APL NanoOil Immersion Cooling Validation
Independent industrial validation of Synano nano-oil in a live CPU server immersion cooling system by APL Group in Germany.
By enhancing a standard PAO dielectric oil with a low concentration of alumina nano-additives, Synano gives data center operators a measurable strategic choice, validated under independent industrial testing by APL Group. In a live CPU server immersion rig, Synano nano-oil cut pump power by 12 to 13% and lowered CPU and heatsink temperatures by around 0.5 °C versus the same base oil, across every flow rate tested. Operators can use this margin two ways: take up to 12 to 13% lower pumping OPEX, or convert the thermal headroom into 30 to 50 kW of additional compute per MW without changing their hardware. Thermal resistance improved 3 to 4%, the fluid proved chemically stable and fully removable, and post-test analysis found no nanoparticles left in the system, marking the second industrial-scale proof-of-concept for Synano’s nano-additive technology.
[ Project ]
COOLING CHALLENGE
Modern AI chips and servers generate heat loads that traditional air cooling can no longer handle. Air cooling now consumes up to 40% of data center energy and is reaching its limit for next-generation hardware, throttling performance and raising costs. Racks that used to draw 5 to 10 kW now demand 30 to 50+ kW, and AI/HPC rack power is projected to climb past 200 kW per rack by 2030.
Immersion cooling, submerging servers directly in dielectric fluid, has emerged as the necessary next step. But conventional dielectric oils are approaching their own limits: lower thermal conductivity, limited heat capacity, and high viscosity translate into reduced thermal performance and higher pumping power. Before committing to a new fluid class, operators need a straightforward answer: does a nano-enhanced oil deliver real, measurable gains in a real system, and does it leave the hardware intact?
Nanoparticle additives have shown thermal and efficiency benefits, but no industrial-scale validation existed for Synano’s drop-in nano-additive technology in a live data center immersion environment before 2025. Synano set out to close that gap with an independent study quantifying thermal performance, pump-power efficiency, and material compatibility, all measured against the same base oil under identical conditions.
Industry
Data Centers / Immersion
Product
Synano Nano-Oil
Location
Landau in der Pfalz, Germany
Year
2025/2026
SOLUTION
Synano formulated and supplied the nano-oil used throughout the study. The base fluid was a synthetic, single-phase PAO (polyalphaolefin) dielectric oil already adopted for immersion cooling of high-density electronics. Synano’s nano-additive combined a base carrier fluid, alumina (Al₂O₃) nanoparticles, and stabilizers to hold the particles in homogeneous suspension. The additive was blended into the base oil at an 80:20 ratio to produce Synano nano-oil. Nanoparticle diameter ranged from 100 to 200 nm, and concentration was kept deliberately low, optimized against the fluid specifications defined by the OCP community for single-phase immersion fluids. Even at this low loading, the nano-oil delivered similar thermal conductivity, lower viscosity, and higher dielectric strength, heat capacity, thermal expansion, and density than the base oil. The additive is non-hazardous and biodegradable.
Automobil-Prüftechnik Landau GmbH APL, an independent German test house specializing in thermal management and coolant performance, ran the validation on its immersion cooling rig. The setup placed an ATX motherboard with an Intel Core i7-12700K processor in a tank fully filled with dielectric oil, instrumented with thermocouples and sensors. The tank fed a liquid-to-liquid heat exchanger that rejected heat to a water loop, with a flow meter and pump circulating the oil. A finned copper heatsink sat on top of the CPU, with a thermocouple placed 0.2 mm above the CPU surface at the base of the heatsink.
Each test held the inlet temperature at 40 °C and mainboard power at approximately 137 W, while the fluid passed through the tank at controlled volume flows. Four relationships were measured: fluid flow rate against heatsink temperature, CPU package temperature, CPU clock speed, and fluid outlet temperature. Each series ran three single runs at defined flow rates; each fluid was held at steady state for roughly one hour per flow rate before logging CPU temperature, frequency, outlet temperature, and pump power. Critically, the base oil was measured both before and after the nano-oil run, and the two baselines had to agree to validate measurement quality. Working-fluid purity was held above 99% (m/m) through sequential flushing, with sampling and ICP analysis after each cycle.
RESULTS
The nano-oil outperformed the base oil on thermal performance, pump efficiency, and stability, consistently and across every flow rate tested.
- Thermal performance improved across the board. The nano-oil lowered heatsink temperatures by around 0.5 °C at almost all flow rates, with CPU package temperatures up to 0.5 °C cooler. The expected inverse relationship between flow rate and temperature held throughout, and the nano-oil sat below the base oil at every point. Calculated at the heatsink-to-fluid interface, thermal resistance improved by 3 to 4% from base to nano-oil. These are modest absolute numbers by design: this was a deliberately low nanoparticle concentration formulation, and higher loadings increase the thermal effect further.
- Compute headroom increased. The i7-12700K’s 3,800 MHz base frequency could be overclocked to 3,845 MHz on the base fluid; with nano-oil it reached roughly 3,850 MHz. The added ~5 MHz of sustained clock speed confirms the more effective heat removal translates directly into higher usable performance headroom.
- Pump power dropped 12 to 13%. Because of its lower viscosity, the nano-oil cut the pump power needed to maintain a given flow rate by 12 to 13% on average across all flow rates, reaching up to 17% at lower flow rates. That reduction flows straight through to operating costs.
- The fluid was stable, clean, and fully removable. The second base-oil baseline, run after the nano-oil, matched the first within experimental error, confirming repeatability and ruling out system drift. ICP analysis showed nearly all elements below detection limits; aluminum appeared in the nano-oil at a consistent 65 to 68 mg/kg, confirming the intentional additive, with no increase after testing. After flushing, no nanoparticles remained in the system, and the materials showed full compatibility with no reactions or adverse interactions.
- The result present the following choice for operators. Holding flow and compute constant returns the efficiency case: about 12% lower pump power, near 15.8 MWh in annual energy savings per MW. Holding flow constant and pushing more compute uses the thermal margin to support roughly 30 kW of extra compute per MW while still saving on pump power. Holding pump power constant and raising flow safely lifts CPU input power and unlocks around 50 kW of extra compute per MW. The same fluid, deployed three different ways, without having to change the existing infrastructure.
TECHNICAL AND PERFORMANCE IMPLICATIONS
- 3 to 4% lower thermal resistance at the heatsink-to-fluid interface, base to nano-oil
- Around 0.5 °C lower CPU package and heatsink temperatures, consistent across all flow rates tested
- 12 to 13% lower pump power on average, up to 17% at lower flow rates, from the nano-oil’s lower viscosity
- Up to 15.8 MWh in annual energy savings per MW from reduced pumping power
- +30 to 50 kW of additional compute capacity per MW, depending on operator strategy with no infrastructure change
- +5 MHz higher sustained CPU clock versus the base-fluid overclock, confirming added thermal headroom
- Chemically stable and fully material-compatible, with no nanoparticles left after flushing, ICP-confirmed
- Achieved at a low nanoparticle concentration; higher loadings increase the thermal gain further
- Drop-in formulation without hardware changes or system modifications
CONCLUSION
The APL study delivered an independent industrial-scale dataset on Synano nano-oil in a live CPU server immersion system, measured directly against the same base oil. The result is consistent. The nano-oil ran cooler, pumped easier, and held compute longer, at every flow rate, while staying chemically stable and leaving the hardware clean on removal.
For data center operators, the implication is a choice rather than a single number. The same drop-in fluid can be tuned for efficiency, taking up to 12 to 13% off pump power for roughly 15.8 MWh of annual energy savings per MW, or for revenue, converting thermal headroom into 30 to 50 kW of additional compute per MW. On a single 1 MW facility, the reclaimed headroom is worth up to 105,000 dollars a year in added compute revenue, or around 15.8 MWh in energy savings, with no extra spend on cooling. Applied across an industry adding on the order of 1,000 MW of capacity, the same per-MW gain scales to roughly 105 million dollars per year in additional value, and it was achieved with a deliberately low nanoparticle concentration. Higher-loaded formulations raise the thermal ceiling further.
This is the second industrial-scale proof-of-concept for Synano’s nano-additive technology. The validated next steps are: Confirm the commercial scenarios under live increased-compute and increased-flow conditions, optimize formulations for forced-convection, high-flow regimes, and scale from single-CPU testing to high-density AI/HPC and rack-level deployments. The physics is proven. The next step is scaling it.
[ Contact ]
Interested in improving compute efficiency?
Synano offers collaborative testing and design support for data center operators, immersion cooling infrastructure partners, and OEMs seeking validated performance improvements through nano-enhanced thermal fluids: