Efficient formation and identification of single emitters in 4H-SiC following maskless heavy ion implantation

AI summary

Overview

Research area: Quantum photonics and semiconductor defect physics — specifically the deterministic creation and rapid screening of single-photon-emitting colour centres in silicon carbide (4H-SiC) for room-temperature quantum technologies.

Technical level: Intermediate. The paper assumes familiarity with confocal photoluminescence microscopy, second-order autocorrelation, and optically detected magnetic resonance (ODMR), but its central arguments about yield, dose optimisation and screening strategy are accessible to readers with a general photonics or materials background.

One-sentence scope: The paper reports maskless focused-ion-beam implantation of bismuth and tin into 4H-SiC to form arrays of single oxygen-vacancy-related emitters, quantifies the formation yield as a function of dose and annealing temperature, and proposes a tiered measurement protocol that identifies emitter type (PL4/PL5/PL6) without cryogenic spectroscopy.

What This Paper Is About

Scalable quantum devices built from single photon emitters in SiC require two things that are currently awkward: placing emitters at chosen positions without a lithographic mask, and then working out quickly which species of emitter each bright spot actually is. Previous work on heavy-ion implantation into 4H-SiC used high-energy, broad-area implants and ensemble measurements, and identifying emitter type has typically relied on time-consuming low-temperature photoluminescence.

This paper addresses both problems together. The authors use a focused beam of heavy ions (Bi and Sn) to write 10 × 10 arrays of isolated implant spots, anneal the samples across a range of temperatures, and then build a statistical model of single-emitter occupancy to predict the dose that maximises the usable single-emitter count. They then correlate four room-temperature-accessible measurement channels to separate PL4, PL5 and PL6 centres without needing a cryostat.

Key Contributions

1. Demonstration of maskless single-emitter formation by heavy-ion FIB implantation in 4H-SiC. Arrays of 10 × 10 spots were written with Bi (²⁰⁹Bi) and Sn (¹²⁰Sn) ions at 25 keV (singly charged) and 50 keV (doubly charged), with an optimum annealing window of 900–1000 °C yielding up to 18% of implanted sites hosting exactly one emitter.

2. A maximum-likelihood Poisson occupancy model that predicts the optimal implantation dose. Fitting single/multiple/unsuccessful spot statistics gives a formation probability α̂ of (8.1 ± 1.3) × 10⁻⁴ emitters per ion for Bi²⁺ and (4.5 ± 0.9) × 10⁻⁴ for Sn²⁺, from which the dose maximising single-emitter count (1/α̂) and the dose maximising the single-minus-multiple difference (1/2α̂) are extracted.

3. A tiered characterisation scheme for rapid emitter identification. Intensity thresholding on the PL map, then a fast polarisation measurement, then g⁽²⁾(τ) confirmation, then ODMR for type assignment — with low-temperature photoluminescence reserved only for cases where PL4 identification is needed.

4. Correlated fingerprints of PL5 and PL6 centres. Pairwise scatter plots of saturation count rate, degree of polarisation (DOP), ODMR frequency and zero-phonon line (ZPL) wavelength show distinct clustering, with working separation thresholds of 75 kcps (saturation) and 0.3 (DOP).

Main Findings

• Optimal annealing window of 900–1000 °C. At the best conditions the maximum single-emitter fraction is 18% (Bi²⁺) and 11% (Sn²⁺), with single-emitter yields of η = 0.09% (Bi²⁺) and 0.05% (Sn²⁺) per implanted ion. Below 900 °C vacancies lack mobility to form stable optically active complexes; at 1100 °C the arrays are almost completely lost, attributed to emitter annihilation or aggregation into larger non-optically-active clusters.

• Positioning accuracy of 146 nm (RMS) for Bi²⁺, measured as deviation of fitted emitter positions from the idealised grid. The simulated ion straggle for Bi²⁺ at 25 kV anode voltage is 3.7 nm and is therefore negligible, so the scatter is dominated by the implantation process rather than ion range straggle. The authors note that the 14 nm FIB imaging resolution is not a good proxy for implant accuracy — this is contrasted with a previously reported masked approach (50 nm lithographic holes) that gave 53.6 nm RMS, where the spread was dominated by hole size.

• Dose optimisation from the Poisson model. The single-emitter probability distribution peaks at 1235 ions/spot (Bi²⁺) and 2210 ions/spot (Sn²⁺); the difference between single- and multiple-emitter probabilities peaks at 620 and 1105 ions/spot respectively. Fit quality was confirmed with P-values of 0.24 (Bi²⁺) and 0.98 (Sn²⁺). The study itself remained in the low-dose regime, reaching mean site occupancy λ = 0.16 (Bi²⁺) and 0.09 (Sn²⁺) at 200 ions/spot.

• 60 isolated single emitters characterised across all Bi²⁺ and Sn²⁺ dose arrays annealed at 900 °C, each showing a distinct ODMR resonance and/or zero-phonon line. In the Bi²⁺ 200 ions/spot array (the highest-yielding), 5 of the 18 g⁽²⁾-confirmed single emitters were PL6 and 2 were PL5; in the corresponding Sn²⁺ array, 1 of 6 single emitters was PL6 and no PL5 centres were identified.

• PL6 ODMR frequency scatter of 4.7 MHz around a mean of 1351.7 MHz (only single-dip emitters used, to avoid misassignment). PL5 showed a dip-1 mean of 1339.9 MHz with standard deviation 10.1 MHz and a dip-2 mean of 1372.2 MHz with standard deviation 2.8 MHz. The asymmetry between the two PL5 dips is interpreted as the two transitions responding differently to local strain, and the overall scatter is attributed to strain from dense heavy-ion collision cascades that is not fully healed at 800–1000 °C.

• Low-temperature spectroscopy is unnecessary for PL6 identification. Only PL6 consistently produced signals across all four metrics (ODMR, ZPL, POL, SAT) under the chosen conditions. LTPL contributes mainly to identifying PL4, and saturation count rate alone cannot separate PL4 from other types. PL7 fell below the applied count-rate threshold and PL8 was not resolved due to lower spin contrast at the chosen laser power and integration time; only one PL4 ODMR signal was observed because of low room-temperature spin contrast.

• Benchmarking against prior formation routes. A focused helium ion beam study (10 × 10 array, 300 ions/spot) showed ODMR and low-temperature PL spectra in 98% of locations, but PL6 in only 5 locations with just 1 designated a single emitter, and no PL5 recorded — a single-emitter yield of ~0.007%, below this work. Masked carbon implantation reached ~0.7% per-ion yield at 20 ions/spot, with ~14% overall single-emitter site fraction and per-site fractions of 7% PL5 and 1% PL6. The authors note that species-specific yields remain at the few-percent level for all approaches in which no oxygen is supplied, consistent with the recent assignment of PL5–8 to oxygen-vacancy complexes.

Methodology in Plain English

Writing the arrays. A focused ion beam tool (Ionoptika QOne) running a eutectic BiSn liquid metal alloy ion source was used at the UK National Ion Beam Centre. A Wien filter picks out the desired mass-to-charge state, so either Bi or Sn, singly or doubly charged, can be selected — and even a specific isotope (¹²⁰Sn, ²⁰⁹Bi). A fixed 25 kV anode voltage gives 25 keV for singly and 50 keV for doubly charged ions. An electrostatic chicane removes neutral atoms that would otherwise land uncontrolled. The beam was pulsed with a 10 ns-rise-time electrostatic blanker to deliver a target ion count per spot, with uncertainty set by Poisson statistics and beam-current stability (current change before and after implantation was typically under 2%). Nine identical samples were implanted with 10 × 10 arrays at doses from 10 to 200 ions/spot.

Annealing and cleaning. Eight samples were annealed in a tube furnace under argon from 600 °C to 1100 °C with a 2-hour dwell and 5 °C/min ramp; one was kept unannealed as a reference. Samples were then cleaned in an IPA ultrasonic bath (20 min) and Piranha acid (15 min, 3:1 concentrated sulfuric acid to 30% hydrogen peroxide).

Finding the emitters. Photoluminescence mapping used a home-built confocal microscope with 940 nm off-resonant excitation, chosen to suppress competing silicon-vacancy (V_Si⁻) emission. The detection path (950 nm dichroic, 1000 nm long-pass, 1326 nm short-pass) covers the room-temperature phonon sidebands of PL1–8 while rejecting V_Si⁻. Spots off the implantation grid were discarded as presumably naturally occurring emitters; distorted spot shapes were treated as multiple emitters. Brightness selection used a fixed threshold of >40 kcts/s at 5 mW excitation.

Confirming "single". Each surviving spot went to a Hanbury-Brown–Twiss measurement; g⁽²⁾(0) < 0.5 defined a single emitter. Count-rate histograms then allowed the authors to compute how many genuine singles survive as the intensity threshold is raised — quantifying the trade-off between yield and confidence.

Modelling the dose. Count rates of confirmed singles set an adaptive threshold for counting multi-emitter spots, splitting each array into singles, multiples (≥2) and failures. Each spot was treated as an independent trial with a Poisson mean λ = αd scaling linearly with dose, and α was fitted by maximum likelihood.

Identifying the type. Four measurements were added per candidate: zero-field ODMR over 1300–1400 MHz (Lorentzian fits, with M = 1 for PL6 and M = 2 for PL5), low-temperature PL at 4 K over 1030–1140 nm with an 80 GHz resolution spectrometer (910 nm excitation), polarisation dependence measured as (I_max − I_min)/(I_max + I_min), and a saturation count-rate measurement. Literature values (PL4: 1334 MHz, 1078 nm, no polarisation dependence; PL5: 1344 and 1375 MHz, 1041.9 nm, polarisation-dependent; PL6: 1352 MHz, 1037.7 nm, no polarisation dependence) provided the reference grid, and pairwise scatter plots revealed the clustering that makes rapid assignment possible.

Why This Matters

Impact on research. The work turns heavy-ion implantation from an ensemble-level materials study into a positioned, single-emitter fabrication technique, and it does so with a maskless FIB process compatible with existing nanophotonic and pin-diode integration. The Poisson occupancy model gives other groups a transferable way to choose doses rather than guess them, and the tiered characterisation scheme removes the cryostat from the critical path for PL6 — the emitter most often compared to the NV⁻ centre in diamond, combining long spin coherence, near-infrared emission in the second biological window, and high ODMR contrast at room temperature.

Real-world applications:

• Quantum sensing. Near-surface, room-temperature spin-active emitters with high ODMR contrast are directly relevant to nanoscale magnetometry and thermometry.
• Quantum communication and photonic integration. Deterministic placement of bright single-photon sources inside waveguides is a prerequisite for on-chip photonic circuits.
• Bio-imaging and metrology. PL6 emission sits in the second biological window, making it attractive for sensing in aqueous and biological environments.
• Process metrology for semiconductor fabs. The yield-versus-dose and yield-versus-temperature curves, plus the finding that FIB imaging resolution underestimates implant scatter, are practical inputs for anyone specifying a fabrication flow.

Industry relevance. SiC is a wafer-scale industrial semiconductor already served by mature processing. The use of a commercial FIB tool with a liquid metal alloy source, an electrostatic chicane for neutrals, and standard tube-furnace annealing means the process is compatible with existing foundry-adjacent infrastructure rather than requiring bespoke equipment. The authors' own comparison table against helium-FIB and masked-carbon routes gives process engineers a basis for choosing between throughput, per-ion yield, and positional accuracy.

Future Directions

• Directly boosting species-specific yield. The authors propose maskless array generation using oxygen-containing gas species delivered through the Ionoptika QOne duoplasmatron source, which should raise PL5 and PL6 yields while retaining positional control — a direct response to the assignment of PL5–8 as oxygen-vacancy complexes.
• Operating at higher dose. Because the study deliberately stayed in the low-dose regime (λ = 0.16 for Bi²⁺), reaching larger numbers of single emitters requires larger doses under the same annealing conditions, at the cost of more multiply occupied sites. The model predicts where that trade-off peaks but the prediction is untested experimentally here.
• Reducing residual strain. ODMR scatter of 4.7 MHz (PL6) and up to 10.1 MHz (PL5 dip 1) is attributed to strain from dense collision cascades surviving annealing, with prior work reporting complete recovery only at substantially higher temperatures. Whether higher-temperature or alternative annealing can narrow these distributions without destroying the emitters is open.
• Extending the scheme to PL7 and PL8. These oxygen-vacancy family members were not detected under this study's measurement parameters — PL7 fell below the count-rate threshold and PL8 was unresolved due to lower spin contrast at the chosen laser power and integration time — leaving their inclusion in a rapid screening protocol an open question.

Target Audience

This paper is most useful to researchers in quantum photonics and defect-based quantum technology who are building fabrication flows around SiC colour centres, and to ion-beam and nanofabrication specialists interested in what focused heavy-ion implantation can and cannot deliver in terms of placement accuracy. It also suits materials scientists studying implantation damage and annealing recovery in 4H-SiC, and semiconductor process engineers evaluating whether maskless FIB writing is a viable route to integrated quantum devices. Readers wanting only the screening protocol can take the tiered scheme and thresholds directly; readers wanting the formation physics will find the dose-yield model and the annealing window most valuable.

Authors’ abstract

Single photon emitters in silicon carbide (SiC) are a leading platform for scalable quantum technologies. Recent interest has focused on oxygen-vacancy-related emitters, which show exceptionally high optical brightness and strong spin readout contrast. One barrier to scalable quantum devices based on these emitters is the challenge of maskless formation and rapid identification. Here, we demonstrate the formation of isolated bright single emitters in 4H-SiC, using low-energy maskless implantation of heavy ions bismuth and tin. Following annealing, up to 18% of implanted sites host a single emitter, with optimal yields achieved at annealing temperatures of 900-1000 degrees C. Occupancy statistics are modelled to estimate the implantation dose that maximises single-emitter yield. We introduce a tiered characterisation scheme, where a simple intensity threshold isolates single-emitter candidates, confirmed through photon correlation measurements, after which correlations between polarisation, saturation count rate and magnetic resonance frequency assign emitter type. It is shown that time-consuming low-temperature spectroscopy is not necessary to distinguish emitter types. Together, maskless heavy-ion implantation and selective screening offer an efficient route to forming and rapidly identifying near-surface single emitters for room-temperature quantum technologies such as quantum sensing.