Quantum Photonic Sources & Demultiplexing

• Quantum photonic sources are deterministic single-photon emitters (e.g., InAs QDs) offering high brightness (~21.5%) and near-ideal purity (g^(2)(0)≈0.005).
• Demultiplexing networks use cascaded Mach–Zehnder interferometers for GHz-speed, low-loss routing with insertion loss as low as 0.8 dB per switch and 96.2% switching fidelity.
• The integrated system leverages heterogeneous integration of quantum dots and LNOI circuits to scale photonic quantum processors for advanced computing and communication tasks.

Quantum photonic sources and demultiplexing networks represent a technologically advanced intersection of single-photon generation, integrated nanophotonics, and high-speed switching, enabling scalable architectures for photonic quantum computing and multi-photon quantum networks. These systems typically comprise deterministic emitters—most often semiconductor quantum dots (QDs) in engineered nanostructures—coupled directly to reconfigurable or passive photonic circuits, facilitating temporal-to-spatial or spatio-temporal routing of photonic qubits with high fidelity, low loss, and multi-GHz speed. This article provides a comprehensive overview of key device principles, performance metrics, architectural implementations, and comparative platform analysis, with a particular focus on the high-speed lithium-niobate-on-insulator (LNOI) quantum processor interfaced to solid-state quantum-dot sources (Sund et al., 2022).

1. Deterministic Quantum Photonic Sources: Quantum Dot Emitters

Semiconductor quantum dots embedded in nanophotonic structures function as nearly ideal two-level emitters under pulsed resonant excitation, supporting on-demand single-photon generation with high brightness, purity, and indistinguishability. The architecture of (Sund et al., 2022) employs a self-assembled InAs QD located centrally in a single-sided photonic-crystal waveguide, terminated by a “mirror” hole array for enhanced emission directionality and a shallow-etched grating coupler for fiber-mode interfacing. A distributed Bragg reflector (DBR) below the membrane shapes the far-field emission profile for efficient vertical extraction.

Performance metrics:

• Brightness: Defined as the ratio of fiber-coupled photons to laser triggers, $$\eta = \frac{P_{\rm coupled}}{P_{\rm excited}} \approx 21.5\%$$ (15.6 million photons/s at 72.6 MHz pulse rate).
• Purity: $\displaystyle g^{(2)}(0)=0.005\pm0.001$, corresponding to 99.5% single-photon emission probability—nearly the quantum limit.
• Indistinguishability: Measured by Hong–Ou–Mandel visibility, $V_{\rm TPI}=94.5\%\pm1.7\%$ (two-photon interference), with negligible degradation from on-chip waveguide coupling ($V_{\rm HOM}=92.7\%$).
• Emission bandwidth: Set by radiative lifetime $T_1\sim1$ ns, yielding $\Delta\nu\approx160$ MHz.

Deterministic operation is achieved via resonant pulsed excitation, ensuring each laser pulse yields a single photon emission. The integrated device supports direct, low-loss interfacing to LNOI photonic circuits at 900–950 nm, with waveguide loss $\leq0.84$ dB/cm and grating coupler loss $-3.4$ dB. The quantum dot–guided mode interaction is captured in the rotating-wave Hamiltonian,

$$H_{\rm int} = \hbar\,g\left(a^\dagger \sigma^- + a\,\sigma^+\right),$$

where $a$ annihilates a waveguide photon and $\sigma^-$ represents the QD dipole transition. The coupling rate $g$ follows a full vectorial mode-overlap integral.

2. Integrated Photon Demultiplexing Networks

Temporal-to-spatial demultiplexing networks map the sequential emission from a single source into multiple spatial channels, constructing multi-photon resources from single-photon streams. In (Sund et al., 2022), a $1\times4$ demultiplexer is realized by cascading three Mach–Zehnder interferometers (MZIs) to perform binary tree routing. Each MZI applies a rapid electro-optic phase shift using the linear Pockels effect in lithium niobate: $\Delta n(E) = -\frac{1}{2} n^3 r_{33} E,$ with $r_{33}=30$ pm/V and $n=2.2$, giving $V_\pi\approx4.5$ V for modulators of length $L=1.25$ mm and voltage-length product $V_\pi L \approx 0.6$ V·cm. The measured bandwidth $f_{3\,\mathrm{dB}}\approx6.5$ GHz supports GHz-class switching. The unitary transformation per MZI is

$$U_{\rm MZI}(\phi) = \frac{1}{\sqrt{2}} \left( \begin{matrix} 1 & i \ i & 1 \end{matrix} \right) \left( \begin{matrix} e^{i\phi} & 0 \ 0 & 1 \end{matrix} \right) \frac{1}{\sqrt{2}} \left( \begin{matrix} 1 & i \ i & 1 \end{matrix} \right),$$

with the routing path selected by phase settings $\{\phi_1,\phi_2,\phi_3\}$ per switching event.

Key network metrics:

• Per-channel insertion loss: Input grating (3.4 dB) + output grating (3.4 dB) + two MZIs (2×0.8 dB) = $8.4$ dB total.
• Extinction ratio per MZI: 21 dB.
• Crosstalk: $-14.2$ dB channel isolation.
• Switching fidelity: 96.2% per four-photon sequence.
• Network fidelity: $$F = \frac{1}{N}\left|\mathrm{Tr}\left(U_{\rm ideal}^\dagger U_{\rm exp}\right)\right|, \qquad F_{\rm demux} \approx 0.962.$$

3. Switching Physics, Losses, and Scaling

The electro-optic phase modulation in LNOI permits sub-nanosecond, GHz-speed switching with minimal transmission penalty ($0.8$ dB per MZI), significantly outperforming carrier-depletion SOI ($\gtrsim1$ dB), piezo-optomechanical ($>3$ dB), or MEMS-based ($>\mathrm{MHz}$ speed) platforms. Channel isolation is achieved via MZI symmetry and electrode design, suppressing unintended routing to $<-14$ dB.

The scaling behavior of demultiplexers is dependent on switch fidelity and loss accumulation. For $N$ output channels, photon coincidence rates scale as

$$R_N = R_{\rm rep} \cdot (\eta_{\rm source} \cdot \eta_{\rm demux} \cdot \eta_{\rm det})^N.$$

In the benchmark device, this yields observable four-photon rates at high fidelity; larger networks would require additional switch stages but maintain low insertion loss per routing event if MZI depth remains quasi-logarithmic in $N$.

Comparison table (selected metrics):

Platform	Modulation Speed	$V_\pi L$ (V·cm)	Ins. loss (per MZI)	Source $\eta$ (%)	$g^{(2)}(0)$	Indistinguishability (%)
LNOI-QD (Sund et al., 2022)	6.5 GHz	0.6	0.8 dB	21.5	0.005	94.5
Si carrier	$\sim$ GHz	few	1–2 dB	1–2*	0.02–0.05	80–85
SiN piezo	120 MHz	50	>3 dB	–	–	–

*SPDC typical, not QD.

4. Platform Comparison and Performance Benchmarks

The LNOI-QD hybrid demonstrates an order-of-magnitude gain over SPDC-based sources in brightness and purity. Whereas traditional parametric sources yield $\eta\sim1$–2%, $g^{(2)}(0)=0.02$–0.05, and lower indistinguishability, deterministic QDs in engineered nanophotonic structures routinely provide $\eta\gtrsim20\%$, $g^{(2)}(0)\sim0.005$, and $V_{\rm HOM}>90\%$. Loss per switch is sub-dB (0.8 dB), supporting scalable architectures with total network loss $<10$ dB up to four channels. Network fidelity ($>96\%$) and crosstalk ($-14$ dB) favor application in high-visibility quantum interference experiments and multi-photon protocols.

Performance advantages:

• Sub-GHz to multi-GHz reconfiguration speed for real-time demultiplexing.
• Low per-channel loss ($<1$ dB) enables multi-photon rates orders of magnitude above passive splitter trees or inefficient SPDC sources.
• High indistinguishability preserved through all on-chip processes.

5. Theoretical Frameworks for Coupling and Switching

The photonic coupling of QD emitters proceeds from rigorous quantum-optical Hamiltonians. The Jaynes–Cummings model in the rotating-wave approximation is applicable for QD–cavity or QD–waveguide systems,

$$H_{\rm int} = \hbar g \left(a^\dagger \sigma^- + a \sigma^+\right),$$

with $g$ calculated from spatial mode overlap as above. Electro-optic switching is described by field-induced refractive index change, with the required voltage and modulator design derived from material Pockels coefficients and the overlap integral of microwave and optical fields. Bandwidth limitations arise primarily from velocity mismatch and electrode RC constant, setting the attainable $f_{3\,\mathrm{dB}}$.

6. Technological Implications and Scalability

The convergence of deterministic single-photon sources and integrated GHz-switching networks marks a pivotal advance for scalable quantum photonic platforms. The heterogeneous integration strategy—merging III–V quantum emitters with thin-film lithium niobate—enables high-brightness, high-fidelity, actively reconfigurable on-chip quantum state generation and distribution. This approach is extensible to tree-like demultiplexing architectures, universal multi-mode photonic circuits, and complex quantum network protocols, supporting the demands of boson sampling, quantum simulation, and linear-optics quantum computation.

A plausible implication is that with continued advances in switch miniaturization and network optimization, multiplexed quantum-dot sources interfaced to high-speed low-loss LNOI switches will provide the backbone for near-term fault-tolerant photonic quantum processors and distributed quantum networks, overcoming longstanding limitations of probabilistic parametric sources and passive splitter architectures (Sund et al., 2022).

7. Summary of Key Results and Outlook

The integrated quantum processor described in (Sund et al., 2022) achieves:

• Deterministic quantum-dot emission with $\eta=21.5\%$, $g^{(2)}(0)=0.005$, and two-photon visibility $94.5\%$.
• On-chip demultiplexing by cascaded MZI switches operating at 6.5 GHz, with $0.8$ dB per switch insertion loss and fidelity $0.962$.
• Total network loss of $\sim8.4$ dB, channel crosstalk suppressed to $-14.2$ dB, switching fidelity $96.2\%$ for four-photon routing.
• Comparative performance exceeding parametric sources and alternative integrated switching platforms (Si, SiN, MEMS).
• Architectural scalability to larger $N$ leveraging low-loss, high-speed MZI binary trees.

These features collectively establish a robust experimental foundation for scalable quantum photonic circuits based on solid-state emitters and integrated lithium niobate, opening pathways to multi-photon quantum information processing, quantum communication, and photonic quantum computing.