Coolant Compatibility: Glycol, Water and Corrosion
Short answer: For most liquid-cooled heat sinks, use 25–50% propylene or ethylene glycol in deionized water with a validated inhibitor package, and keep the loop to one wetted metal family wherever possible. Pure deionized water cools best but is aggressive to aluminum; pure glycol is safe but roughly 15–20% worse on heat transfer than a 50/50 mix. Aluminum heat sinks paired with copper cold plates or brass fittings will corrode at the joint unless the coolant carries a corrosion inhibitor and the loop is electrically isolated. In a mixed-metal loop, expect 0.05–0.2 mm of galvanic metal loss per year at the copper–aluminum interface without inhibition.
Why coolant compatibility decides whether your heat sink survives
A heat sink is a passive part until you put fluid through it. Once coolant enters a cold plate, extruded tube, or skived-fin channel, the metal becomes part of an electrochemical system. Thermal performance is the easy half of the problem. The hard half is chemistry: pH drift, dissolved oxygen, ion migration, and the galvanic couple formed when two dissimilar metals share the same conductive fluid.
In practice, most field failures we see in liquid-cooled electronics are not "the heat sink was too small." They are pitting under a fitting, white aluminum hydroxide sludge blocking a microchannel, or a pinhole leak at a brazed joint after 18 months. All three trace back to coolant selection and material pairing decisions made at the drawing stage.
This article covers what actually matters for B2B buyers specifying heat sinks and cold plates: glycol concentration, water quality, corrosion mechanisms, inhibitor packages, and the material combinations that work without exotic chemistry.
Glycol vs water: which coolant performs better?
Water is the best common heat transfer fluid. Its thermal conductivity, specific heat, and viscosity make it roughly 2–3x more effective than glycol by volume in a typical cold plate. Glycol exists in the loop for freeze protection and inhibitor carriage, not for cooling.
As glycol concentration rises, three things happen at once:
- Thermal conductivity drops
- Viscosity rises, so pump power increases for the same flow
- Freeze point falls (the reason you added it)
The table below shows typical indicative values for a 50/50 ethylene glycol/water mix versus plain water at 25 °C. Treat these as design guidance, not datasheet values — every commercial coolant differs.
| Property (25 °C, indicative) | Deionized water | 50/50 EG/water | 50/50 PG/water |
|---|---|---|---|
| Thermal conductivity (W/m·K) | ~0.60 | ~0.40 | ~0.38 |
| Specific heat (kJ/kg·K) | ~4.18 | ~3.30 | ~3.50 |
| Dynamic viscosity (mPa·s) | ~0.89 | ~3.5 | ~5.0 |
| Freeze protection | None | to about −37 °C | to about −30 °C |
| Relative heat transfer | 1.00 | ~0.70 | ~0.68 |
The practical takeaway: if the equipment never sees sub-zero ambient and never ships unheated, run 20–30% glycol rather than 50%. You keep most of the freeze margin you actually need, retain better heat transfer, and still carry enough inhibitor to protect the metals.
When pure water is acceptable
Sealed, single-metal, indoor loops with a proper deionizing bypass and an oxygen scavenger can run near-pure water successfully. This is common in high-performance computing and some laser systems. It requires disciplined commissioning and periodic conductivity checks. If your end user will top up the loop with tap water, do not design for pure water.
Glycol type: ethylene vs propylene
Ethylene glycol (EG) transfers heat slightly better and costs less. Propylene glycol (PG) is far less toxic, which matters for food-adjacent, medical-adjacent, and consumer-accessible equipment. PG is also more viscous and slightly more prone to forming organic acids as it degrades, so it needs a better inhibitor package. Neither is "corrosion-proof" — both become acidic over time without buffering.
How does galvanic corrosion start in a heat sink loop?
Galvanic corrosion needs three things: two dissimilar metals, an electrolyte bridging them, and an electrical path. A copper cold plate bolted to an aluminum heat sink with coolant flowing through both satisfies all three instantly.
The driving force is the electrode potential difference. Copper sits high on the galvanic series; aluminum sits low. In a conductive fluid, the aluminum becomes the anode and sacrifices itself. The result is pitting, not uniform thinning — which is why failures appear suddenly.
| Metal pair in a wetted loop | Galvanic risk | Practical mitigation |
|---|---|---|
| Aluminum + aluminum (same alloy family) | Low | Inhibitor + pH buffer |
| Aluminum + copper / brass | High | Full inhibitor package, isolate fittings, avoid bare Cu/Al contact |
| Aluminum + stainless steel 304/316 | Moderate | Inhibitor; keep 316 preferred |
| Copper + brass + stainless | Low–moderate | Standard inhibitor package |
| Aluminum + nickel plating | Moderate | Ensure plating is pore-free; add inhibitor |
| Aluminum + anodized aluminum | Low | Anodizing is a barrier, not a seal — keep inhibitor |
Two rules cut most of the risk. First, keep the wetted path to one metal family. Second, if you must mix, isolate the mechanical joint with non-conductive washers or bushings and let the coolant's inhibitor package handle what remains.
Why anodizing is not a corrosion solution
Type II and Type III anodizing on an aluminum heat sink creates a dielectric oxide layer, but that layer is thin, brittle, and porous at thread roots and cut edges. Coolant will eventually reach bare aluminum at a machined face or a fitting thread. Anodizing helps dielectric isolation; it does not replace an inhibitor. See our notes on anodized versus bare finishes for where each finish belongs.
Material selection: what to specify for a wetted heat sink
The material decision is made before the coolant decision, because coolant chemistry is chosen to suit the metals — not the other way around.
- Aluminum 6063 / 6061 — the default for extruded and skived heat sinks and many cold plates. Light, extrudable, good thermal conductivity (~200 W/m·K). Requires an inhibitor package in any mixed loop.
- Copper C1100 / C1020 — best thermal conductivity (~390 W/m·K) and the standard for high-flux cold plates and microchannels. Chemically stable in most coolants but drives galvanic attack on aluminum.
- Copper with aluminum fins — a common cost compromise. Works only with a full inhibitor package and careful isolation. Document it as a mixed-metal loop so the end user does not refill with plain water.
- Stainless 304/316 — used for fittings, tubes, and manifolds. 316 is preferred in chloride-bearing or long-life loops.
- Nickel plating — used on copper to reduce copper ion release and improve wear resistance in pump-adjacent parts.
For a deeper comparison of cold plate material choices, see liquid cold plate materials.
Brazing, soldering and flux residue
A cold plate is only as clean as its last process step. Flux residue from brazing or soldering is a corrosion accelerator: it holds moisture, releases chlorides, and creates local pH cells. Specify post-braze cleaning with a validated rinse and a conductivity check on the final rinse water. If you cannot verify cleanliness, do not put the part into a long-life sealed loop.
Coolant chemistry: inhibitors, pH and water quality
A finished coolant is a system, not a single ingredient. The main components and what they do:
| Component | Function | Typical target |
|---|---|---|
| Glycol | Freeze protection, base fluid | 20–50% by volume |
| pH buffer (e.g. borate, phosphate) | Hold pH in the protective range | pH 7.5–9.5 |
| Aluminum inhibitor (silicate, carboxylate) | Passivate aluminum surfaces | Per supplier spec |
| Copper inhibitor (tolyltriazole, benzotriazole) | Protect copper and brass | 50–200 ppm typical |
| Oxygen scavenger | Reduce dissolved O₂ | Per supplier spec |
| Biocide | Prevent biofilm and sludge | Per supplier spec |
| Defoamer | Prevent pump cavitation | Trace |
Water quality matters as much as the additive package. Deionized or softened water is standard. Tap water brings chloride, sulfate, and hardness ions that consume inhibitors and accelerate pitting. If the loop will be topped up in the field, specify a top-up fluid that matches the factory fill exactly — mixing incompatible inhibitor chemistries can precipitate solids that block microchannels.
Design rules that reduce corrosion risk
Chemistry handles the fluid. Geometry handles the rest. A few design choices make a wetted heat sink far more forgiving:
1. Avoid crevices and dead zones. Stagnant fluid under a gasket or in a blind tapped hole becomes an oxygen-depleted anode. Machine through-holes where possible.
2. Keep flow velocity in range. Below roughly 0.5 m/s, particulates settle; above roughly 2–3 m/s, erosion-corrosion accelerates, especially on copper elbows.
3. Use compatible fittings. Brass fittings on an aluminum manifold are a galvanic couple. Choose aluminum or plastic fittings for aluminum bodies, or isolate with a dielectric bushing.
4. Specify surface finish on wetted channels. A smoother channel reduces deposit adhesion, but do not polish to a mirror — some surface roughness helps inhibitor films anchor.
5. Design for drain and flush. A loop that cannot be fully drained cannot be properly serviced.
These are the same discipline points we apply when machining CNC-machined heat sinks with integrated fluid paths, and they matter just as much on simpler extruded heat sinks used with tube-and-fin liquid loops.
Immersion and single-phase fluid choices
Immersion cooling replaces the sealed loop with a tank of dielectric fluid. The corrosion question changes shape: you are no longer managing glycol degradation, but you are managing fluid aging, material compatibility of every component in the tank, and elastomer swelling. Our overview of immersion cooling covers where that approach fits.
Testing and validation before you commit
Do not accept a coolant recommendation on paper alone. A short, cheap validation program catches most problems:
- Static immersion test — coupons of each wetted metal in the candidate coolant at 60–80 °C for 500–1000 hours. Weigh before and after; look for pitting under magnification.
- Galvanic couple test — aluminum and copper coupons electrically bonded in the same fluid. Measure the galvanic current; a rising trend indicates inhibitor depletion.
- Loop flush and conductivity check — verify final rinse water conductivity before fill.
- Accelerated thermal cycling — 500–1000 cycles between operating and ambient to expose seal and joint fatigue. Our thermal cycling notes describe how to structure this.
- Fluid analysis at intervals — pH, glycol concentration, inhibitor reserve, and metal ion concentration (copper and aluminum in ppm).
Track copper and aluminum ion levels over time. Rising aluminum with stable copper points to a mixed-metal loop losing its anode. That is your early warning, months before a leak appears.
Sourcing considerations for wetted heat sinks
Because the wetted path is a chemistry problem, the supplier's process control matters as much as the drawing. Ask for:
- Material certificates with alloy and temper
- Braze or weld procedure and post-process cleaning records
- Final rinse conductivity data
- Flatness and channel dimension reports on sealing faces
BQUQ runs four production lines in one Dongguan factory under ISO9001 — CNC machining to ±0.005 mm, metal stamping, custom springs, and heat sink production — which means a wetted cold plate, its mounting bracket, and its spring-loaded retention hardware can be quoted and produced as one package instead of three. Standard quotes return in 12 working hours, and MOQ is flexible for prototype and pilot builds. Browse the full heat sink range or send drawings to sc@bquq.com.
Frequently Asked Questions
Q: Can I use plain tap water in an aluminum heat sink loop?
A: No, not for anything beyond a short test. Tap water carries chloride, sulfate, and hardness ions that break down the protective oxide on aluminum and consume corrosion inhibitors quickly. If you need a water-only loop, use deionized water with an oxygen scavenger and a conductivity monitor, and keep the wetted path to a single metal family.
Q: What glycol concentration gives the best balance of protection and cooling?
A: For most indoor equipment, 25–30% glycol by volume is the practical sweet spot. It provides meaningful freeze margin, carries a full inhibitor package, and retains noticeably better heat transfer than a 50/50 mix. Reserve 50% for equipment that ships or stores in sub-zero conditions without drainage.
Q: Why does my copper cold plate corrode the aluminum heat sink it is bolted to?
A: Because coolant bridges the two metals and creates a galvanic cell. Aluminum becomes the anode and pits. Fix it with a full inhibitor package that includes a copper-specific azole, dielectric isolation at the mechanical joint, and by avoiding bare copper-to-aluminum contact anywhere in the wetted path.
Q: How often should coolant be replaced in a sealed loop?
A: Typical service intervals run 12–24 months for inhibited glycol in a clean, sealed, single-metal loop, but the real answer comes from fluid analysis. Test pH, glycol concentration, inhibitor reserve, and metal ion levels at 6 and 12 months. Replace when pH drifts outside 7.5–9.5 or inhibitor reserve falls below the supplier's minimum.
Q: Does anodizing an aluminum heat sink make it safe with copper in the same loop?
A: No. Anodizing is a thin dielectric barrier that helps with electrical isolation but is porous at cut edges, threads, and machined sealing faces. Coolant will reach bare aluminum at those points. You still need a corrosion inhibitor package and, ideally, a single-metal wetted path.
Related Resources
- About BQUQ and our Dongguan manufacturing footprint — /about/
- Full heat sink product range, including liquid cold plates — /heat-sinks/
- Extruded aluminum heat sink profiles — /extruded-heat-sinks/
- CNC-machined heat sinks and cold plates — /cnc-machined-heat-sinks/
- Industry trends in thermal management — /industry-dynamics/
- Technical articles and engineering guides — /bquq-blog/
- Frequently asked questions — /faq/
- Case studies — /case/
- Contact the engineering team — /contact/
Authored by the BQUQ Engineering Team. BQUQ (Dongguan) runs CNC machining (±0.005 mm), metal stamping, custom springs, and heat sink production in one ISO9001 factory. Source-direct from Dongguan, China — quote in 12 hours: sc@bquq.com | WhatsApp +86 13713157787 | www.bquq.com
