Why Sub-THz Matters for 6G
The ITU-R IMT-2030 framework targets peak data rates of 200 Gbps and user-experienced rates of 1 Gbps everywhere. Achieving these numbers within the constraints of current FR1 (<7.125 GHz) and FR2 (24.25–71 GHz) spectrum is physically impossible — there simply is not enough contiguous bandwidth. The sub-THz band from 100 GHz to 300 GHz offers tens of gigahertz of contiguous spectrum, making it the primary candidate for 6G extreme-capacity links.
3GPP has begun studying this range under the working designation FR3, with initial channel model extensions documented in TR 38.901 Section 7.6.1 (frequencies above 100 GHz). The ITU World Radiocommunication Conference 2023 (WRC-23) identified several bands between 102.2 GHz and 275 GHz for possible IMT use, setting the regulatory groundwork.
Sub-THz Spectrum Landscape
The sub-THz range is not uniformly usable. Atmospheric absorption creates deep nulls that segment the spectrum into usable transmission windows separated by absorption peaks.
| Window | Frequency Range | Usable Bandwidth | Primary Absorber | One-Way Attenuation |
|---|---|---|---|---|
| W-band | 75–110 GHz | ~35 GHz | O₂ (weak) | <1 dB/km |
| D-band low | 110–140 GHz | ~30 GHz | Minimal | 0.5–2 dB/km |
| D-band high | 140–170 GHz | ~30 GHz | H₂O (weak at 166 GHz) | 1–4 dB/km |
| G-band window | 200–240 GHz | ~40 GHz | Between H₂O peaks | 2–8 dB/km |
| Upper sub-THz | 270–300 GHz | ~30 GHz | H₂O rising | 5–15 dB/km |
The 183 GHz water vapor absorption line creates a deep null exceeding 100 dB/km attenuation, effectively splitting the sub-THz range. Similarly, the 60 GHz O₂ absorption band (57–64 GHz) attenuates signals by ~15 dB/km, though this sits below the sub-THz range proper.
D-Band: The Sweet Spot
The D-band (110–170 GHz) has emerged as the most practical near-term sub-THz target. Atmospheric attenuation stays below 4 dB/km across most of the band, semiconductor technology (InP HBT, SiGe BiCMOS) can generate sufficient power, and 30+ GHz of contiguous bandwidth is available.
Path Loss at Sub-THz Frequencies
Free-space path loss (FSPL) scales with the square of frequency, per the Friis equation:
`
FSPL(dB) = 20·log10(f) + 20·log10(d) + 20·log10(4π/c)
`
At sub-THz frequencies, this baseline loss is severe. The extended 3GPP channel models in TR 38.901 add stochastic large-scale and small-scale fading components calibrated against measurements up to 150 GHz.
| Frequency | FSPL at 10 m | FSPL at 100 m | FSPL at 1 km |
|---|---|---|---|
28 GHz (FR2) | 71.4 dB | 91.4 dB | 111.4 dB |
140 GHz (D-band) | 85.3 dB | 105.3 dB | 125.3 dB |
300 GHz | 91.9 dB | 111.9 dB | 131.9 dB |
Adding atmospheric absorption and realistic NLoS conditions (per TR 38.901 Table 7.4.1-1 UMi street canyon model extended), total path loss at 300 GHz over 100 m exceeds 130 dB in typical urban environments.
Worked Example 1: Link Budget at 140 GHz (D-Band)
Calculate the achievable SNR for a 140 GHz link over 50 m in an indoor corridor (LoS):
`
Transmit power (Ptx): 10 dBm
Tx antenna gain (1024 elements): 30 dBi
Rx antenna gain (256 elements): 24 dBi
FSPL at 140 GHz, 50 m: 20·log10(140×10⁹) + 20·log10(50) + 20·log10(4π/c)
= 102.9 + 34.0 + (-147.6) [using consolidated form]
= 20·log10(140e9) + 20·log10(50) - 147.6
= 102.9 + 34.0 - 147.6
Alternatively: FSPL = 92.4 + 20·log10(f_GHz) + 20·log10(d_km)
= 92.4 + 20·log10(140) + 20·log10(0.05)
= 92.4 + 43.0 + (-26.0) = 109.4 dB
Atmospheric absorption (0.05 km × 2 dB/km): 0.1 dB
Total path loss: 109.5 dB
Rx noise (BW = 10 GHz, NF = 8 dB): -174 + 10·log10(10×10⁹) + 8
= -174 + 100 + 8 = -66 dBm
Received signal: 10 + 30 + 24 - 109.5 = -45.5 dBm
SNR: -45.5 - (-66) = 20.5 dB
`
At 20.5 dB SNR with 10 GHz bandwidth and 64-QAM (spectral efficiency ~5.5 bps/Hz with LDPC coding), the achievable rate is 10 × 5.5 = 55 Gbps per polarization, or ~110 Gbps with dual-polarization MIMO.
Worked Example 2: Array Size for 300 GHz Outdoor Link
Target: compensate the additional 6.6 dB FSPL penalty of 300 GHz vs 140 GHz at equal distance.
`
FSPL difference: 20·log10(300/140) = 20·log10(2.14) = 6.6 dB
Array gain scales as: G = 10·log10(N) per element (for isotropic)
Additional elements needed: 10^(6.6/10) = 4.57× more elements
If 140 GHz uses 1024 elements → 300 GHz needs ~4,685 elements
`
However, at 300 GHz the wavelength is 1 mm, so element spacing (λ/2 = 0.5 mm) allows 4,096 elements in a 32 mm × 32 mm aperture — physically compact despite the large element count.
Antenna Array Requirements
Sub-THz communications demand massive antenna arrays to overcome path loss. The array gain compensates for the frequency-squared penalty, and the small wavelength enables physically compact arrays.
| Parameter | FR2 (28 GHz) | D-Band (140 GHz) | Upper Sub-THz (300 GHz) |
|---|---|---|---|
| Wavelength (λ) | 10.7 mm | 2.14 mm | 1.0 mm |
| Element spacing (λ/2) | 5.35 mm | 1.07 mm | 0.5 mm |
| Typical array size | 256 elements | 1,024 elements | 4,096+ elements |
| Array aperture (square) | ~86 × 86 mm | ~34 × 34 mm | ~32 × 32 mm |
| Array gain | ~24 dBi | ~30 dBi | ~36 dBi |
| 3 dB beamwidth | ~6.4° | ~1.8° | ~0.9° |
The extremely narrow beamwidths (<2°) at sub-THz frequencies create beam alignment challenges. Initial access and beam tracking require hierarchical codebook designs specified in extensions to TS 38.214 Section 5.1.5 beam management procedures.
Achievable Data Rates
Theoretical peak rates scale directly with bandwidth and spectral efficiency:
`
Rate = BW × SE × MIMO_layers × (1 - overhead)
`
With 20 GHz contiguous bandwidth at D-band, 256-QAM (SE ≈ 7.4 bps/Hz with rate-0.93 LDPC), 2 polarization layers, and 10% overhead:
`
Rate = 20 × 7.4 × 2 × 0.9 = 266.4 Gbps
`
Practical demonstrations have achieved a fraction of this theoretical maximum, but results are rapidly improving.
Real-World Research and Demonstrations
Samsung Research — 140 GHz D-Band Demo
Samsung's 6G research division demonstrated a 140 GHz wireless link achieving 6.2 Gbps throughput over a 2 m indoor distance in 2023. The system used a 256-element phased array at the transmitter and a 64-element array at the receiver, with 2 GHz bandwidth and 16-QAM modulation. Samsung targets outdoor D-band prototypes with adaptive beamforming by 2027, aiming for >50 Gbps at distances up to 100 m.
NTT DOCOMO — 150 GHz OTA Transmission
NTT DOCOMO and NTT Corporation achieved 100 Gbps over-the-air transmission at 150 GHz in a controlled indoor environment in 2024. The demonstration used 23 GHz of contiguous bandwidth with a custom InP-based MMIC transmitter outputting 10 dBm. The receiver employed a 1,024-element antenna array with digital beamforming. DOCOMO considers D-band backhaul for dense urban small cells as the first commercial sub-THz application, targeting 2030 deployment aligned with IMT-2030.
University of Oulu 6G Flagship
The 6G Flagship program at the University of Oulu, Finland — a major EU-funded 6G research initiative — has published channel measurement campaigns at 140 GHz and 300 GHz in indoor, outdoor, and industrial environments. Key findings from their measurement campaigns (reported under the Hexa-X-II project) include:
- RMS delay spread at
140 GHzindoor:5–15 ns(LoS),20–80 ns(NLoS) - Ricean K-factor:
8–15 dBin LoS corridors, confirming strong direct path - Cluster angular spread:
2–8°at300 GHz, significantly narrower than mmWave
These measurements feed directly into ITU-R channel model development for the IMT-2030 evaluation framework, documented in ITU-R M.2412 successor reports.
Key Challenges and Research Directions
| Challenge | Current State | Required Breakthrough |
|---|---|---|
| PA output power | 5–15 dBm (InP), 0–5 dBm (CMOS) | >20 dBm with >10% PAE |
| Phase noise | -85 dBc/Hz at 1 MHz offset | <-95 dBc/Hz for 256-QAM |
| Beam tracking latency | ~1 ms | <100 µs for mobile UEs |
| Cost per array | >$1,000 (research grade) | <$50 for consumer devices |
| Standardization | TR 38.901 extensions, study items | Full FR3 specification (Rel-21+) |
Power amplifier efficiency remains the dominant bottleneck. Current InP HBT PAs achieve 15 dBm at 140 GHz with 5–8% power-added efficiency (PAE), while CMOS solutions lag by 10–15 dB in output power. 3GPP RAN1 is expected to begin formal FR3 study items no earlier than Release 20 (anticipated ~2028).
Key Takeaway: Sub-THz communications from
100 GHzto300 GHzunlock tens of gigahertz of contiguous bandwidth, enabling 6G peak rates beyond100 Gbps. The D-band (110–170 GHz) offers the best near-term balance of bandwidth, atmospheric transparency, and semiconductor maturity. Real demonstrations from Samsung (6.2 Gbps at 140 GHz) and NTT DOCOMO (100 Gbps at 150 GHz) prove feasibility, but practical deployment demands breakthroughs in PA power, beam tracking, and array cost — placing commercial sub-THz systems on a 2030+ timeline aligned with IMT-2030.