Tag Archive for: SiC

fonones y electronesApplication Notes

Phonon Engineering and Advanced Thermal Management in Power Electronics

Engineering Insight · Thermal Management

Phonon Engineering and Advanced Thermal Management in Power Electronics

From phonon-scattering physics to next-generation TIMs, substrate ceramics and active thermal control in SiC/GaN converters

ZT > 1.5
Skutterudite at 800 K
2–5×
Life via active ΔTj control
10–30%
ΔTj reduction, active TM
50 kW
Waste heat at 1 MW / 95%

Executive summary

Thermal management, together with reliability, is the factor that limits power density and useful life of power-electronics converters. The transition to wide-bandgap semiconductors (SiC and GaN) does not solve this problem — it shifts it: more efficient devices generate less heat, but over smaller die areas they produce more concentrated heat fluxes. The bottleneck, once again, is how to evacuate that heat without penalising reliability or introducing thermo-mechanical failure mechanisms.

Contemporary solid-state physics offers a set of conceptual tools — phonon-scattering engineering, Phonon-Glass Electron-Crystal (PGEC) materials, nanostructured superlattices — that have traditionally been developed to improve the thermoelectric figure of merit ZT. The underlying principles, however, are the same ones that govern the design of any advanced cooling system and any waste-heat recovery strategy. This article synthesises the academic state of the art and translates it into the decisions that a designer of industrial and railway inverters will be making in the coming years.

1. The thermal problem of power electronics

The junction temperature Tj of a power semiconductor simultaneously dominates electrical performance, useful life and reliability. The rule every power electronics engineer knows is brutal: for every 10 °C increase in Tj, the life of the power module drops to roughly half by the Arrhenius law; and for every 30 % increase in the thermal cycle amplitude ΔTj, the wire bonds of the module lose a factor of 3 to 5 in cycles to failure according to the Coffin-Manson model.

The most efficient reliability lever is not selecting better components, but reducing both Tj and ΔTj. And that reduction depends on the quality of the entire thermal chain — junction to case, case to heatsink, heatsink to ambient. Each of those links is a thermal resistance Rth in series, and improving it requires understanding what carries heat and what scatters it. Better cooling is not a fluid-mechanics problem solvable with bigger fans; it is, ultimately, a problem of how phonons propagate, scatter and dissipate.

2. Heat carriers — phonons and electrons

In any solid, heat is transported by two parallel mechanisms: the diffuse flow of free electrons (relevant in metals and degenerate semiconductors) and the propagation of crystal-lattice vibrations, quantised as phonons (dominant in insulators and non-degenerate semiconductors). Total thermal conductivity decomposes as κ = κe + κL.

2.1 The kinetic approximation and its limits

The classical phenomenological estimate is κ = (1/3)·C·v·l, where C is specific heat capacity, v the mean carrier velocity and l the mean free path between collisions. A rigorous analysis starts from the Boltzmann transport equation in the relaxation-time approximation. Electronic thermal conductivity is κe = (1/T)·[K₂ − K₁²/K₀], where the generalised transport integrals Kn capture the interplay between electrical conductivity σ, Seebeck coefficient α and κe. Thermoelectric design is, in essence, a vector optimisation problem.

2.2 Failure of the Wiedemann-Franz law

The Wiedemann-Franz law states κe/σT = L₀ = constant. It holds when electrons scatter through elastic (horizontal) processes. At intermediate temperatures, vertical inelastic processes appear: electrons relax through collisions with low-energy phonons that change carrier energy without significantly altering wave vector k. The net effect is that κe drops more aggressively than σ, and the actual Lorenz number falls below L₀.

For an inverter designer, this subtlety is tangible: simple kinetic thermal models overestimate the heat-transport capability of electrons at intermediate temperatures. In copper or aluminium DC bus bars operating in the 80–150 °C range typical of a railway inverter, vertical processes affect the real thermal balance — not an academic correction.

3. Lattice thermal conductivity and phonon scattering

In semiconductors and ceramics — the materials that make up the thermal path of a power module, from the SiC chip to the AlN substrate via the SAC305 solder — thermal conductivity is dominated by κL.

3.1 The Callaway model and Umklapp processes

The Callaway model integrates two types of phonon-phonon process. Normal (N) processes conserve crystal momentum and produce no direct thermal resistance; Umklapp (U) processes involve total momentum “wrapping around” outside the first Brillouin zone and are the primary source of intrinsic thermal resistance at temperatures comparable to or above θD.

Slack derived the U-process relaxation rate as τU⁻¹ ≈ (ℏγ²/Mv²θD)·ω²·T·exp(−θD/3T), where γ is the Grüneisen parameter — quantifying crystal anharmonicity — and M is the mean atomic mass. Design consequence: materials with high γ show high κL degradation with temperature; materials with high M scatter phonons at lower frequencies. This is why Bi₂Te₃ (high M) outperforms Si (high κ) as a thermoelectric despite similar crystal structures.

4. Phonon-Glass Electron-Crystal (PGEC) materials

The PGEC paradigm, formulated by G. A. Slack in the early 1990s, summarises the goal of modern thermoelectric engineering: an ideal material should transport electrons like a perfect crystal (high mobility, high σ) and transport heat like an amorphous glass (low κL). Forcing coexistence requires structures where unit-cell complexity acts as a selective phonon filter.

4.1 Skutterudites — the rattler mechanism

Skutterudites (general formula MX₃, M = Co, Rh, Ir; X = pnictogen) contain large icosahedral cages that host guest atoms. When a high-mass electropositive atom is introduced (lanthanides such as La, Ce, Yb; alkaline earths such as Ba), the guest is weakly bound and vibrates with anomalously large amplitude at a characteristic Einstein temperature θE — it literally rattles within the cage. This resonant scattering strongly suppresses phonons near the rattling frequency, reducing κL by up to an order of magnitude without proportionally affecting σ.

State of the art: double-filled skutterudites (Yb + Ba) have reached ZT > 1.5 at 800 K, compared with ZT ≈ 0.3 for the unfilled binary skutterudite — a 5× improvement obtained by engineering the same crystal with guest-atom insertion alone.

4.2 Clathrates

Clathrates (Types I, II, III, defined by cage geometry) operate under a similar principle with a more rigid and structurally complex host lattice. The reduction of κL is exacerbated by intrinsic guest disorder and the multiplicity of scattering centres in the unit cell — typically an order of magnitude lower κL than the host material without guests.

5. Nanostructuring and superlattices

The second great lever for reducing κL without destroying σ is nanostructuring. The fundamental idea is to introduce artificial boundaries at a scale that limits the phonon mean free path (λph typically 100–300 nm at room temperature) without significantly affecting that of electrons (λdB ~10 nm in degenerate semiconductors). The result is selective scattering: phonons blocked, electrons passing.

5.1 Kapitza resistance and coherent transport

At any interface between two materials with different vibrational properties — for example, Si/SiGe in a superlattice — phonons encounter thermal boundary resistance (Kapitza resistance, RB). The physical origin is the disparity between the dispersion relations ω(q) of the two materials. In superlattices with periods of a few nanometres, a qualitatively new phenomenon appears: the superlattice creates a new folded Brillouin zone with mini-gaps that block certain frequency ranges — a phononic crystal by analogy with photonic crystals.

5.2 Nanowires and low dimensionality

Reducing geometric dimensionality limits the phonon mean free path through surface scattering, especially when surfaces are rough. An emblematic case: graphite in 3D bulk form shows κ up to 2,000 W/m·K in the basal planes, but individual nanowires can show κ reduced by orders of magnitude through rough-surface scattering.

6. Practical implications for the inverter designer

The design-level view of the thermal chain (thermal budget, Foster/Cauer modelling, cooling hierarchy) is covered in our article Thermal Design in Power Electronics. Here we focus on what the physics of phonon transport adds to those decisions.

6.1 Thermoelectric recovery of waste heat

A 1 MW traction inverter at 95 % efficiency dissipates 50 kW of heat — comparable to an industrial electric heater. Advanced thermoelectric materials with ZT > 2 would open the possibility of recovering a non-negligible fraction as usable electricity, powering the converter’s own auxiliaries or returning it to the DC bus. On the 2028–2035 horizon, this represents a credible path to raising system efficiency above the 99 % plateau.

6.2 Advanced thermal interface materials (TIM)

The TIM layer between the power module and the heatsink is the most critical link in the thermal chain. Thermal conductivities by technology:

TIM technology κ (W/m·K) Notes
Standard silicone grease 1–3 Outgassing risk in railway (optical, sensor environments)
Advanced ceramic pads 6–10 Current industrial standard
Phase-change metallised pads (PCM) >12 Premium PSU standard in railway programmes — no silicone migration
Dispersed graphene TIM ~30 Near-term commercial — qualification ongoing
Gallium-based liquid metals 50–80 Al compatibility challenges; high-power prototypes
Vertically aligned CNT >100 Research frontier — not yet at industrial maturity

Premium PSU already standardises PCM pads in railway programmes where silicone is problematic (cabins with optical sensors, instrumentation equipment).

6.3 Substrate ceramics — the battle for high κ

The DBC (Direct Bonded Copper) substrate determines how much of the heat generated in the junction reaches the heatsink. Options ordered by increasing κ:

Substrate κ (W/m·K) CTE (ppm/K) Positioning
Al₂O₃ (alumina) 24–26 7.0 Cost-optimised industrial applications
Si₃N₄ (silicon nitride) 70–90 3.2 Gaining share — superior mechanical strength, severe thermal cycling
AlN (aluminium nitride) 150–180 4.5 Dominant in railway traction, industrial SiC
Diamond DBC >1,000 1.5 Research / aerospace — cost barrier for industrial use

AlN is the dominant option in railway-traction modules and industrial SiC. Si₃N₄ is gaining share because it combines acceptable κ with superior mechanical strength — particularly important where DBC fatigue from thermal cycling is the dominant failure mechanism.

6.4 Active thermal management — the lever available today

While advanced materials mature, the lever with the best cost/benefit ratio available today is software-based active thermal management. Work by Andresen and Liserre (Microelectronics Reliability, 2014) demonstrates ΔTj amplitude reductions of 10–30 % through three control strategies:

Three control strategies for active thermal management

  • Adaptive switching-frequency modulation — fsw is reduced under transient conditions to lower switching losses and relieve the most stressed module.
  • Loss redistribution among parallel arms — the controller identifies the module with the highest Tj and shifts current towards less stressed ones.
  • Real-time Tj estimation — from Vce_sat and current, with compensation by directly measured case temperature.

ΔTj-per-cycle reductions of 10–30 % translate, via Coffin-Manson, into 2–5× life-extension factors. This requires no novel materials and no manufacturing-process changes: it is implemented entirely in controller firmware. At Premium PSU we incorporate it progressively into higher-power platforms where the MTBF return justifies the additional control complexity.

7. Synthesis and outlook

Phonon-scattering engineering has proved to be the most effective route to overcome the limits of thermoelectric efficiency and, by extension, to rethink the foundations of thermal management in power electronics. Integrating the models of Klemens, Slack, Callaway and Pohl, the design of next-generation materials and architectures pivots on three pillars:

Pillar Mechanism Horizon
Selective phonon scattering Rattler atoms (skutterudites), cage disorder (clathrates), nanostructured interfaces 2026–2030 — TIM and substrate integration
Phononic bandgap engineering Superlattice folded Brillouin zones, coherent phonon blocking 2030–2035 — module-level integration
Waste-heat recovery Advanced ZT > 2 thermoelectrics, on-board TEG for auxiliaries 2030–2035 — system level

For the power electronics designer, the practical translation is clear: the physics developed to improve ZT in thermoelectrics is exactly the physics that governs next-generation TIMs, advanced substrates and active thermal management strategies. Understanding phonon transport is not an academic luxury — it is the foundation of the next decade of power-density and reliability improvements in power electronics.

About Premium PSU

Premium PSU (Premium SA) designs and manufactures custom power supplies and power converters in Barcelona, with over 40 years of experience in railway, energy, industrial, data centre and defence applications.

Talk to our engineering team

Discuss your power conversion requirements with Premium PSU.

Contact us

datacenterApplication Notes, Data Center

Engineering the Power Stack for AI Data Centers

Engineering Insight · AI Data Centres & HPC

Engineering the Power Stack for AI Data Centers

Custom mission-critical power electronics for AI factories, HPC clusters and hyperscale data centres — >97% efficiency, >1 kW/L density, MTBF >10⁶ h

40+
Years in mission-critical power
>900
Custom designs delivered
>97%
Efficiency target
>10⁶ h
MTBF baseline

Three barriers between AI and reliable power

Generative AI and HPC workloads are creating unprecedented demand for power infrastructure. Rack densities are growing 10× in five years. Hyperscale operators demand Titanium-class efficiency. AI factories tolerate zero downtime. Three barriers define the engineering problem — and all three must be solved simultaneously.

Efficiency: >97%

Titanium-class efficiency at every conversion stage. Every percentage point saved at 1 MW means 10 kW less heat — and a real capex reduction in cooling infrastructure.

Power density: >1 kW/L

Rack density is growing from 10–15 kW/rack to 200+ kW/rack by 2028. WBG semiconductors (SiC, GaN) and planar magnetics are the only viable path to meet these targets.

Reliability: >10⁶ h MTBF

AI factories run 24/7 with zero tolerance for downtime. Physics-of-Failure design, RAMS analysis and thermal budget management are non-negotiable engineering requirements.

Mission-critical power engineering, now for AI infrastructure

Premium PSU is not entering the data centre market from zero. We are transferring over 40 years of engineering experience from railway traction (EN 50155, EN 61373), defence and grid infrastructure — sectors where failure is simply not an option — into a market that now demands exactly those same standards.

Capability Focus What it means in practice
WBG semiconductors SiC · GaN · Ga₂O₃ Baliga FOM-driven device selection. SiC for high-blocking-voltage stages; GaN for high-frequency switching. Gate-driver design, EMI management and long-term reliability qualification in-house.
Thermal engineering RAMS · phonon engineering 55% of power electronics failures are thermally driven. We apply Coffin-Manson fatigue modelling, thermal budget allocation and advanced substrates (Si₃N₄, AlN) from prototype to production.
Digital control FPGA · SoC · predictive control High-bandwidth digital loops, model-based design (MiL→SiL→HiL). Real-time telemetry via PMBus, SNMP, Modbus and REST. Full DCIM integration out of the box.
Passive components Magnetics · capacitors Custom inductors and transformers with nanocrystalline/ferrite cores. Elimination of electrolytic capacitors in critical paths — historically 35% of field failures.

CPD-Ready 125V DC Kit — from pilot to hyperscale

Turn-key power infrastructure for AI compute, designed for native OCP Open Rack v3 (ORv3) architecture and full DCIM integration. Available in four power configurations — 20, 40, 60 and 72 kW — to match rack density targets from edge to AI factory scale.

  • N+1 redundancy: hot-swap modules with seamless failover — zero downtime during maintenance windows.
  • OCP ORv3 compliant: native support for the Open Compute Project Open Rack v3 architecture.
  • Unified telemetry: SNMP / Modbus / REST — full integration with DCIM platforms.
  • Pre-validated EMC: CE, IEC 62368-1, UL ready — integrated EMC-checked BOM.
  • WBG switching stages: SiC-based conversion for >95% efficiency at full load.
  • Co-development ready: custom configurations of voltage, form factor, firmware and interface.

Standard catalogue models for data centre architecture

Model DC application Key adaptation Power range
CKR-3600 Modular 48V front-end rectifier ≥97% eff., OCP ORv3 hot-swap, PMBus/SNMP Up to 3.6 kW/module
CRS-2000 / 3000 DC-DC mid/high power (48V ↔ 380V) PMBus telemetry, >1 kW/L density 2–3 kW
ODS-1500 / 3000 AC/DC isolated front-end, edge Modular scaling, >95% peak, Titanium 1.5–3 kW
EPS-200 / 2000 DC UPS — edge and critical rooms Li-ion modular, bidirectional, SNMP 200 W – 2 kW
ODX / OVX-6000 Resilience inverters and microgrid 19″ rack, remote control + telemetry Up to 6 kW

Over 40 years engineering power where failure is not an option

Our engineering credentials come from the most demanding industrial applications. Every methodology we apply in AI data centre projects — Physics-of-Failure, RAMS, thermal budget analysis — has been validated under real-world mission-critical conditions:

  • Railway traction and high-speed rail: converters and inverters qualified to EN 50155, EN 61373 and EN 45545-2, including cold-start designs for −40 °C arctic operation and cross-border certification management. The same thermal and reliability design methodology applies to AI rack power.
  • Urban rail systems: full RAMS analysis and EN 61373 vibration qualification, with rapid prototyping cycles.
  • Grid infrastructure: modular inverter deployments across distribution networks. Grid-grade reliability and modular architecture are directly applicable to DC microgrid and AI campus designs.

Power electronics for AI data centres — technical Q&A

What power electronics does Premium PSU offer for AI data centres?

Premium PSU offers custom power electronics for AI data centres, including the CPD-Ready 125V DC Kit (20/40/60/72 kW, OCP ORv3 compliant), modular 48V rectifiers (CKR-3600), DC-DC converters (CRS-2000/3000), AC/DC isolated front-ends (ODS-1500/3000) and DC UPS systems (EPS-200/2000). All products feature SiC/GaN WBG switching for >97% efficiency, PMBus/SNMP/REST telemetry and N+1 redundancy design.

Is Premium PSU OCP ORv3 compliant?

Yes. The CPD-Ready 125V DC Kit is designed for native OCP Open Rack v3 (ORv3) architecture, supporting hot-swap modules, N+1 redundancy and full DCIM integration via PMBus, SNMP and REST interfaces. Premium PSU actively monitors OCP specifications and roadmaps to maintain alignment.

What is the MTBF of Premium PSU power supplies for data centres?

Premium PSU standard platforms achieve MTBF >10⁶ hours, validated through Physics-of-Failure (PoF) design methodology and RAMS (Reliability, Availability, Maintainability and Safety) analysis applied from day one of each project. Thermal budget allocation, Coffin-Manson fatigue modelling and elimination of electrolytic capacitors in critical paths are standard practice.

Can Premium PSU co-develop custom power solutions for AI infrastructure?

Co-development is at the core of how Premium PSU works. We assign dedicated engineers, run co-simulations, and optimise thermal and electrical budgets together with the customer. An NDA is signed on day one. Prototype delivery in under 14 weeks is our demonstrated standard. Custom configurations include voltage levels, form factors, firmware, communication interfaces and mechanical integration.

What efficiency class do Premium PSU AI data centre products achieve?

Premium PSU products target ≥97% efficiency at rated load, meeting Titanium-class requirements (80 PLUS Titanium, ≥96% at 50% load, ≥94% at 20% load). WBG SiC-based switching stages and planar magnetics are the primary enablers. Efficiency is guaranteed across the full operating temperature range for data centre environments.

What certifications do Premium PSU AI data centre products carry?

Products are certified or in certification for IEC 62368-1, CE (EU), UL (North America) and Energy Star. The CPD-Ready 125V DC Kit also targets 80 PLUS Titanium qualification. EMC compliance (CISPR 32, IEC 61000) is validated in-house at Premium PSU’s EMC laboratory in Barcelona.

About Premium PSU

Premium PSU (Premium SA) designs and manufactures custom power supplies and power converters in Barcelona, with over 40 years of experience in railway, energy, industrial, data centre and defence applications.

Talk to our engineering team

Discuss your power conversion requirements with Premium PSU.

Contact us

protosApplication Notes

The First Prototype Admits No Second Chance

Engineering Insight · Simulation & Test

The First Prototype Admits No Second Chance

Why we simulate before soldering: the integrated simulation, quantitative reliability and accelerated test chain behind custom power electronics delivered on time.

Premium SA
Born in Barcelona, Powering the World

In power electronics for critical sectors — railway, defence, energy, digital infrastructure — the custom project does not allow the iterative cycle of the catalogue product. Production runs are short. Deadlines are contractually binding. The first prototype delivered to the customer is validated against the entire contract, not against a simplified preliminary version.

The consequence is structural: success at the first prototype is not a desirable goal, it is the condition that makes the business model work. When a railway customer needs a converter operating at −40 °C, or an AI application requires dissipating more than 700 W/cm² locally, the design cannot be closed “by eye”. Whether a project delivers at the first prototype or iterates three times is decided before the first PCB is manufactured.

The constraint that defines the sector

Typical production runs in custom railway or defence projects sit between 50 and 500 units. The product learning curve runs out before the cost of a partial redesign is covered, so the catalogue business model — amortising iteration over thousands of subsequent units — does not apply. Every deviation detected when validating the first prototype has three possible outcomes, all costly: redesign with replanning, acceptance of contractual penalties, or batch rejection with replacement.

The metric Premium SA uses internally to evaluate design-flow maturity is the first-pass approval rate of DVT (Design Verification Testing). When stages are shortened due to schedule pressure, the indicator drops rapidly and drags timelines and margins with it.

Four fault vectors, four models

Pre-hardware verification is structured in four layers covering the dominant fault vectors in power electronics.

Electrical modelling: SPICE, PLECS and Simulink

SPICE for detailed switching analysis, and PLECS or Simulink for full-system simulation, allow verification — before manufacturing the first PCB — that the closed loop is stable, that waveforms do not exceed semiconductor margins, and that losses are consistent with the available dissipation. For SiC or GaN topologies, accurately capturing packaging parasitic inductance is decisive: a use case of the European ALL2GaN research project has documented how TOLL versus PG-DSO-20 packaging conditions snubber sizing in 3 kW/48 V totem-pole rectifiers achieving 97.5% efficiency and 40 W/in³ density.

Thermal CFD simulation

From chip to system. Real density does not come from table coefficients. For local heat fluxes above 700 W/cm² in AI applications and current-generation OCP servers, integrated liquid microchannels with dielectric coolant, in fraction-of-a-millimetre channel geometries, are already production-ready technology.

EMC pre-compliance simulation

The chamber costs thousands of euros and calendar days; discovering a radiated-emissions failure with the product already wired is one of the most damaging scenarios in a custom project. Modern solvers allow prediction of compliance with CISPR 32, EN 50121-3-2 (railway) or MIL-STD-461 (defence) before manufacturing. The Premium SA criterion: a design margin of at least 6 dB with respect to the applicable limits.

MATLAB / Simulink for digital control

Verification of control-loop stability and automatic C code generation. For SIL2/SIL3 under EN 50128 or ASIL-B/C under ISO 26262, traceability between safety requirement, model and generated code is a normative requirement.

Beyond simulation: reliability by design and physical testing

Reliability is not only verified on the bench; it is designed. Quantitative reliability analysis — the D4R block in Premium SA’s D2x methodology — makes it possible to commit to contractual MTBF figures backed by model and evidence. Tools: Telcordia SR-332, MIL-HDBK-217F, Siemens SN29500 and FIDES, depending on the customer and sector; FMEA with quantitative RPN thresholds; FTA with PFH calculation for SIL3. We explain this approach in more depth in our white paper on reliability in critical power electronics.

Physical testing closes the loop: DVT on three units (all must pass without rework), HALT to identify operational and destructive margins, and HASS as a production filter for each manufactured unit. The HALT+HASS combination does not replace DVT; it complements it — DVT verifies the design, HALT identifies the margins and HASS ensures that each unit stays within them.

The first prototype as the threshold of the business model

The chain described — four layers of simulation, three quantitative reliability tools, two levels of accelerated testing, all structured by the D2x methodology with V-Model and quantitative stage-gates (D2W, D2F, D2C, D4R, D4Co) — is Premium SA’s operational answer to a structural constraint: the first prototype must work. The investment has a measurable return: every hour of correct modelling is an hour not spent — multiplied — on physical bring-up, production redesign or warranty.

For the customer engineer evaluating suppliers, the relevant question is not whether the supplier knows how to design; it is whether the supplier knows how to verify. In custom power electronics for critical sectors, this is not an efficiency improvement: it is the condition that sustains the project.

Normative references

Standard Scope
Telcordia SR-332 · MIL-HDBK-217F · SN29500 · FIDES 2022 MTBF / reliability prediction methods
IEC 61508 · EN 50128 Functional safety; PFH <10⁻⁷/h for SIL3
ISO 26262 Automotive functional safety (ASIL)
CISPR 32 · EN 50121-3-2 · MIL-STD-461 EMC: multimedia equipment, railway rolling stock, defence

Talk to our engineering team

Planning a custom converter for railway, defence, energy or digital infrastructure? Talk to our engineers about how the simulation and test chain is applied to your project from the first design gate.

Contact Premium SA

About Premium SA

Premium SA (Premium PSU) is a Barcelona-based designer and manufacturer of custom power supplies and power converters for railway, energy, industrial, data centre and defence applications, with over 40 years of experience since its foundation in 1981. Born in Barcelona, Powering the World.

© 2026 Premium SA · Barcelona, Spain · Born in Barcelona, Powering the World