BNSF Railway and Vehicle Projects Develop Experimental Hydrogen Fuel-Cell Switch Locomotive; Locomotive-to-Grid Application Explored
Honda to Launch Civic Hybrid in India in 2008

Two Research Groups Demonstrate High-Performance Thermoelectric Capability in Silicon Nanowires

False-color image of Caltech silicon nanowire thermoelectric device. The central green area is the Si nanowire array, which is not resolved at this larger magnification. The four-lead yellow electrodes are used for thermometry. The thermal gradient is established with either of the two Joule heaters (the right-hand heater is colored red). Click to enlarge.

Thermoelectric materials—materials that can convert heat into electricity—are theoretically promising for applications such as waste heat recovery from combustion engines. (Earlier post.)

Now, two research groups working independently at Caltech and UC Berkeley/Lawrence Berkeley National Laboratory (LBNL) have shown that the thermoelectric properties of silicon—a material that can be processed on a large scale but has poor thermoelectric properties—can be vastly improved by structuring it into arrays of nanowires and carefully controlling nanowire morphology and doping. Reports on both sets of research are in the 10 January issue of the journal Nature.

The efficiency of thermoelectric materials depends on the thermoelectric figure of merit ZT of their material components, which is a function of the Seebeck coefficient, electrical resistivity, thermal conductivity and absolute temperature.

Over the past five decades, note both teams in their papers, it has been challenging to increase ZT > 1, since the parameters of ZT are generally interdependent. Although nanostructured thermoelectric materials have delivered ZT of greater than 1, the materials (Bi, Te, Pb, Sb, and Ag) and processes used are not often easy to scale to practically useful dimensions. The developments with silicon could change that, however.

The two groups measured ZT (thermoelectric figure of merit) values near 1 for silicon nanowires at and below room temperature. Both groups found dramatically lower thermal conductivity values. Intriguingly, the Caltech group found a greatly enhanced Seebeck coefficient due to a 1-D phonon drag effect.

UC Berkeley/LBNL. The Berkeley/LBNL team, led by Arun Majumdar and Peidong Yang, used a unique “electroless etching” method by which arrays of silicon nanowires are synthesized in an aqueous solution on the surfaces of wafers that can measure dozens of square inches in area. The technique involves the galvanic displacement of silicon through the reduction of silver ions on a wafer’s surface. Unlike other synthesis techniques, which yield smooth-surfaced nanowires, this electroless etching method produces arrays of vertically aligned silicon nanowires that feature exceptionally rough surfaces. The roughness is believed to be critical to the surprisingly high thermoelectric efficiency of the silicon nanowires.

The rough surfaces are definitely playing a role in reducing the thermal conductivity of the silicon nanowires by a hundredfold, but at this time we don’t fully understand the physics. While we cannot say exactly why it works, we can say that the technique does work.

—Arun Mahumdar

Bulk silicon is a poor thermoelectric material at room temperature, but by substantially reducing the thermal conductivity of our silicon nanowires without significantly reducing electrical conductivity, we have obtained ZT values of 0.60 at room temperatures in wires that were approximately 50 nanometers in diameter. By reducing the diameter of the wires in combination with optimized doping and roughness control, we should be able to obtain ZT values of 1.0 or higher at room temperature.

—Peidong Yang

The Berkeley Lab researchers will be studying the physics behind this phenomenon to better understand and possibly manipulate it for even further improvements. They will also concentrate on the design and fabrication of thermoelectric modules based on silicon nanowire arrays. Berkeley Lab’s Technology Transfer Department is now seeking industrial partners to further develop and commercialize this technology.

Caltech. The Caltech researchers, led by James Heath, used a method developed in Heath’s labs to construct nanowires with cross-sectional areas of 10 nm x 20 nm and 20 nm x 20 nm. By varying the nanowire size and impurity doping levels, they achieved ZT values representing an approximately 100-fold improvement over bulk Si over a broad temperature range, including ZT of approximately 1 at 200 K (-73° C).

Optimizing materials for cooling or heat recovery applications involves a tricky trade-off of several different parameters, including the electrical conductivity and the thermal conductivity. We find that we can greatly drop the thermal conductivity in these nanowires without affecting the other parameters, and this leads to dramatic improvements in the thermoelectric efficiency.

—James Heath

An additional parameter that the researchers were surprised to see improved in the nanowires is the thermopower, which is the amount of voltage generated in a material for a given thermal gradient. The improvement likely arises from a phenomenon known as phonon drag, which comes when the sound-carrying vibrations in the atomic lattice of the nanowires are not in thermal equilibrium with the current carrying electrons.

We find that for ultrathin nanowires the electrons drag certain sound waves along with them as they move down the nanowire. This extra heat from the sound is enhancing the thermoelectric efficiency.

—Jamil Taher-Kheli, contributing author

Although silicon nanowires are still about a factor of two less efficient than the most efficient known thermoelectric materials, the researchers are optimistic that further improvements in the materials will soon be made.

Our theoretical models indicate that a number of exciting avenues are available to significantly improve the efficiency. However, even at their current efficiencies, these nanowires already outperform many commercially available systems, and so could potentially find near-term applications.

—William A Goddard, director of the Materials and Process Simulation Center, and a contributing author




Interesting news, "super" ideas turning up everywhere...

"“It’s like a conventional heat engine,” explains Paul Werbos, program director at the National Science Foundation, which has provided funding for JTEC. “It still uses temperature differences to create pressure gradients. Only instead of using those pressure gradients to move an axle or wheel, he’s using them to force ions through a membrane. It’s a totally new way of generating electricity from heat.”"


Reclaiming heat from a hybrid engine would be very nice. An FFV hybrid with heat recovery could go a long way towards using less energy.

Jim G.

I'd personally go so far as to characterize thermoelectrics as the holy grail of energy, as heat is so easy to come by and yet is often considered waste. Of course the efficiency has been the critical problem; this specific research is quite incredible for the first generation. Thermionic cooling systems for microprocessors are another somewhat parallel technology that is already being investigated in the computer industry.

Jaroslaw Berezowski

Another interesting thing that has possibility not to being a scam:

Harvey D

Let's not jump to conclusion too fast. This process is stil a factor of two below existing thermoelectric efficiency. Its got a long to go before it can be useful.

Efficiently transforming waste heat, without external power, into useful electrical energy is an old dream. I wish them the best.


"phonon drag, which comes when the sound-carrying vibrations in the atomic lattice of the nanowires are not in thermal equilibrium with the current carrying electrons."

So, the extra heat from out-of-phase resonance increases thermoelectric efficiency. Reminding of the very interesting area of plasma physics called thermoluminescence.


Er, sorry, "sonoluminescence," is what I meant. And the Powerchips people are doing fantastic thermoelectric work that is off the shelf now.


JX Crystals has had thermo PV chips for a while. I do not think that their efficiency versus cost is very good however.


I'd like to add these guys to the fray:

They claim to have used a semiconductor in place of a vacuum in a traditional thermionic energy conversion device. Some of the material combinations they use include HgCdTe, PbSnTe, InSb. They are currently achieving ~20% conversion efficiency. They claim, however, that 50% ideal Carnot efficiency is theoretically achievable.

I have often mused that we could use Big Coal to get us off oil using such technology. Here's how:

-A future car is proposed that uses a solid state energy conversion device, an onboard zeolite oxygen separation unit, and a micro furnace. In wheel electric motors are used.
-Coal rods are used (coal has ~2/3 the gravimetric and 3/4 the volumetric energy density of gasoline) in place of a liquid fuel. CO2 emissions might be higher, but NOX and CO emissions would be eliminated (more important than CO2, IMHO).
-A small battery would be onboard to provide regenerative braking capabilities
-Big Coal would use revenues from the sale of (more) coal to fund building long life li-ion batteries (such as those from Altair Nano). These batteries could be swapped (easily) for the coal conversion hardware and could be leased to end users.
-Big Coal could use revenue from leasing said batteries to replace/supplement declining coal fired electricity revenues.



I think the JX wafers use IR from high grade heat sources like stoves that glow red. I remember seeing an IR picture of a BMW Turbosteamer demo and the engine and exhaust looked like good candidates for this.

What might be more useful would be a lower grade heat recovery in large quantity. The hot coolant that comes out of an engine block and the heat around the exhaust manifold might be good.



Your proposal suggests that Big Coal view itself as an energy company instead of a coal resource. This does makes sense especially assuming that during the electrification of transport the cost of LiIo batteries is the highest hurdle.

Without the zeolite separation unit it would be smart for coal to invest in the E-storage units that will demand more of their product.



Your proposal suggests that Big Coal view itself as an energy company instead of a coal resource. This does makes sense especially assuming that during the electrification of transport the cost of LiIo batteries is the highest hurdle.

Without the zeolite separation unit it would be smart for coal to invest in the E-storage units that will demand more of their product. The conversion of coal fired plants to NG, H2 or some other fuel is still a challenge.

The comments to this entry are closed.