Skip to content

Optical probing of phononic properties of a tin-vacancy color center in diamond

FieldValue
JournalPhysical Review Research
Published2026-04-02
AuthorsCem Güney Torun, Joseph H. D. Munns, Franziska Marie Herrmann, Viviana Villafane, Kai Müller
DOI10.1103/swkr-r6mb

A frequency-domain coherent population trapping measurement determined a 31(5) ps orbital relaxation time for a single tin-vacancy center at 4 K. The same analysis gave a 958(138) ns orbital excitation time. An independent D-transition linewidth analysis returned 30.6(2) ps. The paper therefore resolves orbital dynamics that cannot be timed directly with standard photodetectors.

Emitter E1 supported the temperature-dependent spectroscopy in IIa-type (100) electronic-grade single-crystal CVD diamond. Tin implantation used 35 MeV (^{120}\mathrm{Sn}) ions at a nominal fluence of (10^9) ions/cm², with a simulated depth of 4.35 μm. A 2100 °C, 7.9 GPa treatment for about 20 min followed. A 4.35 μm-radius solid immersion lens was then FIB-milled around E1.

After optical power broadening and a phononic linewidth component were subtracted, E1’s low-temperature linewidth was 37.3(9) MHz. The lifetime-limited linewidth was 32.0(8) MHz, and a separate decay fit gave 5.0(2) ns. A tenfold higher-dose area showed a 7.5(3.0) GHz FWHM spread in emitter central frequencies.

Emitter E2 supported CPT in electronic-grade single-crystal CVD diamond with {100} faces. Implantation used 400 keV tin ions at (5 \times 10^{10}) atoms/cm² and a simulated depth of about 100 nm. Annealing at 1050 °C for about 12 h at (P < 7.5 \times 10^{-8}) mbar preceded fabrication of a 240 nm-diameter nanopillar.

Linewidth spectroscopy locates the phonon regimes

Section titled “Linewidth spectroscopy locates the phonon regimes”

Photoluminescence excitation spectra covered 4.5 to 40 K by scanning a resonant laser and counting phonon-sideband photons. Data from 4.5 to 32 K used Voigt fits, while the 36 and 40 K points used Lorentzian fits. Higher-temperature points came from PL under nonresonant excitation. The analysis subtracted the 32.0(8) MHz natural linewidth and corrected measured power broadening.

Adding higher-temperature points to the C-transition fit gave the best fit at 24 K, where (R^2) was 0.98. Below this cutoff, direct single-phonon transitions dominate. Above it, two-phonon dephasing produces the observed cubic trend.

The extracted C-transition coefficient was (\alpha_+=(2\pi)^{-3}\times3.8(3)\times10^{-9}\ \mathrm{GHz}^{-2}). The D-transition coefficient was (\alpha_-=(2\pi)^{-3}\times7.0(6)\times10^{-9}\ \mathrm{GHz}^{-2}). The two values differ by about a factor of two. Reported high-temperature cubic fits gave (A_+=2.52(8)\times10^{-4}\ \mathrm{GHz/K^3}) and (A_-=3.8(1)\times10^{-4}\ \mathrm{GHz/K^3}).

The same spectroscopic data gave zero-phonon-line wavelengths of 619.058 nm for the C transition and 620.111 nm for the D transition. The D transition had been difficult to observe because its lower orbital state has a very low thermal population.

A diamond thermal-expansion model did not reproduce the measured wavelength shifts. The phonon model instead gave (A_C=4.2(7)\times10^{-6}\ \mathrm{nm/K^2}) and (B_C=2(6)\times10^{-11}\ \mathrm{nm/K^4}). For the D transition, (A_D=1.4(6)\times10^{-6}\ \mathrm{nm/K^2}) and (B_D=0(6)\times10^{-11}\ \mathrm{nm/K^4}). The quadratic term was larger. The authors linked this behavior to E-symmetric phonons below 110 K.

Two resonant lasers formed a lambda scheme through the shared excited state. At 4 K, the C laser stayed on resonance while the D laser scanned its transition. Blue light at 450 nm reinitialized the charge state during each scan. The fluorescence dip records coherence between the two orbital ground states.

Nine power configurations were measured in both scan directions, producing 18 datasets. Two low-power datasets were excluded because their signals were too small, leaving 16 fits. Each profile was fitted by numerically integrating a Lindblad master equation until steady state. A 1 μs integration interval was sufficient.

The fit varied the C and D Rabi frequencies and the orbital relaxation rate (\gamma_-). The upward rate (\gamma_+) was tied to (\gamma_-) by the Boltzmann relation at 3.86 K and a ground-state splitting of 831 GHz. Pure dephasing (\gamma_d) was fixed at zero because single-phonon processes dominate below 24 K.

Across the accepted fits, the mean orbital excitation time was 958(138) ns. The mean orbital relaxation time was 31(5) ps. The spectral dip was transformed into its conjugate time variable, giving picosecond resolution without timing the event directly.

The upper orbital branch contained less than 0.1% of the population at 4 K. Direct D-transition excitation required 800 nW, about 8,000 times the roughly 100 pW used for the C transition. That scan gave a total D-transition FWHM of 5.406(43) GHz. Subtracting lifetime, power, and estimated charge-noise contributions yielded a D-transition relaxation time of 30.6(2) ps. Agreement with the CPT result supports the 31(5) ps value.

The extracted D-transition coefficient agrees with the previously reported value ((2\pi)^{-3}\times7.51\times10^{-9}\ \mathrm{GHz}^{-2}). The 31 ps relaxation is about ten times shorter than the electronic jitter of standard avalanche photodiodes. It is also at the resolution limit of state-of-the-art superconducting nanowire detectors.

The factor-of-two difference between (\alpha_+) and (\alpha_-) remains unexplained. The C-transition cubic model misses data between 24 and 50 K, so the appendix selected a combined (T^3) and (T^5) form instead. Model sensitivity placed the visible effect of pure dephasing at 149(84) ps.

The 31 ps value was inferred from CPT rather than timed directly. Laser power drifts of 10% to 50% over roughly 3 h prevented direct power-to-Rabi calibration. The dye laser also drifted by up to ±40% in 30 min. Sample motion reached about 100 nm per hour, while fluorescence fell by about 10% per hour.

The CPT result also depends on fixing pure dephasing at zero and deriving the upward rate from the Boltzmann relation. The D-linewidth estimate assumes that both transitions couple to external charge noise in the same way.

The 31 ps lifetime of the upper orbital branch makes that state unsuitable as a practical qubit. The lower spin branch remains relevant because a closed cycle of spin-preserving orbital transitions limits (T_2^*) to 958 ns at 4 K when phonons dominate. This bound assumes that spin-bath and gate-error contributions are smaller.

Using the single-phonon relation, the authors estimated an orbital excitation time of 91(8) ms at 1.8 K. This is about one order of magnitude above the highest recorded SnV spin coherence time of 10 ms. They therefore conclude that phonons will not limit spin coherence at 1.8 K or below.

The weak D transition favors detuned CPT over resonant probing, particularly when charge-state stability is poor. On (100) material, the C and D dipoles are orthogonal, so optical polarization and crystal orientation require careful control. Material selection should also preserve narrow linewidths and low inhomogeneous spectral spreading for photon applications.

For this SnV photonics application, the relevant catalog option is 6CCVD’s MPCVD-grown single-crystal diamond. The paper requires a {100}-terminated substrate for tin implantation, annealing, and 180 to 340 nm nanopillar fabrication, while the catalog lists plates and wafers up to 125 mm with thickness from 0.1 to 500 µm. Confirm orientation and process requirements through the diamond materials catalog, and use the engineering support page for custom fabrication support.

View Original Abstract

The coherence characteristics of a tin-vacancy (SnV) color center in diamond are investigated through optical means, including linewidth broadening effects and coherent population trapping (CPT) between the ground-state orbital levels. Spectral analysis is required as due to the large spin-orbit splitting of the orbital ground states, thermalization between the ground states occurs at rates that are impractical to measure directly in the time domain. First, by implementing a temperature-dependent linewidth broadening measurement, including the challenging-to-measure D transition, phononic coupling coefficients are determined. These measurements are performed on an emitter with a lifetime-limited linewidth and atomlike properties, making the measurement representative for high-quality SnVs. Next, a CPT-type experiment is carried out to independently analyze thermal decoherence processes at 4 K. The spectral information is transformed into its conjugate variable time, providing picosecond resolution and revealing an orbital depolarization timescale of <a:math xmlns:a=“http://www.w3.org/1998/Math/MathML”> <a:mrow> <a:mo>∼</a:mo> <a:mn>30</a:mn> <a:mrow> <a:mspace width=“0.28em”/> <a:mi>ps</a:mi> </a:mrow> </a:mrow> </a:math> . Consequences of the investigated dynamics are then used to estimate spin dephasing times limited by thermal effects.