Metasurface-controlled high-speed tunable external cavity lasers

TL;DR

The paper addresses the need for fast, compact tunable lasers for free-space communication and sensing by integrating a resonant electro-optic metasurface as the external mirror of an external-cavity diode laser. The approach achieves single-mode, wideband spectral control with a drastically reduced external cavity footprint, enabling voltage-driven frequency tuning with high phase sensitivity. Key results include frequency tuning rates up to $70~\mathrm{THz/s}$ and frequency excursions up to $110~\mathrm{MHz}$ at $10~\mathrm{V}$, plus electro-optic amplitude modulation exceeding $100\%$ near laser threshold; spectral stability is maintained across a broad current range due to resonance locking. These findings establish metasurface-controlled external cavities as a scalable, low-footprint platform for high-speed optical communications, ranging, and spectroscopy, with future potential to embed metasurfaces on the laser facet and to explore alternative EO materials for further performance gains.

Abstract

Tunable lasers are essential for optical communication, spectroscopy, and precision sensing, where flexible and fast control of the laser wavelength is needed. However, conventional tunable laser systems often rely on mechanical actuation, which limits their tuning speed, stability, and repeatability. Alternative tuning methods, such as adjusting the temperature of the gain medium or the injection current, are either slow or suffer from unwanted coupling between frequency and intensity. Electro-optic modulators implemented after the laser offer fast frequency tunability but these components are bulky and prone to high insertion losses. Here, we introduce an external cavity semiconductor laser that achieves ultrafast frequency and amplitude tunability within a compact footprint without any moving parts. The laser employs a resonant electro-optic metasurface as the external mirror, providing narrowband feedback that enables simultaneous single-mode operation and voltage-controlled tuning across a wide range of currents. Leveraging a narrow linewidth design, we demonstrate mode-hop-free linear frequency tuning with rates of up to 70 THz/s, and frequency excursions reaching 110 MHz under a 10 V peak RF drive, clearly surpassing rates attainable through mechanical actuation. Moreover, by biasing the laser near its threshold, our platform enables intensity modulation with an efficiency above 100% at a peak voltage of 4 V. These results highlight the platform's potential for applications requiring high-speed and mode-hop-free tunable lasers, such as free-space optical communication, laser ranging, and high-resolution spectroscopy.

Figures (17)

• Figure 1: Working principle of the metasurface assisted tunable external cavity.a. The injection current $I_{\mathrm{in}}$ supplies carriers to the gain medium of the laser diode. The back facet has reflection $r_1$, while the front facet is anti-reflection coated with $r_2 \ll r_1$. An EO crystal is placed in the external cavity and modulates the optical path length via the Pockels effect. An applied electric field to the EO crystal induces a refractive index change in the crystal, which results in a laser frequency shift. The optical field $E_i$ inside the cavity is partially transmitted, modulated twice by the EO material, and diffracted by the grating as $r_{\mathrm{ext}}e^{2i\varphi}E_i$, where $\varphi$ is the EO phase shift induced by the modulator. The first-order diffraction provides coherent feedback to the cavity, while the zeroth-order serves as the laser output. The frequency shift scales inversely with the total cavity length $L_{\mathrm{ext}}$, and directly with the EO crystal length $L_{\mathrm{eom}}$ and material parameters of the crystal, such as refractive index and electro-optic coefficient. b. Schematic of the metasurface-integrated external-cavity diode laser. The emitted light is collimated and reflected back by the electro-optic metasurface, establishing external optical feedback. The reflection from the metasurface is frequency-selective, exhibiting a resonance around $\nu_{\mathrm{res}}$, where it reaches a maximum value $r_{\mathrm{max}} \gg r_1$. The reflected beam, $r_{\mathrm{ext}}(\nu) E_i$, re-enters the laser cavity and is reflected by the internal mirror ($r_1$), completing the round-trip feedback path. The metasurface consists of interdigitated gold electrodes and a nonlinear optical polymer, supporting guided-mode resonances. The electro-optic response is activated by periodic poling, producing alternating domains with electro-optic coefficients $+r_{\mathrm{eo}}$ (purple) and $-r_{\mathrm{eo}}$ (orange), aligned through molecular dipole orientation. This configuration enables dynamic modulation of the metasurface reflection via an applied RF field $V_m$. c. The simulation illustrates the resonant behavior of the electro-optic metasurface. The upper graph shows the reflection amplitude, while the lower graph displays the corresponding phase response. The maximum reflection ($r_{\mathrm{max}} = 0.375$) occurs near $\nu_{\mathrm{res}} = 193.41~\mathrm{THz}$, corresponding to a wavelength of $\lambda_{\mathrm{res}} = 1549~\mathrm{nm}$. The metasurface geometry used in the simulation features a pitch size $p = 1.04~\mu\mathrm{m}$ and a nonlinear material layer thickness $T_{\mathrm{JRD1}} = 1.7~\mu\mathrm{m}$. Applying a voltage to the metasurface modifies its refractive index which leads to the shift of the resonance, as illustrated by semi-transparent curves. d. Without optical feedback, the lasing threshold is determined solely by the length and gain bandwidth of the laser diode. When the EO metasurface provides frequency-selective feedback, the effective gain threshold decreases near the resonance, resulting in a lower required injection current for lasing. e. A reduced gain threshold lowers the injection current from $I_{\mathrm{th},i}$ to $I_{\mathrm{th}}$, but also decreases the output power $P_{\mathrm{out}}$ due to modified external cavity conditions.
• Figure 2: Metasurface-controlled laser diode feedback.a. To characterize the solitary laser diode, the optical spectrum is sent directly to an Optical Spectrum Analyzer (OSA), and the output power is monitored with a power meter. The electro-optic metasurface is then placed in the external cavity and aligned such that the electric field inside the cavity is oriented perpendicular to the gold stripes of the metasurface. b. The reflectance spectrum of the metasurface is plotted. The maximum reflectance, $R_{\mathrm{max}} = 0.28$, occurs at the resonance wavelength, with a linewidth of $\delta\lambda_{\mathrm{res}} = 7~\mathrm{nm}$. Fabry–Pérot oscillations from the substrate are visible in light purple, while the filtered reflectance—removing these oscillations—is shown in darker purple for clarity. c. The resulting LI (light–current) curves are compared for the two configurations. The solitary diode shows a lasing threshold of $I_{\mathrm{th,sd}} = 22~\mathrm{mA}$ (gray curve). Once optical feedback is established, the lasing threshold is reduced to $I_{\mathrm{th}} = 13.4~\mathrm{mA}$. The step-like behavior observed in both LI curves originates from mode hopping, which is characteristic of multi-mode laser operation. d. Spectral response with and without feedback. (i) The emission spectrum of the solitary laser diode, denoted $S_{\mathrm{sd}}$, is recorded using an Optical Spectrum Analyzer. Below threshold (light gray), at $I_{\mathrm{in}} = 5~\mathrm{mA}$, and above threshold, at $I_{\mathrm{in}} = 25~\mathrm{mA}$, the laser exhibits spontaneous and stimulated emission, respectively. The gain spectrum is approximately $80~\mathrm{nm}$ wide, with a noticeable blue shift of $\sim 32~\mathrm{nm}$ as the current increases. (ii) The emission spectra in the presence of feedback, denoted $S_{\mathrm{out}}$, are shown both below threshold at $I_{\mathrm{in}} = 5~\mathrm{mA}$ (light blue) and above threshold at $I_{\mathrm{in}} = 14~\mathrm{mA}$ (dark blue). Below threshold, the resonance of the metasurface is observable in transmission. Above threshold, lasing occurs predominantly around the resonance, while off-resonant modes are suppressed. The internal cavity mode spacing is marked as $\mathrm{FSR}_i$, and the difference in intensity between the main lasing peak and the adjacent internal sidebands is quantified as $\mathrm{SMSR}_i^+$ and $\mathrm{SMSR}_i^-$. A zoomed-in view of the dominant lasing peak reveals the external cavity mode structure, with a free spectral range of approximately $\mathrm{FSR}_e = 5~\mathrm{GHz}$. The relative strength of these external sidebands is characterized by $\mathrm{SMSR}_e^+$ and $\mathrm{SMSR}_e^-$. Supplementary Information Note \ref{['Supp:sub:sbr']} presents the evolution of the internal sideband ratio $\mathrm{SMSR}_i = \min(\mathrm{SMSR}_i^+, \mathrm{SMSR}_i^-)$ and the external sideband ratio $\mathrm{SMSR}_e = \min(\mathrm{SMSR}_e^+, \mathrm{SMSR}_e^-)$ as a function of injection current. Notably, $\mathrm{SMSR}_i$ remains stable and below $-40~\mathrm{dB}$ above threshold, while $\mathrm{SMSR}_e$ remains below $-20~\mathrm{dB}$ across a wide current range, until disrupted by mode hopping due to the continuous blue shift.
• Figure 3: Ultrafast frequency modulation of a laser diode with an electro-optic metasurface. a. An RF source applies a triangular voltage waveform with peak amplitude $V_m$ and frequency $\nu_{\mathrm{mod}}$ to the metasurface, dynamically tuning its reflection. This modulation alters the lasing frequency of the diode laser. The output beam is combined with a frequency-stable reference laser in a heterodyne detection setup, enabling extraction of the instantaneous frequency $\nu_{\mathrm{inst}}(t)$ through RF beatnote analysis. The modulation induces a frequency excursion $\Delta \nu$ centered around a carrier frequency $\nu_0$. The detailed setup is presented in Supplementary Information Figure \ref{['s-fig:HD_eo_setup']}. b. Spectrograms of the heterodyne beatnote signal under triangular RF modulation at three different frequencies: $50~\mathrm{kHz}$, $250~\mathrm{kHz}$, and $1~\mathrm{MHz}$. All measurements were performed at a fixed RF amplitude of $V_m = 10~\mathrm{V}$ and injection current $I_{\mathrm{in}} = 16~\mathrm{mA}$. Each panel shows ten consecutive modulation periods. An average frequency excursion of approximately $100~\mathrm{MHz}$ is observed across all modulation rates. The color scale indicates optical power, calibrated to the output power from the laser’s LI curve. The spectrograms capture the time-resolved evolution of the laser frequency enabled by the electro-optic metasurface. Even at the highest modulation rate, the excursion remains visible, demonstrating the fast response of the metasurface-enabled tuning. At higher modulation frequencies, however, the intrinsic time–frequency resolution trade-off of spectrogram analysis limits the ability to fully resolve the instantaneous frequency. c. Heterodyne spectra measured at a fixed modulation frequency $\nu_{\mathrm{mod}} = 1~\mathrm{MHz}$ for injection currents $I_{\mathrm{in}} = 16~\mathrm{mA}$, $20~\mathrm{mA}$, and $24~\mathrm{mA}$. The traces illustrate how the frequency excursion and heterodyne signal amplitude evolve with injection current. As the current increases, the frequency excursion gradually narrows. The amplitude of the beatnote is calibrated using the laser’s LI curve. d. Frequency excursion versus modulation frequency. The left vertical axis (purple) shows the peak-to-peak frequency excursion $\Delta \nu_{pp}$, with the shaded region representing the standard deviation across modulation cycles. The right axis (black) shows the root-mean-square (RMS) deviation from an ideal triangular waveform. As $\nu_{\mathrm{mod}}$ increases, both the frequency excursion and the waveform fidelity degrade, illustrating the trade-off between modulation speed and dynamic range. e. Frequency excursion as a function of injection current $I_{\mathrm{in}}$ at fixed modulation frequency $\nu_{\mathrm{mod}} = 1~\mathrm{MHz}$. The shaded area indicates the standard deviation of the measured excursion. The frequency excursion remains relatively stable around $90~\mathrm{MHz}$ for moderate injection currents, with a slight decrease observed at higher current levels.
• Figure 4: Experimental setup for amplitude modulation.a. After optical feedback is established between the laser diode and the electro-optic metasurface, an RF voltage with amplitude $V_m$ at $\nu_{mod} = 50~\mathrm{MHz}$ is applied using an arbitrary waveform generator. This signal is also sent as a reference to the lock-in amplifier. The detailed setup is presented in Supplementary Information Figure \ref{['s-fig:HD_eo_setup']}. After coupling the output light from free space into a fiber using fiber coupler FC1 and directing it to photodiode PD1, the detected signal appears as $V_0 + \Delta V$, consisting of a DC component $V_0$ and an RF component $\Delta V$ oscillating at $\nu_{mod}$. In the homodyne detection path, the lock-in amplifier measures the modulated signal $\Delta V$, while the DC component $V_0$ is recorded from the DC output of PD1 using a voltmeter. b. Lock-in measurements of the modulation. The plots show the variation of the detected voltage $\Delta V$ as a function of injection current $I_{\mathrm{in}}$ and RF amplitude $V_m$, measured at a modulation frequency of $\nu_{mod} = 50~\mathrm{MHz}$. The bipolar colorbar indicates the magnitude and sign of $\Delta V$. c. Measured modulation efficiency $\Delta I / I_0$ as a function of injection current $I_{\mathrm{in}}$ for various RF amplitudes $V_m$ at $\nu_{mod} = 50~\mathrm{MHz}$. The efficiency peaks near the laser threshold, exceeding $100\%$ for $V_m > 4~\mathrm{V}$. Inset: Zoomed-in view showing the absolute voltage variation $\Delta V$ (orange), the DC component $V_0$ (gray), and the corresponding modulation efficiency $\Delta I / I_0$ (blue, right axis) at $V_m = 5~\mathrm{V}$. Mode-hopping events are also visible as discrete jumps in the traces. For background on the threshold behavior, see Supplementary Information Figure \ref{['s-fig:sbr']}. d. Maximum modulation efficiency plotted for two regimes: injection currents near threshold $I_1 = 5~\mathrm{mA}$ (gold) and well above threshold $I_3 = 20~\mathrm{mA}$ (gray). The results show that the maximum modulation increases approximately linearly with the RF amplitude $V_m$ in both regimes. e. Oscilloscope time-trace measurements at fixed voltage ($V_m = 5$ V) and varying currents, as indicated by the arrows in panel b. The vertical axis shows the RF modulation superimposed on the DC signal $V_0$. Traces are shown near threshold at $I_\mathrm{in} = 13.8~\mathrm{mA}$, slightly above threshold at $I_\mathrm{in} = 14.2~\mathrm{mA}$, and well above threshold at $I_\mathrm{in} = 20~\mathrm{mA}$. As the output power increases, the modulation depth decreases due to the faster growth of the DC component $V_0$ compared to the modulated signal $\Delta V$.
• Figure S1: Nonlinear activation. a. The IVT curves of the poling. During the poling and after reaching the glass transition temperature, the current conducted by the polymer between the two electrodes reaches $\approx\;300$nA. After tuning off the heater, the current starts to go down. b. DC measurement before and after poling. An increase of current is observed after the poling. c. $\&$ d. AFM measurements of the poled and unpoled devices. A change in the topography of the polymer is observed after the poling. Red and blue cut lines show the height of the not-poled and poled devices, respectively.