Millimeter Wave Specific Challenges mmWave Design Challenges Informational

How do I select between LTCC, HTCC, and organic substrates for a millimeter wave module?

Q: Which substrate is best for 77 GHz automotive radar?

LTCC is the traditional choice (used by Infineon, NXP, and Texas Instruments for 77 GHz radar from 2010-2020): the ceramic provides excellent stability over the automotive temperature range (-40 to +125°C), low loss at 77 GHz, and integration of the antenna in the package (AiP). However: organic (eWLB - embedded wafer-level ball grid array) is gaining traction for next-generation 77 GHz radar (2022+). The antenna is formed in the redistribution layer of the wafer-level package, eliminating the separate ceramic substrate. Cost is lower and the form factor is more compact. The trend: high-volume automotive radar is moving from LTCC to organic/WLP as the packaging technology matures.

Q: Can I use standard FR-4 as the carrier for an LTCC or organic AiP module?

Yes, the AiP module is typically surface-mounted on a standard FR-4 motherboard. The mmWave signals are processed entirely within the AiP module (antenna, PA, LNA, transceiver). Only low-frequency signals (IF, baseband, control, power) cross the boundary between the AiP module and the FR-4 motherboard. These low-frequency connections are not affected by the FR-4 loss tangent. This architecture is used in 5G smartphones: the Qualcomm QTM525 mmWave AiP module mounts on the phone main PCB (FR-4 or FR-4 variant). The module handles all 28/39 GHz RF functions internally, connecting to the baseband modem via low-frequency IF or digital interfaces.

Q: What about glass substrates?

Glass is an emerging substrate technology for mmWave packaging: Dk = 4-6 (depending on glass composition). Loss tangent = 0.001-0.005 at 30 GHz. CTE = 3-8 ppm/°C (tunable by glass composition, can closely match silicon). Ultra-smooth surface (< 1 nm roughness, vs 1-3 um for LTCC/organic): minimizes conductor surface roughness loss at mmWave. Very high dimensional accuracy (no shrinkage during processing, unlike LTCC which shrinks 12-16% during firing). Through-glass vias (TGVs): laser-drilled vias with 20-50 um diameter. Glass panels can be large (500 × 500 mm) for high-volume, low-cost production. Status: glass substrates are in early commercialization for 5G AiP and mmWave radar (2025+). Companies like Corning, AGC, and Schott are supplying mmWave-grade glass panels.

The three main substrate technologies for mmWave modules offer different tradeoffs in electrical performance, integration capabilities, and cost: (1) LTCC (Low Temperature Co-fired Ceramic): a multilayer ceramic substrate co-fired at 850-900°C with embedded conductors (silver or copper), resistors, and capacitors. Dielectric constant: Dk = 5-9 (depending on formulation). Loss tangent: 0.001-0.005 at 30 GHz (low loss). CTE: 4-7 ppm/°C (good match to GaAs: 5.7, Si: 2.6). Layer count: 10-50+ layers. Line resolution: 50-100 um (adequate for mmWave). Embedded passives: yes (inductors, capacitors, resistors). Cavity capability: yes (for die attach within the substrate). Cost: moderate ($5-50 per module depending on complexity). Best for: highly integrated mmWave modules with embedded antennas, filters, and interconnects (e.g., automotive radar 77 GHz modules, 5G AiP). (2) HTCC (High Temperature Co-fired Ceramic): co-fired at 1500-1600°C with refractory metal conductors (tungsten or molybdenum). Dk = 9-10 (alumina-based). Loss tangent: 0.0002-0.001 (very low loss). CTE: 6-7 ppm/°C. Conductor: tungsten has higher resistivity than silver/copper (LTCC), causing higher conductor loss at mmWave. Cost: high (custom tooling, small volumes). Best for: high-reliability military and space modules where the alumina substrate provides excellent thermal conductivity and long-term stability. Less common for commercial mmWave due to high cost and high conductor loss. (3) Organic (laminate) substrate: high-performance PCB laminate (Rogers, Panasonic Megtron, Isola). Dk = 3.0-3.8. Loss tangent: 0.001-0.004 at 30 GHz. CTE: 10-17 ppm/°C (much higher than GaAs/Si). Layer count: 4-20+ layers. Line resolution: 50-75 um (HDI process). Cost: low ($0.50-5 per module in volume). Best for: high-volume consumer 5G modules (smartphones, routers) where cost is the primary driver. The higher CTE mismatch is managed through underfill and compliant bump structures.

Category: Millimeter Wave Specific Challenges

Updated: April 2026

Product Tie-In: mmWave Components, Substrates, Packaging

mmWave Substrate Selection

The substrate choice fundamentally determines the cost, performance, reliability, and integration level of a millimeter-wave module.

Electrical Performance Comparison

(1) Transmission line loss at 28 GHz (50-ohm microstrip, 100 um wide trace): LTCC (Dk = 7, Ag conductor): 0.15-0.3 dB/cm. HTCC (Dk = 10, W conductor): 0.3-0.6 dB/cm (tungsten resistivity is 3× silver). Organic (Dk = 3.5, Cu conductor): 0.2-0.5 dB/cm (depends on laminate quality). At 77 GHz (automotive radar): losses increase by approximately 1.5-2×. (2) Interconnect density: LTCC: via diameter 75-150 um, via pitch 200-400 um. Supports high via density for vertical interconnects. Organic: via diameter 50-100 um (laser drilled), via pitch 100-300 um. HDI organic substrates match or exceed LTCC via density. HTCC: via diameter 100-200 um (punched), via pitch 300-500 um. Lower density. (3) Embedded components: LTCC excels: capacitors (MIM layers), inductors (spiral traces on multiple layers), resistors (buried resistive paste), and waveguide structures (SIW in multilayer ceramic). Organic: embedded capacitors (thin laminate layers), embedded resistors (resistive foil), but lower component density and value range than LTCC. HTCC: similar to LTCC but with higher processing temperature constraints on embedded materials.

Integration Capability

(1) Antenna-in-Package (AiP): LTCC: ideal for AiP because the ceramic provides a stable, low-loss dielectric for antenna elements. The multilayer structure allows 3D antenna designs (stacked patches, slot-coupled feeds, horn antennas in cavity). Many commercial 77 GHz radar modules use LTCC AiP (Infineon, NXP). Organic: increasingly used for 5G AiP (Qualcomm QTM525/527). The organic laminate provides the large area needed for phased-array antennas at low cost. The lower Dk (3.5 vs 7) results in larger antenna elements but wider bandwidth (BW ∝ substrate thickness / sqrt(Dk)). (2) System-in-Package (SiP): LTCC: excellent for SiP with multiple die (MMIC PA, LNA, transceiver), embedded filters, and antenna in a single hermetic package. Organic: good for SiP at lower cost. The die are flip-chip mounted on the organic substrate, with redistribution layers (RDL) for signal routing. Used in high-volume 5G front-end modules. (3) Hermetic sealing: LTCC: inherently hermetic (ceramic is impervious to moisture). Suitable for military and space applications. Organic: not hermetic. Must be sealed with an overmold or encapsulant. Moisture absorption over time can degrade performance (Dk change, delamination).

Cost and Volume

(1) LTCC: setup cost: $10K-50K (screen printing + firing tooling). Per-unit: $5-50 (depending on size and complexity). Best for: medium volume (1K-100K units/year) and high-integration applications. (2) HTCC: setup cost: $50K-200K. Per-unit: $20-200. Best for: military and space (small volume, high reliability). (3) Organic: setup cost: $2K-20K (PCB tooling). Per-unit: $0.50-5 (in volume). Best for: high volume (100K-100M units/year) consumer applications. The cost advantage of organic substrates is dramatic at high volumes, which is why the 5G smartphone industry has driven the development of high-performance organic AiP modules.

Substrate Material Properties

LTCC: Dk=5-9, tan δ=0.001-0.005, CTE=4-7
HTCC: Dk=9-10, tan δ<0.001, CTE=6-7
Organic: Dk=3-3.8, tan δ=0.001-0.004, CTE=10-17
Line loss @28GHz: 0.15-0.6 dB/cm
AiP BW ∝ thickness/√Dk

Common Questions

Frequently Asked Questions

Which substrate is best for 77 GHz automotive radar?

LTCC is the traditional choice (used by Infineon, NXP, and Texas Instruments for 77 GHz radar from 2010-2020): the ceramic provides excellent stability over the automotive temperature range (-40 to +125°C), low loss at 77 GHz, and integration of the antenna in the package (AiP). However: organic (eWLB - embedded wafer-level ball grid array) is gaining traction for next-generation 77 GHz radar (2022+). The antenna is formed in the redistribution layer of the wafer-level package, eliminating the separate ceramic substrate. Cost is lower and the form factor is more compact. The trend: high-volume automotive radar is moving from LTCC to organic/WLP as the packaging technology matures.

Can I use standard FR-4 as the carrier for an LTCC or organic AiP module?

Yes, the AiP module is typically surface-mounted on a standard FR-4 motherboard. The mmWave signals are processed entirely within the AiP module (antenna, PA, LNA, transceiver). Only low-frequency signals (IF, baseband, control, power) cross the boundary between the AiP module and the FR-4 motherboard. These low-frequency connections are not affected by the FR-4 loss tangent. This architecture is used in 5G smartphones: the Qualcomm QTM525 mmWave AiP module mounts on the phone main PCB (FR-4 or FR-4 variant). The module handles all 28/39 GHz RF functions internally, connecting to the baseband modem via low-frequency IF or digital interfaces.

What about glass substrates?

Glass is an emerging substrate technology for mmWave packaging: Dk = 4-6 (depending on glass composition). Loss tangent = 0.001-0.005 at 30 GHz. CTE = 3-8 ppm/°C (tunable by glass composition, can closely match silicon). Ultra-smooth surface (< 1 nm roughness, vs 1-3 um for LTCC/organic): minimizes conductor surface roughness loss at mmWave. Very high dimensional accuracy (no shrinkage during processing, unlike LTCC which shrinks 12-16% during firing). Through-glass vias (TGVs): laser-drilled vias with 20-50 um diameter. Glass panels can be large (500 × 500 mm) for high-volume, low-cost production. Status: glass substrates are in early commercialization for 5G AiP and mmWave radar (2025+). Companies like Corning, AGC, and Schott are supplying mmWave-grade glass panels.

Keep Reading