RF & mmWave Hybrid Assembly for 5G and Defense Systems

Introduction: Why RF Hybrids Matter Now

The global shift to 5G New Radio (NR) has created an immediate need for RF front-end modules operating at 28 GHz, 39 GHz, and beyond—frequencies that demand fundamentally different assembly approaches than sub-6 GHz electronics. Simultaneously, defense systems are moving from monolithic microwave integrated circuits (MMICs) toward highly integrated RF system-in-package (SiP) solutions that pack multiple active and passive functions into a single hybrid module.

What makes RF assembly different from conventional hybrid microelectronics is the critical role of electromagnetic wave propagation. At low frequencies, a circuit board is just a mechanical support structure. At 24 GHz and above, the substrate becomes part of the electrical circuit—its dielectric constant, loss tangent, and thickness directly determine impedance, insertion loss, and radiated emissions. Avia that is acceptable at DC becomes a catastrophic radiating stub at 77 GHz automotive radar.

Market drivers are accelerating demand. The 5G RF front-end module market exceeded $6 billion in 2025 and is projected to grow at 14% CAGR through 2030. Defense radar and electronic warfare systems are investing heavily in active electronically scanned arrays (AESA), each of which contains hundreds of T/R modules, each built as an RF hybrid. Quality and yield in these assemblies directly determine system cost and performance.

RF Fundamentals in Hybrid Circuits

Successful RF hybrid design requires fluency in a set of high-frequency concepts that rarely appear in lower-frequency analog or digital design.

Impedance Matching and the 50Ω Standard

RF circuits are designed around a characteristic impedance standard—almost universally 50 Ω for most RF and microwave applications, with 75 Ω used in some video and cable applications. Impedance mismatches cause signal reflections measured by the reflection coefficient (Γ) or its logarithmic expression, return loss (RL). Return loss is calculated as:

RL (dB) = -20 × log10(|Γ|)

A return loss of 10 dB means approximately 10% of the incident power is reflected back toward the source. For most RF hybrids, a return loss exceeding 15 dB (VSWR < 1.4:1) across the operating band is the minimum acceptable target; critical path components often require 20 dB or better.

S-Parameters

Scattering parameters (S-parameters) are the standard language for characterizing RF networks. The four key S-parameters are:

• S11 / S22: Input and output return loss (reflected power at ports 1 and 2)
• S21 / S12: Forward and reverse transmission (signal transfer through the network)
• Insertion Loss (S21): Power lost in the transmission path, measured in dB. Every 1 dB of insertion loss in a transmit path represents ~20% reduction in effective radiated power.

Skin Effect and Conductor Roughness

At high frequencies, current density concentrates near the surface of conductors—a phenomenon called the skin effect. The skin depth (δ) decreases with frequency:

δ = sqrt(2 / (ω·μ·σ)) ≈ 0.066 / sqrt(freq_GHz) mm for copper

At 10 GHz, skin depth in copper is approximately 0.66 μm; at 77 GHz it drops to 0.24 μm. This means conductor surface roughness directly impacts insertion loss. Standard electrodeposited (ED) copper with RMS roughness of 2–3 μm can add 0.3–0.5 dB/cm loss at 77 GHz compared to 0.1 dB/cm with reverse-treated (RT) copper at 1 μm roughness. For demanding mmWave applications, annealed copper or silver-filled conductors are preferred.

Substrate Selection for RF Applications

The substrate material sets the foundation for every RF hybrid's electrical and thermal performance. Material selection is a multi-variable trade-off between dielectric constant, loss tangent, thermal conductivity, and cost.

CTE matching is a frequently overlooked consideration when mounting silicon or GaN dice. Silicon has a CTE of approximately 2.6 ppm/°C, while GaAs is 5.9 ppm/°C and GaN is approximately 5.6 ppm/°C. Alumina at 6.5 ppm/°C provides reasonable CTE match to GaAs, but Aluminum Nitride at 4.5 ppm/°C is better matched to silicon and GaN. CTE mismatch during thermal cycling drives die crack failures—a particularly severe problem in automotive and aerospace applications.

Transmission Line Design

The choice of transmission line topology is the most fundamental electrical design decision in an RF hybrid. Each structure has distinct advantages, limitations, and frequency ranges where it excels.

Microstrip

Microstrip is the most common transmission line structure in hybrid microelectronics—a signal trace on the surface of a dielectric substrate with a ground plane on the bottom. Impedance is determined by trace width (W), substrate thickness (h), and dielectric constant (εr):

Z0 ≈ (87 / sqrt(εr + 1.41)) × ln(5.98h / (0.8W + t))

Thin substrates give higher impedance but increase coupling to the ground plane and radiation losses. Thick substrates support lower impedance but can excite surface waves at mmWave frequencies. For 50 Ω microstrip on alumina (εr = 9.6) at 1 mil (25 μm) thickness, trace width is approximately 1.1 mil—impractical for standard photolithography. At 5 mil thickness, width becomes ~12 mil—a more manufacturable dimension. This is a fundamental tension in mmWave microstrip design.

Coplanar Waveguide (CPW)

CPW places the signal trace between two ground coplanar conductors on the same surface as the signal, with a ground plane on the bottom. The key advantage is that all connections are on one surface—no via holes are needed for grounding. This makes CPW far superior at frequencies above 30 GHz where via inductance becomes significant.

CPW requires air bridge crossovers where signal lines cross each other, to suppress slotline modes that would otherwise propagate on the ground plane. Without proper grounding and air bridges, CPW can excite unwanted parallel-plate modes through the substrate, creating severe crosstalk and radiation.

Stripline

Stripline embeds the signal conductor between two ground planes, fully enclosed in the dielectric. This provides excellent shielding and eliminates radiation losses, making it ideal for high-power applications or signal routes that must not couple to external fields. The trade-off is that stripline requires multi-layer structures with via holes to connect to the inner layer—adding manufacturing complexity and via discontinuities.

mmWave Considerations: Up to 110 GHz

Operating above 30 GHz introduces a new set of failure modes and design constraints that are not present at lower frequencies.

Via Stubs and Resonance

A through via that is not backdrilled acts as a quarter-wave stub. At lower frequencies, this is irrelevant—the stub is electrically short. At 77 GHz, a 250 μm thick alumina substrate via is approximately an electrical quarter-wavelength, causing a resonance notch of 3–5 dB in the return loss. Mitigation strategies include:

• Back-drilling or blind vias to eliminate the stub length
• Via-in-paste technology to fill vias with conductive paste before firing, eliminating the air column
• Electroless nickel immersion gold (ENIG) plating to reduce via diameter and increase resonance frequency beyond the operating band

Ground-Signal-Ground (GSG) Probe Launch

Characterization and testing of mmWave circuits requires GSG coplanar probe launches. The probe pad geometry must transition smoothly to the transmission line structure with minimal impedance discontinuity. Key guidelines:

• Probe pad pitch must match standard 100 μm or 150 μm pitch probes
• Ground pad width should be ≥ 3× signal pad width for consistent 50 Ω CPW
• Dielectric stress relief slots around pad openings prevent cracking under probe contact force
• Launch transition length should be ≥ 3× substrate thickness for smooth mode conversion

Skin Depth at mmWave Bands

Dispersion in Microstrip at 60+ GHz

Microstrip is an inhomogeneous transmission line—part of the field is in air, part in the dielectric. This causes dispersion: the effective dielectric constant changes with frequency, causing velocity variation. At frequencies above 60 GHz, dispersion causes pulse spreading and phase distortion in modulated signals. CPW is less dispersive because the fields are more evenly distributed in the dielectric and air regions. For ultra-wideband 5G systems operating at 57–71 GHz (the 60 GHz Wi-Fi band), dispersion management becomes a critical design constraint.

Thermal Management in RF Hybrids

RF power devices generate significant heat, and the thermal resistance from junction to case directly determines whether a module survives in deployed conditions.

GaN vs GaAs Power Dissipation

Gallium Nitride (GaN) HEMT devices achieve power densities of 5–10 W/mm of gate width—roughly 5× higher than equivalent GaAs pHEMT devices. This makes GaN attractive for high-power T/R modules but creates thermal challenges. Junction temperatures must remain below 200°C for reliable long-term operation, and every 10°C increase in junction temperature approximately doubles the failure rate (Arrhenius behavior). A GaN power amplifier dissipating 10W from a 3mm gate-width device requires thermal conductivity exceeding 100 W/m·K at the die-attach interface.

Die Attach Selection for Thermal Performance

Thermal Via Design

Thermal vias under high-power dice should be designed with:

• Via diameter 100–150 μm with pitch ≤ 300 μm (thermal via grid density of at least 20 via/inch)
• Via fill with silver-glass or copper paste for thermal conductivity of 10–20 W/m·K (vs ~0.3 W/m·K for unfilled through-hole)
• Thermal clearance ring of ≥ 50 μm around die attach footprint to avoid die tilt from protrusions

Military vs Commercial RF Hybrids

The divide between military/aerospace-grade and commercial RF hybrids is not just a quality level—it spans the entire supply chain, material selection, screening flow, and qualification regime.

MIL-PRF-38534 Requirements

MIL-PRF-38534 is the performance specification for hybrid microcircuits used in military and aerospace applications. It defines:

• Design and manufacturing controls required for a Qualified Manufacturer List (QML) source
• Hermetic seal requirements (leak rate ≤ 1 × 10⁻⁸ atm·cc/s for ceramic packages)
• Wire bond diameter limits and peel strength minimums (MIL-STD-883 Method 2011)
• Die attach shear strength minimums (MIL-STD-883 Method 2019)
• Active screening flows including burn-in (100 hours at 125°C) and temperature cycling

Known Good Die (KGD) Screening

Military hybrids typically use Known Good Die—bare dice that have passed electrical and burn-in screening before they are assembled. KGD programs add cost but prevent assembly-level yields from being devastated by bad dice. Key KGD screening includes:

• Wafer-level burn-in at elevated temperature
• On-wafer RF S-parameter verification
• Visual inspection to MIL-STD-750 inspection conditions
• Particle impact noise detection (PIND) to detect loose particles in packaged dice

Radiation Hardness and TID

Space and some defense environments require total ionizing dose (TID) hardness of 100 krad(Si) or greater. GaN HEMTs are inherently more radiation-hard than GaAs due to their wider bandgap, but RF passive components (thin-film resistors, MIM capacitors) must be designed with radiation-tolerant materials. Silicon dioxide dielectrics in capacitors are particularly susceptible to trapped charges from ionization events, which shift capacitance values and increase leakage current.

Conclusion and Key Takeaways

RF and mmWave hybrid assembly is not an evolution of conventional hybrid electronics—it is a distinct engineering discipline requiring careful co-design of electrical, thermal, and materials science considerations from the earliest stages. The most common root causes of RF hybrid field failures are preventable: via stubs at mmWave frequencies, inadequate thermal management under high-power GaN dice, and insufficient KGD screening leading to infant mortality in the field.

Key design rules to follow:

• Use CPW rather than microstrip above 30 GHz to avoid via stub resonances
• Specify RT copper or silver conductors for applications above 50 GHz
• Select substrate materials with CTEs within 1–2 ppm/°C of the die material
• Use AuSn or sintered silver die attach for GaN power devices exceeding 5W dissipation
• Implement thermal via grids with > 20 vias/inch pitch and filled via technology
• For military applications, engage a MIL-PRF-38534 qualified manufacturer from the prototype stage—not just the production stage

Emerging trends to watch: 6G research is pushing toward sub-THz frequencies above 100 GHz, which will further amplify conductor roughness and dispersion concerns. Flat-panel phased arrays for SATCOM are driving demand for low-cost multi-layer RF modules, pushing LTCC and embedded active technologies. And GaN-on-diamond technology, which achieves thermal conductivity of 600+ W/m·K in the substrate, is beginning to enter production for military radar T/R modules.

When engaging a contract manufacturer for RF hybrid work, verify their RF measurement capabilities (network analyzer to at least 67 GHz, preferably 110 GHz), their thermal simulation tools, their experience with your specific frequency band and power level, and their quality system certification (AS9100 for aerospace, MIL-PRF-38534 for defense). A qualified manufacturer should be able to provide process capability studies (Cpk ≥ 1.33) on critical parameters such as line width, impedance, and die attach void percentage.