The Long Journey
When we began our thermoelectric research 18 years ago, the field was dominated by a particular orthodoxy: bismuth telluride (BiTe) was the gold standard for mid-range applications. Lead telluride (PbTe) was known to have superior theoretical performance, but its brittleness and difficulty in manufacturing made it seem impractical for commercial deployment. Everyone in the industry "knew" that BiTe was the right material for real-world applications, and that significantly better performance was probably impossible.
We didn't accept that consensus. That decision – to question the orthodoxy and pursue higher-performance materials despite the technical barriers – defined the next 18 years of our work.
The Early Years: Understanding the Fundamentals
The first years were foundational work. We focused on understanding why thermoelectric materials behaved as they did. We needed to understand the relationship between crystal structure, dopant levels, phonon scattering, and electrical properties. We ran thousands of experimental iterations, measured zT values (the figure-of-merit for thermoelectric efficiency) across different temperatures, and documented what actually moved the needle.
This phase taught us an important lesson: most improvements in thermoelectric materials are incremental and interdependent. You can't optimize for high electrical conductivity alone; you must simultaneously manage thermal conductivity and Seebeck coefficient. They're coupled through the underlying physics. Small improvements in each parameter compound multiplicatively into larger system improvements.
By year 3, we had achieved modest improvements over the BiTe baseline. Nothing dramatic, but enough to convince us that the path was real, not theoretical.
The Energy-Sorting Barrier: A Different Kind of Breakthrough
The architectural breakthrough that defines MicroPower's IP came not from a chemical trick at the interface but from the physics of how electrons move inside the chip itself. In a normal thermoelectric, hot electrons flow from the hot side to the cold side – but slow, low-energy electrons flow back the other way and largely cancel them out. The useful current you get out is a fraction of what the temperature gradient could in principle drive.
The team that came to MicroPower from ENECO had spent years working on this problem with Peter Hagelstein at MIT. The eventual answer was an MBE-grown energy-sorting barrier layer inside the chip – a thin structure that sorts electrons by energy, letting the hot ones through to the cold side and blocking the ohmic backflow. The same temperature gradient drives a much larger open-circuit voltage and short-circuit current. The physics was published in Applied Physics Letters in July 2002 and developed in detail in Journal of Applied Physics in May 2005, with foundational US Patent 6,396,191 covering the device structure. The enhanced-voltage signature was independently replicated by NIST, Bechtel-Bettis, and Lockheed Martin KAPL in 2001–2002 on the original HgCdTe and InSb material systems.
Bringing that physics from HgCdTe and InSb (which work at modest temperatures and degrade above ~240°C) onto PbTe-group materials better matched to industrial waste-heat temperatures (300–550°C) was the multi-year engineering pursuit. Sputtering could not hold the composition or thickness control required, which is what made Molecular Beam Epitaxy the production-relevant deposition method.
On modern PbTe chips, the demonstrated effect is 1.5–1.8× chip-level power-density enhancement per barrier, with multiple barriers stackable on a single chip. Importantly, the barrier layer is part of MicroPower's post-funding production roadmap – it is not in current shipped modules. The 14% module efficiency at 550°C demonstrated to date was achieved on the strength of the underlying PbTe/TAGS material system and the high-temperature contact and thermal-interface structures, informed by MicroPower's early ARL collaboration and substantially evolved internally since. Reintroducing the barrier layer multiplies power density on top of that 14% baseline.
Why Material Choice Matters
The journey through different material systems – from BiTe to PbTe to TAGS alloys – taught us that material selection is foundational. Some materials have inherently better thermoelectric properties. Some are more chemically stable at high temperature. Some are easier to manufacture. Some are more cost-effective.
You can't engineer your way around fundamental material limitations. If a material has poor thermal properties, no amount of structural optimization will fix it. You have to choose the right material first. That seems obvious in retrospect, but in an industry wedded to BiTe, it was a contrarian position.
Over the years, we explored multiple material families. We found that each had strengths and limitations. PbTe had superior high-temperature performance but brittleness challenges. TAGS alloys showed excellent stability and good performance across a wide temperature range. Skutterudites had interesting promise but manufacturability challenges. Understanding these trade-offs allowed us to make deliberate choices rather than defaulting to "what everyone else uses."
We did not abandon BiTe – we returned to it for the role it does well. Today MPG ships BiTe formulations for the low-temperature tail (cooling-mode applications, including cascade cooling to below −150°C) and PbTe/TAGS for the 300–1000°C high-temperature power band.
The Compounding Effect of Incremental Improvement
What we learned is that thermoelectric device efficiency improvement is rarely about a single breakthrough. It's about hundreds of small improvements, each of which moves the needle by 1-3%, that compound over time. Better material properties, optimized doping, refined manufacturing processes, improved device architecture, better interface design – each contributes modestly, but together they create dramatic change.
When we started, we were achieving 5-6% conversion efficiency at 400°C. Today, we're reliably hitting 14% at the same temperature range. That's more than a 2x improvement. It didn't come from a single genius insight. It came from 18 years of disciplined, patient optimization across every component of the system.
This teaches an important lesson for anyone pursuing deep-tech innovation: patience is essential. The field rewards those willing to commit to long development cycles and incremental improvement. Quick wins are great, but lasting competitive advantage comes from sustained optimization.
The Organizational Discipline Required
Perhaps the most important lesson is organizational. An 18-year research program requires maintaining focus, securing sustained funding, and retaining technical talent across multiple product cycles. It requires saying "no" to distracting opportunities and staying committed to a long-term vision.
We've been fortunate to work with partners and investors who understood that deep-tech doesn't have quick payoffs. We've been fortunate to retain a core team committed to the mission. And we've been disciplined about not abandoning the path when progress felt slow.
For entrepreneurs and researchers pursuing next-generation energy technologies, that's perhaps the most important message: expect the journey to be long. Thermoelectric efficiency improvement has been a 18-year marathon, not a sprint. But the destination – 14% conversion efficiency, demonstrated reliability, validated applications – is worth it.
We're not done optimizing. The work continues. But we're confident that the patient, disciplined approach to materials science and device engineering will keep pushing the boundaries of what's possible.