of the hydrogen-containing group is doubled and the combined frequency is different. The greater the difference in chemical composition, the stronger the characteristic differences in the map. The use of simple peak position identification can identify different varieties of Chinese medicine. The peak position identification method is mainly used to analyze different substances with large differences in composition. This method is intuitive and simple, but it has no power to identify samples with similar properties. Therefore, other methods, such as chemometric methods, must be used for identification.

Pattern recognition was introduced into the field of chemistry in the late 1960s. It is based on a very intuitive basic assumption, that is, "things are clustered together." It is believed that samples with similar properties are located in the pattern space similarly, and they form "clusters" in space. The pattern recognition method has obvious advantages. It does not require the prior knowledge required by mathematical models and is rarely good at dealing with complex things and multivariate data. In actual work, we often encounter the problem that only the type or grade of the sample needs to be known, and the number of components contained in the sample and its content are not required. At this time, the pattern recognition method needs to be applied. Pattern recognition is mainly used for qualitative analysis of spectra.

NIR

1.CO2

3.PH

1010

In the calibration process, the number of standard samples directly affects the accuracy of the analysis results. Too few numbers are not enough to reflect the normal distribution of the sample groups being tested, too much data and too much work. In addition, when selecting a sample for chemical analysis, not only the sample composition content and gradient, but also the physical, chemical, growth area, variety, growth conditions, and botanical characteristics of the sample should be considered to improve the calibration effect and make the calibration curve have A wide range of applications. Multiple samples can be calibrated according to specific screening principles in order to improve the calibration effect and test accuracy. Generally speaking, a single type of pure sample is relatively easy to calibrate due to its stable nature and relatively small amount of chemical information, such as corn, wheat, soybean and other pure samples. The mixed sample sample information is complex and will cause a variety of Overlapping group spectral peaks, difficult information analysis, and difficult calibration, such as various full-value feeds, compound feeds, and concentrated feeds in livestock production.

Problems with near-infrared spectroscopy

Near-infrared spectroscopy analysis technology is relatively mature at this stage. Various types and types of near-infrared analyzers are available in the market, but the price of analytical instruments is relatively high, especially the Fourier transform type (such as the American Nicolet company), Raster scanning type (Foss company in Denmark) and other high-precision analyzers are unbearable for ordinary commercial users and cannot be promoted on a large scale. So how to reduce the cost of instrument development and maintain sufficient analytical accuracy is one of the main issues that researchers are concerned about.

The filter-type near-infrared spectrum analyzer has relatively low cost, and the calibration model has good long-term stability, and the operation and maintenance of the instrument are convenient. However, due to the limited filters, it is difficult to cope with complex samples. The optimization of the calibration model and the transfer of the calibration model between instruments of the same model have always been a hot topic of discussion among NIR workers.

In addition to the quantitative analysis accuracy of the near-infrared spectral analysis instrument, in addition to its own signal-to-noise ratio and stability, the accuracy of the reference physicochemical analysis method also directly affects the accuracy of the measurement results given by the calibration model, so how to further improve the physical and chemical reference analysis The accuracy of the method to improve the correlation between the near-infrared spectral absorbance data measured by the instrument and the physicochemical analysis values needs to wait for the development of the science and chemistry department.

Although the development of chemometrics successfully explains the correlation between the wavelength model information of the calibration model and the chemical information of the substance, and the pre-processing of the calibration data improves the stability and accuracy of the model, it is not the same as directly using the spectral data Log ( The result calculated by 1 / R) has a small improvement in comparison accuracy and greatly increases the complexity of data processing, so only through further research and understanding of the mechanism of matter and light interaction can fundamentally solve the problem between substance component spectra. And the interference of external factors on the calibration model.