Ultra-sensitive all-fibre photothermal spectroscopy with large dynamic range

Abstract

Photothermal interferometry is an ultra-sensitive spectroscopic means for trace chemical detection in gas- and liquid-phase materials. Previous photothermal interferometry systems used free-space optics and have limitations in efficiency of light-matter interaction, size and optical alignment, and integration into photonic circuits. Here we exploit photothermal-induced phase change in a gas-filled hollow-core photonic bandgap fibre, and demonstrate an all-fibre acetylene gas sensor with a noise equivalent concentration of 2 p.p.b. (2.3 × 10(-9) cm(-1) in absorption coefficient) and an unprecedented dynamic range of nearly six orders of magnitude. The realization of photothermal interferometry with low-cost near infrared semiconductor lasers and fibre-based technology allows a class of optical sensors with compact size, ultra sensitivity and selectivity, applicability to harsh environment, and capability for remote and multiplexed multi-point detection and distributed sensing.

Figure 1. PT-induced phase modulation in a HC-PBF.

(a) Modulated pump beam (λ_pump) and constant probe beam (λ_probe) are counter-propagating in a fluid-filled HC-PBF. The pump and probe may also be arranged to co-propagate within the HC-PBF. (b) Processes involved in producing phase modulation in a HC-PBF.

Figure 2. Experimental set-up for gas detection with 10-m-long HC-PBF.

The splitting ratios of FC1 and FC2 are, respectively, 80/20 and 50/50, which approximately balanced the powers in the two arms of the interferometer. The optical path lengths of the two arms are also balanced to minimize the laser phase to intensity noise conversation. Filter 1 is used to filter out the residual pump and Filter 2 to minimize the effect of EDFA's ASE noise. The splitting ratio of FC3 is 90/10. Output from PD1 passes a low-pass-filter (LPF) and is used for interferometer stabilization. Output from PD2 contains the PT-induced phase modulation signal. The driving current of the DFB is modulated at 50 kHz by use of the internal signal generator of the lock-in. Inset: scanning electron microscopy (SEM) image of the HC-1550-02 fibre's cross-section. DAQ, data acquisition; DFB, distributed feedback laser (the pump); ECDL, external-cavity diode laser (the probe); EDFA, erbium-doped fibre amplifier; FC1-FC3, fibre couplers; OC, optical circulator; PD1-PD2, photo-detectors; PZT, piezoelectric transducer.

Figure 3. Experimental results for 10-m-long HC-PBF.

(a) Second harmonic lock-in output (signal) when pump laser is tuned across the P(9) line of acetylene at 1,530.371 nm. (b) Second harmonic lock-in output when the pump wavelength is fixed to 1,530.53 nm. Black line: pump power is ∼15.3 mW. Red line: pump power is zero (off). The measurements were conducted consecutively for the two pump power levels but the results are displayed in the same panel for easy comparison. (c) Second harmonic signal and the s.d. of the noise as functions of pump power level. Error bars show the s.d. from five measurements and the magnitudes of the error bars are scaled up by 10-fold for clarity reason. The mean probe power level on PD2 is ∼50 μW. The gas concentration used is 10 p.p.m. acetylene balanced by nitrogen, and the experiments were conducted at room temperature. The amplitude of wavelength modulation was set to ∼2.2 times of the absorption linewidth. The time constant of the lock-in amplifier is 1 s with a filter slope of 18 dB Oct⁻¹, corresponding to a detection bandwidth of 0.094 Hz.

Figure 4. Second harmonic signal as function of gas concentration.

(a) Second harmonic lock-in output signal when pump laser is tuned across the P(9) line of acetylene at 1,530.371 nm for 50, 100, 200, 400 p.p.m. acetylene concentration. (b) Second harmonic signal (peak-to-peak value) as function of gas concentration. Error bars in the horizontal axis are based on the accuracy of the two mass flow controllers we used to prepare different gas concentrations. Error bars in the vertical axis show the s.d. from five measurements and the magnitudes of the error bars are scaled up by 10-fold for clarity reason. The pump power in the hollow-core was estimated to be ∼25 mW and the mean probe power level on PD2 is ∼200 μW. The time constant of the lock-in amplifier is 1 s with a filter slope of 18 dB Oct⁻¹, corresponding to a detection bandwidth of 0.094 Hz. The length of the sensing HC-PBF is 0.62 m.

Figure 5. Shot-noise limit in terms of NEC and NEA for different pump and probe power levels.

The pump wavelength is tuned to the P(9) absorption line of acetylene at 1,530.371 nm and the length of HC-PBF is 10 m. Detection bandwidth is 0.094 Hz, corresponding to 1 s time constant of the lock-in with a filter slope of 18 dB Oct⁻¹. Probe wavelength λ_probe=1,556.59 nm, v_probe=c/λ_probe. Detector quantum efficiency η=0.8, Planck constant h=6.626 × 10⁻³⁴ J s.