Color center in β-Ga₂O₃ emitting at the telecom range

Color centers, i.e., light-emitting point defects and impurities embedded in a semiconductor host, are currently attracting great attention^1–5 as a promising system for quantum sensing, quantum information, and communication technologies. Due to strong localization of the bound electrons, the color centers possess atom-like electronic states that can be addressed optically. When combined with appropriate spin properties of the electronic states, they can act as solid-state qubits with optical initialization and read out of the electron spin, atomic-scale sensors and also single-photon emitters. Among color centers, several classes of defects and impurities are considered as favorable candidates for the aforementioned applications, including vacancy-related complexes and rare-earth ions in diamond and other wide bandgap semiconductors.

In addition to these model systems, transition-metal (TM) ions, robust and easily controllable dopants in semiconductors, have recently emerged as another viable contender for solid-state qubits.^6–12 Since intra-center electronic transitions within TM color centers are confined to d-orbitals of the ions, they are largely decoupled from the host lattice. Therefore, information about their properties is transferable between different host materials and can be understood with guidelines from the Tanabe–Sugano theory.^13,14 In addition, the 3d electronic states are affected by local symmetry, coordination, and crystal field experienced by the ion, which could be used to fine-tune its electronic structure. Furthermore, some of the TM ions emit at telecom wavelengths, which makes them very attractive for quantum communication technologies due to low attenuation loss within this spectral region. Recent studies indeed show that several TM ions can be utilized as optically active solid-state qubits, including Cr⁴⁺, V⁴⁺, V³⁺, and Mo⁵⁺ in SiC and GaN.^6–11 

Among important advantages of solid-state qubits in a semiconductor host is the possibility to integrate them in traditional microelectronic and photonic structures and thus to take full advantage of the mature semiconductor technology. Here, wide bandgap semiconductors are especially attractive since the wide bandgap can assure one of the key requirements for a semiconductor host: absence of interference between optical transitions of the color center with electronic states of the host.^1,3,4 One of such technologically important wide bandgap semiconductors is β-Ga₂O₃. This material is currently attracting wide-spread research attention due to its electronic properties attractive for numerous applications, e.g., in power electronic devices, solar-blind UV photo-detectors and emissive displays, and the availability of affordable native substrates.^15–20, β-Ga₂O₃ could be easily doped with TM impurities, which introduce electronic levels within the bandgap affecting electrical and optical properties^21–23 that in turn may also provide additional functionalities.^24–26 For example, Cr³⁺ is a well-known color center giving rise to the bright intra-center emission within the red spectral range,^27–30 whereas Ti³⁺ in β-Ga₂O₃ has been suggested as a potential spin-bus system for quantum information processing.³¹ 

In this work, we report on a TM color center in β-Ga₂O₃, which emits at the telecom range and has the electronic structure suitable for quantum information applications. The center is responsible for a sharp emission at 1.317 μm with a weak phonon sideband, beneficial for single-photon emission. Based on the performed magneto-photoluminescence (PL) measurements complemented by time-resolved PL and PL excitation spectroscopies, we establish the electronic structure and spin properties of this TM ion as well as its possible chemical identity.

A variety of β-Ga₂O₃ crystals were studied in this work, including commercial Fe-doped and undoped crystals from Tamura as well as Czochralski-grown crystals doped with Cu, Ni, Cr, and Mn,^32–34 see also Sec. S1 of the supplementary material. Magneto-PL measurements were performed in a backscattering geometry using a LHe-cooled cold-finger cryostat operating between 7 and 300 K, equipped with a superconducting magnet (up to 5 T). The magnetic field direction was aligned parallel to the light excitation and collection axis. The excitation light was produced by either a solid-state 660-nm laser or a tunable Ti-sapphire laser, whereas the emitted PL was dispersed using a grating monochromator and detected using an InGaAs charge-coupled device (CCD). Transient PL and PL excitation (PLE) measurements were performed using an optical parametric oscillator (OPO) laser as an excitation source with a laser pulse width of about 5 ns and a repetition rate of 20 Hz. The excitation power was kept constant at all excitation wavelengths. The transient PL signals were detected using a liquid-nitrogen-cooled Ge-detector through a high-resolution double grating spectrometer and recorded with a digital oscilloscope. The time resolution of the setup was 8 μs.

A representative PL spectrum of the uncovered color center is shown in Fig. 1(a). At 15 K, it contains two very narrow zero-phonon lines at ∼1316 nm (0.9415 eV) and 1314 nm (0.9436 eV) accompanied by a weak phonon sideband. Below we will refer to these lines as line 1 and line 2, respectively. With the exception of Ni-doped Ga₂O₃, both lines can be detected in all investigated samples with the highest intensity found in Fe-doped and Cu-doped samples (see also Sec. S2 of the supplementary material). The narrow linewidth suggests that the monitored emission stems from an internal transition, most likely involving a TM impurity. High-resolution PL measurements reveal that line 1 in fact consists of three components, which are clearly resolved at 5 K but become broadened with increasing temperature—see Fig. 1(b). Temperature (T) increase, however, does not affect the relative intensities among these components, which proves that they must originate from zero-field splitting (ZFS) of the ground state. On the other hand, the intensity of line 2 increases at elevated temperatures. As line 2 always accompanies line 1 with its intensity following a well-defined thermal distribution between them at a given temperature, it most likely originates from a higher-lying excited state of the same center. PL transient measurements reveal a mono-exponential PL decay of line 1 with a long lifetime (τ) of 2.3 ms, as shown in Fig. 1(c), typical for a spin-forbidden electronic transition.^27,35–37

In view of a noticeable deviation of the measured electron g-values from 2.0023 for a free electron, a sizable contribution from a non-s-like orbital state and the associated anisotropy is expected for the ground state. This should lead to a polarization effect as indeed observed experimentally. This can be seen from Fig. 3, which shows results from polarization-resolved PL measurements performed when the light propagating direction (defined by k) is aligned along the a^* crystallographic direction that is orthogonal to both c and b crystallographic axes, and also Sec. S3 of the supplementary material. In the geometry of kǁa^*, the high (low) energy component of line 1 is polarized along the c (b) crystallographic axis, whereas the central component is weak due to its preferential linear polarization close to the a axis. The polarization measurements also reveal a zero-field splitting of line 2, which is identical to that of line 1 but with a reversed polarization order of the PL components. The same ZFS (δ = 0.57 meV) of lines 1 and 2 unambiguously confirms our assignment that the monitored radiative transitions involve the same spin-triplet ground state. Therefore, line 2 stems from a higher-lying spin-singlet excited state separated from the lowest spin-singlet excited state by ZFS (Δ) of 2.5 meV as illustrated in the inset in Fig. 3. These two excited states should differ in their orbital states to account for the observed opposite ordering of their PL components.

Based on these results, we can now suggest a possible electronic configuration of the revealed color center. Substitutional TM ions in β-Ga₂O₃ could in principle reside on either a tetrahedral Ga(1) site with fourfold coordination or a sixfold-coordinated octahedral Ga(2) site, though the Ga(2) site is usually energetically preferable.^33,38–41 Therefore, their electronic states are often analyzed with the help of the so-called Tanabe–Sugano diagrams. These diagrams envisage splitting of electronic states of different TM ions under the influence of a perfect octahedral (or tetrahedral) crystal field, which is described in terms of the crystal field strength Dq and the Racah parameters B and C that characterize the inter-electron repulsion effects. Based on these diagrams, only two electronic configurations with a spin-triplet ground state, namely, a 3d⁸ ion on the octahedral site and a 3d² ion on the tetrahedral site,⁴² could explain our experimental data. According to the Tanabe–Sugano diagram [see Fig. 4(a)], under the action of an octahedral crystal field in the 3d⁸ configuration or a tetrahedral crystal field in the 3d² configuration, the ³F ground state and ¹D first excited states of such ions will split into three spin triplets (³A₂, ³T₂, and ³T₁) and two spin singlets (¹E and ¹T₂), respectively. Consequently, the ground state of the ion becomes ³A₂, whereas the origin of the first excited state depends on the strength of the crystal field and is ¹E for high crystal fields, i.e., when Dq/B > 1.6. The remaining degeneracy of both ground and excited states can further be lifted due to a spin–orbit interaction and also when the symmetry of the crystal field is reduced. For a point defect in a crystal of monoclinic symmetry, such as β-Ga₂O₃, the S = 1 (³A₂) ground state and S = 0 (¹E) excited state will experience zero-field splitting into three and two sublevels, respectively, resulting in the electronic structure shown in the inset of Fig. 3. The complete lifting of the twofold orbital degeneracy of the ¹E excited state under a crystal field of monoclinic symmetry or lower should result in two sublevels of orthogonal orbital characters, explaining the reversed polarization ordering between their optical transitions to the same ground state seen in Fig. 3. Moreover, since the ¹E→³A₂ transitions are spin-forbidden, the corresponding emission will have a very long radiative lifetime, as indeed observed experimentally—see Fig. 1(c).

From Fig. 4(a), the splitting between the ¹E and ³A₂ states is only weakly dependent on the crystal field strength. On the other hand, the energy separation between the higher-lying excited states and the ³A₂ ground state is a strong function of Dq/B and, therefore, could be used to roughly estimate this parameter. To experimentally determine energy positions of the higher-lying excited states of the revealed color center, we perform PLE measurements. It is found that the PLE spectrum contains several bands centered at around 1.66, 1.85, and 3.0 eV—see Fig. 4(b) (the blue line). The relatively large linewidth of these bands is likely caused by splitting of the involved states due to spin–orbit and crystal field interactions, which often could not be resolved experimentally.^27,28,30,43 Taking into account that PLE spectra of TM color centers are usually dominated by spin-allowed optical transitions^{27,28,30,43,44} involving the electronic states shown by the blue solid lines in Fig. 4(a), we can tentatively assign the 1.66, 1.85, and 3.0 eV PLE bands to the ³A₂→³T₂(F) ³A₂→³T₁(F) and ³A₂→³T₁ (P) absorption, respectively. (The label in parentheses denotes the parental electronic state of an isolated ion.) For an octahedral configuration of a 3d⁸ ion, the best agreement between the energy levels predicted by the Tanabe–Sugano diagram and the experimentally measured PL and PLE spectra is then obtained for Dq/B = 2.2 [indicated by the dashed vertical line in Fig. 4(a)] and B = 500 cm⁻¹. The obtained values are reasonable and are within the range of Tanabe–Sugano parameters previously reported for TM ions in solid-state materials.^{27,28,30,44,45}

In summary, we have uncovered a color center in β-Ga₂O₃ emitting a sharp PL emission at 1316 nm. Based on its long radiative lifetime (2.3 ms at 5 K) and also its electronic and spin configurations, we attribute this emission to the spin-forbidden ¹E→³A₂ transition within a 3d⁸ TM ion that is a common contaminant in β-Ga₂O₃. The following electronic, optical, and spin properties of this color center are attractive for applications as an optically addressable spin qubit or a single-photon emitter: (i) emission within the telecom O-band; (ii) a strong zero-phonon line with only a weak phonon sideband; (iii) an S = 1 ground state that experience a zero-field splitting; and (iv) optically addressable higher-lying excited states over a wide spectral range. Based on our results, the 1316-nm color center is likely a complex involving a sixfold coordinated Cu³⁺ ion. Further theoretical and experimental studies are required to unambiguously determine the chemical origin of the revealed color center and also to evaluate its spin dynamics, important for understanding its potential for quantum technologies.