When a photonics team evaluates thin-film lithium niobate for a new design, the platform selection decision is often presented as a single choice: TFLN. In practice there are two architecturally distinct approaches — monolithic LNOI (lithium niobate on insulator, where the LN film is the sole waveguide material) and TFLN-on-SiN (where a silicon nitride layer provides the primary optical confinement and the LN film adds electro-optic or nonlinear-optical functionality) — and the right choice depends on which device performance metric is the binding constraint in your system. This article compares the two platforms on waveguide physics, electro-optic performance, nonlinear efficiency, foundry availability at 4 and 6 inch, and typical lead times from a sourcing standpoint.

Terminology First: LNOI and TFLN Are the Same Material

LNOI and TFLN are used interchangeably in the literature to describe the same substrate: a sub-micron lithium niobate film bonded onto a silicon dioxide cladding over a silicon handle. The distinction that matters for platform selection is not the substrate but the waveguide architecture built on top of it. In this article, "LNOI" refers to a monolithic platform where the waveguide ridge is etched directly into the LN film; "TFLN-on-SiN" refers to a hybrid platform where the LN film interacts with a pre-patterned SiN waveguide below or adjacent to it.

Monolithic LNOI: When Electro-Optic Performance Is the Priority

In a monolithic LNOI waveguide, the optical mode is confined primarily within the LN film. This maximises the overlap integral between the optical field and the electro-optic perturbation applied by the RF electrodes — which directly determines the modulation efficiency, expressed as the Vπ·L product. For X-cut LNOI waveguides with optimised electrode geometry, Vπ·L values below 2 V·cm are routinely demonstrated, enabling Mach-Zehnder modulators with Vπ below 2 V and 3dB electro-optic bandwidths exceeding 70 GHz in packaged modules, and exceeding 100 GHz in research devices.

The principal limitation of monolithic LNOI is propagation loss. Dry etching of lithium niobate — necessarily through physical ion milling rather than selective chemical etching — produces waveguide sidewalls with roughness levels that scatter light out of the guided mode. Best-in-class foundries achieve propagation losses below 0.1 dB/cm on ridge waveguides; less mature processes show 0.3–1 dB/cm. For a 3 cm long modulator, even 0.1 dB/cm loss is acceptable (0.3 dB on-chip). For high-Q resonators or delay lines where light must travel several centimetres or more, 0.1 dB/cm is not nearly low enough.

TFLN-on-SiN: When Propagation Loss Is the Binding Constraint

Silicon nitride waveguides fabricated with low-pressure chemical vapour deposition (LPCVD) and a careful thermal annealing schedule can achieve propagation losses below 0.001 dB/cm — two orders of magnitude lower than the best LNOI etch quality achievable today. This is the defining advantage of the TFLN-on-SiN platform: by guiding the optical mode primarily in the SiN and coupling it evanescently into or through a thin LN film deposited above, you access the LN electro-optic and nonlinear coefficients while still benefiting from the low-loss SiN waveguide backbone.

The trade-off is electro-optic efficiency. Because the optical mode extends predominantly into the SiN rather than the LN, the overlap with the electro-optic perturbation is reduced. Vπ·L products in TFLN-on-SiN hybrid modulators are typically 2–5x higher than in equivalent monolithic LNOI devices, meaning higher drive voltages for the same electrode length. For applications where propagation loss is not the binding constraint — wideband modulators for coherent transceivers, for instance — this makes TFLN-on-SiN a less efficient choice than monolithic LNOI.

Where TFLN-on-SiN excels is in applications requiring simultaneously low loss and electro-optic or nonlinear-optical functionality: integrated optical frequency comb generators (microcombs), optical parametric oscillators on chip, electro-optic frequency synthesisers, and narrow-linewidth laser stabilisation loops. These devices require cavity Q factors that only SiN can provide at the required optical power levels, while the LN layer enables the fast electro-optic control or phase-matched nonlinear interactions that SiN alone cannot deliver.

Nonlinear Efficiency Comparison

For second-order nonlinear optical applications — second-harmonic generation (SHG), optical parametric amplification (OPA), and difference-frequency generation (DFG) — monolithic LNOI with periodic poling (PPLN) provides higher nonlinear efficiency than TFLN-on-SiN because the mode is more tightly confined within the χ(2) active material. Phase-matched SHG conversion efficiencies above 1000% W⁻¹cm⁻² have been reported in PPLN waveguides on LNOI. TFLN-on-SiN nonlinear devices see a reduced interaction, though the benefit of lower resonator loss can partially compensate this in a resonantly enhanced geometry.

For third-order (χ(3)) nonlinear applications — soliton frequency combs, four-wave mixing, stimulated Raman — SiN alone is often the better platform, and TFLN-on-SiN adds complexity without a clear efficiency gain for χ(3) processes.

Foundry Availability and Lead Times at 4–6 Inch

The practical foundry landscape as of mid-2026 differs meaningfully between the two platforms. Monolithic LNOI at 4-inch is a commercially established product with multiple active suppliers globally, including both established Western vendors and an expanding Chinese supply base. Six-inch monolithic LNOI is available from a narrower set of foundries with verified uniformity across the larger wafer area. Lead times for monolithic LNOI at 4-inch from a mature foundry are approximately 8–16 weeks for a standard process run; 6-inch adds lead time variability due to tighter yield constraints.

Attribute	Monolithic LNOI	TFLN-on-SiN
Propagation loss	0.05–1 dB/cm (process-dependent)	<0.01 dB/cm (SiN quality)
EO efficiency (Vπ·L)	<2 V·cm (best-in-class)	2–5x higher than LNOI
χ(2) nonlinear efficiency	Higher (mode in LN)	Reduced (mode in SiN)
Best-fit applications	Wideband modulators, PPLN converters	Microcombs, OPOs, EO synthesisers
4-inch foundry availability	Broad (multiple suppliers)	Moderate (specialised foundries)
6-inch foundry availability	Narrower, longer lead time	Limited, early adoption
Typical 4-inch lead time	8–16 weeks	16–28 weeks
CMOS-tool compatibility	Dedicated LN tooling required	SiN CMOS-compatible; LN steps dedicated

TFLN-on-SiN hybrid wafers — where SiN has been patterned and the LN film deposited or bonded on top — require additional process steps and specialised heterogeneous integration capability. The foundry set offering this as a standard product is smaller, and lead times of 16–28 weeks for a 4-inch run are typical at an established heterogeneous integration foundry. Non-standard stack geometries or custom SiN layer thickness specifications add further lead time.

The sourcing implication is that TFLN-on-SiN development timelines are longer not because the physics is more complex, but because the supplier base is less mature and the verification chain is longer. A deep-tech sourcing engagement that maps the qualified foundry set for your specific film stack, confirms available wafer diameter, and verifies propagation loss data on the SiN layer before LN integration can compress programme timelines by eliminating foundry qualification iterations. The sourcing process for a hybrid platform needs to account for verification at each layer, not just the final assembled wafer.

单片LNOI与TFLN-on-SiN是两种架构不同的薄膜铌酸锂平台，平台选择应由系统中的主要约束决定。

单片LNOI将光学模式约束在LN薄膜内，最大化RF-光学重叠因子，从而实现最低Vπ（Vπ·L < 2 V·cm）和最宽电光带宽（封装后 >70 GHz）。传播损耗受干法刻蚀质量限制，优质代工厂可达0.05–0.1 dB/cm。适合宽带调制器、IQ调制器和PPLN非线性波导。4英寸供应商最多，交货周期8–16周；6英寸供应商较少。

TFLN-on-SiN以氮化硅波导为主要导光层（LPCVD SiN可实现 <0.01 dB/cm超低损耗），LN薄膜通过倏逝场耦合提供电光或非线性功能。电光效率比单片LNOI低2–5倍，但高Q谐振腔的低损耗特性对微梳、光参振荡器和窄线宽激光稳频应用至关重要。4英寸需异质集成专用代工厂，交货周期16–28周；6英寸处于早期阶段。

选型原则：传播损耗是主要约束→TFLN-on-SiN；调制带宽/Vπ是主要约束→单片LNOI。深度技术采购可通过提前筛选已验证代工厂、确认可用晶圆尺寸并核验逐层性能数据来压缩项目周期。

單片LNOI與TFLN-on-SiN是兩種架構不同的薄膜鈮酸鋰平台。單片LNOI將光學模式約束在LN薄膜內，最大化電光重疊因子，實現Vπ·L < 2 V·cm和封裝後 >70 GHz帶寬，但傳播損耗受乾法刻蝕限制（0.05–1 dB/cm）。TFLN-on-SiN利用LPCVD氮化矽的超低損耗（<0.01 dB/cm），適合高Q諧振腔應用，但電光效率比單片LNOI低2–5倍。供貨方面：單片LNOI 4英寸供應商最多（交貨8–16週）；TFLN-on-SiN需異質整合代工廠（交貨16–28週）。選型由系統主要約束決定：損耗主導選TFLN-on-SiN，調制效率主導選單片LNOI。深度技術採購可通過提前篩選已驗證代工廠來壓縮項目周期。