Er³⁺-Doped TiO₂ Thin Films: Quantum Photonics

• Er³⁺-doped TiO₂ thin films integrate trivalent erbium ions into titanium dioxide to achieve telecom-band quantum emitters with long optical and spin coherence times.
• Advanced deposition techniques such as MBE, ALD, and PLD enable precise control over dopant placement and phase selection, which are critical for optimizing optical transitions and defect properties.
• The compatibility with CMOS and nanophotonic device integration paves the way for scalable on-chip quantum memories and single-photon sources in quantum photonics applications.

Er$^{3+}$-doped TiO$_2$ thin films are a class of materials in which trivalent erbium ions are incorporated into the lattice of titanium dioxide, with the aim of producing optically and magnetically coherent atomic-scale quantum emitters in a technologically scalable, CMOS-compatible form factor. The defining appeal of this system lies in the shielded 4$f$ orbital transitions of Er$^{3+}$, which provide long optical and spin coherence times and operate in the telecom C-band (∼1520–1533 nm). TiO$_2$ in its various crystalline phases (anatase/rutile), featuring low intrinsic nuclear spin density and high refractive indices, serves as a versatile host enabling photonic integration, quantum memory, and on-chip single-photon sources. Material realization of these films spans a range of epitaxial, polycrystalline, and amorphous growth methods, with precise control over dopant distribution, local environment, and structural defects dictating the quantum-relevant properties and device utility.

1. Thin Film Growth Techniques and Dopant Incorporation

Film synthesis approaches for Er$^{3+}$:TiO$_2$ include molecular beam epitaxy (MBE), atomic layer deposition (ALD), and pulsed laser deposition (PLD) (Singh et al., 2022, Ji et al., 2023, Hammer et al., 5 Nov 2025). Growth protocols are tailored for phase selection (anatase vs. rutile), dopant placement, and minimization of interfacial and bulk defect densities.

Growth Methodologies

• MBE: Utilizes titanium tetraisopropoxide (TTIP) as Ti precursor, O$_2$ flow for oxidation, and effusion cell–delivered Er. Substrate temperature (480–850 °C) and O$_2$ partial pressure (∼10$^{-9}$–10$^{-6}$ Torr) tune crystal phase and grain structure (Singh et al., 2022, Martins et al., 29 Sep 2024).
• ALD: Alternating TTIP and H$_2$O (or O$_3$) pulses at $T_s ≃ 120$–350 °C, with Er(thd)$_3$ or cyclopentadienyl-Er precursors for precise, atomic-scale delta-doping (Ji et al., 2023, Ji et al., 4 Jun 2024).
• PLD: Deposition onto III-V substrates (e.g., GaAs, GaSb) at $T_{\rm grow}=390$–565 °C, with As-capping and oxygen-deficient buffer strategies for interface engineering and phase control (Hammer et al., 5 Nov 2025).

Dopant Placement and Concentration

• Uniform Doping: Achieved by co-deposition across the entire film thickness (typical concentrations: 10–5000 ppm).
• Delta-Doping (δ-Doping): 1–10 nm Er-rich layers sandwiched between undoped TiO$_2$ to confine dopants and localize optical emission (Ji et al., 4 Jun 2024).
• Substitutional Yield: Highest for low-dose, annealed or as-grown films (up to 40%), with Er$^{3+}$ preferentially occupying Ti$^{4+}$ lattice sites (Phenicie et al., 2019).

Interfaces and Buffer Layers

• Use of undoped TiO$_2$ buffer/cap layers (10–60 nm) significantly mitigates spectral diffusion and inhomogeneous broadening by spatially isolating Er$^{3+}$ from defect-rich interfaces (Singh et al., 2022).

Growth Method	Doping Profile	Achievable Er
MBE	Uniform/δ	10–5000
ALD	Uniform/δ	<1–39,200
PLD	Uniform/δ	3000

2. Crystallography, Phase Engineering, and Defects

The optical and spin properties of Er$^{3+}$ emitters are strongly phase- and site-dependent.

Crystal Phase and Epitaxy

• Anatase (A-TiO$_2$): Stabilized by low-temperature growth (T ≈ 390–500 °C), As-capping or appropriate substrate/buffer selection. Polycrystalline on Si yields grain size 10–30 nm; epitaxial on LaAlO$_3$, SrTiO$_3$ (Singh et al., 2022, Hammer et al., 5 Nov 2025).
• Rutile (R-TiO$_2$): Formed at higher T (≥ 450 °C), via laser annealing, or in single-crystal films on r-sapphire. Grains reach 50–90 nm after post-growth annealing or local laser conversion (Phenicie et al., 2019, Sullivan et al., 2023).
• Phase-localization: Focused laser annealing enables diffraction-limited rutile regions (diameter ≈0.45 μm) in an anatase host, with deterministic spatial addressability (Sullivan et al., 2023).

Defect Chemistry and Site Occupancy

• Substitutional Er$^{3+}$: Occupancy of the Ti$^{4+}$ site (octahedral, D$_{2h}$ or D$_{2d}$ symmetry) is established by ESR/XAS/EXAFS; full first-shell O coordination (N$_{\rm Er–O}$ ≈ 6.4) with Δd (expansion) = 0.28 Å, matching ionic radii (Martins et al., 29 Sep 2024).
• Charge Compensation: Substitutional Er$^{3+}$ (vs. Ti$^{4+}$) induces oxygen vacancies V$_{\rm O}^{2+}$ for local charge neutrality; inferred as accompanying defect peaks in O K-edge XAS (Martins et al., 29 Sep 2024).
• Extended Defects: High-$x$ films ($>$200 ppm) exhibit increased Er-vacancy cluster formation, extended strain fields, and suppressed O(2p)–Ti(3d) hybridization, contributing to non-radiative decay (Martins et al., 29 Sep 2024).
• Interfacial SiO$_x$: Si substrates typically develop a ∼1–4 nm amorphous SiO$_x$ at the film interface, influencing PL uniformity and local disorder (Singh et al., 2022, Dibos et al., 2022).

3. Optical and Spin Coherence Properties

The critical quantum attributes of Er$^{3+}$ in TiO$_2$ include narrow optical and spin transitions, long fluorescence lifetimes, and sensitivity to phase and defects.

Optical Transitions and Linewidths

• C-band Emission: $^4$I$_{13/2}\rightarrow {}^4$I$_{15/2}$ at λ$_{\rm rutile}$ ≈ 1520 nm, λ$_{\rm anatase}$ ≈ 1533 nm. Phase transition (anatase → rutile) shifts Z$_1$–Y$_1$ emission by Δλ = 13 nm (ΔE ≈ 0.9 meV) (Sullivan et al., 2023).
• Inhomogeneous Linewidths (Δν$_{\rm inh}$): Lowest values found in implanted rutile (0.46 GHz), buffered/anatase on Si (5.2 GHz), and ALD/low-density films (as low as 44 GHz) (Phenicie et al., 2019, Singh et al., 2022, Ji et al., 2023). Broader lines (50–79 GHz) arise in high-$x$, polycrystalline or defect-rich environments (Martins et al., 29 Sep 2024, Ji et al., 2023).
• Spectral Diffusion (Δν$_{\rm SD}$): Minimized to 180 MHz with buffer/cap engineering; charge noise and interface defects are dominant sources in thin films (Singh et al., 2022).
• Crystal Field Splitting and Branching: Dominant decay Y$_1\rightarrow$Z$_1$ (∼90% of decay processes), with minor population in alternative branches (Phenicie et al., 2019).

Spin Properties

• ESR g-Factors: g$_{zz}$(B||c) = 14.30, g$_{xx}$(B||a) = 1.63; probed for site verification (Phenicie et al., 2019).
• Spin linewidths (Δν$_{\rm spin}$): 20 MHz in low-dose rutile, corresponding to $T_2^* = 1/(\pi \Delta\nu_{\rm spin})\sim 15$ ns (Phenicie et al., 2019).
• Hyperfine Coupling (for $^{167}$Er): A$_{zz}$ = 1503 MHz (Phenicie et al., 2019).

Lifetime and Coherence Trade-offs

Material/Structure	Δν$_{\rm inh}$ (GHz)	Δν$_{\rm SD}$ (MHz)	$T_1$ (ms)
Implanted rutile bulk	0.46	—	5.25
ALD/Anatase/SiO$_2$	44	—	1.72
Poly-anatase/Si, buffered	5.2	180	1.1
Epitaxial rutile/sapphire	50	—	2.1

Smaller grains, higher Er concentration, and proximity to strained or defective interfaces degrade both $T_1$ and Δν$_{\rm inh}$ due to elevated $k_{\rm nr}$ and spectral diffusion.

4. Nanophotonic and Device Integration

The high refractive index and CMOS compatibility of TiO$_2$ support integration with Si-based nanophotonics, permitting scalable photonic quantum devices.

On-chip Integration

• Films are grown or transferred directly onto SOI wafers; device stacks typically consist of 15 nm undoped buffer / 1–10 nm Er-doped middle layer / 15 nm undoped cap (Ji et al., 4 Jun 2024, Dibos et al., 2022).
• Surface roughness control ($<$0.5 nm RMS) is essential for achieving high-Q photonic structures and minimizing scattering loss (Ji et al., 2023).
• Novel interface engineering (As-capping, oxygen-deficient buffers, MCIA modeling) enables direct growth on III-V substrates (GaAs, GaSb), supporting hybrid quantum photonic integration (III-V emitters + rare earth quantum memories) (Hammer et al., 5 Nov 2025).

Photonic Crystal Cavities and Purcell Enhancement

• 1D photonic crystal cavities (PCCs) in Si device layer, overlaid with Er:TiO$_2$, achieve Q-factors $>$5×10$^4$ and mode volumes %%%%70$_2$71%%%% (Dibos et al., 2022).
• Purcell enhancement of up to 200–460 for Er$^{3+}$ lifetimes observed, reducing $T_1$ to sub-10 μs regimes and driving single-photon emission rates above 10 MHz for isolated emitters (Ji et al., 4 Jun 2024, Ji et al., 2023).
• Focused laser annealing allows for submicron spatial control of emitter phase/resonance, supporting deterministic placement at waveguide/resonator anti-nodes (Sullivan et al., 2023).

Single-Ion Addressability

• Delta-doped films and low-concentration ALD methods yield single Er$^{3+}$ ions per cavity mode volume. Photoluminescence excitation (PLE) scans reveal narrow ($<$200 MHz) single-ion lines with g$^{(2)}(0)=0.29$ (background-corrected $<$0.05), confirming true single-photon emission (Ji et al., 4 Jun 2024).

5. Defect Control, Limitations, and Optimization Strategies

The interplay between defect chemistry, phase, and device architecture sets clear performance bounds and optimization routes.

Defect Sources and Impact

• Oxygen Vacancies (V$_{\rm O}$): Inherent to charge compensation for Er$^{3+}$ substitution, these introduce mid-gap states, non-radiative decay channels, and broaden inhomogeneous linewidths (Martins et al., 29 Sep 2024).
• Extended Strain Fields: Higher doping increases lattice distortions, with experimental evidence from EXAFS/Ti K-edge amplitude reductions (Martins et al., 29 Sep 2024).
• Interfacial Defects: SiO$_x$ formation and proximity to substrate/air interface increase charge noise, worsen spectral diffusion, and degrade $T_2$ (Singh et al., 2022).

Mitigation and Engineering

• Lower Dopant Densities: Reducing Er directly narrows Δν$_{\rm inh}$ and lengthens $T_1$ (Singh et al., 2022, Martins et al., 29 Sep 2024).
• Buffer and Cap Layers: Increasing thickness of undoped TiO$_2$ layers isolates Er from defects, reducing both $\Delta\nu_{\rm inh}$ and $\Delta\nu_{\rm SD}$ (down to 5.2 GHz and 180 MHz, respectively) (Singh et al., 2022).
• Post-Growth Annealing: High-T anneals (800–1000 °C) heal lattice damage, increase substitutional yield, and revert broadened lineshapes (Phenicie et al., 2019).
• Phase Targeting: Rutile with D$_{2h}$ symmetry yields inversion-symmetric sites, suppressing first-order DC Stark shifts and thus reducing sensitivity to external electric-field noise (Phenicie et al., 2019).

6. Quantum Photonics Applications and Prospects

The unique quantum-optical attributes of Er$^{3+}$:TiO$_2$ thin films enable advanced device concepts.

• Quantum Memory: Millisecond-scale lifetimes and telecom-compatible optical transitions facilitate on-chip quantum memories for repeater nodes (Ji et al., 4 Jun 2024, Singh et al., 2022).
• Single-Photon Sources: Isolation of single Er ions with confirmed antibunching establishes a platform for deterministic, narrow-linewidth telecom-band sources suitable for quantum networks (Ji et al., 4 Jun 2024).
• Purcell-Enhanced Photon Interfaces: Engineered high-Q, low-mode-volume cavities produce MHz-rate photon emission from individual ions (Ji et al., 2023).
• Hybrid Integration: Direct epitaxy of Er:TiO$_2$ on III-Vs (GaAs, GaSb) creates chip-scale nodes with both quantum dot and rare-earth functionalities (Hammer et al., 5 Nov 2025).
• Spectral Multiplexing and Tuning: Local phase conversion allows for 13 nm emission tuning (C-band), enabling multiplexed quantum channels (Sullivan et al., 2023).

Limitations and Open Challenges

• Achieving homogeneous linewidths near the ∼kHz radiative limit in device-relevant geometries remains an unsolved problem, limited by residual charge noise and extended defects (Ji et al., 4 Jun 2024, Martins et al., 29 Sep 2024).
• Spectral diffusion and slow stochastic wandering are reduced but not eliminated by current interface and buffer layer approaches.
• Strategies such as active/reversible phase cycling, further reduction of dopant density, and exploration of alternative host matrices (e.g., TiO$_2$ polymorphs, perovskite oxides) are under investigation for performance gains (Sullivan et al., 2023, Martins et al., 29 Sep 2024).

7. Outlook and Future Directions

Optimization of Er:TiO$_2$ thin films has rapidly progressed toward scalable quantum photonic devices, with demonstrations of single-ion addressability, nanocavity integration, and robust phase/selectivity engineering (Ji et al., 4 Jun 2024, Ji et al., 2023, Sullivan et al., 2023, Hammer et al., 5 Nov 2025). Future directions include:

• Realizing reversible and reconfigurable emitter arrays via in-situ phase control (Sullivan et al., 2023).
• Minimizing spectral diffusion through defect/spacer engineering, substrate choice, and improved crystallinity (Singh et al., 2022).
• Coupling to spin degrees of freedom for quantum networking and memory, leveraging the $I=0$ nuclear spin environment of Ti and O (Phenicie et al., 2019).
• Extending integration beyond Si to compound semiconductors, enabling monolithic hybrid quantum and classical photonic circuits (Hammer et al., 5 Nov 2025).

A well-controlled balance among Er$^{3+}$ optical density, structural quality, defect suppression, and photonic device engineering will define the next advances in on-chip quantum information science with Er:TiO$_2$ thin films.