Comparison of different technologies
Table of ContentsBasic knowledge about VIS NIR spectroscopy
- What is spectroscopy
- The 4 steps of spectroscopic measurements
- What is chemometrics
- How to create a chemometric model for Material Sensing
SILICON, INGAAS AND SENORICS
NIR spectroscopy with and without conventional dispersive elements
The main difference between the commercially available spectroscopic technology and our innovative technology is that, up to now, a dispersive element was necessary to separate the spectral components of the reflected or transmitted light. With our technology it is possible to do NIR spectroscopy without a conventional dispersive element and pave the way for new applications.
Dispersive elements as longstanding tools for spectroscopic technology
Dispersive elements used in spectroscopy have been known for decades and one classic example is a prism; typically, a triangular shaped object made from glass. If white light enters the surface of the prism, it will be refracted and dispersed into its spectral components. All in all, leading to a rainbow-like splitting of the incoming white light as illustrated in the sketch. If you would like to learn more about this phenomenon, it is described by Snell’s law.
Necessary components to perform spectroscopic measurements
To obtain the spectroscopic information of an object, the following 4 components are required:
- This can be a broad band light source which radiates light and interacts with the object of interest. It could also be a light source that only covers a certain spectral range, for example near-infrared light when using NIR spectroscopy.
Object of interest
- The object interacts with the light, leading to absorption, reflection and transmission.
- In spectroscopy we need to analyze spectrally resolved information of reflected or transmitted light. Therefore, the light needs to be split into its spectral components and directed to a detection unit where it is measured.
- A detector receiving the reflected or transmitted light and converting it into an electronic signal which can be measured precisely. Analyzing the spectral components enables us to quantify properties of the object, such as the ingredients and composition of the material.
Spectroscopy without conventional dispersive elements
In contrast to the established methods, we invented a sensor which consists of a spatial arrangement of detectors that are sensitive to specific wavelengths. The main advantage is that the wavelength selection and measurement happen within the same device. Therefore, our solutions are much smaller and more robust than conventional spectrometers. Our sensors do not need any moving optic components, which is one of the main reasons why our technology is the ideal solution for small and mobile spectroscopy applications.
Paving the way for a new level of spectroscopy
With our sensor solutions and the omission of a conventional dispersive element, the foundation for a new era of Material Sensing applications is laid, taking spectroscopy out of the laboratory and bringing it into people’s pockets.
Full spectrum of lab spectrometers versus 16 pixels of Senorics hardware
Have you ever wondered how measurements carried out by a lab spectrometer differ from those carried out by our SenoCorder Solid? Or how a small chip can perform spectral measurements the same way that a conventional lab spectrometer can? To answer these questions, we will use this article to compare a lab spectrometer with the Senorics sensor.
How a lab spectrometer records a spectrum
In labs worldwide, most spectrometers used in the VIS and NIR range are either based on silicon or InGaAs. Silicon based spectrometers cover a wavelength range from 200nm to 1100nm whereas InGaAs based spectrometers cover a range from 800nm to 1700nm (in some cases up to 2500nm).
Both types of spectrometer provide high resolution spectral analysis up to 1nm. To come to such detailed insights, the dispersive elements in the lab spectrometers must be adjusted precisely and continuously. This results in the devices being quite sensitive to vibrations and shocks, as well as being rather large and bulky. On the bright side however, measurements are very precise, the resolution can be set exactly as needed (ranging from 1nm to 10nm) and you get a spectrum with the desired level of detail.
If you were to take pasta to a lab asking for spectral measurements, the diagram you receive would look something like this:
This graph shows the spectrum of pasta in the wavelength range from 1100nm to 1800nm with a resolution of 2nm. This means that there is a measuring point every 2nm, resulting in the 350 measuring points shown in the diagram.
To create a better understanding of what this means, we have come up with a simple analogy that will be referred to again at other points in this article. If you were to draw a birch tree in detail, you would draw roots, trunk, bark, and leaves as well as the leaf veins, and detailed levels of texture. In this case, the above graph, showing 350 points of measurement, is the detailed drawing.
How our sensor records a spectrum
The technology of our chip is based on thin film organic semiconductors arranged in a stack configuration and requires no dispersive elements. It is a small, light and robust sensor solution that can be implemented into everyday devices or used in the form of a handheld device.
The chip inside has 16 wavelength channels called pixels. Each pixel covers a certain wavelength in the VIS or NIR range (450 nm to 1800 nm). The coverage of wavelengths can be adjusted depending on the application.
Using our SenoCorder Solid, you can measure the pasta on your own by placing the device on the pasta and immediately beginning to measurement.
Our Software SenoSoft outputs this diagram:
It shows the measurement of the same pasta as in the previous diagram, but this time measured with our SenoCorder Solid. The wavelength range is 1100nm to 1800nm and the resolution is 45nm, resulting in 16 measuring points.
Coming back to the example with the birch tree. If this graph is our drawing of the birch tree, it would be a sketch that shows little detail except for the most important characteristics such as the overall shape of the tree, with a simple representation of the trunk, branches, and leaves.
Now that we have covered the technical background, it is time determine whether 16 pixels are enough to get a reliable result.
Are 16 pixels enough? Lab spectrometer vs. Senorics technology
Comparing the two diagrams, it becomes clear that the lab spectrometer provides significantly more measuring points than the SenoCorder Solid (350 vs. 16). Another difference is that the measured values of the SenoCorder Solid differ slightly from those of the lab spectrometer.
However, it also becomes obvious that the 16 measuring points represent the most important characteristics. That is to say that the SenoCorder Solid can qualitatively reproduce the spectrum.
Let’s come back to the birch tree again, thinking of this information:
- It is a tree.
- It has a white bark.
- It has many small and serrated leaves.
You would immediately know that this must be a birch tree. You don’t get information about the height of the tree, whether it’s healthy or not, or what kind of insects are living in the bark – questions that might be important for a biologist or an ecological scientist. But you get the information that it is a birch tree, which is sufficient for everyday life. It is the same with this spectrum. A few measuring points covering the most important features are enough for most everyday applications.
In daily life you might ask yourself questions like: “Is this pasta gluten free or not?”, “Is this textile polyester or silk?”, “Is this drug genuine or counterfeit?”. Answering these questions often requires less than 16 measuring points, depending on the spectrum of the material.
In daily life you ask questions like “Is this pasta gluten free or not?” “Is this textile polyester or silk?” “Is this a genuine drug or fake?” To answer these questions you often need even less than 16 measuring points – depending on the spectrum of the material.
So, coming back to our question of whether 16 pixels are enough, the answer is yes. In most everyday applications they are perfectly sufficient.