GaSb-based material technology is an emerging technology platform for beyond-telecom optoelectronic devices that open and enable new application opportunities. Or, as the title of the article states – paves the way towards optical sensing revolution. The fundamental cornerstones of the technology attractiveness are: 1) accessible spectral range 1.5 – 4 µm, which includes strong and specific combination and fundamental absorption bands of molecular ro-vibrational transitions, allowing for spectroscopic sensing; 2) efficient room-temperature operation of light sources and detectors (see Table 1 for comparison); 3) high-volume scalable manufacturing technology; 4) possibility to integrate with SOI-based photonic integrated circuits. Spectral range, spanning beyond telecom wavelengths, has a huge potential for various optical sensing applications by means of tunable laser absorption spectroscopy due to existence of strong ro-vibrational molecular absorption bands. (figure 1) . Particularly interesting is the range from 1.5 to 2.5 µm, where strong first-overtone and combination absorption bands of lactates, ethanol, glucose, urea and other important bio-molecules appear and light is still able to penetrate through skin and reach the interstitial fluid and first dense capillary network . This opens an opportunity for non-invasive blood-metabolite monitoring by optical means for a plurality of purposes, especially in the health and wellness market segments. For example, lactate concentration rise in human blood is an early predictor of sepsis , which is responsible for 1 out of 5 hospital deaths globally , moreover lactate level dynamics in blood during sports activities provide information of athlete’s fatigue , overtraining and physical form, thus is of great interest for professional and amateur athletes. Glucose is extremely critical for type-I and type-II diabetes management, however also important for general diet monitoring and customization. Finally, ethanol (blood alcohol) is also of great interest either for drinking-and-working (driving) prevention. In addition to blood metabolites, measuring concentration in real time in fluids carries is of particular interest in industrial and agricultural sectors. For example, real-time milk analysis shows great application potential . Here, the spectral range allows to measure milk fat content, protein content, urea content, etc., which, in turn can be used as a tool for monitoring and diagnosing herd immunity in real-time and helping increase or, at least sustain, milk production efficiency. Moreover, one can very easily apply spectroscopic sensing techniques in plastic sorting industry. The spectral range, accessible by antimonide-based devices, allows non-destructive, real-time sensing of different plastic types and can be easily implemented in industrial sorting lines.
In order to access the abovementioned consumer and industrial market segments, the technology has to be mature, scalable, energy-efficient and devices have to exhibit the desired performance. Table I summarizes the key device performance parameters of different technological approaches to access the desired spectral band.
Whereas Table II compares the maturity of manufacturing technology, needed to reach the desired spectral band.
Summarizing data from the two tables, it is clear why Sb-based material technology is the material of choice and that type-I (or standard quantum well) technology is the path to go for volume energy consumption-sensitive applications. The latter is mainly due to light generation due direct interband optical transition in the quantum wells with type-I band alignment to generate a photon. This type of technology is the most common one and has the advantage of mature design, high-gain due to largest possible overlap between the electron and hole wave-functions and allows direct wafer-level testing of main electrical, optical and structural parameters such volt-amperic characteristic, electroluminescence, photoluminescence and HRXRD (strain, composition, layer thickness) before the wafer is taped-in for complete frontend technology to make device. This provides reliable early control of manufacturing yield and allows pre-selection of wafers that are allowed to proceed for further processing. Emission wavelength control is performed by adjusting the composition of the quantum well and barrier layers as well as the width of the quantum well layer. Typical laser structures contain 1-3 quantum-wells, therefore, the entire device epitaxial structure is composed of mainly bulk layer making the epitaxy process less complex and less sensitive to deviations during the manufacturing process.
Commercial, at a large-scale, optoelectronic devices based on type-I optical transition are available for near infrared and telecom wavelengths based on GaAs and InP substrate platform. Most common substrate size is 3-inch, however recent datacom technology developments are ongoing on 6-inch GaAs substrates, primarily for low-cost VCSEL applications. Longer wavelength GaSb technology for laser diodes is primarily developed using a multi-wafer Molecular Beam Epitaxy (MBE) technology for 7 x 3-inch or 4 x 4-inch configurations. Wafer manufacturers, such as Wafer Technology, over recent years have matured GaSb substrate technology up to 6-inch diameter, whereas MBE tool manufacturers, such as Veeco (USA) and Riber (France) have made considerable improvements in new component designs and chamber geometries to increase group-III flux uniformity and temporal stability, which is crucial for multi-wafer growths of very thick heterostructures. Experience obtained at Brolis over the last 8 years of operation growing GaSb-based optoelectronic devices in multi-wafer mode and close cooperation with MBE tool manufacturers shows that all needed components to facilitate large scale production already exist as of today.
Moreover, since the world-first demonstration of widely-swept hybrid GaSb/SOI laser in 2016 , Brolis has achieved considerable progress in integrating GaSb type-I light emitters with SOI photonic integrated circuits and have recently demonstrated in the lab the full monolithic laser spectrometer, based on 4 ultra-widely swept hybrid lasers and GaSb-based photodetectors. The spectrometer is based on purely single-mode (>20 dB SMSR) laser line, which is rapidly electrothermally swept (up to 1 kHz sweep rates) using a Vernier effect in two coupled micro-ring resonators, with slightly different diameters, realized in SOI circuit. The laser spectrometer chip contains four such widely tunable lasers, based on slightly shifted gain peak in each GaSb light source and is able to achieve 120 nm tuning per gain chip and a combined tuning range of over 400 nm (Figure 2). Thermally tuning one of the micro-rings allows spectral hopping across the free spectral range of the ring (mode hops of, typically, 4-5 nm), whereas tuning both in a certain fashion allows achieving, virtually, continuous tuning across entire spectral range. (Figures 3 and 4). Figure 5 shows the microscope picture of the first hybrid sensor, realized at Brolis. Despite the “clumsy” look of a “first transistor”, the sensor is functional, has four in-house manufactured GaSb-based photodetectors integrated and one GaSb-based light emitter coupled to the external SOI cavity. The ultimate footprint of such a spectroscopic sensor is expected to be 2 x 3 x 1 mm, which is fully compatible with all current consumer and industrial platforms.
Current state-of-the art performance of Brolis GaSb/SOI technology includes a combined four-laser sweep across 480 nm with output powers around 1 mW and uncooled detector technology, allowing to achieve responsivities of 1.5 A/W, D* values of 2-3×1010cmHzW and NEP values of 2 ×10-12WHz. The latter combination already has sufficient performance for a plurality of applications – from fluid to gas analysis by tunable laser absorption spectroscopy and there still is plenty of room for improvement.
The most (in)famous application or the “holy grail” of spectroscopic sensing is glucose sensing for diabetes management applications. The need and market of such a non-invasive optical sensor that could track glucose levels in blood is clear and does not need additional marketing. However, the complexity of sensing glucose comes from skin (tissue matrix), which has a great variability from person to person and day to day. Even having overcome most of those challenges – the reality of glucose monitoring involves highly regulated environment and even having such a technology ready, could take up to a decade to enter the market. Even though the “holy grail” of optical sensing seems mythical and hard to achieve, the application field of transdermal spectroscopic sensors does not stop there. Lactates, a biomarker for anaerobic performance and fatigue is of great importance in endurance and power sports. A possibility to monitor lactate dynamics during the exercise is a dream-come-true for many professional athletes, coaches and amateur-sportsmen. Monitoring lactate in real-time would allow a sportsman to precisely adjust training loads, avoid overtraining and injury and increase performance. Having the spectroscopic sensor in the integrated chip form would allow integrating it into a wearable device for factor and accessing the consumer segment, making such monitoring of sports performance highly individualized and continuous. Another important molecule to consider would be ethanol (or alcohol in blood). Figure 6 shows a measured reflection spectrum through skin (transdermal measurement) of tissue containing ethanol. Prevention of drinking and driving, drinking and working is critical from a regulatory point of view to avoid accidents. However, easily accessible self-monitoring is no less critical. Ability to tell early on and allow to change the decision whether one can work or drive is much more important beforehand rather than catching someone “in the act”. Again, having such a device integrated in a wearable form or a sensor integrated in an already existing wearable/smartwatch platform, would allow one easily and continuously monitor alcohol content and prevent unwanted events. Moreover, in the age of IoT, this could be integrated with critical piece of machinery (car or industrial critical tools) to prevent the operator from using it while not in desired condition. The technological complexity of measuring all the above-mentioned molecules trans-dermally is similarly complex and requires long and expensive research and development effort, meaning that entry into the market in the final sensor form could be in a 3 – 5 years perspective, at best.
However, as explained in section II, the technology for spectroscopic sensing already exists today. And prior to conquering complex transdermal sensing, immediate use in hand-held fluid spectroscopic sensing can be deployed. For example, ethanol sensing in fluids is very important in brewing, where it is critical to tell ethanol content precisely in the mixture as well as the possibility to tell apart different alcohols (see figure 1) and control their presence during the process. Such spectroscopic sensors can be easily integrated into already existing industrial machinery or can be used a hand held spectrometer for home or small-scale brewing as well.
In a similar fashion, ethanol content sensor in fuel or other fluid matrix can also be realised, ether in handheld shape or integrated into already existing industrial machinery.
Another potential application that we envision is related to plastic sorting, either for recycling or other purposes. Different plastics have strong and clearly distinguishable spectral features in the range from 1.5 – 2.5 micrometers (see figure 7), that enable the use of such a laser spectrometer in sorting lines and contesting the currently used expensive SWIR camera technology.
Agricultural sector also opens potential opportunities for spectroscopic sensing. For example, milk constituents, such as fats, proteins, urea, etc. have significant absorption features in the spectral range from 1.5 to 2.5 micrometers. Monitoring content of in real-time and tracking the trends per individual cow as well as herd-wise has great importance. For example, monitoring milk fat, milk protein, somatic cell count and urea content and, respective, trends with time would allow one to adjust nutrition and get an early warning on incoming mastitis and allow for preventative actions rather than prescribing antibiotics at later stage and temporarily removing the cow from milk production. Again, such a piece of sensing hardware can easily be integrated into most of already existing milking machinery without any additional effort.
The list of applications can be expanded further to other fluid, solid state or gas sensing applications. The important take-away is that current spectroscopic sensor or, in other words, a laser-spectrometer on chip is a generic piece of hardware that can be scaled to consumer scale applications as well as can serve niche specific market segments, utilizing the same widely-swept hybrid GaSb/SOI lasers and uncooled GaSb-based detector technology.