PCB & Electronic Substrates — Material Selection Guide

Selection Criteria That Actually Drive Outcomes #

Procurement teams sourcing PCB substrates from China typically compare three parameters: Tg, Dk, and price. Those three variables will get you to a shortlist. They will not get you to the right material.

The real selection framework requires six criteria, evaluated in a specific order. Sequence matters because some properties are application-dependent (Dk/Df only matters above roughly 1 GHz), while others are universal risks regardless of application — z-axis CTE, CAF resistance, and lot-to-lot consistency chief among them.

The mistake we see most often in incoming qualification is treating the material grade label as a proxy for performance. “FR4” covers a range from Tg 130°C to Tg 180°C materials, with z-axis CTE anywhere from 50 ppm/°C to 80 ppm/°C. Those are not the same material. A PO that specifies “FR4 per IPC-4101” without a slash sheet designation is underspecified, and any compliant-looking COA from a Chinese supplier will still leave you exposed.

Head-to-Head Comparison — Six Criteria, Five Substrate Classes #

The table below compares the five substrate classes most commonly sourced from Chinese manufacturers for multilayer PCB production. Values reflect typical production lot ranges from our audit database, not datasheet maxima.

Z-axis CTE and CAF resistance values are drawn from incoming inspection records across our supplier qualification program — not from manufacturer datasheets, which consistently report best-case figures.

For most industrial multilayer designs operating below 500 MHz, high-Tg FR4 per IPC-4101/26 is the right call. The z-axis CTE is meaningful, the Dk/Df is adequate, and the cost premium over standard FR4 is recoverable in reduced via-failure rates over product lifetime. We push buyers away from standard FR4 (IPC-4101/21) for any design exceeding 6 layers or 250 thermal cycles in service — the z-axis CTE gap between 80 ppm/°C and 50 ppm/°C accumulates into barrel crack failures that a standard COA test won’t predict.

For signal-integrity-critical designs above 5 GHz, the hydrocarbon high-speed laminates are the only class with Df values below 0.010 that are achievable in production. PTFE composites deliver the best RF performance but introduce serious fabrication complexity — drilling parameters, desmear chemistry, and press cycles all require process qualification that many EMS providers in China have not completed for these materials.

The area where opinion genuinely divides is the 1–5 GHz band. Some design teams specify high-speed hydrocarbon at 1 GHz as a margin strategy. Others use high-Tg FR4 with impedance tuning and accept slightly wider tolerance stacks. Both approaches work. The high-speed laminate delivers tighter impedance control; the FR4 approach costs less and sources from a larger supplier base. Which one is right depends on whether your design has margin at all — and that requires running loss budget numbers, not reading a datasheet.

The Overlooked Variable — Prepreg-to-Core CTE Matching #

Standard substrate comparisons focus on core laminate properties. The parameter that frequently breaks multilayer builds is the CTE mismatch between the core laminate and the prepreg used for bonding — and this is almost never specified at the purchasing stage.

When core and prepreg come from different production lots or, worse, different suppliers, z-axis CTE delta can reach 15–20 ppm/°C even within nominally compatible material families. Under IPC-TM-650 method 2.4.24, the acceptable CTE for z-axis in standard multilayer is ≤70 ppm/°C below Tg — but that threshold assumes matched systems. A mismatched core/prepreg combination can produce a stack that passes individual laminate tests and still delaminates at the press stage.

We log this under our MS-11 material system verification step, which requires that core and prepreg be co-qualified from the same supplier lot before a new material combination enters our approved vendor list. The gap in most procurement workflows is that core and prepreg are purchased as separate line items, often from different suppliers or different reorder cycles, with no formal review of system compatibility.

One scenario worth flagging: a buyer qualifies a high-Tg FR4 core from a Shenzhen-area CCL manufacturer based on strong incoming inspection results. Six months later, a prepreg shortage forces a supplier substitution on the prepreg side only. The core COA still passes. The press yield drops from 96% to 87% over the next three production months before the root cause is identified. This is not a hypothetical — CTE-mismatch delamination at the core/prepreg interface is a recurring entry in incoming quality logs across EMS operations we work with, particularly when Chinese suppliers source their resin systems from a secondary compounder during capacity crunches.

Implementation Notes — Incoming Inspection and Qualification Steps #

After selecting a substrate class, the qualification sequence matters as much as the spec itself.

For any new Chinese CCL supplier, the minimum incoming qualification battery under what we call our IQ-3 baseline protocol is: peel strength per IPC-TM-650 method 2.4.8, Tg by DSC per IPC-TM-650 method 2.4.25, z-axis CTE per IPC-TM-650 method 2.4.24, and moisture absorption at 24h/23°C per IEC 62326-4. Dk/Df testing by the split-post dielectric resonator (SPDR) method should be added for any material targeting above 1 GHz.

Pass/fail thresholds we use in production qualification:

• Peel strength: ≥1.0 N/mm (1 oz Cu) at room temperature; ≥0.8 N/mm after thermal stress at 288°C/10s
• Tg by DSC: within ±5°C of nominal grade designation
• Z-axis CTE below Tg: ≤70 ppm/°C for standard FR4; ≤55 ppm/°C for high-Tg grades
• Moisture absorption: ≤0.35% for FR4 grades; ≤0.20% for high-speed hydrocarbon grades
• Dk tolerance: ±0.05 from lot average; any lot exceeding ±0.10 is returned regardless of certificate status

The last point on Dk tolerance is where most buyers underspecify. A COA that reports a single Dk value with no lot variation data is not useful for impedance-controlled designs. Request three consecutive lot test reports before qualification, not just a single sample certificate. If a supplier cannot produce three-lot Dk data, that is a qualification gap — not a price negotiation.

Red flags in early shipments that trigger hold-and-retest in our program:

• Copper surface showing oxidation or patchy discoloration on receipt
• Thickness variation exceeding ±10% across a panel (measure at 5 points)
• Any report of delamination blistering during solder float test at 288°C
• COA Tg values that are round numbers (e.g., exactly 170°C, 180°C) — real DSC results are not round

Run the full IQ-3 battery on the first three incoming lots before releasing any material to production. After that, spot-test at AQL 2.5 per ANSI/ASQ Z1.4 for dimensional and visual parameters, with full Tg and CTE testing on any lot that triggers a visual alert or arrives from a supplier following a production line change notification.

Most Western buyers do not realize that GB/T 4722 and GB/T 4724, the Chinese national standards governing electrical laminate and prepreg, allow dimensional tolerances that are wider than the equivalent IPC-4101 slash sheet requirements. A substrate that is fully compliant with SAC GB/T standards may not meet your IPC-specified engineering drawing. This is not a quality failing on the supplier’s part — it is a standards gap that the buyer needs to specify around explicitly in the PO.

Practical Guidance for Buyers #

When sourcing PCB substrates from China, the first specification to request is not Tg — it is the slash sheet designation under IPC-4101 and three consecutive lot COAs showing Dk measured by SPDR method. Tg is easy to state on a certificate; lot-to-lot Dk variation is harder to manage and far more predictive of signal integrity yield. A supplier that cannot provide multi-lot Dk data has not been producing to impedance-controlled quality standards, regardless of what the single-sample certificate shows.

The specific risk scenario to watch for is prepreg substitution during periods of raw material shortage — particularly Q3 and Q4 when Chinese resin suppliers tend to run allocation constraints. A core laminate that qualified cleanly can still cause delamination failures if the prepreg changes supplier or resin lot without notification. Require your Chinese CCL supplier to notify you of any resin system change 30 days in advance as a contractual term, not a preference.

Before committing to production volume, insist on a press qualification run: minimum 20-panel multilayer build using production-intent stack-up, with cross-section analysis at 5 locations per panel and solder float test at 288°C for 10 seconds per IPC-TM-650. Suppliers who resist this step — or offer lab-fabricated samples instead of production-line builds — represent elevated risk regardless of their paper qualifications.

What to Specify in Your PO (Checklist)

• IPC-4101 slash sheet number (e.g., /26 for high-Tg FR4) — not just “FR4”
• Nominal Dk at 10 GHz with ±0.05 lot tolerance requirement
• Tg by DSC method with ±5°C acceptance band
• Z-axis CTE ≤55 ppm/°C below Tg (for high-Tg grades)
• Prepreg supplier identity and resin system designation — co-qualified with core
• Three consecutive lot COAs required on first order; single COA per lot thereafter
• 30-day advance notification required for any resin or prepreg system change
• Incoming inspection right reserved: AQL 2.5 for dimensional, full thermal testing on flagged lots

Frequently Asked Questions

Is FR4 suitable for 5G antenna board applications?
Standard FR4 is not. At 5G sub-6 GHz frequencies, the Df of 0.020–0.025 for standard FR4 generates insertion loss that most antenna designs cannot budget for. High-speed hydrocarbon laminates with Df below 0.007 are the practical minimum for this application.

What Tg do I need for lead-free assembly?
For lead-free reflow with peak temperatures at 260°C, specify a minimum Tg of 170°C. Mid-Tg FR4 at 150°C is marginal — it will survive single reflow cycles but shows accelerated CAF and delamination risk in double-sided assemblies with two reflow passes.

Can I mix core and prepreg from different Chinese suppliers to reduce cost?
It depends on whether you have co-qualification data for the specific combination. Mixing suppliers without system-level CTE verification is where delamination failures at the press stage originate. The cost saving is real but small; the yield risk is not.

How do I verify that a Chinese supplier’s FR4 is actually halogen-free?
Request XRF test data for chlorine and bromine, not just a declaration on the COA. Halogen-free per IEC 61249-2-21 requires Cl ≤900 ppm, Br ≤900 ppm, and total halogens ≤1500 ppm. We have seen COA declarations of “halogen-free” on materials that failed XRF at incoming — the declaration is not the same as the test result.

What AQL level should I use for CCL incoming inspection?
For dimensional and visual parameters, AQL 2.5 per ANSI/ASQ Z1.4 is our standard. For thermal and electrical parameters (Tg, Dk), we test at 100% on the first three lots from a new supplier, then move to spot-testing unless an alert is triggered.

For related sourcing guidance, see our evaluation resources on pump-valve-seals and fluid control materials and conductive and functional materials for electronic applications.

Published by sinoraw.com Technical Team | Request a sourcing consultation