IGBT Cold Plates: A Design and Selection Guide for IGBT and SiC Power Modules

Introduction

EV traction drives, solar and energy storage power conversion systems (PCS), wind converters, industrial motor drives—they’re all heading the same direction: more power in the same enclosure. As power density climbs, losses concentrate in fewer square centimeters, and the cooling margin keeps shrinking.

Silicon carbide makes the problem sharper. For the same current rating, a SiC MOSFET die is far smaller than a silicon IGBT die, so every square centimeter has to shed more heat. An IGBT cold plate that was “good enough” for an older design can quietly eat your junction temperature headroom once you drop in a SiC module—and that shows up later as shorter power-cycling life.

This guide looks at the problem from a cold plate manufacturer’s bench. We’ll cover where the cold plate sits in a power module’s thermal path, how IGBT and SiC requirements differ, how to choose between cooling architectures, and the design, selection, and validation details that actually move the needle. If you’re new to the hardware itself, start with what a cold plate is and how it works.

What Is an IGBT Cold Plate?

An IGBT cold plate is an aluminum or copper plate with internal coolant channels. An IGBT or SiC power module’s baseplate mounts to it through a thermal interface material (TIM). Module losses flow into the plate, and circulating coolant carries them away, keeping junction temperatures within limits and multiple modules evenly cooled.

Key takeaways

• Size an IGBT cold plate from the module’s maximum junction temperature and worst-case losses, not from plate surface temperature.
• SiC modules concentrate heat in smaller dies, so heat-transfer enhancement should sit directly beneath the die footprint.
• The mounting surface must meet the module maker’s flatness and roughness spec, typically around 50 μm per 100 mm and Rz ≤ 10 μm.
• Indirect cooling (flat baseplate + TIM) suits most industrial equipment; pin-fin direct cooling is common in EV traction inverters.
• Define coolant, inlet temperature, pressure drop budget, and validation tests before the first prototype.

Where an IGBT Cold Plate Sits in a Power Module’s Thermal Path

Heat travels from the die to the coolant through three main thermal resistances:

The cold plate directly sets Rth(s-f) and indirectly affects Rth(c-s) through the quality of its mounting surface. Rth(j-c) belongs to the module; no cold plate can change it.

That’s why APALTEK engineers set cold plate performance targets on a junction-temperature basis rather than looking only at plate surface or case temperature. We start from the module’s maximum allowable junction temperature and its worst-case losses, subtract the temperature rise inside the module and across the interface, and work backward to the thermal resistance the cold plate has to hit. In simplified form, for one module dissipating P watts:

Rth(s-f) ≤ (Tj,max − T_inlet − margin) / P − Rth(j-c) − Rth(c-s)

Here, Rth(s-f) is referenced to the coolant inlet temperature. This quick budget ignores the coolant’s temperature rise along the plate and uneven heat distribution among dies, so a real design confirms it with CFD and testing. For definitions, units, and test boundary conditions, see our guide to cold plate thermal resistance; for the step-by-step math, see liquid cold plate design calculations.

One exception: modules with pin-fin baseplates that are cooled directly have no Rth(c-s) term. Their thermal stack runs straight from junction to fluid (j-f)—more on that below.

How Do IGBT and SiC Cold Plate Requirements Differ?

Rising heat flux is well documented. Researchers have reported that IGBTs in hybrid vehicles already run at heat fluxes of roughly 100–150 W/cm², with values expected to exceed 500 W/cm² as current and switching frequency increase. And for SiC, one of the hurdles to volume production is the cooling challenge created by smaller die sizes.

For the cold plate, that translates into two practical rules:

1. Aim the heat-transfer enhancement at the heat sources. SiC concentrates heat in a smaller footprint. Placing pin fins or turbulators directly beneath the die footprint is usually more effective than spreading them evenly across the whole plate—and it keeps pressure drop in check.
2. Interface and flatness issues get amplified. For the same interface resistance, higher heat flux means a bigger temperature rise across that interface. When you move from IGBT to SiC, re-evaluate mounting-surface quality and TIM selection instead of carrying over the old spec.

Which Cooling Architecture Fits Your Power Module?

There are three main ways to liquid-cool an IGBT or SiC power module: indirect cooling through a cold plate, direct cooling of a pin-fin baseplate, and double-sided cooling. The right choice depends on heat flux, cost, and how often modules need to be serviced.

Indirect Cooling: Flat-Baseplate Module + TIM + Cold Plate

This is the most common setup in industrial equipment. The module’s flat baseplate gets a layer of thermal grease or phase-change material and is bolted to the cold plate. The module and the coolant loop stay completely separate, so you can swap a module without opening the liquid circuit, and one cold plate can carry several modules plus other components. The trade-off is one extra interface resistance.

Direct Cooling: Pin-Fin Baseplate Module + Open-Cavity Cooler

A pin-fin baseplate is a module baseplate with an array of small pins on its underside. In direct cooling, those pins extend into a cooler cavity sealed with an O-ring, so coolant flows directly over the fins and the TIM disappears. Researchers note that liquid-cooled SiC modules commonly use pin-fin baseplates with ceramic substrates, and the approach is especially common in EV traction inverters.

It has limits, though. A Danfoss thermal engineer has pointed out that pin-fin designs develop a temperature gradient along the flow direction and add cost: the coolant warms as it moves downstream, so downstream dies run hotter. The cooler’s cavity, seal groove, and flow path also have to match one specific module precisely, which limits reuse.

Double-Sided and Integrated Cooling

Some automotive modules use double-sided cooled packages, with heat paths above and below the die and cold plates clamping the module from both sides. These designs can deliver the highest thermal performance but place the highest demands on tolerances, clamping force, and system integration. They tend to appear in high-volume automotive platforms.

Key Design Decisions for IGBT and SiC Cold Plates

General cold plate design—heat load, materials, channels, flow, pressure drop, and CFD simulation—is covered in our cold plate design guide. Here we focus on issues specific to power module projects.

Mounting Surface: Follow the Module Maker’s Spec

Module makers publish specific requirements for the heatsink mounting surface. Fuji Electric’s IGBT module mounting instructions, for example, call for a surface roughness of Rz ≤ 10 μm and flatness within 50 μm per 100 mm; going beyond those limits raises contact resistance. Requirements also vary by package. For its HybridPACK 2 automotive module, Infineon adds a separate spec for the sealing surface, with roughness Ra ≤ 1.6 μm—a detail that matters whenever a module seals directly against the cooler.

So the flatness, roughness, thread locations and depths, and mounting torque on a cold plate drawing should come from the official mounting instructions for the exact module in your project. For the cold plate manufacturer, that means finish-machining the mounting surface after welding or brazing; otherwise, distortion from joining can push flatness out of tolerance.

Flow Layout: Put the Enhancement Under the Dies

Heat inside a power module isn’t spread evenly; it’s concentrated at the IGBT and diode dies. Channel geometry makes a big difference. A CFD study of a cold plate for three PrimePack3 IGBT modules found that, compared with straight fins, staggered pin fins and oblique fins cut thermal resistance by 46.5% and 60.1%, respectively.

Enhanced structures also add pressure drop, so in practice they’re concentrated beneath the die footprint, while non-heated areas keep low-resistance passages.

Multi-Module Layout: Control Phase-to-Phase Temperature Differences

A three-phase inverter often mounts three or more modules on one cold plate. If coolant flows through them in series, each downstream module sees warmer coolant. The phases end up at different temperatures, so their devices age at different rates.

Common fixes include parallel flow distribution, extra enhancement downstream, and optimized inlet and outlet locations. Done well, the spread can be very small: a Chalmers University of Technology study with Volvo Cars used an optimized rectangular pin layout to keep the temperature difference among three SiC power modules below 2°C in a three-phase inverter.

Double-Sided Mounting: Cooling More Than the Modules

Power modules aren’t the only heat sources in an inverter. DC-link capacitors, busbars, and gate-driver power supplies generate heat too. Many designs mount the modules on one face of the cold plate and capacitors or other components on the other, using a single plate for several heat loads. Define the location and power of those backside loads early, so their heat doesn’t stack on top of the channels beneath the modules.

Coolant and Material Compatibility

EV drive systems typically run water-glycol coolant, often at higher inlet temperatures than data center liquid cooling loops. The cold plate material has to be compatible with the rest of the loop: mixing copper and aluminum components without corrosion inhibitors invites galvanic corrosion. Lock down the coolant type, glycol concentration, and inlet temperature at the start of the project.

Pressure Drop Budget

Cold plate pressure drop has to match what the system pump can deliver. In EVs, the same coolant loop often serves the motor, onboard charger, and other components, leaving a limited pressure drop budget for the inverter cold plate. The design with the lowest thermal resistance isn’t necessarily the right one; you need the best balance between thermal resistance and pressure drop.

Operating Environment

Solar, storage, wind, and charging equipment often runs outdoors or semi-outdoors, so condensation, dust, and temperature cycling matter. Automotive and rail equipment also faces constant vibration. These conditions drive sealing design, fitting selection, surface treatment, and mounting method.

Which Manufacturing Process Is Best for a Power Module Cold Plate?

There’s no single best process: embedded tube is the low-cost option, friction stir welding suits mid- to high-volume aluminum plates, and vacuum brazing or brazed copper fits the highest heat flux. Each process strikes a different balance between thermal performance, cost, and volume. For how each one works, see our overview of cold plate manufacturing processes.

APALTEK builds cold plates with vacuum brazing and brazed copper construction, processes well suited to complex internal fin structures and the high heat flux of modern power modules.

For SiC projects with especially high heat flux, microchannel structures are another option. Learn more in what is a microchannel liquid cold plate.

How to Choose an IGBT Cold Plate by Application

For more on specific industries, see APALTEK’s power and utilities, EV and EV charging, and energy storage application pages.

What to Send Your IGBT Cold Plate Supplier

Our custom cold plate RFQ guide covers the general checklist. For power module projects, these inputs matter most:

• Module part number or package: including baseplate type (flat or pin fin)
• Loss data: worst-case losses per module (W), ideally split between IGBT/MOSFET and diode dies
• Temperature limits: maximum allowable junction temperature and the margin you want to keep
• Coolant conditions: coolant type, glycol concentration, inlet temperature, available flow rate
• Pressure drop limit: the maximum allowable drop across the cold plate
• Mechanical constraints: envelope, inlet/outlet locations and fitting types, module layout, any components on the back side
• Operating environment: outdoor, vehicle, or enclosure; vibration and condensation conditions
• Volumes: prototype quantity and expected annual volume

Validating an IGBT Cold Plate Before Production

Agree on validation items with your supplier early in the project. Common tests include:

• Thermal performance: heater blocks simulate module losses while temperature rise and thermal resistance are measured at the target flow rate and inlet temperature
• Pressure drop: measured across a range of flow rates
• Leak testing: pressure-decay or helium leak testing
• Pressure/burst testing: confirms the safety margin at the system’s maximum pressure
• Environmental reliability: temperature cycling, vibration, and other tests chosen for the application

At APALTEK, validation items are defined around each customer’s requirements rather than a fixed test list. See our testing and validation capabilities for more.

How APALTEK Supports IGBT and SiC Cold Plate Projects

APALTEK is a liquid cold plate manufacturer with engineering capabilities spanning thermal design, mechanical design, simulation, prototyping, and volume production. Every cold plate is customized to the customer’s modules, losses, and mounting conditions, not picked from a catalog as a “close enough” fit. For paid prototype projects, we typically provide thermal simulation at no extra cost, so channel and structural choices are confirmed before the first part is built.

FAQ

Can an IGBT cold plate be used with a SiC module?

Not necessarily. SiC dies are smaller and run at higher heat flux, so the original cold plate’s enhancement zones may not line up with the new die locations, and mounting requirements may differ. When you change modules, re-evaluate the cold plate against the new module’s losses and package.

Should an IGBT cold plate be aluminum or copper?

Aluminum is lighter and less expensive, and it handles many industrial and automotive projects. Copper conducts heat better and suits very high heat flux or tight spaces. Either way, check compatibility with the other materials in the coolant loop.

How flat does an IGBT cold plate need to be?

It depends on the module, so always follow the maker’s mounting instructions. As a reference point, Fuji Electric specifies flatness within 50 μm per 100 mm and surface roughness Rz ≤ 10 μm for the heatsink mounting surface of its industrial IGBT modules.

Do I still need thermal grease with a cold plate?

Yes, for flat-baseplate modules. The TIM fills microscopic gaps between the baseplate and the cold plate, but too thick a layer increases thermal resistance, so follow the module maker’s recommended material and thickness. Directly cooled pin-fin modules don’t need thermal grease.

Is pin-fin direct cooling better than a cold plate?

Pin-fin direct cooling eliminates the interface resistance and typically delivers higher thermal performance, one reason it’s common in EV traction inverters. Cold plates are more versatile, make module replacement easier, and keep costs more predictable, which suits industrial and energy equipment. The right choice depends on heat flux, cost, and serviceability requirements.

What coolant should an IGBT cold plate use?

Industrial and data center systems often use deionized water or water-glycol; automotive drive systems typically use water-glycol. What matters most is compatibility with the cold plate material and the rest of the loop, with inlet temperature and concentration defined during design.

Start Your Cold Plate Evaluation With Module Data

The right IGBT or SiC cold plate doesn’t come out of a catalog. It’s worked out step by step from the module part number, loss data, and coolant conditions.

If you’re developing an inverter, converter, or charger, send APALTEK’s engineering team your module part numbers, worst-case losses, coolant type and inlet temperature, and mechanical constraints. We’ll get back to you within one business day to start evaluating your cold plate design. Request a quote, or browse our cold plate products.