Sub-second spin and lifetime-limited optical coherences in ¹⁷¹Yb³⁺:CaWO₄

Abstract

Optically addressable solid-state spins have been extensively studied for quantum technologies, offering unique advantages for quantum computing, communication, and sensing. Advancing these applications is generally limited by finding materials that simultaneously provide lifetime-limited optical and long spin coherences. Here, we introduce ¹⁷¹Yb³⁺ ions doped into a CaWO₄ crystal. We perform high-resolution spectroscopy of the excited state, and demonstrate all-optical coherent control of the electron-nuclear spin ensemble. We find narrow inhomogeneous broadening of the optical transitions of 185 MHz and radiative-lifetime-limited coherence time up to 0.75 ms. Next to this, we measure a spin-transition ensemble line width of 5 kHz and electron-nuclear spin coherence time reaching 0.15 seconds at zero magnetic field between 50 mK and 1 K temperatures. These results demonstrate the potential of ¹⁷¹Yb³⁺:CaWO₄ as a low-noise platform for building quantum technologies with ensemble-based memories, microwave-to-optical transducers, and optically addressable single-ion spin qubits.

The authors identify ¹⁷¹Yb³⁺:CaWO₄ as a promising material for quantum technological applications, revealing lifetime-limited optical coherence of 0.75 ms and spin coherence of 0.15 s, highlighting its potential for quantum memories and transducers.

Introduction

Solid-state optically addressable spins^1–3 have been used to demonstrate a variety of applications, including quantum state storage^4–6, spin-spin entanglement⁷, control of nuclear spin clusters⁸, and quantum sensing⁹. Among the most popular platforms are optically active defects in diamond¹⁰, semiconductor quantum dots¹¹, vacancies in silicon¹², silicon carbide¹³ and hexagonal boron nitride¹⁴. Despite significant progress, the search for defects in a solid-state material simultaneously exhibiting lifetime-limited optical and long spin coherences remains ongoing.

Rare-earth (RE) ion-based defects are particularly attractive due to their unique electronic configuration, leading to narrow optical transitions and long coherence times in many different crystalline host materials^15–21. Ensembles of such defects have been used for quantum memories⁴. Ensembles of rare-earth ions have been extensively explored as quantum memories for photonic⁴ and microwave qubits²² and optical-to-microwave transduction^23,24. Single emitters coupled to integrated nanophotonic cavities^25–30 are actively explored as they enable single-shot spin readout^31–33, addressing of nearby nuclear spins^34–36, spin-photon entanglement³⁷, and entanglement between remote ions³⁸.

CaWO₄ has recently drawn a lot of attention as an appealing host material because of its low magnetic impurity background compared to conventional yttrium-containing hosts. The remaining magnetic background is dominated by the ¹⁸³W isotope (low natural abundance of 14.3%) and paramagnetic impurities such as iron and other RE ions³⁹. The non-polar defect site symmetry yields a first-order insensitivity to electric fields. These material properties combined have enabled the demonstration of spin coherence times up to 23 ms for small ensembles (<1 ppb) in Er³⁺:CaWO₄²⁰ as well as two-photon interference and spin-photon entanglement using a single Er³⁺ ion in CaWO₄ coupled to a heterogeneous photonic crystal cavity³⁷.

In this work, we investigate a previously unexplored host-ion combination: ¹⁷¹Yb³⁺ doped into CaWO₄. Our results establish ¹⁷¹Yb³⁺:CaWO₄ as a unique material platform possessing radiatively limited optical coherences and sub-second electron-nuclear spin coherence, even at appreciable doping concentrations. Among the trivalent RE species, ¹⁷¹Yb³⁺ is the only paramagnetic isotope to possess a nuclear spin of I = 1/2. The hybrid electron-nuclear spin states can be used as highly coherent states, while a low I ensures simple spin manipulation. Using high-resolution optical spectroscopy, we extract the so-far unknown excited state Hamiltonian parameters of ¹⁷¹Yb³⁺ in CaWO₄. Having complete knowledge of the energy levels and the optical transitions allows us to realize efficient all-optical spin initialization and control. In ensembles, we demonstrate optical coherence times reaching 0.75 ms, nearly limited by radiative decay, and realize all-optical spin control with spin coherence times up to 0.15(1) s. Coherence properties are achieved at zero magnetic field and with temperatures up to a few K. Compared to other solid-state emitters possessing electron spins, including defects in diamond, silicon and other RE ion-doped systems, our results represent a substantial advance in both optical and spin coherence times, even at elevated dopant concentrations.

Results

High-resolution spectroscopy of the ground and excited level structure

A CaWO₄ crystal doped with ¹⁷¹Yb³⁺ ions was grown using the Czochralski method with a nominal concentration of 20 ppm, see the Supplementary Information for more details. In this work, we study the tetragonal S₄ site in which the Yb³⁺ replaces a Ca²⁺ (Fig. 1(a)). This dopant’s optical transition connecting ²F_7/2(0)→²F_5/2(0) is centered around 973.16 nm (vac). The concentration of ions in the S₄ site was measured to be ≈5 ppm using electron paramagnetic resonance, with an isotopic purity of 96%. The high-resolution optical absorption spectrum at 4 K showed a series of narrow peaks spanning several GHz (Fig. 1(b)). The narrow optical inhomogeneous broadening of 185 MHz indicates the high quality of the material and low distortion of the crystalline site induced by dopant ions. This allows us to optically address each ground state level separately (Fig. 1(b)).

Fig. 1. Energy level diagram and magnetic field dependence of ¹⁷¹Yb³⁺:CaWO₄.

a CaWO₄ crystal structure, with a ¹⁷¹Yb³⁺ dopant replacing a Ca²⁺ ion at a site of S₄ point symmetry. b High-resolution absorption spectra of ¹⁷¹Yb³⁺:CaWO₄ at 4 K for light electric field polarized along the c axis (E∥c) or perpendicular (E⊥c) to the c axis. The labeling of the absorption peaks corresponds to the energy level diagram (c). Absorption peak C corresponds to Yb³⁺ isotopes with zero nuclear spin. Absorption peak D only appears for E⊥c. c Low-field energy level diagram for the ²F_7/2(0)→²F_5/2(0) transition of ¹⁷¹Yb³⁺:CaWO₄ at 973.1 nm. Energy splittings in the ground and excited states are determined using the ground state and extracted excited-state hyperfine tensors. The transitions corresponding to the observed absorption spectrum in (b) are shown with solid lines. d Absorption spectra of ¹⁷¹Yb³⁺:CaWO₄ for varying magnetic field strengths applied perpendicular to the crystalline c-axis. The polarization of the incident light is E⊥c. Dashed lines show the results of the fit to the model given by Eq. (1). Purple lines denote energy levels of the ¹⁷¹Yb³⁺ ions, and blue lines denote energy levels of the I = 0 isotopes. The color scale is linear.

¹⁷¹Yb³⁺ carries an electron spin S = 1/2 and nuclear spin I = 1/2, and both its ground and excited states spin levels can be described using the effective Hamiltonian

$${\mathcal{H}}^{g/e}=\mathbf{I}\cdot {\mathbf{A}}^{g/e}\cdot \mathbf{S}+{\mu }_{\mathrm{B}}\mathbf{B}\cdot {\mathbf{g}}^{g/e}\cdot \mathbf{S}-{\mu }_{\mathrm{n}}\mathbf{B}\cdot {\mathbf{g}}_{\mathrm{n}}\cdot \mathbf{I},$$ 1

where g^g/e, g_n and A^g/e are the coupling tensors of the electronic Zeeman, nuclear Zeeman, and hyperfine interactions, respectively, and μ_B and μ_n are the electronic and nuclear spin magnetons. The superscript indicates the ground or excited state manifold. The S₄ point symmetry of the tetragonal occupation site significantly simplifies the spin interaction, giving a symmetry axis parallel to the c-axis of the crystal. As a result, the uniaxial interaction tensors consist of one parallel and two identical perpendicular components $A_{\parallel }^{g/e}$, $A_{\perp }^{g/e}$ and $g_{\parallel }^{g/e}$, $g_{\perp }^{g/e}$, for the A and g tensors, respectively. The nuclear Zeeman interaction g_n tensor is considered to be isotropic with g_n = 0.987 for ¹⁷¹Yb³⁺ nuclear spin⁴⁰.

We identified the energy level diagram for the ²F_7/2(0) and ²F_5/2(0) levels using the high-resolution optical spectroscopy with different polarizations of the incident light (Fig. 1(c)). We extract the hyperfine tensors in the ground⁴¹ and excited states: $A_{\parallel }^g/h=0.787\left(1\right)$ GHz, $A_{\perp }^g/h=3.08384\left(1\right)$ GHz and $A_{\parallel }^e/h=-2.878\left(2\right)$ GHz, $A_{\perp }^e/h=2.734\left(2\right)$ GHz. To complete the optical spectroscopy, we determine the excited state g^e-tensor by fitting the magnetic field dependence of the energy levels to the model given by Eq. (1). We sweep the magnetic field by varying the current through a set of Helmholtz coils mounted parallel (perpendicular) to the crystalline c-axis (Fig. 1(d)). The extracted values are $g_{\parallel }^e=-1.452\left(1\right)$ and $g_{\perp }^e=1.362\left(1\right)$, qualitatively matching previous results⁴². In non-zero magnetic fields, symmetry-forbidden optical transitions become apparent as the symmetry of the site is reduced and additional spin mixing occurs due to the combined hyperfine and Zeeman interactions. Surprisingly, to fully explain the optical absorption spectrum, we need to assume a higher D_2d point symmetry, leading to S₄ site symmetry under a perturbation⁴³. This is discussed in more detail in the Supplementary Information.

At zero magnetic field, the eigenstates of the ground state manifold can be expressed as

$${∣1⟩}_g=\left({∣\uparrow \Downarrow ⟩}_g-{∣\downarrow \Uparrow ⟩}_g\right)/\sqrt{2},{∣2,3⟩}_g={∣\uparrow \Uparrow ⟩}_g,{∣\downarrow \Downarrow ⟩}_g,{∣4⟩}_g=\left({∣\uparrow \Downarrow ⟩}_g+{∣\downarrow \Uparrow ⟩}_g\right)/\sqrt{2},$$ 2

and the excited state manifold as

$${∣1,2⟩}_e={∣\uparrow \Uparrow ⟩}_e,{∣\downarrow \Downarrow ⟩}_e,{∣3⟩}_e=\left({∣\uparrow \Downarrow ⟩}_e+{∣\downarrow \Uparrow ⟩}_e\right)/\sqrt{2},{∣4⟩}_e=\left({∣\uparrow \Downarrow ⟩}_e-{∣\downarrow \Uparrow ⟩}_e\right)/\sqrt{2},$$ 3

where $∣\uparrow ⟩=∣m_S=+1/2⟩$, $∣\downarrow ⟩=∣m_S=-1/2⟩$ represent the electron spin and $∣\Uparrow ⟩=∣m_I=+1/2⟩$, $∣\Downarrow ⟩=∣m_I=-1/2⟩$ the nuclear spin projections. Two out of four electron-nuclear hyperfine states are completely entangled in their electronic and nuclear components (Fig. 1(a)). The average magnetic moment for these states is zero ($⟨i∣\mathbf{S}∣i⟩=⟨i∣\mathbf{I}∣i⟩=0$), rendering the energy levels insensitive to external magnetic fields in first order. As a result, the ${∣1⟩}_g-{∣4⟩}_g$ spin transition connecting two hybridized states is considered a clock transition. Similarly, optical transitions such as ${∣4⟩}_g-{∣4⟩}_e$ are also first-order protected from magnetic noise (Fig. 1(a)).

The combination of large hyperfine splittings and narrow optical lines allows for optical initialization of the whole spin ensemble into a desired ground state level. By addressing the ${∣4⟩}_g$ and ${∣2,3⟩}_g$ states via the A and B transitions (Fig. 1(b)), we prepare ¹⁷¹Yb³⁺ ions into the ${∣1⟩}_g$ state with high efficiency. Efficient optical pumping is possible thanks to strongly suppressed spin-phonon relaxation processes at temperatures below 2 K. We measure a spin ensemble recovery time of more than 3 hours at 50 mK. Together with a relatively short excited state lifetime of $T_1^o=0.385$ ms, we achieve more than 99% initialization efficiency for the whole ¹⁷¹Yb³⁺ spin ensemble under the experimental conditions specified in the Supplementary Information (Section IV.D), where optical pumping was performed repeatedly until a steady-state population distribution was reached.

Spin coherence

Next, we implement all-optical coherent spin control to measure the spin coherence properties of the ¹⁷¹Yb³⁺ ensemble. We use an optical Λ system, consisting of transitions A and E, to address the ${∣1⟩}_g-{∣4⟩}_g$ spin transition at 3.083 GHz (Fig. 1(a, b)). To drive the Raman transition, we use an electro-optic modulator to generate two spectral tones separated by the ${∣1⟩}_g-{∣4⟩}_g$ energy splitting. Using this control, we perform a spin echo experiment consisting of a sequence of π/2 and π pulses separated by a τ₁₂ delay (Fig. 2(a)). With a final probe pulse addressing the ${∣4⟩}_g$ state using the A transition, we observed the spin echo by detection of the coherent Raman scattering⁴⁴. First, we varied the frequency separation of the two laser tones and detected the spin echo signal at a fixed delay to extract a spin inhomogeneous linewidth of 5(1) kHz (Fig. 2(b)). The observed spin inhomogeneous broadening is one of the narrowest reported for paramagnetic solid-state ensembles, less than 48 kHz in ¹⁷¹Yb³⁺:YVO₄³² and 0.5 MHz in ¹⁷¹Yb³⁺:Y₂SiO₅⁴⁵. Then, by varying the delay between two pulses, we measured a spin coherence time up to $T_2^s=0.15\left(1\right)$ s (Fig. 2(c, d)). The measured spin coherence time is the longest achieved at zero magnetic fields among RE ions, even those possessing nuclear spin (15 ms in ¹⁵¹Eu³⁺:Y₂SiO₅⁴⁶).

Fig. 2. All-optical coherent spin control.

a Experimental pulse sequence used for all-optical detection of spin coherence. First, we polarize the spin ensemble of ¹⁷¹Yb³⁺ ions into the ${∣1⟩}_g$ state by optical pumping around the ω_A and ω_B frequencies. We then use optical pulses with frequency ω_A and ω_E to induce a Raman transition between the ${∣1⟩}_g$ and ${∣4⟩}_g$ spin states. The spin echo sequence with τ₁₂ delay between the first two spin rotations ends with a pulse at the ω_A frequency used for detecting the spin echo signal. b Measurement of the ${∣1⟩}_g$ - ${∣4⟩}_g$ transition illustrating the inhomogeneous broadening at zero magnetic field with Δ₁₄/2π detuning around ω_A − ω_E = 3083.85 MHz. A Gaussian fit gives a FWHM of ${\mathrm{\Gamma }}_{\text{inh}}^s/2\pi$= 5(1) kHz. The shoulder on the main peak is presumed to arise from fast spin flip-flop dynamics depending on the detuning, which modifies the measured lineshape; further measurements at different doping concentrations are needed to confirm this effect. c Spin echo intensity as a function of the delay τ₁₂ between the two spin rotations, at zero magnetic field. The solid line is the fit giving $T_2^s=0.15\left(1\right)$ s. d Spin coherence time as a function of the temperature, explained by the model based on flip-flops. The dashed line represents the measured 2T₁ limit given by the spin-lattice interaction, explained in the Supplementary Information. The spin echo decays were fitted with a pure exponential function; no stretch factor was needed, suggesting that the dominant dephasing mechanism remains spin flip-flop dynamics.

This long spin coherence time is governed by several features. First, the nuclear spin density of CaWO₄ is one of the lowest for crystalline materials and is dominated by the ¹⁸³W isotope⁴⁷. Additionally, ¹⁸³W has a rather low g_n = 0.2356, resulting in weak interactions with ¹⁷¹Yb³⁺ spins. Second, the ${∣1⟩}_g-{∣4⟩}_g$ transition is protected by the clock condition, i.e., it is only second-order sensitive to external magnetic noise. Third, the spin-spin dynamics between ¹⁷¹Yb³⁺ ions are suppressed due to the polarization into the ${∣1⟩}_g$ state. We measured the temperature dependence of the coherence time up to 3 K, while repeating the spin polarization in ${∣1⟩}_g$ via optical pumping (Fig. 2(d)). The measured spin coherence time is relatively constant up to 1 K and decreases to 10 ms at 3 K. We attribute this drop to ¹⁷¹Yb³⁺ spins experiencing flip-flop processes that exchange their spin states due to long-range magnetic dipole-dipole interaction¹⁹. More specifically, at 1K, the optical pumping efficiency starts to drop due to the stronger spin-phonon relaxation, which prevents the complete emptying of the ${∣2,3⟩}_g$ states. At 3 K, the spin-lattice relaxation rate increase to ≥2 s⁻¹, resulting in  ≈ 10% of the population staying in ${∣2,3⟩}_g$ states after the initial polarization. This intensified flip-flop process between ¹⁷¹Yb³⁺ ions affects the spin coherence in several ways (Fig. 3(a))⁴⁸. The spin coherence is directly affected due to an increase in spin flip-flops involving spins in ${∣2,3⟩}_g$ state, which is efficiently quenched by the optical pumping at lower temperatures. Indirectly, the magnetic field noise is increased due to increased spin flip-flops of ¹⁷¹Yb³⁺ spins near the probed ensemble. However, this effect is assumed to be small due to the low magnetic field sensitivity of the states and the modest concentration of ¹⁷¹Yb³⁺ spins.

Fig. 3. Optical coherence of ¹⁷¹Yb³⁺:CaWO₄.

a Left: flip-flops between the addressed ions and the neighboring spins directly contributes to a reduction in spin coherence. Middle: indirect contribution to decoherence due to flips of neighboring spin pairs (light red) producing magnetic noise for the probed ions (dark red). Right: spin dynamics rates for different transitions are highly transition dependent. Flip-flop rates with states involving ${∣2,3⟩}_g$ are two orders of magnitude larger than for ${∣1⟩}_g$ - ${∣4⟩}_g$ transition. b Experimental pulse sequence used for photon echo (PE) measurements. It consists of spin polarization by optical pumping using ω_E and ω_F frequencies (see Fig. 1(a)). Optical pulses resonant with the D transition are then used for the photon echo (PE) sequence with a τ₁₂ delay between the two pulses. c Measurement of the PE intensity at the lowest temperature (50 mK) as a function of the delay τ₁₂ after polarizing the spin ensemble of ¹⁷¹Yb³⁺ ions into ${∣4⟩}_g$ state (red) or reshuffling population between all the spin states (blue). Solid lines are exponential fits, giving optical coherence times $T_2^o=0.75$ ms and $T_2^o=0.54$ ms, respectively. d Absorption profile after polarizing the spin ensemble into ${∣4⟩}_g$ state (solid line) or reshuffling population between all the spin states (dashed line). e Optical coherence time as a function of temperature while reshuffling the spin populations (blue diamond). The optical coherence time when polarizing spins into the ${∣4⟩}_g$ state is enhanced (red circles) and approaches the 2T₁ limit (dashed line). The solid and dash-dot lines show predictions from the model described in the Supplementary Information.

Lifetime-limited optical coherence

In this section, we measure the optical coherence and subsequently use the optical coherence measurements to provide more insight into the fast spin-spin dynamics. The optical coherence time $T_2^o$ of the ${∣4⟩}_g-{∣4⟩}_e$ transition is measured through a standard photon echo technique using a single laser tone resonant with the D transition (Fig. 3(a)). Without optical pumping, an optical coherence time $T_2^o$ up to 0.54 ms was measured, surpassing the excited state lifetime $T_1^o=0.385$ ms (Fig. 3(b,c)). Next, using optical pumping, we initialize all the population in the ${∣4⟩}_g$ state (Fig. 3(d)). In this manner, the optical coherence time increases to $T_2^o=0.75\left(2\right)$ ms, which is nearing the 2T₁ = 0.77 ms limit (Fig. 3(e)).

We attribute the decrease in the optical coherence time without optical pumping to spin-spin dynamics via magnetic dipole-dipole interactions involving the ${∣2,3⟩}_g$ state, described above. Due to the high anisotropy of the g tensor, the spin flip-flop processes are strongly state-dependent, and the flip-flop rate is proportional to g⁴¹⁹. Only the ${∣1⟩}_g-{∣4⟩}_g$ spin transition is solely coupled via g_∥ (since only ${⟨1∣}_g{S}_z{∣4⟩}_g$ is non-zero, see Section 5 in the Supplementary Information), all other spin transitions involve g_⊥ which is almost four times higher. Hence, the flip-flop rates involving the ${∣2,3⟩}_g$ are faster by a factor $g_{\parallel }^4/g_{\perp }^4\approx 200$ and greatly reduce the spin coherence time $T_2^s$ (Fig. 3(a)). We estimate the flip-flop rates involving the ${∣2,3⟩}_g$ states to reach 10³ s⁻¹, leading to millisecond spin lifetimes¹⁹. Removing the population from ${∣2,3⟩}_g$ states is thus crucial to suppress the direct contribution from spin flip-flop dynamics (Fig. 3(a)). Above 2 K, the additional decrease of the optical coherence can be explained by phonon scattering in the excited state with a T⁹ dependence (Fig. 3(e)). This optical coherence time of 0.75 ms is one of the longest observed for paramagnetic solid-state emitters, comparable to ¹⁷¹Yb³⁺:Y₂SiO₅ (1 ms⁴⁹) and nuclear spin ions in Eu³⁺:Y₂SiO₅, (2.6 ms⁵⁰).

Discussion

We stress that even if the energy levels are insensitive to magnetic fields in the first order, the dipole moments of their transitions remain strong and correspond to electron spins. Such that, ${⟨1∣}_g{\mu }_{\mathrm{B}}\mathbf{g}\cdot \mathbf{S}{∣4⟩}_g={\mu }_{\mathrm{B}}{g}_{\parallel }^g$, the spin transition can be driven using an AC magnetic field along the c-axis. The optimal orientation of the AC drive for all other spin transitions is perpendicular to the c-axis, with the transition dipole moment proportional to $g_{\perp }^g$. We emphasize that even under the clock condition, microwave driving can still be efficient and fast, making this system compatible with interfacing with superconducting circuits^20,51.

Due to these strong transition dipole moments, electron spin-like fast spin-spin dynamics are present between ¹⁷¹Yb³⁺ ions. As we demonstrate in this study, such dynamics can be suppressed at low temperatures by polarizing the whole spin ensemble via optical pumping. However, applications involving large ensembles must have long-lived auxiliary spin states storing part of the population. These states must be free from fast spin-spin processes to prevent coupling the spin population to other states. One way to suppress the fast spin-spin dynamics is to work with lower doping concentrations. For example, lowering the concentration to 0.5 ppm will suppress the fast flip-flop rate down to 10² s⁻¹ level, while still exhibiting significant optical absorption. To reach the regime where decoherence is limited by the W nuclear spin bath, concentrations on the order of a few hundred ppb are expected to be required.

Another solution is to perform additional engineering of the material, and to controllably broaden the spin inhomogeneous linewidth without affecting the optical absorption. An example of this approach has been explored through co-doping of rare earth ions^52,53.

Lowered doping concentrations would enable the observation of a single ¹⁷¹Yb³⁺ in CaWO₄ and are expected to yield even longer spin and optical coherence times. Promisingly, a 0.7 ppb natural concentration of Er³⁺ allowed for 23 ms spin coherence times at high magnetic fields²⁰. The use of zero-field clock transitions, reported in our work, will strongly enhance coherences, while integration with nanophotonic structures will improve the properties of a spin-photon interface^37,38. Several photonic integration strategies are under development, including hybrid nanophotonic platforms where photonic structures are either transferred onto the bulk crystal³⁷ or directly fabricated within it³⁸. Additionally, recent work on the growth of epitaxial CaWO₄ thin films will enable future monolithic photonic and electronic integration⁵⁴. Importantly, CaWO₄ offers a unique advantage to realize a completely nuclear-spin-free environment through proper isotopic purification of Ca and W.

In conclusion, we have experimentally investigated the novel material ¹⁷¹Yb³⁺:CaWO₄ and reached lifetime-limited optical coherence and spin coherence times up to 0.15(1) s, facilitated by the clock transitions at zero magnetic fields. This represents an order-of-magnitude increase compared to the previously studied ¹⁷¹Yb³⁺:Y₂SiO₅, which exhibited spin coherence times up to 10 ms under similar conditions²¹. The enhanced performance of ¹⁷¹Yb³⁺:CaWO₄ can be attributed to the lower density of nuclear spins in the CaWO₄ host crystal. Combined with a reduced concentration of ¹⁷¹Yb³⁺, this suggests that even longer coherence times can be achieved than those measured in the present work. This material provides a set of exclusive properties such as optically resolved hyperfine transitions, the ability for all-optical initialization of the spin ensemble, and long coherence times, making it a unique resource for quantum networking and information applications. These features make ¹⁷¹Yb³⁺:CaWO₄ appealing for both ensemble-based applications, where collective ensemble coupling is beneficial, and integrated spin-photon interfaces based on single emitters. These applications include optical quantum memories^4,55, coupling to superconducting circuits²⁰, integrated spin-photon interfaces^37,38, and optical-to-microwave quantum transducers²⁴.

Methods

Sample preparation

Samples were cut along the a (b) and c crystallographic axes. Several samples with 1 cm and 1.5 cm sizes along the c-axis were prepared, and polishing was done to get optically transmitting surfaces along several directions. For optical experiments, two samples with different sizes (9.2, 9.7, 9.9) and (15, 4.1, 4.8) mm in the (a, a, c) frame were prepared. For the electron paramagnetic resonance (EPR) study, a (1.5, 2.2, 2.5) mm sample weighing 63 mg was used.

Optically detected spin echo

We measured the spin coherence time $T_2^{\text{s}}$ through optical detection of a spin echo in a Hahn echo sequence using all-optical spin manipulation via Raman transitions through the excited state¹⁶. All spin echo measurements reported in this paper were carried out on the ${∣1⟩}_g\to {∣4⟩}_g$ transition (3083.85 MHz) of the tetragonal S₄ site. Without compensating for the Earth’s magnetic field, a spin coherence time of 30 ms was measured. A permanent magnet was used to compensate for the external magnetic field and optimize the coherence properties by maximizing the spin echo signal at long delays.

The duration of optical pulses in the spin echo sequence was 2 μs and was optimized using the spin echo intensity. All pulses had a Gaussian envelope shape. Optical pulses of 200 μs duration were used to resolve the spin linewidth, which was found to be 5(1) kHz FWHM. The detected signal was analysed with a Fast Fourier Transform, and the area of the peak corresponding to the beat with the local oscillator ω_LO at ∣ω₀ − ω_LO + Δ_AOM∣ was acquired at various time delays τ₁₂. The coherence time $T_2^{\text{s}}$ was then extracted by fitting the peak area decays as $E_{\mathrm{echo}}\left(\tau \right)=E_0\exp \left(-2{\tau }_{12}/T_2^{\text{s}}\right)$.

Photon echo

The optical coherence time $T_2^{\text{o}}$ was measured in an analogous way by preparing a photon echo sequence on a transition between a ground and an excited state. We used the heterodyne signal to measure the echo intensity detected at ∣ν_LO − ν₀∣ to improve the detection sensitivity. The duration of the optical pulses was about 1 μs. The optical coherence time $T_2^{\text{o}}$ was extracted by fitting the photon echo area using $E_{\mathrm{echo}}\left(\tau \right)=E_0\exp \left(-2{\tau }_{12}/T_2^{\text{o}}\right)$.

Author contributions

The crystal growth was done by A.T., P.L., and P.G. Initial optical spectroscopy was performed by A.T. and P.G. EPR measurements were performed by S.B. and A.M.K. Optical characterization under a magnetic field was done by E.G., S.H., and E.L. with support from A.F. All the photon echo and optically detected spin echo measurements were carried out by A.T. with support from P.G., and the results were interpreted by A.T. and P.G. with contributions from F.C. and D.S. The manuscript was mainly written by A.T. and P.G., with contributions from all the authors.

Peer review

Peer review information

Nature Communications thanks Zong-Quan Zhou, who co-reviewed with Xiao Liu, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data from the plots within this paper and other details of this study are available from the corresponding author upon request and are also openly available at https://figshare.com/s/b735d51cdc8e5945c990.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70534-9.