Basic knowledge about VIS NIR spectroscopy

Basic knowledge about VIS NIR spectroscopy

What is spectroscopy

Spectroscopy is commonly described as the study of the interaction between matter and electromagnetic radiation. More generally speaking, what you retrieve when using spectroscopy is the response of a system or object to some sort of excitation.

The operation of spectroscopy

Let us explain the operation of spectroscopy with an example. Think of a guitar as the system or object of interest (1). When you play it, the guitar strings become excited (2) and deliver a response signal – the music (3).

This signal can be detected by a microphone (4) and furthermore analyzed by software or other analysis units. This is done to represent the recorded notes and their intensity in form of a diagram – the response or spectrum of your guitar (5).

The combination of a detection device and analysis unit is called a spectrometer. In this respect, your ear is a spectrometer as it detects acoustical waves and, in collaboration with your brain, converts them into meaningful information – the sound you hear.

Another example could be the human eye. It converts the interaction of visible electromagnetic radiation with an object, like the interaction of sunlight with a tree, into a meaningful picture that can be visualized by your brain.

The meaning of wavelength in spectroscopy

As you might have guessed, our eye is limited to the visible part of the electromagnetic spectrum. What your eye does is split up the incoming light, which consists of many colors, into the three fundamental colors: red, green, and blue. This allows your brain to identify the colors of the object you are looking at.

Each color is mathematically represented by a wavelength number. The wavelength number is defined as the distance between two consecutive peaks of an electromagnetic wave traveling in space.

For example, blue color has a wavelength of around 450 nanometers, green around 600 nm and red roughly 700 nm. Your eye can detect wavelengths in the range of ~400 to ~780 nm, which is commonly called the visible range (VIS). If your eye were to receive all of the radiation in that range at once, you would only see white light.

Identifying the optical fingerprint

In the VIS range you get basic information about the color of your object of interest, but there is much more possible information than just that part of the spectrum.

When you extend the observation range into the infrared (IR) region above 780 nm, you will be able to detect “fingerprints” of materials, giving you information about their composition. By using these unique optical fingerprints, it is possible to identify, qualify and quantify different kinds of liquids and solids.

This part of the spectrum cannot be seen by your eye but can be made visible using a technical spectrometer.

The 4 steps of spectroscopic measurements

You may have heard that there are different kinds of spectroscopy, for example NIR spectroscopy and UV spectroscopy. However, all of them essentially consist of the same four steps. We will explain these steps to you in a simplified way to give you a better understanding of spectroscopy.

Step 1 - The illumination of a sample

In spectroscopy, a sample is illuminated with a light source. A common lightbulb, as shown in the illustration, emits mostly visible light. But there is also light that cannot be seen by the naked eye, for example near infrared light, which our technology is based on the analysis of.

Step 2 - The reflection and absorption of incoming light

The light is then either absorbed, transmitted, or reflected by the object. As you can see in the illustration, not all the light rays are being reflected by the apple, a part of the light is always absorbed. When using spectroscopy, we differentiate between transmission and reflection spectroscopy.

In transmission measurements, the light transmitted through the sample is detected. For liquids, a glass cuvette can be used. In reflection measurements, the light hits the sample and is reflected, as is done with our SenoCorder Solid.

Step 3 - The detection of transmitted and reflected light

The reflected and transmitted light contains information about the object’s material composition. This is because the incoming light intensity is changed through the interaction with the object. The gained information is also referred to as the “optical fingerprint”, unique for each material.

To record the optical fingerprint, it needs to be detected and converted into electrical signals. This process is shown in the illustration with the sensor receiving the information and converting it into a spectrum, available in digital form for later analysis.

Step 4 - The identification and quantification of materials

The recorded fingerprints are usually interpreted by relevant software algorithms and used to either identify or quantify the material, depending on the use case. A good example is the identification of textiles or fraud detection in various industries e.g., differentiating between leather and pleather. Quantitative results can be for example the sugar content in food or moisture levels in cardboard.

Learn more

If you would like to get more in-depth information, head to our knowledge base. There you can find articles such as the comparison of Senorics and other technologies, as well as more information about possible applications and industries.

What is chemometrics

From a purely technical point of view, our sensor hardware delivers a spectrum of each sample you measure. So, you basically get a data stream of 16 values, representing intensities at 16 wavelengths. This is – of course – not what you finally need, because you would like to identify the material or material composition and not interpret a spectrum.

To come to this final goal, the sensor needs to be trained and “learn” which spectrum belongs to which material. So, it can afterwards identify new materials based on this training. In technical terms, this process is called calibration and is nothing else than a piece of software, relating reference values to measured spectra. The underlying models and algorithms belong to the field of chemometrics.

How to create a chemometric model for Material Sensing​

Chemometrics is the science that deals with the extraction of information from chemical data. At Senorics we use chemometrics to evaluate spectroscopic data and build so called predictive chemometric models.

These models are used to align measured spectra with reference values and enable the output of interpretable results. The result serves as an answer to a certain issue. For example, a chemometric model could tell whether a spectrum is representing silk, viscose or neither of the two when used for authenticity evaluation.

This process includes two phases: calibration and application. During the calibration process, the models are trained by connecting spectra with corresponding reference values. During the application process, precise answers to specific issues, covered during the models training, can be identified.

In this article, we would like to give you a better understanding of the creation of chemometric models based on the graphic below.


Selecting appropriate reference samples

At first, representative samples according to your problem of interest need to be selected (1). For example, if you want to identify whether fabrics from the wholesaler are made of silk or viscose, a sufficient number of silk and viscose samples is required to serve as a reference. A sufficient number in this case would be 80 samples for each material. However, the number heavily depends on the scenario and can be higher or lower depending on its complexity.

Undertaking spectral measurements

Next, spectroscopic measurements must be taken of each sample (2b). This can be done with our SenoCorder Solid, for example. Through that process spectral data is obtained. If you would like to learn more about the basics of spectroscopy, you can find the corresponding article here.

Obtaining reference values

In addition to undertaking spectral measurements, the reference values for the selected samples need to be provided (2a). This is done to make sure that they really are what they appear to be e.g., that the silk sample really is made from silk and not viscose or a mixture of other materials. Those reference values can be obtained directly from reliable suppliers when ordering the samples or from an external lab through quality assessments of the provided samples.

Combining the data

The reference values and the spectral data are then combined to train a chemometric model (3).


The chemometric model is then added into SenoCloud and used to identify materials and material compositions. Thus, an unknown sample (4) can be measured (5) and information about its properties received (6). As per our example, a scarf can be measured and SenoSoft will immediately display whether it is made from silk or viscose. In technical terms, the model will look for the best fit of the new spectra to the known spectra from the sensor training and picks the associated reference value to identify the material class / composition of the new sample.

In this process, the chemometric model aligns data of new spectral measurements with known reference values to output an interpretable result. It is customized depending on its application, meaning there is no general chemometric model but rather a different model for each use case. Unknown samples can only be interpreted correctly if they are the same type of sample as the ones used as reference during the calibration. For example, if you use silk and viscose reference samples, the model can only identify other samples made of silk or viscose, it will not be able to identify e.g., polyester samples.