---
title: Multi-Channel Microwave-to-Optics Conversion
url: https://www.emergentmind.com/topics/multi-channel-microwave-to-optics-conversion
type: topic
---

# Multi-Channel Microwave-to-Optics Conversion

Searching arXiv for the core and contextual papers cited in the article.
Multi-channel microwave-to-optics conversion denotes the coherent transduction of more than one microwave frequency channel, mode, or spectral slice into optical outputs, typically at telecom wavelengths, by means of electro-optic, optomechanical, acousto-optic, magnonic, or atomic interactions. Across the literature, the phrase covers several non-equivalent regimes: true simultaneous multi-channel conversion, multimode conversion through several internal resonances, and channelization architectures that sequentially extract narrowband microwave information without destroying the original optical modulation. Recent work has made this distinction explicit by contrasting traveling-wave waveguide systems that support continuous phase matching and cascaded channelization with resonant cavity systems that remain fundamentally single-resonance limited on the microwave side [2509.10052], [2307.07875], [2005.12763].

## 1. Scope and classification

A useful classification, suggested by the published literature, separates multi-channel microwave-to-optics conversion into three categories. The first is **true simultaneous multi-channel conversion**, in which several independent microwave channels are converted in parallel within one device. The clearest reported example is a traveling-wave thin-film lithium niobate converter that demonstrates simultaneous operation of nine conversion channels in a single device, with an optical bandwidth exceeding 40 nm and a microwave operating bandwidth of about 250 MHz [2509.10052]. Room-temperature \(^{87}\)Rb vapor has also been used for simultaneous conversion of a multi-channel input microwave field to corresponding optical channels under frequency-division multiplexing control [2305.19221].

The second category is **multimode conversion**, in which multiple internal resonances participate coherently, but the system is not organized as a multiplexed parallel converter. A representative case is the YIG magnon-cavity hybrid in which the Kittel mode and higher-order magnetostatic modes both mediate microwave-to-optical conversion, with the YIG sphere size and position controlling whether the response is Kittel-dominated or genuinely multi-mode [2007.15299]. Integrated waveguide cavity optomagnonics in YIG likewise uses several forward volume magnetostatic standing-wave modes to create a broadened multimode conversion response rather than isolated parallel channels [2005.06429].

The third category is **channelization or non-destructive tap architectures**, in which multiple resonant sections sequentially interrogate a broadband microwave photonic signal without erasing its intensity envelope. In an electrically interfaced Brillouin-active AlN-on-silicon waveguide, cascaded electro-optomechanical segments with distinct resonance frequencies implement a multi-channel microwave photonic filter/channelizer, but the immediate experimental readout is optical-to-microwave rather than a direct simultaneous set of microwave-to-optical outputs [2307.07875].

| Regime | Representative work | Reported status |
|---|---|---|
| Simultaneous multi-channel conversion | [2509.10052], [2305.19221] | Nine channels in one device; simultaneous multi-channel atomic FDM |
| Multimode conversion | [2007.15299], [2005.06429] | Several internal magnon modes contribute coherently |
| Channelization / tap architectures | [2307.07875] | Cascaded resonant segments extract narrowband channels |

A recurrent misconception is that any device with several resonances is already a multi-channel converter. The cavity electro-optic lithium-niobate whispering-gallery transducer, for example, is explicitly described as a single microwave mode plus single optical pump/signal mode converter, despite its few intrinsic resonances and sideband-selective mode structure; it does not demonstrate multiplexing across multiple microwave modes or parallel multi-mode optical readout [2005.12763].

## 2. Physical mechanisms and transduction formalisms

The field spans several interaction mechanisms. In traveling-wave acousto-optic systems, microwave drive excites phonons via piezoelectric IDTs and the phonons scatter guided light through Brillouin interactions. In the AlN-on-silicon Brillouin-active waveguide, the relevant process is forward Brillouin scattering in an intramodal configuration: an RF drive excites an acoustic mode near \(\Omega_0/(2\pi)\sim 3.63~\mathrm{GHz}\), and the resulting phonon field phase-modulates the optical carrier, generating sidebands \(\omega_n=\omega_0+n\Omega_0\) [2307.07875]. The modulation depth is written as
\[
\beta = \frac{2g|b|L}{v_g},
\]
and the sidebands follow
\[
\tilde{a}_n(L,t)=i^n a_0 J_n(-\beta)e^{-in\Omega_0 t}.
\]
Because the process is phase modulation rather than amplitude modulation, the optical intensity envelope is preserved [2307.07875].

The thin-film lithium-niobate traveling-wave converter uses a different Brillouin geometry: microwaves are coupled into traveling phonons by an IDT, and the phonons convert to optical photons through backward Brillouin scattering in a hybrid photonic-phononic waveguide [2509.10052]. The phase-matching relations are
\[
k_{\mathrm{s}}=\beta-k_{\mathrm{p}}, \qquad \omega_{\mathrm{s}}=\omega_{\mathrm{p}}+\Omega,
\]
with mismatch
\[
\Delta\beta = \beta - k_{\mathrm{s}} - k_{\mathrm{p}}.
\]
Its internal efficiency is modeled by
\[
\eta_{\text{int}} \approx e^{-\alpha_b L} g^2 L^2 P \left|\mathrm{sinc}\left(\Delta\beta L/2\right)\right|^2
\]
in the small-signal limit [2509.10052].

Resonant electro-optic systems instead rely on Pockels three-wave mixing among a microwave mode, an optical pump mode, and an optical signal mode. In the lithium-niobate whispering-gallery transducer, the interaction Hamiltonian is
\[
\hat{H}_\text{int}=\hbar g (\hat{a}_e\hat{a}_p\hat{a}_o^\dagger+\hat{a}_e^\dagger\hat{a}_p^\dagger\hat{a}_o),
\]
which becomes a beam-splitter interaction under strong optical pumping [2005.12763]. Mechanically mediated electro-opto-mechanical systems use two beam-splitter-like interactions, microwave-to-mechanics and optics-to-mechanics, with the mechanics as the intermediary [1310.5276], [1812.07588], [2406.02704].

Magnonic converters use collective spin excitations rather than phonons. In YIG-based devices, microwave photons hybridize with Kittel or magnetostatic magnon modes, while the optical field couples through the Faraday effect or cavity optomagnonic interactions [1601.03908], [2007.15299], [2005.06429]. Atomic converters use nonlinear wave mixing in vapor media. In room-temperature \(^{87}\)Rb, the generated optical signal obeys \(\omega_S=\omega_P+\omega_I\), and the large Doppler width supports channel tunability and simultaneous conversion of multichannel microwave input fields [2305.19221].

## 3. Traveling-wave and continuum-mode architectures

Traveling-wave architectures are fundamentally different from cavity transducers because they replace discrete optical resonances with a continuum of optical modes. The AlN-on-silicon membrane device of [2307.07875] is a suspended membrane containing a central single-mode silicon ridge waveguide for telecom light, surrounded by a phononic crystal that confines a high-\(Q\) acoustic mode and flanked by slim interdigitated transducers that piezoelectrically drive the phonons. The acousto-optic interaction region is only \(160~\mu\mathrm{m}\), so the phase-matching mismatch remains small, \((\Delta q L\ll 1)\), and the optical interaction is intrinsically broadband in wavelength because the guide is not an optical cavity [2307.07875].

Experimentally, this platform produced a maximum first-sideband modulation efficiency of \(-16.32\) dB at \(3.63\) GHz for an incident microwave power of \(8.92\) dBm, a half-wave voltage \(V_\pi = 8.13 \pm 0.48~\mathrm{V}\), and \(V_\pi L = 0.1301 \pm 0.0077~\mathrm{V\cdot cm}\). The optical phase-modulation bandwidth exceeded 100 nm with less than 1 dB variation over that span. In reverse operation, the peak optical-to-microwave conversion efficiency was \(-59.24 \pm 0.24\) dB at an optical pump power of \(16.12\) dBm, while the peak microwave-to-optical conversion efficiency was \(-54.16 \pm 0.24\) dB at \(20.61\) dBm [2307.07875].

Its channelizer exploits cascaded electro-optomechanical waveguide segments with distinct acoustic resonance frequencies \(\Omega_n\) and linewidths \(\Gamma_n\), tuned geometrically through membrane width \(W\) and the number of IDT teeth. The cascaded optical field is written as
\[
a_\mathrm{out}=a_\mathrm{in}\exp\!\left[-i\sum_n \beta_n \cos(\Omega_n t)\right].
\]
Measured microwave channel center frequencies were \(3.746\) GHz, \(3.654\) GHz, and \(3.631\) GHz for membrane widths of \(16.15~\mu\mathrm{m}\), \(15.43~\mu\mathrm{m}\), and \(14.70~\mu\mathrm{m}\), with RF channel bandwidths from 7 MHz to 13 MHz [2307.07875]. The authors state that low optical propagation loss would in principle allow cascading up to 100 channels with acceptable phase mismatch and minimal total loss [2307.07875].

The later thin-film lithium-niobate hybrid photonic-phononic waveguide extends the traveling-wave concept to direct simultaneous multi-channel microwave-to-optics conversion [2509.10052]. The device is a straight TFLN waveguide of length \(1.28~\mathrm{mm}\), width \(1.2~\mu\mathrm{m}\), and film thickness \(220~\mathrm{nm}\), with a fan-shaped IDT of bandwidth 225 MHz launching a quasi-Rayleigh acoustic mode. It converts about 9 GHz microwaves to 1550 nm telecom light with 2.2% internal efficiency and \(2.4\times 10^{-4}\) system efficiency at room temperature, and it supports nine simultaneous conversion channels. Dual-channel operation used 100 GHz DWDM spacing, while the nine-channel demonstration used 50 GHz comb spacing and nine RF tones spaced by 4.8 MHz [2509.10052].

These two waveguide systems illustrate different consequences of traveling-wave physics. Forward Brillouin scattering yields amplitude-preserving phase modulation and therefore non-destructive taps and channelizers [2307.07875]. Backward Brillouin scattering with continuous phase matching yields wavelength-multiplexed parallel conversion channels in a single straight waveguide [2509.10052].

## 4. Resonant cavity and mechanically mediated converters

Resonant cavity converters remain central because they set many of the efficiency, bandwidth, and added-noise benchmarks, even when they are not multi-channel. The 2013 membrane-based electro-opto-mechanical converter used a superconducting LC resonator at about 7.1 GHz, a Fabry–Perot optical cavity at about 282 THz, and a stoichiometric Si\(_3\)N\(_4\) membrane as the shared intermediary. It demonstrated reversible, coherent, bidirectional conversion with a measured maximum apparent efficiency of \(0.178 \pm 0.011\), an inferred gain-free conversion efficiency of \(0.086 \pm 0.007\), and a matched conversion bandwidth of 30–32 kHz, but its several mechanical resonances were alternative conversion modes rather than multiplexed simultaneous channels [1310.5276].

Quantum-noise-limited operation has subsequently been pursued in several cavity and cavity-like platforms. A GaAs electro-opto-mechanical nanobeam initialized near its quantum ground state achieved \(n_\text{th}=0.90 \pm 0.01\), optomechanical cooperativity \(C\approx 1.7>1\), and verified phase preservation through the full RF-to-mechanical-to-optical chain, but the electromechanical loading stage remained the efficiency bottleneck [1812.07588]. Thin-film GaP demonstrated pulsed microwave-to-optics conversion with \(n_{\mathrm{th}}=0.55 \pm 0.05\) for 40 fJ optical pulses, pulsed cooperativity \(C=1.74\), and total conversion efficiency \(6.8\times 10^{-8}\), again as a single-channel device [2107.04433]. A continuous silicon electro-optomechanical platform later reported external efficiency \(2.2\%\) with input-referred added noise \(0.94\) and bandwidth \(88.9 \pm 2.1\) kHz in microwave-to-optics conversion, or \(0.47\%\) efficiency with added noise \(0.58\) at the lowest-noise operating point [2406.02704].

Electro-optic cavity converters show similar narrowband selectivity. The lithium-niobate whispering-gallery converter operated in a millikelvin environment with microwave mode occupancy \(0.025 \pm 0.005\) noise photons at low pump power, reached maximum total bidirectional conversion efficiency \(\eta_\text{tot}=3.16\times10^{-4}\), and had bandwidth rising to 10.68 MHz at highest pump power; yet it explicitly did not demonstrate multi-channel conversion [2005.12763]. The pulsed AlN microring electro-optic transducer reached highest observed conversion efficiency \(2.4\times10^{-5}\) at 4.6 dBm pulsed drive while maintaining \(\bar{n}_\mathrm{e}=0.09\pm0.06\), but it likewise remained a single microwave mode to single optical signal mode platform [2010.11392].

Taken together, these resonant devices establish that high efficiency, low added noise, and ground-state operation have often been achieved first in single-channel settings. This suggests that, in the present literature, multi-channel capacity and quantum-grade noise performance have usually been optimized separately rather than simultaneously.

## 5. Magnonic and atomic routes to multimode and frequency-division multiplexing

Magnonic systems provide a distinct route in which multiple spin-wave modes define the channel structure. In a 3D cavity containing a YIG sphere, coherent conversion from microwave to optical wave was extended from the Kittel mode to higher-order magnetostatic modes. For a 0.45-mm sphere, the system behaved mainly as a Kittel transducer with cavity frequency \(\omega_c/2\pi = 10.632~\text{GHz}\), Kittel coupling \(g_K/2\pi=28.6~\text{MHz}\), and cooperativity \(\mathcal{C}_K=132\), with resonant conversion efficiency \(\eta_t \approx 8.45\times 10^{-11}\). Larger spheres activated additional magnetostatic modes, but the resonant efficiencies decreased to \(5.12\times 10^{-12}\) for 0.75 mm and \(3.46\times 10^{-12}\) for 1.0 mm, reflecting the weak magnon-optical coupling [2007.15299]. By shifting the sphere into a less uniform cavity field, the higher-order magnetostatic-mode coupling increased to \(g_M/2\pi=25~\text{MHz}\), and the total resonant conversion efficiency reached \(1.02\times 10^{-11}\) [2007.15299].

Integrated cavity optomagnonics in a YIG rib waveguide compresses the mode volume and uses several forward volume magnetostatic standing-wave modes. The device showed up to four magnon resonances, magnon free spectral range of about 7 MHz, conversion bandwidth broader than 50 MHz in the raw response, FWHM conversion bandwidth of 16.1 MHz at 2.5 dBm pump power, and highest on-chip efficiency \(\eta \approx 1.08\times 10^{-8}\) at 8.451 GHz and 4.8 dBm pump power [2005.06429]. The paper attributes its performance to a magneto-optical cooperativity about three orders of magnitude higher than earlier YIG-sphere implementations [2005.06429].

Atomic transducers address multi-channel conversion differently. In room-temperature \(^{87}\)Rb vapor within a microwave cavity, coherent microwave-to-optical conversion mapped a microwave signal to a tunable \(550(30)\) MHz range of optical frequencies, and the Doppler width supported simultaneous conversion of a multi-channel input microwave field to corresponding optical channels [2305.19221]. The same work demonstrated phase-correlated amplitude control of select channels, complete extinction of one channel, and 97% visibility in sideband interference. Its maximum conversion efficiency was \(\eta_0 = 1.5(1)\times 10^{-9}\), and the instantaneous conversion bandwidth was \(910(20)\) kHz, distinct from the much wider tunable optical output range [2305.19221].

A related warm-vapor Rydberg six-wave-mixing converter provided continuous-wave conversion of a 13.9 GHz field to a 776 nm optical signal with 57 dB dynamic range and 16 MHz conversion bandwidth, and it demonstrated simultaneous conversion of two near-optimal microwave tones with coherence \(\mathcal{C}=0.88\), but it did not present a multiplexed multi-channel architecture [2302.08380].

## 6. Metrics, bottlenecks, and scaling directions

The central trade-off in this field is between spectral multiplicity and the traditional transducer figures of merit: efficiency, bandwidth, added noise, and thermal compatibility. Traveling-wave systems offer wide optical acceptance and scalable frequency-division architectures, but current efficiencies can remain modest. In the AlN-on-silicon Brillouin waveguide, the microwave extinction \(\eta_e = 1-|S_{11}(\Omega)|^2\) is only about 2% because of large impedance mismatch between the compact IDT and the 50 \(\Omega\) source, and the measured Brillouin gain is about \(4\times\) smaller than COMSOL simulation because of fabrication imperfections, suboptimal acousto-optic overlap, and reduced photoelastic constants in strained silicon [2307.07875]. In the TFLN traveling-wave converter, acoustic propagation loss \(\alpha_b \approx 2.9\times 10^3~\mathrm{m^{-1}}\), phonon \(Q \sim 2.8\times 10^3\), IDT efficiency \(\eta_{\mathrm{IDT}}=3.1\%\), and optical side-coupling efficiency \(\eta_c = 35\%\) dominate the present performance, while 50 GHz comb spacing in the nine-channel experiment produces adjacent-channel crosstalk because it is narrower than the individual optical bandwidth [2509.10052].

Resonant cavity converters, by contrast, often achieve stronger single-channel conversion or lower noise at the price of narrow bandwidth and limited multiplexing. The lithium-niobate whispering-gallery transducer is microwave-side single-resonance limited, uses only one resonant optical pump/signal pair separated by a fixed free spectral range, and suffers strong heating: at the highest pump power, the dilution refrigerator mixing chamber warms from 7 mK to 320 mK and the superconducting cavity undergoes a superconducting-to-normal transition around \(P_p\approx 0.7\) mW [2005.12763]. The pulsed AlN microring transducer identifies superconductor absorption of stray light scattered from the chip-fiber interface as the dominant fast bath under intense optical drive [2010.11392]. These results show that scaling a narrowband quantum transducer into a multi-channel device is not only a spectral-engineering problem but also a thermal and packaging problem.

Several works identify concrete routes forward. The AlN-on-silicon Brillouin waveguide notes that stronger piezoelectrics such as LiNbO\(_3\) or AlScN, improved impedance matching, and longer interaction lengths could substantially improve performance [2307.07875]. The TFLN traveling-wave converter estimates that with phonon \(Q\) improved to \(10^6\), internal efficiency could exceed 50%, only 15 mW peak pump power might be needed, and more than 70 channels could be supported in one device [2509.10052]. A theoretical route to bandwidth scaling in cavity arrays is the spatially adiabatic optoelectromechanical array, for which the conversion bandwidth can exceed the cavity linewidth and, for symmetric arrays, scales as
\[
\Delta\omega = \left(\frac{4\sqrt{2}}{3} g^2\kappa N\right)^{1/3},
\]
with collective cooperativity \(CN\) governing thermal-noise suppression [1707.03339].

The broader record therefore suggests a convergence of previously separate agendas. Wide optical bandwidth, high power handling, non-destructive channelization, and simultaneous channel multiplicity have emerged most clearly in traveling-wave and atomic systems [2307.07875], [2509.10052], [2305.19221]. Quantum-enabled low-noise operation has been established most clearly in resonant electro-optomechanical and electro-optic systems [1812.07588], [2107.04433], [2406.02704]. Multi-channel microwave-to-optics conversion, in the strongest sense, is now best understood as the effort to combine those two trajectories within architectures that preserve coherence, manage thermal load, and scale in channel count without reverting to one cavity per frequency channel.

Source: https://www.emergentmind.com/topics/multi-channel-microwave-to-optics-conversion