Aluminum PCB & MCPCB Manufacturer China | LED & Power Electronics

Thermal Conductivity: Dielectric Layer vs Aluminum Substrate in Metal Core PCBs

The first specification confusion to resolve when quoting an MCPCB (Metal Core Printed Circuit Board) is which thermal conductivity number the supplier is quoting. Aluminum alloy 6061 has a bulk thermal conductivity of approximately 160 W/m·K. Aluminum 5052 is similar at 138 W/m·K. Chinese aluminum PCB factory sales quotes frequently lead with these numbers. They are almost never the limiting factor in your thermal path.

The limiting thermal resistance is the dielectric layer bonded between the copper circuit layer and the aluminum base — a polymer-ceramic composite 75–150µm thick. Standard dielectric materials (similar to Bergquist GP series or domestic Shengyi MT-80) achieve 1.0 W/m·K. Premium filled dielectrics reach 2.0 W/m·K. High-performance materials targeting <1°C/W thermal resistance in compact LED applications can reach 3.0 W/m·K, at roughly 2–3× the material cost.

Worked example — 5W LED, 10mm² footprint, 100µm dielectric at 1.0 W/m·K:

Rth_dielectric = t / (k × A)
               = 0.0001m / (1.0 W/m·K × 10×10⁻⁶ m²)
               = 10 °C/W

At 5W dissipation, the dielectric layer alone contributes 50°C of junction-to-base temperature rise. Switch to a 2.0 W/m·K dielectric:

Rth_dielectric = 0.0001m / (2.0 × 10×10⁻⁶)
               = 5 °C/W  →  25°C rise at 5W

That 25°C reduction at the junction has a direct impact on LED lumen maintenance. A Cree XHP70.2 LED derated from 85°C junction to 60°C junction (using manufacturer’s L70 life curves) approximately doubles rated L70 lifetime from 50,000 to 100,000 hours.

The aluminum substrate’s 160 W/m·K is effectively irrelevant in this calculation — for a 1mm thick aluminum base, Rth_aluminum = 0.001 / (160 × 10×10⁻⁶) = 0.625°C/W, negligible compared to the dielectric. This means upgrading from 6061 to a more expensive aluminum alloy buys you almost nothing thermally. Spend the budget on dielectric grade instead.

Practical sourcing guidance: always request the dielectric layer thermal conductivity from the material datasheet, not the aluminum substrate value. Ask the factory which dielectric material brand/grade they use (Shengyi, Iteq, EMC, Ventec, or Bergquist/Henkel). Standard domestic dielectrics from Shengyi (MT-80) and EMC (EM-827) are well-characterized at 1.0–1.5 W/m·K and are entirely appropriate for most LED lighting applications. High-performance 2.0–3.0 W/m·K materials from Ventec (VT-4A2) or Bergquist (GP3.0) are typically worth the cost only when the thermal path is tightly constrained and there is no room to increase the footprint.

Our PCB sourcing service qualifies MCPCB suppliers on dielectric material traceability as part of the standard specification review.

Dielectric Thickness and Voltage Isolation Trade-off

Thinner dielectric reduces thermal resistance but reduces the voltage isolation between the copper circuit and the aluminum base (which is typically ground or chassis potential in LED driver and power supply applications).

For a 75µm dielectric, IPC-6012 Class 2 requires a minimum dielectric withstand voltage of 500V DC in production testing. In practice, quality-grade domestic suppliers test at ≥2kV AC (per IPC-TM-650 2.5.7), which provides a comfortable margin for typical 48V DC or 24V AC applications.

For products operating at 230V AC mains (LED drivers, power supplies complying with EN 60335-1 or IEC 62368-1, the isolation requirement is more stringent:

Basic insulation (single fault protection): typically requires a 1.5kV AC dielectric withstand test (IEC 60664-1 for Pollution Degree 2, Overvoltage Category II).

Reinforced insulation (double insulation, no PE on the aluminum chassis): EN 60335-1 requires reinforced insulation equivalent to two layers of basic insulation. This typically means a dielectric withstand test of 3kV AC (twice the basic insulation test voltage plus margin). A 75µm dielectric at 2kV breakdown cannot satisfy this — you need 150µm dielectric tested to ≥3kV.

IPC-2221A creepage and clearance distances also apply to the trace routing on the copper layer, independent of the dielectric thickness. For 230V reinforced insulation on a CTI ≥600 material surface, IPC-2221A requires ≥8.0mm creepage between primary and secondary circuit elements. Verify this in your Gerber layout review before sending for fabrication — a factory will not flag a creepage violation automatically.

Verification incoming QC: for 230V applications, test every panel (or a statistically valid sample per AQL 0.65 for Class 2) at the rated dielectric withstand voltage. Do not rely solely on the factory’s production test data without independent lot verification. Our inspection service includes hipot (dielectric withstand) testing as a standard check on MCPCB lots for power supply applications.

MCPCB vs FR4 + Heatsink vs Ceramic (AlN) Substrates

Three competing approaches cover most LED board and power electronics thermal management requirements. The right choice depends on power density, volume, and budget for your printed circuit boards.

MCPCB: $0.08–0.40/cm² The cost-effective baseline for LED lighting and power modules up to approximately 50W/cm² power density. Single-sided copper circuit only — components mount on top of the copper, aluminum is the heat spreader. Cannot support blind/buried vias or multi-layer routing. For mixed-signal designs with digital control circuitry and power stages, MCPCB forces you to separate the digital and power sections onto different board sections or use a separate FR4 interface board.

FR4 + copper coin insert: $0.15–0.60/cm² Where MCPCB falls short is in designs that need the kind of multi-layer routing a multilayer FR4 PCB provides alongside selective thermal management. A 4-layer FR4 board with copper coin inserts (solid copper cylinders pressed into through-holes beneath high-power components) can achieve thermal conductivity approaching 400 W/m·K at the coin location while retaining standard FR4 dielectric properties for signal routing. Cost is higher than standard FR4 but lower than full MCPCB for boards with mixed thermal requirements. Lead time is longer — coin pressing requires additional tooling and process steps. Not all factories in China offer this capability; it requires qualification before commitment.

AlN ceramic (aluminum nitride): $1.50–4.00/cm² Thermal conductivity of 150–200 W/m·K through the ceramic substrate itself, with no polymer dielectric layer. Suited for power electronics modules (SiC/GaN MOSFETs, IGBT modules) where power density exceeds what MCPCB can handle and where the ceramic can be direct bonded to a copper heat spreader (DBC — Direct Bonded Copper process). AlN is brittle and requires careful mechanical design for mounting. Cost is 5–10× MCPCB. Lead time is 4–6 weeks for custom dimensions.

BeO (beryllium oxide): thermally excellent (250–300 W/m·K) but restricted under EU RoHS and OSHA 1910.1024 (beryllium exposure standard). Do not specify for new designs. Legacy military/aerospace programs only.

Direct Bonded Copper (DBC) on AlN or Al₂O₃: the standard substrate for commercial power modules (Infineon, Mitsubishi, Semikron). 0.3mm copper bonded directly to ceramic at 1,000°C+ in a controlled atmosphere furnace. Thermal resistance from junction to substrate <0.1°C/W for a 10mm² footprint at 3W/m·K effective path. Chinese DBC manufacturers (Natam, IXYS/Littelfuse domestic partners) produce substrates for domestic power module assembly. Minimum order is typically 500 pieces with 6–8 weeks lead time.

The PCB assembly industry page covers qualification requirements for each substrate type in more detail. For a broader fabrication and assembly workflow, see our PCB assembly in China guide.

Chinese Supplier Landscape and Incoming Quality Control

Material supply chain. The dominant domestic MCPCB dielectric material suppliers in China are Shengyi Technology (SY-MTG series, 1.0–3.0 W/m·K), Iteq (IT-80A, 1.0 W/m·K), and EMC (EM-827, 1.0 W/m·K). Shengyi and Iteq are publicly traded and supply to most mid-tier MCPCB fabricators. International materials — Bergquist (now Henkel) and Ventec VT-4A2 — are used by premium Chinese fabricators targeting export markets where material traceability to the original manufacturer’s datasheet is a customer requirement. For applications where the thermal conductivity number on your datasheet must be traceable to a named material, specify the material by brand and grade in your fabrication notes, not just the thermal conductivity value.

Most export-oriented MCPCB factories are concentrated in the Pearl River Delta, within a few hours of Shenzhen and Dongguan. Proximity matters for quick material verification and line audits when you are qualifying a new thermal-management supplier.

Thermal conductivity verification. Factories quote thermal conductivity from the material supplier’s datasheet. For audit purposes, the verification method that matters is laser flash diffusivity measurement (ASTM E1461) on a coupon cut from the production lot. This measures thermal diffusivity directly; thermal conductivity is calculated from diffusivity × density × specific heat. A factory with in-house laser flash equipment (Netzsch LFA or equivalent) can provide lot-level verification data. Most fabricators do not have this equipment — they rely on the material supplier’s incoming QC. An alternative, lower-cost verification is hot disk transient plane source (ISO 22007-2), which works on laminated panels but has more uncertainty on thin dielectric films. For critical applications, request lot certificates from the dielectric material supplier, not just the MCPCB fabricator.

Electrical test for isolation. IPC-6012 Class 2 requires 100% bare board testing. For MCPCB, the relevant test is dielectric withstand (hipot) between the copper circuit and the aluminum base. Standard production test: 500V DC for 5 seconds, zero breakdown events. Request the production test report with the actual test voltage and the serial number or lot number linked to your specific order. For Class 3 (high reliability, aerospace/medical), 100% continuity and isolation testing at ≥1kV is standard.

Peel strength test. The adhesion between the copper foil and the dielectric layer degrades with thermal cycling and poor lamination process control. IPC-TM-650 2.4.8 specifies the test method: a 1-inch-wide copper strip is peeled at 90° at 50mm/min. Minimum acceptable value per IPC-4101 (the laminate specification): 8.8 N/mm for 1oz copper. Chinese factories producing for commodity LED luminaire markets sometimes use dielectric pre-preg with peel strength at the lower end — adequate for static thermal applications but problematic in products subject to mechanical vibration (automotive electronics, industrial). For vibration-exposed applications, specify a minimum peel strength of 10 N/mm in your fabrication specification and include peel strength coupon testing in your incoming inspection plan.

Sourcing notes from the floor

We visited an MCPCB factory in Shenzhen last quarter and checked dielectric material traceability and bare-board hipot testing. During the audit we saw 2.0W/m·K dielectric quotes that actually used 1.0W/m·K material once we checked the incoming certificates. The most common spec mismatch is 75µm dielectric sold for 230V AC LED drivers where 150µm and reinforced creepage are required. Real-world MOQ/price is often 50–100 panels at $0.08–0.40 per cm². Certification gotcha to watch: UL 94V-0 must cover the specific laminate construction; a UL file for FR4 does not extend to aluminum-base boards.