Figure 1. A spectrometer for a mobile smart phone.

Mobile NIR spectroscopy has gathered a lot of interest in recent years. On site and real time measurements of the chemical composition of solid or fluid samples could be applied to identification, authentication or estimation of quality parameters and similar relevant measurement tasks. This measurement technique is particularly useful for, but not limited to samples containing organic compounds.

In order to achieve these aims, NIR absorption spectroscopy as a method for molecular spectroscopy has been established in the early 20th century. The interaction of electromagnetic radiation with matter very often allows for the determination of the chemical composition within certain limits. Absorption of radiation in the sample at specific wavelength or spectral bands is the main mechanism of interaction in NIR spectroscopy. The amount of light absorbed is governed by the famous Lambert-Beer law [1]:

I / I0 = e – N f l

If the parameters of the measurement are known – optical path length (l) and transition parameter (f) – the density (N) of the species correlated to the transition can be calculated directly from knowing incident illumination (I0) and the measurement of the intensity (I) after passing the matter. Knowledge on the specific densities serves the calculation of material properties of the sample.

This method can be used for a broad variety of applications from sorting different materials in recycling processes to the estimation of quality parameters like the ripeness or freshness of fruit. The near-infrared region starts at the boarder to visible light at 780 nm and reaches up to the mid-infrared, which starts at 3000 nm. In this region, the absorption of photons excites overtone and combination band of different materials. These bands show weak absorption strength and broad transitions. Thus, an absorption band has a typical full width at half maximum (FWHM) of about 10 nm.

The weak absorption allows for high penetration depths in the millimeter to centimeter range and thus insight into the object. Due to the broad and overlapping bands, the evaluation is complex, especially on large organic molecules with a wide variety of functional groups (C-C, C-H, C-X single, double and triple bonds). Many applications require complex mathematics called “chemometry” and reference data acquired from previous measurements on model substances. In turn, quantitative analysis can be enabled.

The most common principle is the illumination of the sample with a suitable light source emitting a known “white” spectrum. After interaction with the material of the sample, either in transmission or in diffuse reflection, the light is captured and analyzed by means of spectroscopic equipment. In earlier times, prismbased spectrographs with photographic emulsion to record a whole spectrum at once were used. With the availability of high-quality gratings and very sensitive single pixel photon detectors, scanning spectrometers of the famous Czerny-Turner or Ebert-Fastie type appeared. Modern detectors with a large number of pixels arranged in a line or array configuration allow for the parallel acquisition of spectra with high resolution. Furthermore, Fourier-Transform (FT) spectrometers, formerly prominent in the mid-infrared, entered the NIR range as well. In laboratory use, reliable tools serve many different applications of chemical analysis.

More applications arise in our day-today life, which is mostly based on organic matter. The human body, our food and its origin, medications and cosmetics but also our clothing, furniture and other items are mostly made of carbon-based compounds. The quality in the area of nutrition, health, pharmaceutical and medical items but also the analysis of consumer goods and technical aspects (e.g. plastics, fuel and lubricants) as well as our environment could be improved by a broad use of spectral analysis.

For a long time, samples have been taken and transferred to the lab where large and ultraprecise equipment is used to perform the measurements under precisely controlled conditions. Chemometric models based on reference data are used for the evaluation. For selected applications, this offers an appropriate solution. Others would benefit from real time and on-site capabilities, for example on goods with quickly changing conditions as in food analysis for the determination of freshness.

The transfer from well-established laboratory analysis to mobile use is based on the experience and implementation of analytical models through scientific work since the 1950s. Miniaturized NIR spectrometers have been presented up to now, which allow for the integration into handheld devices or even mobile phones (Figure 1). First approaches were based on detector arrays and fixed grating [1].

Recently, MEMS scanning grating [2] and MEMS based FT spectrometer [3] have started to enter applications. Furthermore, filter-based solutions have been realized using InGaAs or even Si-based detectors [4], considering that Silicon detectors can be used up to 1050 nm only. Besides spectral range, all systems face similar requirements regarding spectral resolution, as well as wavelength stability and reproducibility. Today also the overall size, production cost and power consumption become more and more important.

Unfortunately, a great breakthrough in broad use has been inhibited up to now by the cost of appropriate NIR detectors in InGaAs technology, especially if array detectors are required.

Figure 2. A simple single-axis MEMS scanning mirror.

A new approach to realize compact NIR spectrometers is based on the use of a simple single-axis MEMS scanning mirror (Figure 2). This comes along with some advantages: The production technology can be simplified, as no complex three-dimensional grating structure must be etched into the MEMS plate. Also, the stress in the silicon structures induced through the etch process is reduced. Due to the optical path, the deflection required is only half of a scanning grating for the same spectral range. The grating assembled to the system can be changed and adjusted simply.

The basic idea of a scanning mirror micro spectrometer is well-known. Instead of deflecting or rotating the grating itself, a deflectable mirror is applied to illuminate a fixed grating. The optical path is close to the original design of Czerny and Turner. The light enters the spectrometer through an entrance slit. A first mirror acts as a collimator for the incident light and directs the light towards the scanning mirror plate. Reflected by the scanner, the radiation subsequently impinges on the diffraction grating.

By rotating the mirror plate the angle of incidence on the grating is varied. A part of the spectrally dispersed light is back reflected from the grating onto the scanner and directed onto a second mirror. This second mirror focuses the laterally dispersed light onto an exit slit. The exit slit cuts out a small portion of the spectrum. A single pixel detector is placed behind the slit. By rotating the scanning mirror plate, the spectrum is scanned across the exit slit. A synchronization of the mirror movement and the detector signal readout enables the recording of a full spectrum.

The main characteristic of the original Czerny-Turner approach was the use of two spherical mirrors for collimation and focusing on a symmetric configuration. Since such a scanning spectrometer (monochromator) has a very small field of view given by the area of the exit slit, the dominant optical aberrations are spherical, coma and astigmatism. The symmetric arrangement of the two mirrors is very favorable for the correction of optical aberrations and hence the significance of the Czerny-Turner approach.

The required symmetry can be achieved in several ways. They differ in the orientation of the two mirrors relative to the diffraction grating and are termed W-configuration, folded double-V or crossed configuration. The requirements of a particular design are determined by the intended application and can be very different. Among those requirements are the spectral range, the spectral resolution, signal-to-noise ratio (SNR) and the overall size of the system.

Several different spectrometers based on the Czerny-Turner approach have been developed at IPMS. The folded double-V and the crossed Czerny-Turner configuration allow for the most compact optical ray paths and hence for the smallest spectrometer outlines.

Therefore, they are well-suited for mobile applications. The development process included the optical design and optimization along with the mechanical design as well as the development of the electronics to drive the MEMS device and for the detector readout. Special consideration was given to assembly aspects since the resulting spectrometer size is on the order of (10 × 10 × 6) mm³. The small size along with tight tolerances required an adequate assembly strategy. Both, stacking and the so-called “Place and Bend Assembly” [5, 6], have been tested successfully.

In addition to the above-mentioned considerations the overall cost and availability of optical, mechanical and electronic parts are of importance. To ensure a realistic chance of high-volume production, state-of-the-art pick and place technology to assemble the MEMS and the optical components along with additive manufacturing for some mechanical parts was used.

Figure 3.The optical path inside of a miniaturized spectrometer.

The optical path inside an example of such a miniaturized spectrometer developed at IPMS is shown in Figure 3.

Here, the folded double-V configurations has been implemented. The double-V design offers good performance and sufficient optical throughput. The two mirrors are grouped together on one substrate. In this case the mirror surfaces are off-axis aspherics for good aberration correction which is a slight deviation from the use of spherical surfaces in the original Czerny-Turner setup. The spectral range was 950 nm to 1900 nm whereas a spectral resolution of 10 nm FWHM was achieved.

Figure 4. This “white powder recognition” demonstration setup shows NIR spectroscopy on salt, sugar and flour.

The first application close trial has been conducted. Based on the idea of a food application, the “white powder recognition” demonstration setup was successfully presented showing NIR spectroscopy on salt, sugar and flour (Figure 4).


Part of this work was funded by the Fraunhofer Society under contract # 600 890, MEF “Origami.” The authors gratefully acknowledge the work of Hans-Georg Dallmann in the field of readout and drive electronics and together with Johannes Ziehbarth on the realization of the demo unit.


  2. T. Pügner, J. Knobbe, H. Grüger, “Near-Infrared Grating Spectrometer for Mobile Phone Applications”, Applied Spectroscopy 2016, Vol. 70 (5) 734–745 (2016)
  5. Patent US 12/425,582
  6. Patent DE102016221303.2

This article was written by Dr. Heinrich Gruger, Research Engineer; Jens Knob-be, Research Engineer; and Dr. Tino Pugner, Research Engineer, Fraun-hofer-Institut fur Photonische Mikrosys-teme (Germany). For more information, visit here .