Skip to content

Photorefractive tuning seeded by third-harmonic light in a diamond photonic crystal cavity

FieldValue
JournalOptica
Published2026-08-14
AuthorsJoe Itoi, Elham Zohari, Nicholas J. Sorensen, Sean McNaney, Waleed El-Sayed
DOI10.1364/optica.607145
Citations1

A diamond photonic crystal nanocavity was tuned in place by 20.2(2) GHz, a blue shift that exceeds the cold-cavity linewidth of 17.8(2) GHz. The tuning is seeded by the cavity’s own third-harmonic output at 522 nm, and it corresponds to a fractional refractive index change of -1.05(1) × 10^-4. When the pump power is reduced, the shifted resonance drifts back toward its cold frequency over several tens of hours. The Q-factor and resonance contrast stay unchanged during both the shift and the relaxation.

The cavity was etched from ‘optical grade’ bulk single-crystal diamond using a previously reported quasi-isotropic etching process, with an improved Si-rich SiNx hard mask for better sidewall protection. The device is a suspended nanobeam about 550 nm wide and 30 µm long, perforated by an array of air holes whose diameter and spacing are symmetrically tapered to form a localized mode near 1550 nm. Finite element simulations in COMSOL give a fundamental TE-like mode with an effective mode volume of 0.13 × (λ0/n0)^3 and a diamond confinement factor of 0.84.

Light enters through a fibre taper parked on support beams patterned beside the cavity, which keeps the fibre-to-device separation at roughly 300 to 500 nm and the coupling constant over the measurement. The cavity mode extends about 4 µm along the beam, and the taper transmission was measured at 23% at 1566 nm and 15% at 532 nm. The fibre link between taper and spectrometer transmitted 0.0035 of the green light.

Transmission was measured with a widely tunable continuous-wave laser through the taper and fitted with coupled-mode theory that includes Fano interference. The cold-cavity parameters are listed below.

ParameterValueNotes
Resonance wavelength λ01565.98(1) nmlow input power, taper coupled
Linewidth δλ00.146(2) nmδω0/2π = 17.8(2) GHz
Loaded Q1.07(1) × 10^4Lorentzian fit
Intrinsic Q3.47(9) × 10^4intrinsic plus parasitic loss
Coupling Q3.11(3) × 10^4fibre-taper coupling
Effective mode volume0.13 × (λ0/n0)^3finite element simulation
Q over mode volume10^5 × (λ0/n0)^-3figure of merit

Input power was capped at 75 mW because the fibre taper, not the diamond, is the power-sensitive element. At that input the intracavity photon number reaches 3.5 × 10^6, and the estimated peak intracavity intensity is 56 GW/cm². The authors note this is the largest field intensity reported in a diamond nanophotonic cavity driven by a continuous-wave laser.

Third-harmonic generation from a continuous-wave pump

Section titled “Third-harmonic generation from a continuous-wave pump”

At high power the cavity line shape becomes asymmetric because laser heating red-shifts the resonance through the thermo-optic effect. At an input of about 70 mW the thermo-optic shift is -9.1(2) GHz, which the model converts into a cavity temperature rise of roughly 12 K. The line shape shows no significant hysteresis between forward and backward sweeps, so thermal bistability was not reached at this power.

Three telecom infrared photons are converted into one green photon at 522 nm, visible on a CCD camera when the pump sits near resonance. The green emission scales with the cube of the input power on a log-log plot, which confirms a third-order process. The resulting end-to-end conversion efficiency is 6 · 10^-7 W^-2. This figure is not corrected for poor coupling between the third-harmonic light and the taper’s guided mode.

The output at the taper was about 250 pW, which corresponds to 0.65 nW in the taper immediately after the cavity. The circulating green power inside the device could only be bounded, between 2.8 nW and 484 nW, because the coupling rates at the third-harmonic wavelength were never measured directly. The cavity supports no confined mode at 522 nm, so the green field behaves as a leaky waveguide mode with a simulated Q of 270.

At the highest input powers a shoulder appears on the blue side of the resonance in both the pump transmission and the green spectra. The authors attribute this to optomechanical self-oscillation near the 10 GHz mechanical resonance. Noisy oscillations in the green spectra are traced to etaloning in a beam splitter built for telecom wavelengths.

Separating photorefraction from thermal damage

Section titled “Separating photorefraction from thermal damage”

The blue shift appeared while the third-harmonic measurements were running. The authors measured the cavity resonance over long periods by parking the laser on resonance for five minutes, then taking a fast transmission scan, and repeating this sequence fifty times over roughly 4.5 hours. The total shift reached 20.2(2) GHz, about 4 hours into the sequence, and it exceeds the cold linewidth of 17.8(2) GHz.

The resonance contrast and the line shape stayed constant across the whole series. A shift caused by movement of the fibre taper would also change the linewidth and contrast, so the observed movement is dispersive rather than mechanical. The green output power did not change during this measurement.

Several permanent mechanisms are excluded. The intensity needed to create single vacancies in diamond is three orders of magnitude above the peak field in this cavity. Graphitisation from prolonged heating is ruled out because the 12 K temperature rise is far below what that process requires, and surface graphite would red-shift the resonance and degrade the Q-factor. The fact that the change relaxes shows the cavity returns toward its initial state.

Diamond is centrosymmetric, so its intrinsic second-order susceptibility vanishes. The authors argue that static electric fields from charged defects couple to the bulk third-order susceptibility and produce an effective second-order response, written as 3χ(3)E_DC. This effective response makes the electro-optic effect possible. The supplementary material reports second-harmonic generation near 782 to 783 nm at an input of 51 mW, which supports the same argument.

A two-timescale shift and a slow relaxation

Section titled “A two-timescale shift and a slow relaxation”

The accumulated shift follows a double-exponential form, with a fast and a slow component. The fast and slow contributions are attributed to charge traps near the surface and in the bulk, as has been observed in lithium niobate.

Fit parameterValueRole
Fast shift amplitude C_f/2π8(1) GHzblue-shift component
Fast shift rate Γ_f5(1) hours^-1blue-shift component
Slow shift amplitude C_s/2π18.2(6) GHzblue-shift component
Slow shift rate Γ_s0.48(7) hours^-1blue-shift component
Fast relaxation amplitude C_f(R)/2π5.5(2) GHzred-shift component
Fast relaxation rate γ_f4.3(4) hours^-1red-shift component
Slow relaxation amplitude C_s(R)/2π4.1(1) GHzred-shift component
Slow relaxation rate γ_s0.177(7) hours^-1red-shift component

Relaxation was measured with the laser at about 0.8 mW, scanned across the resonance every five minutes for 36 hours. The resonance red-shifted by 10.2(2) GHz toward the cold frequency but did not return fully after 40 hours. The residual shift is attributed to a leftover space-charge field. Charges that drift out of the mode region cannot be recovered, and electrons trapped in long-lived surface states or deep-level defects may never be released.

Recovering the initial charge environment may require surface passivation, illumination with a higher-energy laser, or thermal annealing, possibly in combination. The parameters of the shift are expected to depend on intracavity power, which was not tested systematically here.

Green light drives the charge redistribution

Section titled “Green light drives the charge redistribution”

A control experiment isolates the role of green light. The authors injected a 520 nm laser at about 3 mW into the taper alongside the high-power infrared pump. The blue shift accelerated whenever the green laser was on, and slowed to a much lower rate when it was switched off. Reapplying the green laser accelerated the shift again. The authors conclude that green light is required for photorefraction in diamond.

The optical grade material contains substitutional nitrogen at about 1 ppm and nitrogen-vacancy centres at about 0.01 ppm. The photoionisation threshold for substitutional nitrogen is 2.2 eV, and one third-harmonic photon carries 2.38 eV, so a single green photon can ionise it. For the NV centre, one green photon populates the excited state 1.945 eV above the ground state, and photoionisation from there can proceed with one green photon or two infrared photons at 0.79 eV each. Direct ionisation across the 5.5 eV bandgap would need three green photons and is considered unlikely.

Freed electrons drift toward regions of low intensity and accumulate there, building the space-charge field that modulates the index through the electro-optic effect. The authors did not observe third-harmonic-induced NV photoluminescence through the taper, which they attribute to two-photon infrared ionisation trapping the population in a dark, non-fluorescent state.

The coupling rates at the third-harmonic wavelength were never measured independently, because no tunable laser was available near 522 nm. The internal green field intensity is therefore unknown, and the ionisation rates that govern charge dynamics cannot be quantified. The effect of the strong infrared field on the photophysics of substitutional nitrogen is also unknown.

All experiments used one photonic crystal cavity, and the incomplete relaxation prevents exactly repeatable measurements. A systematic study across surface terminations was left outside the scope of the work. Measurements were carried out at ambient conditions, where convection dominates heat dissipation.

Consequences for diamond device engineering

Section titled “Consequences for diamond device engineering”

Heating and gas condensation both red-shift a cavity and act on every device on a chip. Photorefractive tuning here shifts in the blue direction, leaves the Q-factor intact, and can be applied to individual cavities on the same chip. An alternating high-power and low-power sequence of 78 mW and 27 mW scans over 18 hours produced blue shifts at high power and red shifts at low power, which the authors present as deterministic in situ tuning.

That control matters for aligning a cavity with a colour centre transition, for Purcell-enhanced coupling to single emitters, and for tuning onto the 1042 nm singlet transition of the NV centre for absorption-based magnetometry. It also opens the way to doubly resonant cavities for four-wave mixing, Brillouin scattering, and Raman scattering.

The green output is itself a result. This is the first demonstration of green light generation in a diamond nanocavity, and the first demonstration of integrated nonlinear optics in the visible in diamond. The conversion efficiency is limited by the modest Q-factor compared with silicon nitride or lithium niobate, yet diamond’s wide bandgap and thermal conductivity allow large field intensities with limited absorption or thermal damage. Because the third-order nonlinearity scales inversely with NV density, ultrapure diamond gives the best conversion efficiency, while NV-rich material offers the photorefractive response and the photoluminescence needed for infrared-excited magnetometry.

The cavity was etched from optical grade bulk single-crystal diamond with substitutional nitrogen near 1 ppm and NV concentration near 0.01 ppm, so the starting stock is a single-crystal plate rather than a polycrystal. The 6CCVD diamond materials catalog lists single-crystal and polycrystalline diamond grown by MPCVD as plates and wafers up to 125 mm, with single-crystal material polished to Ra below 1 nm, which suits the quasi-isotropic undercut etch that produced this suspended nanobeam. Geometry and defect-density requirements for colour-centre work can be raised through engineering support.

View Original Abstract

Single-crystal diamond nanocavities have tremendous potential for use in quantum and nonlinear optical technologies. The ability to precisely control their resonant frequencies is essential for many applications, and in situ tuning is particularly desirable. In this work, we demonstrate deterministic resonance tuning of a diamond nanocavity. We observed a photorefractive effect in concert with the generation of third-harmonic light within the device. This effect blue-shifted the cavity resonance frequency by 20.2 (2) GHz, exceeding the cavity linewidth. The shift corresponded to a fractional change in refractive index of −1.05(1)×10 −4 , and its relaxation occurred over several tens of hours. Although photorefraction is a second-order nonlinear effect and has previously not been observed in diamond owing to its vanishing χ (2) , the observed behavior is consistent with the generation of non-zero χ (2) by electric fields from charged crystal defects. In addition to allowing tuning of diamond cavities for resonant nonlinear and quantum photonics applications, this observation could enable the realization of diamond frequency converters and electro-optical modulators that rely on second-order nonlinearity.