Skip to content

Enhanced stability of a single-longitudinal-mode diamond Raman laser enabled by pump-wavelength locking

FieldValue
JournalOptics Express
Published2026-02-18
AuthorsWenqiang Fan, Hui Chen, Longjie Zhang, Zhilin Yan, Yancheng Guo
DOI10.1364/oe.589418
Citations3

Pump-wavelength locking produced 2.5 W of single-longitudinal-mode first-Stokes light at 1240 nm from 22 W of incident pump. The output linewidth was approximately 2.9 kHz, while power instability stayed below 1.59% and wavelength drift below 87 MHz. A fast pump-frequency loop and a slow cavity-length loop maintained lock during thermally driven drift.

MetricConditionResult
Stokes thresholdIncident 1064 nm pumpApproximately 3.7 W
Stokes output22 W incident pump2.5 W at 1240 nm
Optical conversion efficiency22 W incident pump11.3%
Slope efficiency22 W incident pump13.7%
Stokes linewidthMaximum output, delayed self-heterodyneApproximately 2.9 kHz
Power instabilityDual-loop locking at maximum outputBelow 1.59%
Wavelength driftDual-loop locking at maximum outputBelow 87 MHz

The beam profile had a well-defined Gaussian intensity distribution, and no Brillouin laser emission was detected throughout the measured power range. A Thorlabs SA210-8B scanning Fabry-Perot interferometer confirmed a typical single-longitudinal-mode profile at maximum output.

Linewidth was measured with the delayed self-heterodyne method. Because the linewidth was at the kHz level, the authors used 300 m of short fiber in the partially coherent regime. A conventional long-delay measurement would require tens of kilometers, which introduces substantial 1/f noise and can bias the result. The CDSPST value was 21.0 dB, corresponding to approximately 2.9 kHz.

The seed was a narrow-linewidth 1064 nm fiber laser. An electro-optic modulator generated modulation sidebands, and an optical amplifier raised the pump to 30 W with a 6 kHz linewidth. An optical isolator blocked back-reflections from reaching the seed source.

A 160 mm focal-length lens mode-matched the pump waist at the diamond to the cavity’s fundamental mode. The V-shaped cavity used two concave mirrors with 50 mm radii and one flat mirror. M1 reflected 96% at 1064 nm and more than 99.9% at 1240 nm. M2 reflected more than 99.9% at 1064 nm and transmitted 0.6% at 1240 nm. M3 reflected more than 99.9% at both wavelengths.

The gain crystal measured 2 × 2 × 7 mm³ and was mounted on a thermoelectric cooler. A half-wave plate set pump polarization for maximum Raman gain. A photodetector after M3 collected the transmitted pump and its modulation sidebands. The signal was mixed with a local oscillator in a Moku:Pro frequency locker to produce a Pound-Drever-Hall detuning error signal.

Why pump locking replaced temperature control

Section titled “Why pump locking replaced temperature control”

In preliminary tests, the frequency locker drove the cavity PZT to maintain resonance. The transmitted pump and error signals jumped, and the transmitted pump often undershot.

Raman output heated the diamond and increased its optical path length. Because diamond has high thermal conductivity, the thermoelectric controller responded with a delay. The delayed control produced temperature and cavity-length oscillations. PZT nonlinearity, hysteresis, and mechanical jitter added further disturbances. The resonator therefore spent much of the test in a poor resonance state.

When the locker instead drove pump frequency, the transmitted pump remained high for most of the test. Occasional signal and voltage jumps occurred, but the system did not lose lock. The diamond temperature still varied during this test. The authors therefore used pump-wavelength locking for the remaining experiments.

The fast pump loop had a control bandwidth of approximately 1 kHz. The slow PZT auxiliary loop had a control bandwidth of approximately 100 Hz.

With active diamond temperature control removed, temperature rose slowly during locking. The required pump tuning voltage rose with the accumulated optical path length. The voltage was amplified by a factor of 5, but the frequency locker’s output range was limited to ±1 V. At saturation, the loop could no longer track cavity drift and lost lock.

Over 20 minutes, the output experienced three mode hops. The longest continuous mode-hop-free period was 10 minutes, with frequency drift below 645 MHz. For comparison, the free-running pump drifted about 300 MHz over 30 minutes. A HighFinesse WS-7 wavemeter recorded the output with ±2 MHz absolute accuracy and 0.1 MHz resolution.

The authors attributed mode hops to cavity-length drift and competition between adjacent longitudinal modes. They propose improving cavity Q factor and controlling the oscillator free spectral range.

The dual-loop architecture assigns rapid corrections to pump wavelength and slow corrections to cavity length. The fast PID loop modulates pump laser current, and the slow PID loop regulates PZT voltage. The Pound-Drever-Hall error signal anchors the pump wavelength to a cavity resonance.

Thermal drift then requires only slow PZT correction, rather than rapid PZT motion. At maximum output, the resulting Raman laser had power instability below 1.59%. Its wavelength drift remained below 87 MHz. The authors report an approximately tenfold improvement over pump-frequency locking alone.

Comparison with earlier diamond Raman lasers

Section titled “Comparison with earlier diamond Raman lasers”

An earlier external-cavity diamond Raman laser kept the Stokes wavelength within 80 MHz over tens of seconds. The paper notes that stability degraded as pump power increased because of diamond temperature fluctuations.

A 2016 volume Bragg grating coupled cavity produced 0.5 W at 1480 nm with 40 MHz frequency stability over 90 seconds. The current 2.5 W source at 1240 nm had drift below 87 MHz with dual-loop control.

The 87 MHz result is not a direct improvement over the earlier 40 MHz result because the architectures and test durations differ. The paper’s long-term stability claim is instead supported by the tenfold gain over pump-only locking in the same system.

The authors propose a wavelength-scalable route based on diamond’s 1332.3 cm⁻¹ Raman shift and broad transmission from ultraviolet to far-infrared. The paper does not test another output wavelength.

For laser engineers, the main design requirement is to separate correction bandwidths. Fast pump tuning must handle rapid disturbances, while the slow PZT loop must cover accumulated thermal drift. In this setup, active thermoelectric control worsened stability rather than improving it.

System designers also need tuning headroom beyond the observed ±1 V saturation point. Cavity Q and free spectral range require attention because drift-induced mode competition caused mode hops.

The authors expect longer continuous lock durations after further control optimization and diamond-temperature feedforward. Those improvements remain proposed rather than demonstrated.

The paper identifies precision spectroscopy, remote sensing, and quantum optics as target fields. The study does not test those applications or report locked operation at an incident pump power above 22 W. Underlying data are not public, although the authors offer them on reasonable request.

The 1240 nm Raman laser requires a diamond gain crystal, and the cavity used one measuring 2 × 2 × 7 mm³. 6CCVD’s diamond materials catalog includes MPCVD-grown single-crystal and polycrystalline diamond, with thickness from 0.1 to 500 micrometers and substrates up to 10 mm. The paper’s gain-element envelope is within the stated 10 mm substrate limit, and its geometry can be reviewed through engineering support for custom fabrication support.

View Original Abstract

To overcome the insufficient long-term stability of conventional Raman lasers, we demonstrate a single-frequency diamond Raman laser enabled by a pump-wavelength locking strategy. A three-mirror V-shaped cavity pumped by a 1064 nm fiber laser (6 kHz linewidth) provides single-longitudinal-mode (SLM) output at 1240 nm using Pound-Drever-Hall (PDH) stabilization. We further find that eliminating active temperature control of the diamond crystal suppresses rapid thermal perturbations and improves short-term stability. To compensate thermally induced slow cavity-length drifts, we implement a dual-loop feedback architecture that employs a fast PID loop acting on the pump wavelength and a slow PID loop driving a cavity piezoelectric transducer (PZT). Specifically, the fast loop tunes the pump wavelength via laser-current modulation, while the slow loop stabilizes the cavity length by regulating the PZT voltage. With 22 W pump power, the system produces 2.5 W SLM output with a linewidth of ∼2.9 kHz. The locked Raman laser exhibits an output power instability below 1.59% and a wavelength drift below 87 MHz, demonstrating markedly improved long-term operational stability. Unlike mainstream cavity-length locking techniques, our approach exploits the PDH error signal to directly lock the pump wavelength, providing a high-performance and wavelength-scalable route to highly stable single-frequency laser sources.