Storage of telecom-band time-bin qubits in thin-film lithium niobate
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.
Kai: I'm Kai.
Mira: And I'm Mira, here with Lev.
Kai: We're looking at the paper: "Storage of telecom-band time-bin qubits in thin-film lithium niobate".
Mira: Here's the gist of the paper. Storage of telecom-band time-bin qubits in thin-film lithium niobate is demonstrated by realizing the first on-chip quantum memory using erbium ions doped TFLN The developed quantum memory achieves a storage time of 400 ns with an efficiency of 1.95% ± 0.04% The multimode capability is demonstrated by successfully storing four temporal modes Furthermore, single-photon-level coherent pulses are encoded into time-bin qubits and stored with a fidelity of 96.8% ± 0.3%, surpassing the classical limit achievable by measure-and-prepare strategy Device Fabrication and Preparation The quantum memory is implemented in an Er3+ doped waveguide fabricated on a commercially available x-cut TFLN wafer with a thickness of 500 nm, doped with 0.5 mol% Er3+ ions The Er3+:TFLN layer was bonded onto a 4.7 µm silica layer, which was grown on a 525 µm thick silicon substrate The waveguide has a length of 5 mm, with light coupled in and out through two grating couplers The coupling efficiency of the grating coupler is about 12.5% at 1531.6 nm Spectroscopic Characterization To achieve a long optical coherence time, the device is operated at cryogenic temperatures to suppress phonon-induced decoherence, and a strong magnetic field is applied to slow the Er3+ spin-flip process The measured optical T1 is 2.78 ± 0.02 ms, which is consistent with previous reports The measured results are shown in Fig.
Kai: 2(b) A coherence time of 17.48 ± 1.69 µs is obtained from the photon echo measurement The shortest decay time obtained from the fit for the spectral hole lifetime is estimated to be 1.95 ± 1.09 s Atomic Frequency Comb Storage The AFC scheme relies on spectral hole burning techniques to periodically tailor the inhomogeneously broadened absorption profile of Er3+ ions into a series of narrow, equally spaced absorption peaks The comb spacing corresponds to a memory time of τ = 400 ns The observed memory efficiency is 1.95% ± 0.04%, with a signal-to-noise ratio of 56.3 ± 7.0 The AFC protocol inherently supports temporal multimodes, where the maximum number of storable temporal modes is determined by the ratio of the mode duration to the storage time Time Bin Qubit Encoding and Fidelity The input pulses were encoded as time-bin qubits, achieving a memory fidelity up to 96.8% ± 0.3% for an input mean photon number of µin = 0.578 The total conditional fidelity of the retrieved qubit FT is found to be 97.3% ± 0.
Title and authors: Mira: Let's look at the title and authors of this paper, "Storage of telecom-band time-bin qubits in thin-film lithium niobate." The focus here is clearly on using TFLN for a task that hasn't been done before.
Kai: It’s important to see the authors because they are working at the intersection of quantum optics and device physics, which tells you they know how to build things that actually work in a lab setting.
Lev: I’m curious about their specific expertise; if you’re building hardware, it matters if they understand the material limitations of TFLN versus standard silicon or other platforms.
Mira: They are focused on realizing quantum memory in TFLN, which is the material itself that has gained attention for its exceptional optical properties in integrated devices.
Kai: So the core idea is moving from just using light through a waveguide to actually storing a quantum state inside the material using erbium ions.
Lev: I wonder if this realization is scalable beyond this small five millimeter waveguide length they used, because real quantum repeaters need much longer storage times than four hundred nanoseconds.
Mira: The paper points out that the TFLN layer is bonded onto a silica layer which is on top of a silicon substrate, setting up the entire integrated structure.
Kai: And they did use grating couplers for coupling light in and out, which means this memory could potentially be integrated with other components on the same chip.
Lev: Those coupling efficiencies are important; twelve point five percent at one thousand five hundred thirty-one nanometers is a practical number for connecting to standard telecom fibers.
The paper's summary: Kai: Now, looking at the summary of this paper, it really boils down to realizing the first on-chip quantum memory using erbium ions doped TFLN. They are tackling the challenge of making quantum memory a practical building block for a quantum internet.
Mira: They achieved a storage time of four hundred nanoseconds with an efficiency around one point nine five percent and demonstrated that they can store four temporal modes at once, which shows some multimode capability.
Lev: Four hundred nanoseconds is the key metric here; it gives us a baseline for how long we can hold a photon before we have to worry about decoherence or loss.
Kai: They also used atomic frequency comb storage, which relies on spectral hole burning techniques to tailor the absorption profile of those ions into narrow, equally spaced peaks.
Mira: That AFC protocol inherently supports temporal multimodes because the maximum number of storable modes depends on the ratio between that mode duration and the storage time.
Lev: The signal-to-noise ratio they report is fifty six point three plus seven point zero, which tells us how much information we can reliably extract from these stored states.
Kai: And for those single photon pulses encoded as time-bin qubits, the fidelity reached ninety six point eight percent, beating the classical limit of measure and prepare strategies.
Lev: That beating of the classical limit is what makes this result interesting; it means they've managed to use a physical medium to achieve better performance than just trying to track and recreate the quantum state classically.
The paper's improvements: Mira: One big implication they suggest is using AI systems to implement on-chip quantum repeaters by integrating this TFLN photonic quantum memory with low-loss waveguides, high-performance electro-optic modulators, and single photon detectors for chip scale quantum repeaters.
Kai: That sounds like a huge step toward realizing scalable networking; the improvement suggests that the next step is tying this memory into an active system using AI control.
Lev: From my side, if we were to run this on real hardware, we would immediately have to account for the loss from those grating couplers and the coupling efficiency of twelve point five percent.
Mira: The second improvement they highlight is that this new system can achieve long time delays with substantially lower loss and a significantly reduced footprint compared to conventional waveguide delay lines by leveraging their four hundred nanosecond storage time with one point nine five percent efficiency.
Kai: That comparison to traditional delay lines is telling; it implies we can build longer quantum circuits on a much smaller chip because the memory itself is so efficient at holding the state.
Lev: A reduced footprint is crucial for integration, but we need to be careful about how that efficiency holds up when you scale up the size of the Er three plus doped waveguide.
Conclusion: Kai: So to wrap this up, this paper on "Storage of telecom-band time-bin qubits in thin-film lithium niobate" shows we can actually build on-chip quantum memory using erbium ions doped TFLN.
Mira: The key findings are the four hundred nanosecond storage time and that ninety six point eight percent fidelity for encoded pulses, along with the multimode capability they demonstrated.
Lev: What this means for us is that we have a physical mechanism to store quantum information on a chip, which is necessary groundwork for moving toward fault-tolerant quantum hardware.
Kai: I think the integration into chip-scale repeaters via AI control, as suggested in the improvements section, is where the real near-term potential lies for networking.
Mira: And we have to keep in mind their limitation: they only demonstrated storage time of four hundred nanoseconds, and they also showed that one point nine five percent efficiency.
Lev: That low efficiency means any real implementation will need significant overhead just to get the desired performance, so scaling this up is going to be a huge engineering hurdle.
Kai: It’s interesting how they managed to beat the classical limit in fidelity, but we still need to figure out how to make that memory robust enough for larger systems.
Mira: This paper on "Storage of telecom-band time-bin qubits in thin-film lithium niobate" gives us a concrete platform, but the next step is moving from proof of concept to a system that can handle more complex interactions.
Xiao-Jie Wang, *, Yong-Teng Wang, *, Zi-Wei Zhao, Yong-Min Li, Tian-Shu Yang
State Key Laboratory of Quantum Optics Technologies and Devices, Institute of Opto-Electronics, Shanxi University
quant-ph, physics.optics
Submitted: 2026-05-13
Updated: 2026-05-13
Comments: 6 pages, 4 figures
Journal ref: ACS Photonics (2026)
DOI: 10.1021/acsphotonics.6c01734
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
Importance score: 83/100
The gist: Storage of telecom-band time-bin qubits in thin-film lithium niobate is demonstrated by realizing the first on-chip quantum memory using erbium ions doped TFLN The developed quantum memory achieves a…
Key concepts
- Time-Bin Qubits
- These are quantum bits that encode information based on the arrival time of a photon in two different paths or 'bins'. By using these qubits, researchers can store and manipulate quantum information, which is essential for building quantum computers.
- Erbium Ions Doped TFLN
- This is the physical material used to create the quantum memory. Thin-film lithium niobate (TFLN) is a semiconductor wafer doped with erbium ions, which act as the storage medium for storing light and quantum information.
- Atomic Frequency Comb (AFC)
- The AFC scheme uses spectral hole burning techniques to create a series of narrow, equally spaced absorption peaks in the Er3+ ions. This allows for periodic tailoring of the material's absorption profile, which directly dictates the memory time, achieving 400 ns storage.
- Memory Efficiency
- This metric measures how successfully the quantum memory stores and retrieves information. In this experiment, an efficiency of 1.95% means that for every photon stored, only about two photons are retrieved with high fidelity.
Terminology
Summary
Storage of telecom-band time-bin qubits in thin-film lithium niobate is demonstrated by realizing the first on-chip quantum memory using erbium ions doped TFLN The developed quantum memory achieves a storage time of 400 ns with an efficiency of 1. p. 0 In this study, we realized the
first on-chip quantum memory using erbium ions
doped TFLN. The developed quantum memory
achieves a storage time of 400 ns with an efficiency of 1. The
multimode capability is demonstrated by successfully storing four temporal modes. Furthermore, single-photon-level coherent pulses are encoded into time-bin qubits and stored with a fidelity of 96.Where the paper says it
Where the paper says it
Device Fabrication and Preparation
The quantum memory is implemented in an Er3+ doped waveguide fabricated on a commercially available x-cut TFLN wafer with a thickness of 500 nm, doped with 0.5 mol% Er3+ ions The Er3+:TFLN layer was bonded onto a 4. p. 1 5 mol% Er3+ ions. The Er3+:TFLN layer
was bonded onto a 4. 7 µm silica layer, which was grown
on a 525 µm thick silicon substrate. The waveguide has
a length of 5 mm, with light coupled in and out through
two grating couplers, as shown in Fig. The
coupling efficiency of the grating coupler is about 12. 5%
at 1531.Where the paper says it
Where the paper says it
Where the paper says it
Spectroscopic Characterization
To achieve a long optical coherence time, the device is operated at cryogenic temperatures to suppress phonon-induced decoherence, and a strong magnetic field is applied to slow the Er3+ spin-flip process The measured optical T1 is 2.78 ± 0.02 ms, which is consistent with previous reports The measured results are shown in Fig. p. 1 02 ms, as shown
in Fig. 2(b), which is consistent with previous reports
[20, 24, 29]. The relevant energy-level structure
of Er3+ in TFLN used in this study is shown in the inset of Fig. To achieve a long optical coherence
time, the device is operated at cryogenic temperatures
to suppress phonon-induced decoherence, and a strong
magnetic field is applied to slow the Er3+ spin-flip process [24, 28]. To investigate
the ground-state lifetime, we measured the temporal decay of the spectral hole area, as shown in Fig. The
shortest decay time obtained from the fit is 1.Where the paper says it
Where the paper says it
Where the paper says it
Atomic Frequency Comb Storage
The AFC scheme relies on spectral hole burning techniques to periodically tailor the inhomogeneously broadened absorption profile of Er3+ ions into a series of narrow, equally spaced absorption peaks The comb spacing corresponds to a memory time of τ = 400 ns The observed memory efficiency is 1.95% ± 0.04%, with a signal-to-noise ratio of 56. p. 1 The observed memory efficiency
is 1. 04%, with a signal-to-noise ratio of 56.Where the paper says it
Time Bin Qubit Encoding and Fidelity
The input pulses were encoded as time-bin qubits, achieving a memory fidelity up to 96.8% ± 0.3% for an input mean photon number of µin = 0. p. 0 Moreover, the
input signals were encoded as time-bin qubits, achieving a memory fidelity up to 96. 3% for an input
mean photon number of µin = 0. 2% is obtained. Therefore, the total conditional fidelity of the retrieved qubit FT =
1
3
Fel +
2
3
F+−
is found to be 97.Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improved AI systems can implement on-chip quantum repeaters by integrating TFLN photonic quantum memories with low-loss waveguides, high-performance electro-optic modulators, and single-photon detectors for chip scale quantum repeaters.
-
The improved system can achieve long time delays with substantially lower loss and significantly reduced footprint compared to conventional waveguide delay lines by leveraging the realized on-chip quantum memory performance of 400 ns storage time with 1.95% efficiency.
-
The AI system can utilize the AFC protocol for on-demand readout, which is realized by employing a Stark modulated AFC protocol, allowing for demand readout capability while maintaining high storage efficiency and longer lifetime.
Abstract
Integrated photonics has emerged as a promising platform for quantum communication and quantum computation. Thin-film lithium niobate (TFLN) has gained significant attention in this field due to its exceptional optical properties, enabling the realization of numerous integrated photonic devices. However, quantum memory, which serves as a universal building block for the quantum internet, has not yet been demonstrated in TFLN. In this study, we realized the first on-chip quantum memory using erbium ions doped TFLN. The developed quantum memory achieves a storage time of 400 ns with an efficiency of 1.95%, significantly outperforming conventional waveguide delay lines. The multimode capability is demonstrated by successfully storing four temporal modes. Furthermore, single-photon-level coherent pulses are encoded into time-bin qubits and stored with a fidelity of 96.8%, surpassing the classical limit achievable by measure-and-prepare strategy. Our results demonstrate the first on-chip quantum memory for telecom-band time-bin qubits in TFLN, providing a key building block toward integrated quantum registers and repeaters for scalable quantum information processing.
Sources
Related papers
- Computational regimes in matrix-product-state-based quantum trajectory simulations
- Toward chemically accurate quantum simulations on quantum hardware: A quantum-HPC hybrid approach
- Transversal surface-code game with reconfigurable qubits
- Fulqrum: a generalized framework for quantum subspace diagonalization
- Flux-Activated Resonant Control of a Bosonic Quantum Memory
- Machine learning the effects of many quantum measurements