46 papers, 88 rows
- VπL median
- 2.44 V cmn = 66
- Bandwidth median
- ≥ 70 GHzn = 67, 42% bounds
- Electrode median
- 6.4 mmn = 77
Atlas statistics only.
lithium niobate / 08 Industry / 8.4
8.4 Platform comparison from the atlas
L1 One device per mark
Each mark below is one published device; LN and its sibling LT are highlighted. Hover a mark for its paper, values, and evidence file.
SiPh: 0 rows with push-pull VπL
1 row outside the axes (counted, not drawn). Hollow: convention other than push-pull; arrows: bandwidth lower bound; points enter at their year. Defaults: telecom bands, MZM and IQ device rows, push-pull VπL.
atlas eb97f96 · 2026-10-20 · fetched 2026-10-25 github.com/jwt625/eo-modulator-atlas, no license file upstream cross-check: photonics-atlas 26aa340 · 2026-10-23
SiPh: 0 rows with push-pull VπL
1 row outside the axes (counted, not drawn). Hollow: convention other than push-pull; arrows: bandwidth lower bound; points enter at their year. Defaults: telecom bands, MZM and IQ device rows, push-pull VπL.
atlas eb97f96 · 2026-10-20 · fetched 2026-10-25 github.com/jwt625/eo-modulator-atlas, no license file upstream cross-check: photonics-atlas 26aa340 · 2026-10-23
L2 Readings
- TFLN telecom push-pull VπL median 2.44 V cm (n = 66) vs TFLT 2.43 (n = 17); TFLN bandwidth median ≥ 70 GHz (n = 67, 42% lower bounds), maximum 170 GHz derived.
- TFLT papers: 4 (2024), 4 (2025), 14 (2026; 5 of them OFC-tagged). TFLT holds the atlas line-rate maximum (768 Gb/s, Xu et al., OFC 2026 Th4A.2).
- Highest symbol rate per year for TFLN rose from 100 GBd (2018) to 225 GBd (2026); BTO reached 256 GBd (2025), SiPh 200 GBd (2025), InP 200 GBd (2026), polymer 224 GBd (2026).
- Footprint: median electrode length TFLN 6.4 mm, TFLT 7.5, InP 4, BTO 1.38, SiPh 0.9, polymer 0.28 mm.
- Wafers: 50 of 56 TFLN/TFLT papers that name a supplier name NanoLN (TFLN 41 of 43, TFLT 9 of 13).
- Integration: in 2026, 3 of 36 TFLN/TFLT papers report a foundry-native process, 3 bonded-heterogeneous, 2 micro-transfer-printed, and 10 have O-band rows (2022: 0 foundry-native, 1 O-band of 17).
Neighbors and rivals
| Platform | Definition | Papers / rows | VπL median | n | Bandwidth median | n | Bounds | Electrode median | n | Note |
|---|---|---|---|---|---|---|---|---|---|---|
| TFLN | Thin-film LN | 46 / 88 | 2.44 V cm | 66 | >= 70.0 GHz | 67 | 42% | 6.40 mm | 77 | atlas only |
| TFLT | Thin-film LT | 15 / 21 | 2.43 V cm | 17 | >= 85.0 GHz | 10 | 40% | 7.50 mm | 20 | LT is already used in 5G RF filters (EPFL and SIMIT, Nature 2024) reported |
| SiPh | Silicon (carrier depletion) | 14 / 25 | no push-pull rows | 0 | >= 66.0 GHz | 21 | 29% | 0.90 mm | 16 | atlas only |
| InP | InP | 6 / 12 | 0.70 V cm | 3 | >= 67.0 GHz | 9 | 44% | 4.00 mm | 9 | atlas only |
| BTO | Barium titanate | 7 / 9 | 0.72 V cm | 2 | >= 70.0 GHz | 5 | 20% | 1.38 mm | 8 | PsiQuantum's Omega chipset uses BTO switches on SiN C |
| Polymer | EO polymer (incl. SOH/POH) | 12 / 36 | 0.09 V cm | 9 | >= 80.3 GHz | 8 | 13% | 0.28 mm | 34 | atlas only |
46 papers, 88 rows
15 papers, 21 rows
- VπL median
- 2.43 V cmn = 17
- Bandwidth median
- ≥ 85 GHzn = 10, 40% bounds
- Electrode median
- 7.5 mmn = 20
LT is already used in 5G RF filters (EPFL and SIMIT, Nature 2024) reported
14 papers, 25 rows
- VπL median
- no push-pull rowsn = 0
- Bandwidth median
- ≥ 66 GHzn = 21, 29% bounds
- Electrode median
- 0.9 mmn = 16
Atlas statistics only.
6 papers, 12 rows
- VπL median
- 0.70 V cmn = 3
- Bandwidth median
- ≥ 67 GHzn = 9, 44% bounds
- Electrode median
- 4 mmn = 9
Atlas statistics only.
7 papers, 9 rows
- VπL median
- 0.72 V cmn = 2
- Bandwidth median
- ≥ 70 GHzn = 5, 20% bounds
- Electrode median
- 1.38 mmn = 8
PsiQuantum's Omega chipset uses BTO switches on SiN C
12 papers, 36 rows
- VπL median
- 0.09 V cmn = 9
- Bandwidth median
- ≥ 80.3 GHzn = 8, 13% bounds
- Electrode median
- 0.28 mmn = 34
Atlas statistics only.
Telecom bands, MZM and IQ device rows; VπL push-pull only; "≥" where bandwidth lower bounds could raise the median derived. Ruler: 10 mm, ticks every 1 mm.
L3 The sample
- The atlas is a curated sample of 233 papers (web search, author-group follow-up, landmark papers, OFC 2026); it is not a census and per-year counts are not growth rates.
- Bounds are drawn as arrows; VπL comparisons use the push-pull convention only; SiPh has no push-pull telecom rows and is absent from the default VπL view.
- Plotted VπL = reported DC VπL, else Vπ × length / 10 (DC), else reported RF VπL; never converted between conventions.
- Platform mapping: TFLN = lithium niobate rows without titanium in-diffusion; TFLT = lithium tantalate; Other includes bulk Ti:LN, AlGaAs, graphene/2D, GeSi EAM, and the remaining materials.
- Source rights in the atlas: 66 papers open-license, 167 restricted (facts and locators only).