Fig. 1. Conceptual design of the dual-laser sensor based on a graphene D-shaped fiber DBR microcavity. (a) Schematic diagram of the device; graphene is deposited on the D-shaped region of an erbium-doped fiber section, and two Bragg gratings provide high reflection. Gas adsorption on graphene would change the laser frequencies (blue and red curves) distinctly. (b) Optical microscopic images of the graphene-based fiber microcavity. Here the bright green scattering is due to the erbium excitation. (c) Scanning electron microscopic image of the D-shaped fiber, in front view. (d) In situ Raman spectroscopic maps, for the G peak and the 2D peak of the graphene on fiber. Color bar: intensity. (e) Simulated optical mode fields. The left panel and the right panel show the electrical field distributions in TE and TM polarization. The color bar is normalized. (f) Measured resonances of the D-shaped fiber DBR microcavity around 1550 nm; here the red, orange, and gray curves are sech2 fittings. (g) Q factors of the resonant modes before and after graphene deposition.

Fig. 2. Excitation of the orthogonally polarized dual lasers. (a) Correlation of the 980 nm pump power and the total laser power. Laser threshold, 36 mW; dual-laser threshold, 54 mW. (b) Polarization of the laser states. Left panel: for PP=50  mW, the laser is only in the TM polarization. Right panel: for PP=60  mW, there are two lasers; one is in the TM polarization, while the other one is in the TE polarization. (c) Laser spectra for distinct pump powers. (d) Low-frequency beat notes of the microcavity laser device.

Fig. 3. Dual-laser beat note with suppressed common mode noise. (a) Schematic mechanism of the common mode noise cancelation. (b) Setup for verifying that noise of the dual-laser beat is lower than that of independent lasers. LD, laser diode; PD, photodetector; OSA, optical spectrum analyzer; ESA, electrical spectrum analyzer. (c) Optical spectrum containing the microcavity dual lasers and the reference laser. (d) Measured RF spectrum; the fitting curves are in sech2 shape. (e) Linewidth of the beat notes. (f) Frequency noise of the TE and TM laser beat signal. Gray dashed line, β-line; blue dotted line, white noise; green dotted line, flicker noise.

Fig. 4. Measured results of the NH3 gas detection in vacuum. (a) Principle of the gas sensing in our graphene-based microfiber laser cavity. (b) and (c) Measured spectrum and the “concentration–frequency shift” correlation of the dual-laser beat signal, when injecting NH3 with concentration from 0.446 to 4464 nmol/L. (d) and (e) Measured spectrum and the “concentration–frequency shift” correlation of the dual-laser beat signal. Horizontal error bars, uncertainty of the gas concentration; vertical error bars, measurement uncertainty limited by the signal linewidth (930 Hz). (f) and (g) Measured spectrum and the ‘concentration–frequency shift’ correlation of the “fTE and reference” beat signal. Vertical error bars, measurement uncertainty limited by the signal linewidth (2.3 MHz).

Fig. 5. Measured results of the NH3 gas detection in air. (a) Spectrum of the “fTE and fTM” beat note in the air. (b) and (c) Measured spectrum and the “concentration–frequency shift” correlation of the dual-laser beat signal, when increasing the NH3 concentration from 0 to 1000 ppb. (d) and (e) Detailed spectrum and the “concentration–frequency shift” correlation. Horizontal error bars, uncertainty of the gas concentration; vertical error bars, measurement uncertainty (±800  Hz).

Fig. 6. Recoverability and capability for tracing gas–graphene interactions. (a) Blue curve, recoverable frequency shift in periodically injected NH3 gas, with concentration 0–44.64 nmol/L; red curve, high-resolution frequency shift when increasing the NH3 concentration from 0 to 2.23 and 4.46 pmol/L. (b) Lock-in amplified trace and its derivation when keeping the sensor in 4.46 pmol/L NH3 environment stably. (c) Histograms of the lock-in intensity change.