Terahertz Synthetic FM Triplet for Distortion-Free Stabilization and Lamb-Dip Spectroscopy

TL;DR

The study tackles distortion-free stabilization in the terahertz regime where dense molecular spectra cause modulation sideband interference. It introduces a synthetic FM triplet for THz frequency discrimination, implemented with a dual-loop stabilization scheme and THz heterodyne detection, and demonstrates its application to CH$_3$CN rotational transitions, achieving a fractional instability of $1\times 10^{-9}$ at 1 s and enabling Lamb-dip spectroscopy. The results establish acetonitrile as a frequency-agile molecular-reference candidate and outline paths to ultrastable performance (potentially down to $10^{-13}$) with improved seed-laser stabilization, including DWBLs, enabling robust THz molecular clocks and precision spectroscopy of complex hyperfine structures.

Abstract

We demonstrate a distortion-free terahertz frequency stabilization technique using a "synthetic FM triplet" to overcome modulation sideband interference associated with high-density spectral lines in molecular clocks. By applying this method to the rotational transitions of acetonitrile (CH$_3$CN), we successfully generated clean derivative waveforms free from inter-line interference, achieving a fractional frequency instability of $1 \times 10^{-9}$ at an averaging time of $1~\mathrm{s}$. Furthermore, we report the observation of Lamb-dips using this high-fidelity approach. Our results establish acetonitrile as a promising candidate for high-agility molecular clocks and provide a robust solution for precision spectroscopy of molecules with complex hyperfine structures.

Figures (5)

• Figure 1: (a) Schematic layout of THz synthetic FM triplet generation. The solid lines are electrical lines, and dashed lines are optical fibers. DDS, Direct digital synthesizer; $A_v$, Voltage controlled variable attenuator; $\Phi_v$, Voltage controlled phase shifter; AOM, Acousto-optic modulator; VOA, Voltage controlled variable optical attenuator; BPD, Balanced photo detector; PD, Photo detector; LIA, Lock-in amplifier. The gray dashed regions ①-③ correspond to ① a optical power-balance servo loop, ② a feedback loop that maintains the FM triplet, and ③ a THz power-stabilization servo loop.(b) Schematics diagrams of THz synthetic FM triplet generation.
• Figure 2: (a) Calculated absorption strength of acetonitrile rotational transition. The target absorption $(J, K)=(17, 0)\leftarrow(16, 0)$ is highlighted in red. (b) Pressure-dependent derivative spectra of acetonitrile: $0.5~\mathrm{Pa}$ (purple), $0.4~\mathrm{Pa}$ (blue), $0.3~\mathrm{Pa}$ (light blue), $0.2~\mathrm{Pa}$ (green), $0.15~\mathrm{Pa}$ (yellow-green), $0.10~\mathrm{Pa}$ (orange), and $0.05~\mathrm{Pa}$ (red). The modulation frequency was set to $f_m=300~\mathrm{kHz}$. (c) Modulation frequency-dependent derivative spectra of acetonitrile: $900~\mathrm{kHz}$ (purple), $600~\mathrm{kHz}$ (blue), $300~\mathrm{kHz}$ (light blue), $100~\mathrm{kHz}$ (green), and $30~\mathrm{kHz}$ (red). The sample pressure was set to $0.05~\mathrm{Pa}$. (d) Pressure-dependent noise-to-slope ratio measured at several modulation frequencies: 30 kHz (red), 50 kHz (orange), 75 kHz (yellow-green), 100 kHz (green), 200 kHz (teal), 300 kHz (light blue), 600 kHz (blue), and 900 kHz (purple).
• Figure 3: (a) Schematic layout of the THz frequency locking system. Solid lines indicate electrical paths, and dashed lines indicate optical fibers. VCO: Voltage-controlled oscillator, LPF: Low-pass filter, SA: RF spectrum analyzer, LO UTC-PD: Local-oscillator UTC-PD. The LO UTC-PD output frequency was set to $f_{\mathrm{LO}}=311.3$ GHz. The gray shaded regions correspond to the slow and fast laser frequency servos. (b) Schematic diagrams of the frequency tuning mechanism achieved by the VCO and Mixer. (c) Schematics of THz frequency variation and the corresponding linewidth narrowing compensated by the slow and fast servos, respectively.
• Figure 4: (a) Schematic layout of the frequency locking experiment. (b) Derivative absorption signal used for locking. Modulation frequency: 300 kHz; pressure: 0.5 Pa; integration time: 70 ms. (c) IF spectrum of the THz beat note; green: free-running, blue: slow servo only. (d) Zoomed view of the locked spectrum; blue: slow servo only, red: slow and fast servo. (e) Temporal variation of frequency deviation from the mean value. (f) Modified Allan deviation; green: free-running, blue: slow servo only, red: slow and fast servo.
• Figure 5: (a) Schematic layout of Lamb-dip spectroscopy. (b) Pressure-dependent derivative-shaped Lamb-dip spectra of acetonitrile. Traces are vertically offset for clarity: $0.5~\mathrm{Pa}$ (purple, $+9~\mathrm{mV}$), $0.4~\mathrm{Pa}$ (blue, $+7~\mathrm{mV}$), $0.3~\mathrm{Pa}$ (light blue, $+5~\mathrm{mV}$), $0.2~\mathrm{Pa}$ (green, $+3.4~\mathrm{mV}$), $0.15~\mathrm{Pa}$ (yellow-green, $+2.2~\mathrm{mV}$), $0.10~\mathrm{Pa}$ (orange, $+1.2~\mathrm{mV}$), and $0.05~\mathrm{Pa}$ (red, $0~\mathrm{mV}$). The modulation frequency was set to $f_m=30~\mathrm{kHz}$. (c) Modulation frequency-dependent derivative-shaped Lamb-dip spectra of acetonitrile. Traces are vertically offset for clarity: $900~\mathrm{kHz}$ (purple, $+5~\mathrm{mV}$), $600~\mathrm{kHz}$ (blue, $+4.2~\mathrm{mV}$), $300~\mathrm{kHz}$ (light blue, $+3.4~\mathrm{mV}$), $200~\mathrm{kHz}$ (teal, $+2.6~\mathrm{mV}$), $100~\mathrm{kHz}$ (green, $+1.8~\mathrm{mV}$), $75~\mathrm{kHz}$ (yellow-green, $+1.2~\mathrm{mV}$), $50~\mathrm{kHz}$ (orange, $0.6~\mathrm{mV}$), and $30~\mathrm{kHz}$ (red, $0~\mathrm{mV}$). The sample pressure was set to $0.05~\mathrm{Pa}$. (d) Pressure-dependent $|\mathrm{noise}/\mathrm{slope}|$ value measured at several modulation frequencies; filled circles denote Lamb-dip signals and open squares denote transmission signals. Colors indicate the modulation frequency $f_m$ (red: 30 kHz, orange: 50 kHz, blue: 300 kHz, light blue: 600 kHz, cyan: 900 kHz).