One main obstacle in using infrared spectroscopic chemical sensors in out-of-the-lab applications is the lack of suitable spectrometers. Presently, most spectrometers for infrared wavelengths are comparatively large, stationary and expensive. Additionally, the use of massive optical components, like gratings or mirrors, severely limits the time resolution of such systems. With the advent of small, compact micro-electro-mechanical systems (MEMS), it became possible to build pocketsized spectrometers for various spectral ranges, including the near-IR or mid-IR. Due to extremely low moments of inertia, these systems are both highly rugged and can be operated at scanning speeds well beyond 100Hz, resulting in possible time resolutions of 10ms or less. This allows either the measurement of spectral changes at ms time resolution or the co-adding of several hundreds of scans to one spectrum within the time conventionally required to acquire a single scan, thus achieving signal-to-noise ratios equivalent to full-scale laboratory instruments. With one Czerny-Turner MEMS scanning grating spectrometer for the near-IR already approaching product status, research now focuses on other, in particular longer wavelengths and identification of (mass) applications. Additional research focuses on building compact, ultra-fast FT-IR MEMS spectrometers.
Chemical sensors can be divided into two fundamentally different groups: i) indirect sensors, i.e. sensors that sense the – preferably characteristic and analyte-selective – change of some property of a chemo- or bio-selective layer in response to the presence of an analyte, and ii) direct sensors, i.e. sensors that directly detect and evaluate an inherent property of an analyte or analytes. While the first group is still by far the prevalent type of sensors, direct sensors (would) have some significant advantages. By directly evaluating inherent properties, such sensors typically exhibit good substance selectivity and are often capable of multi-analyte sensing. Also, as they are not reliant on a chemical or bio-chemical layer, they are – at least in theory – simpler in design and more robust in operation.
The most important class of direct sensors are optical spectroscopic sensors, from the UV to the mid-infrared. In particular IR sensors have serious potential, as they have excellent inherent analyte selectivity and can be used not only for quantitative multi-analyte detection, but also for qualitative chemical sensing, i.e. the identification of the substances present in a sample. Over the last years, the potential merit of fully spectroscopic IR sensors has clearly been proven. The fact that the practical realisation lags behind is to a large extent due to a lack of suitable, affordable and truly miniaturised spectrometers translating spectroscopic information into useful sensor data. The motivation behind this work was hence the development of infrared broadband spectrometer devices that could be used to replace the previously used laboratory devices in practical chemical sensor applications, ranging from industrial process control and environmental real-time monitoring to autonomous patient surveillance and security monitoring.
For wavelengths from the UV to the shortwave nearinfrared up to ~ 1.0μm, these demands can be satisfied through stationary diffraction gratings in combination with linear detector arrays. In the UV/VIS (200nm – 800nm) a range of compact, sensitive spectrometers with good spectral resolution are available from various suppliers. Such instruments are industrially accepted, widely used and have replaced larger conventional instruments in many applications. Still, for longer wavelengths this type of compact spectrometers suffers from a lack of suitable and affordable detector arrays.
State of the art for wavelengths above 1μm are i) grating spectrometers and ii) Fourier-Transform (FT) spectrometers, both using a single detector element. Even though some efforts have been reported to miniaturise such instruments, the necessity of having massive, moving optical components limits both the acquisition speed and the achievable degree of miniaturisation. This is where micro-electro-mechanical (MEMS) components come in. MEMS components are well suitable for modulating optical radiation, since deflecting or otherwise modulating light does not require much force. Furthermore, the achievable movement of micro-components often is sufficient to achieve the intended (optical) effects.
MEMS components are usually produced using well established semi-conductor production processes allowing for a convenient combination of e.g. semiconductor materials, dielectric layers and metals.
T Sandner, H Schenk
Fraunhofer Institute for Photonic