High-speed and high-sensitivity multi-gas detection based on parallel heterodyne LITES sensor

Rapid and high-sensitivity simultaneous detection of multiple trace gases has become a critical requirement in many frontier fields. In deep-sea origin-of-life studies, tracking subtle concentration variations of methane (CH₄) and hydrogen sulfide (H₂S) and their coupling with microbial activity demands sensors that can simultaneously capture transient signals of multiple species in complex mixtures. In environmental safety monitoring, rapidly distinguishing and simultaneously quantifying multiple hazardous gases, such as nerve agents and toxic industrial chemicals, is essential for risk mitigation. In planetary exploration, the combined detection of CH₄ and acetylene (C₂H₂) in the Martian atmosphere may provide crucial evidence for geological or biological activity; however, both gases exist at extremely low concentrations and must be acquired simultaneously within a limited detection window.

 

Single-gas sensors can no longer meet the demands of these complex scenarios. Existing multi-gas laser absorption spectroscopy methods, such as time-division multiplexing, spatial-division multiplexing, or frequency-division multiplexing, are often constrained by long scanning times, complex system architectures, or incompatibility with detectors operating at fixed resonant frequencies. Therefore, achieving high-speed and high-sensitivity multi-gas simultaneous detection in a single compact system has become a core challenge that urgently needs to be addressed in this field.

 

Recently, Prof. Yufei Ma’s group at Harbin Institute of Technology proposed a parallel heterodyne light-induced thermoelastic spectroscopy (PH-LITES) sensing technique, which synergistically innovates at both the physical and information layers to achieve high-speed and high-sensitivity multi-gas simultaneous detection. This work, titled “High-speed and high-sensitivity multi-gas detection based on parallel heterodyne LITES sensor”, was published in Light: Science & Applications. The overall architecture of the sensor system is shown in Fig. 1. At the physical layer, the research team constructed a collaborative signal enhancement architecture (CSEA). To address the challenges of weak absorption signals and insufficient detector responsivity, two core components were independently developed. First, a cylindrical multi-pass cell (MPC) with an ultra-dense spot pattern was designed, as illustrated in Fig. 2. By leveraging the astigmatism generated by two orthogonally arranged compact cylindrical mirrors, the cell achieved an effective optical path length (OPL) of 38.3 m within a compact volume of only 102.6 mL, yielding a record-high OPL to volume ratio (OPL/V) of 37.4 cm^-2 and significantly enhancing the gas absorption signal. Second, a high-performance four-tine QTF was developed by innovatively increasing the number of tines from the conventional two to four and optimizing the geometrical dimensions, resulting in a 4.34-fold enhancement in output signal amplitude compared to the commercial standard QTF. These two devices work synergistically, laying a solid physical foundation for highly sensitive sensing. High-speed detection capability is enabled by parallel heterodyne modulation, where lasers at different wavelengths share the same dense spot pattern, and a single QTF is excited to generate a composite transient response signal, allowing simultaneous acquisition of spectral information from multiple gases within a short time. At the information layer, a collaborative intelligent processing architecture (CIPA) was constructed. The thermoelastic responses excited by the laser modulation signals on a single QTF naturally superimpose into a composite transient signal. A deep learning model integrating convolutional neural networks (CNN), a hybrid attention mechanism (HAM), and bidirectional long short-term memory (BiLSTM) performs “feature extraction, attention-based enhancement, and temporal modeling” on the composite transient signal at the single QTF output, accurately disentangling the contributions of each gas from the overlapping spectral features and enabling precise mapping from a single QTF output signal to multiple gas concentrations.

 

In experimental validation, distributed feedback (DFB) lasers at 1.65 μm and 1.53 μm were used to target the absorption lines of CH₄ and C₂H₂, respectively. Within a scanning time of only 4 s, the sensor achieved minimum detection limits (MDLs) of 378 ppb for CH₄ and 285 ppb for C₂H₂, with a correlation coefficient (R²) of concentration prediction as high as 0.998. As shown in Fig. 3, the PH-LITES sensor exhibited excellent linear response to gas mixtures at different concentrations. In tests with various concentration ratios, the mean relative errors (MRE) of the predicted concentrations for CH₄ and C₂H₂ were as low as 8.9% and 5.8%, respectively. Compared with other dual-gas sensing techniques, the PH-LITES sensor demonstrates significant comprehensive advantages in both detection sensitivity and response speed.

---

---