06 Mar 2007
Markus Ortsiefer from Vertilas explains why long-wavelength tunable VCSELs are a better option for gas sensing than the lead-salt, quantum-cascade and diode-pumped solid-state lasers that are used today.
Gas sensors are employed for numerous applications, including detecting moisture; monitoring gas leaks and greenhouse gas emissions; searching for toxic and hazardous species; and optimizing combustion processes. Although many of these tasks can be carried out with chemical detectors, there is now a growing interest in optical systems that have several inherent advantages. These include rapid data collection and a non-contact approach that is more suited to the awkward measurement conditions in power stations and chemically reactive environments. In contrast to chemical sensors, which only detect gases at a fixed location, optical techniques can monitor target molecules anywhere within the detection path of the measurement system.
At Vertilas, in Garching, Germany, we have been developing a unique VCSEL structure that can provide the source for a modern sensing technique – tunable diode laser spectroscopy (TDLS). Our lasers cover the 1300–2300 nm wavelength range and have been used to determine the presence of numerous gases. With the tunable laser as a source, individual gases are identified by their unique optical absorption fingerprint. By focusing on a specific wavelength where a gas has a characteristic absorption, the amount of light absorbed reveals the concentration of the gas.
TDLS has frequently been applied to wavelengths outside the range covered by our lasers. By analysing the strong absorption lines in the mid-infrared region (2.5–50 µm) the technique can determine the concentration of common gases such as water vapour, carbon dioxide and ammonia. These lines are associated with so-called rovibronic ground energy transitions, a term that describes the combination of molecular rotations and oscillations. The absorption strength of these transitions enables detection of gas concentrations even in the parts per billion range.
However, the various lasers that are used to probe these gases at mid-infrared wavelengths, including lead-salt, quantum-cascade and diode-pumped solid-state lasers, tend to be either expensive or inconvenient to use. This can be avoided by focusing on alternative transitions in the near-infrared (800–2500 nm). The absorption strength of these transitions is one to two orders of magnitude weaker than that of the rovibronic transitions. However, more reliable, cost-effective and easy-to-use sources are available that draw on the technology associated with lasers used for telecommunications or data storage. Sensing in this spectral range is also an advantage because uncooled high-sensitivity detectors are available that can partly offset the weaker absorption strength of the higher order transitions.
VCSELs versus edge emitters
The first TDLS systems operating in the near-infrared used conventional edge-emitting distributed feedback (DFB) lasers, but improvements to the performance of VCSELs operating at 1.3 µm and above have made a better source for gas sensing (see table for the benefits of VCSELs over DFB lasers for TDLS applications). These VCSELs have a tuning range that can extend to several nanometres, which is much wider than that of a DFB laser. This is because VCSELs have a much smaller active region, so that their internal temperature is much more susceptible to increases in drive current than a DFB laser. The tunability yields a wider range of measurements. For example, pressure-broadened absorption lines can be examined, several lines relating to one species can be observed simultaneously and multiple species can be detected in a gas mixture. The VCSEL wavelength can also be tuned through the absorption line several million times per second, enabling real-time monitoring of rapid combustion processes.
Although GaAs-based VCSELs for optical communication applications are made in their millions &ndash their use in TDLS systems is restricted to the detection of oxygen, which has an absorption line at 760 nm. Emission can be extended to 1300 nm while maintaining an adequate level of performance, but this is still short of the absorption lines for important species such as methane (1651 nm), carbon dioxide (2004 nm), water vapour (1854 nm) and ammonia (1512 nm).
Unfortunately, the performance of the long-wavelength VCSELs has lagged behind that of their short-wavelength counterparts, due to technological difficulties linked with the different material system required. The ternary and quaternary InP-based compounds that are needed to reach the longer wavelengths suffer from a relatively small index contrast, which means that a very large number of mirror pairs are needed to produce the required reflectivity for the laser cavity. Thermal conductivity is also more than an order of magnitude below that of the AlGaAs/GaAs compounds, making thermal management a major problem. Despite efforts to improve thermal dissipation by attaching a heat sink to one of the mirrors, most of the long-wavelength VCSELs fabricated so far have poor performance characteristics.
Success is possible, however, if one turns to novel designs, such as our InP-based VCSELs that feature a buried tunnel junction (BTJ) and dielectric mirrors. We demonstrated this approach in 1999 and since then have improved the performance of our devices and increased their spectral coverage. Our BTJ VCSELs now operate at over 100 °C, have good thermal management properties (see “Coping with heat”) and deliver singlemode output powers of more than 1 mW at 85 °C. They also have a wavelength tuning range of several nanometres, modulation speeds in excess of 10 Gbit/s and are compatible with high-volume manufacturing – the full-wafer processes we use today on 2 inch material can be scaled to larger sizes.
Our BTJ VCSELs are mounted on a TO-header with an integrated thermo-electric cooler and thermistor to monitor and control the laser stage temperature. Temperature stability is demanded by gas sensing applications because the temperature-dependent wavelength is crucial to accurate detection. In a final production step, the header is completed with a cap and a wavelength-specific antireflection coating. The lasers’ gas-sensing capability has been demonstrated by absorption measurements of ammonia, which have shown many fine details (figure 1). We acquired these spectra by operating the laser at –3.4 °C and then 15.2 °C to provide a coarse control of the emission wavelength and using current tuning for fine adjustments.
By adjusting the thickness and composition of the active region and epitaxial and dielectric mirror layers, we produced singlemode BTJ VCSELs operating from 1.3 to 2.05 µm. We also demonstrated the first electrically pumped room-temperature continuous-wave 2.3 µm BTJ VCSEL. This produces 1.47 mW at 0 °C and 0.74 mW at 20 °C, and a tuning range of 4.2 nm at room temperature. It only operates in a multimode fashion that is unsuitable for spectroscopy, but we are developing a singlemode version that could be used to detect carbon monoxide, which has a 2332 nm absorption line. This work will extend the range of our BTJ VCSELs, which can provide a cost-effective, tunable source and offer a good choice for a variety of gas-sensing applications.