High-Frequency Low-Loss M6-M9 Grade Hydrocarbon Resin by Shengquan Group: Lifting AI Server CCL Performance

Meeting the Rising Demands of AI Server Substrates

As AI models grow more sophisticated, massive data centers put heavier requirements on their backbone: copper clad laminate (CCL) for server PCBs. Over the years, engineers noticed old-school resins hit their ceilings—especially as bandwidth rose and power supply noise crept into every corner of the board. In our work at Shengquan Group, plenty of direct dialog with server manufacturers pointed to the same technical roadblock: signal integrity. That's where high-frequency and low-loss resin truly matters. For modern AI servers running 56G, 112G, and 224G channels, our M6-M9 grade hydrocarbon resins have been delivering the kind of loss performance that helps to avoid repeated PCB revisions. We know every CCL batch must tick those critical boxes: low dissipation, stable dielectric, and process reliability—every time.

Our Credentials: Decades on the Polymer Front-Line

Looking back at our production records, Shengquan Group spent decades scaling up phenolic, epoxy, and hydrocarbon chemistry in-house. Once 5G and AI server customers flagged the loss tangent above 10GHz as a pain point, we shifted focus. Building M6 to M9 grades meant iterating resin backbone structures, adjusting molecular weight, and chasing purity standards that older mass-market grades never attempted. In our own labs, we constantly re-traced the link between residual polar groups and insertion loss. It’s tempting to take shortcuts but using highly refined precursor feedstocks and running exhaustive fractional distillation set the groundwork for what we now offer as M6 and M9 hydrocarbon resins.

Performance in the Real World: Why Low Loss Really Matters

People outside the server fabrication floor might see “low-loss” as marketing lingo. Engineers and production managers know otherwise. At 28GHz and above, even a small improvement in dissipation factor (Df) often means the difference between passing or failing bit error rate tests on a dense server board. We routinely track customer feedback and noticed that boards using our M9 grade resin often delivered insertion loss values 10-15 percent lower than previous “mid-end” alternatives. This translates directly to longer trace runs, more I/O density, and the freedom to upgrade existing hardware platforms. It also means less heat and better signal-to-noise through all those ribbon lines in AI server backplanes. As resins go, this becomes critical for 5G base stations, high-performance computing motherboards, and industrial AI vision modules—verticals pushing data every millimeter they can.

Understanding M6-M9: Real Differences Under the Hood

Some ask—what’s so materially different between M6, M7, M8, and M9? In resin production, the little things make the largest impact. M6, for example, targets general high-frequency server substrates, offering low Df around 0.0035 at 10GHz, but also process flexibility for mass production. Move up to M9 and you find tighter molecular control, exceedingly low ionic content, and a loss tangent sinking below 0.0020 at 10GHz. Higher grades offer better performance as signal speeds increase, but also demand stricter discipline around handling and lamination parameters in the field. We learned this by tracking dozens of end-users—OEMs running everything from vast AI inference clusters to telecom switches—who pushed us to close the gap between benchmark specs in the lab and what boards show after multiple reflows and months of use. Each upgrade in grade delivers more than lab values: better heat resistance, lower water absorption, and compatibility with new copper foil roughness targets. Our engineers throughout the supply chain keep a close eye on what happens after resins meet actual copper foil, glass fiber, and solder mask.

Model and Specifications in Focus

Model M6 sits as our entry for cost-performance balance, where OEMs need reliable signal support up to 40Gbps and rolling ramp of PCB volumes. This grade tolerates standard lamination cycles, and can match semi-additive process (SAP) requirements. Many of our volume server manufacturing clients report easier integration with available standard glass fabrics, making M6 ideal for the first wave of AI deployment servers or rapid scale-ups. As the stakes rise, M9 commands attention in boards hungry for the highest speed, tightest loss budget, and strictest halogen-free requests. Proprietary hydrocarbon chemistry and meticulous filtration produce resins with exceptionally low Dk and Df, meeting next-generation CCL demands including finer copper line etch and thinner core prepregs. Each batch runs full dielectric constant mapping at frequencies up to 40GHz, plus outgassing and time-to-delamination tests to ensure reliability even under server farm hot spots.

Moving Beyond Old Trade-Offs: Process Compatibility and Yields

Working directly with CCL manufacturers, we've fine-tuned each M6-M9 batch for compatibility with both standard multi-stage lamination and emerging lower-temperature cycles optimized for high layer-count PCBs. Our technicians collaborate at factory floors to solve issues like copper foil adhesion, resin flow around tight glass weaves, and compatibility with hybrid resin stackups—especially important for boards where mixed-signal and power layers share tight real estate. Many small but significant tweaks only come from years of pressing real panels in real autoclaves. Process yield jumps noticeably with cleaner flow, as shop workers report fewer delamination or voiding issues—even in 20-layer complex server boards with heavy copper pours.

The Manufacturing Difference: In-House from Monomer to Final Granule

Unlike traders or spot-market blenders, we start from the ground up—polymerizing monomers at our own plant, piloting batch reactors, then scaling each successful run to industrial output. This control over raw material quality translates into far better reproducibility. We answer too many queries about why one lot sometimes behaves differently than another between intervals—real production discipline means keeping molecular weight distribution and purity within much tighter windows than contract blending ever allows. Handling tens of thousands of tons annually, our QA teams use GC-MS, FTIR, and advanced chromatography to catch ionic contamination, chain scission, or any out-of-spec moieties before anything ships to CCL partners. This tight process delivers what customers expect: panel-to-panel and lot-to-lot uniformity, and real confidence during qualification cycles for AI server platforms.

The Shift from Traditional Epoxy—Sustainability and Stability Matters Too

Many of our early partners started their journey on standard halogenated epoxy or PTFE-rich resin blends, used to solve low-loss but struggling with process waste and regulatory concerns. CCL manufacturers often juggle new environmental requirements around VOCs, halogen content, and recyclability. Our M6-M9 hydrocarbon platform delivers non-halogen, stable performance, aligning with RoHS and newer EU directives without the need for constant reformulation. Moving away from PTFE not only slashes scrap but also reduces the headaches of volatile filler sourcing and batch-to-batch drift, which shows up with some third-party filled products. Working hand-in-hand with PCB fabricators—especially in high-mix, low-volume settings—our resins cut out some of the surprise process challenges and help meet stricter sustainability metrics.

Use Cases: Why AI and High-Speed Networks Push Resin Development

It's easy to overlook how AI hardware shapes CCL demands. We’ve watched new workloads—large language models, distributed SAR imaging, generative AI—cause huge spikes in board power density and trace counts. Signal paths shrink, more memory banks squeeze onto the same footprint, and backplane connectors stretch data rates beyond previous design rules. We’ve faced these shifts head on, iterating each major M-grade release against feedback from both design engineers and manufacturing managers. For example, the M8 and M9 product families let hardware teams redesign PCB layouts for 112G NRZ/224G PAM4 transceivers, pushing signal retention further. Network switch makers and hyperscale data centers now set eyes on system upgrades every 18-24 months, each time putting pressure on supply partners to deliver greater performance stability under shrinking line pitches and solder mask openings. In many cases, we observed that our resins not only passed velocity factor tests but also survived stress imposed by larger copper ground pours and intense local heating near AI accelerator chips. This shaped our choice of crosslinkers and stabilizers to guard against real-world warpage and micro-cracking.

Differences from Other Hydrocarbon Resins—Lab to Large-Scale Reality

Some customers new to high-frequency hydrocarbon systems ask what really sets one resin apart from another, especially when competing suppliers twist numbers or mix specs across test frequencies. We operate on one guiding rule: only numbers confirmed by third-party CCL board tests gain a place on our datasheets. Our M6-M9 grades run cleaner, with quantifiable, reproducible low Df and low Dk values across longer panel runs, rather than “hero” coupon sweeps. The real test comes after a batch ships—customers regularly report lower scrap rates, fewer void inclusions, and less process deviation compared to previous generations or competitor grades. Even resin flow-out around tight drill hits and microvia plugging improves, helping boost backend yields on high-stack designs. In the field, M8 and M9 resins see broad uptake not just for their low loss, but for the way they withstand repeated soldering cycles, temperature and humidity extremes, and the load imposed by forced air cooling in server racks. Our feedback systems ensure every update gets built back into the next production run, closing the loop between Shengquan’s manufacturing line and the world’s most demanding electronics labs.

Manufacturing Challenges—Solving Real Supply Chain Pressure

During every industry upturn—AI hardware rush, cloud upgrades, 5G rollouts—the hardest part isn’t announcing new resin grades; it's sustaining stable output when demand spikes and raw input markets fluctuate. Our experience as a direct manufacturer comes to the fore under these pressures. We don’t rely on outside blending or farm out intermediate steps: feedstock selection, conversion, and resin granule formulation all stay within our own ISO-certified operations. This means faster corrective action if a feedstock shift threatens to nudge the Dk or flow properties out of bounds. We regularly invest in backup reactors, bulk silos, and redundant QA lab lines, enabling uninterrupted shipments even as orders triple during server ramp cycles. And because we own the chemistry start to finish, we control not just grade specs, but also the resin modification tools needed to custom-balance prepreg glass content, resin flow, or cure profile for partner CCL plants.

Collaborating Upstream—Direct Feedback from PCB and CCL Producers

Much of our resin innovation comes from day-to-day problem solving with upstream partners—those who run the CCL presses, resin application tanks, and pre-preg lines. Our technical service staff meet on factory floors across Asia, Europe, and North America, monitoring not just lab coupons but actual server panel yields. In recent years, several large partners highlighted new requirements for anti-migration during fine line etching, or tighter stacking tolerances as server board layer counts crept past 24 or 32. We train production teams in how best to handle and process our hydrocarbon grades, and use this plant-level feedback to revise handling protocols and fine-tune cross-link density in new product generations. More than a few M9 upgrades came directly from a customer’s late-night process engineer requesting deeper lamination robustness or more forgiving plug-filling during dense via runs.

Focusing on Real Reliability—From Lab Test to Long-Term Workloads

Paper specs often miss the pressures faced by boards mounted in high-density server racks. Thermal expansion, outgassing, and repeated hot/cold cycles eventually reveal what resin choices mean for board stability. Our reliability team runs hundreds of accelerated aging and peel strength tests every cycle, mimicking actual server workloads—constant temperature swings, local burnout near high-current modules, and long hours under full rack loading. Results show higher grades—especially M8 and M9—pass peeling, solder float, and time-to-delamination tests better than legacy filled epoxies or “off-brand” hydrocarbons. This long-term resilience becomes a real value multiplier: server builders see lower field failure rates, fewer costly RMAs, and less downtime carrying lost inventory. Our approach stays grounded in one goal: support real, operating tech—not just surface numbers—across decades of field operation.

Why Our Experience Matters—Decades in Polymer, Not Trading

In the market’s rush for AI-ready CCL, too many spot players cast wide nets for “high-frequency” resin, chasing short-term wins. By contrast, Shengquan’s core business stayed rooted in direct synthesis, continuous improvement, and closed-loop customer support. This hands-on, chemical manufacturing approach means our engineering teams cut no corners and live with the results through every batch, every shipment, and every customer claim. Building a name over decades in phenolic, epoxy, high-performance hydrocarbon resin involves not just meeting immediate needs, but setting up reliable, auditable systems that outlast product cycles. Our hard-won expertise drives consistent improvement: whether it’s updating batch traceability, doubling down on raw material auditing, or scaling up new reactor designs as hardware standards shift.

Facing the Next Wave—Keeping AI Hardware Ahead of Tomorrow’s Demands

No resin batch runs on autopilot. With each new standard—PCIe 6.0, next-gen 5G, higher AI acceleration loads—customers return looking for that edge that time-tested manufacturing can deliver. At Shengquan Group, we continue refining our hydrocarbon resin pipeline, investing in analytics to track board performance not just in the factory, but in racks and under workloads. This approach lets us spot yield drift, field issues, or rare process variations and feed the right fixes into the next process run. New challenges keep coming: denser component footprints, stricter VOC restrictions, thinner copper and glass tapes, even more convoluted stack-up schemes. Tackling these means more than “checking the box”—it requires resin manufacturing with skin in the game, eye on the details, and deep understanding of both the chemistry and the practical world of AI server board manufacturing. We stand ready to answer the challenge, batch after batch.