Electronics, industrial equipment, and automobiles release vast amounts of unused heat into the environment. If this waste heat could be converted directly into electricity, it could power IoT sensors that run without an external power supply, and lead to new power-generation technologies that use energy more effectively. One approach under active study is thermoelectric conversion, which exploits the Seebeck effect: a voltage appears when a temperature difference is applied across the two ends of a material. Practical use, however, still faces challenges in both the physical properties of the materials and the manufacturing process.
One is a trade-off specific to thermoelectric materials. The Seebeck coefficient, which determines the voltage generated by a temperature difference, and the electrical conductivity, which indicates how easily current flows, generally move in opposite directions as carrier density changes. Raising carrier density to increase conductivity tends to reduce the Seebeck coefficient. The other challenge is manufacturing temperature. Materials that need high-temperature sintering or crystallization to achieve good thermoelectric performance are difficult to integrate directly onto heat-sensitive flexible substrates or semiconductor devices with fine electronic circuits already formed on them.
A research team at Tohoku University, consisting of Junya Kurahashi (a graduate student at the time of the research), Assistant Professor Yat Sou, and Professor Yuji Sutou, has addressed these challenges by adding titanium (Ti) to chromium nitride (CrN) thin films. According to a research paper published on October 12, 2026 in the Journal of Materials Chemistry A, adding a small amount of Ti to CrN films made without heating the substrate raised the room-temperature power factor (PF) about 250-fold compared with undoped films. Improving the electrical properties of a thermoelectric material without relying on high-temperature processing is an important step toward thin-film power generation that can be built into heat-sensitive electronic devices.
What Does a 250-Fold Improvement Mean? The Physical Definition of Power Factor
It is worth being careful about what the "250-fold" figure represents. What improved is not the efficiency of converting heat into electricity itself, but the "power factor (PF)," a performance metric for thermoelectric materials. The two are closely related, but they are not the same physical quantity.
The power factor PF is defined as the product of the Seebeck coefficient \(S\) and the electrical conductivity \(\sigma\). Because conductivity is the reciprocal of electrical resistivity \(\rho\), it can be written as follows:
\[ PF = S^2 \sigma = \frac{S^2}{\rho} \]
The Seebeck coefficient \(S\) is expressed in units such as V K⁻¹, and the electrical conductivity \(\sigma\) in units such as S cm⁻¹. The Seebeck coefficient is squared because the open-circuit voltage produced by a temperature difference is proportional to it, and the maximum power that can be drawn from the circuit is proportional to the square of that voltage. PF therefore serves as an index of a thermoelectric material's electrical performance in delivering power.
The metric that evaluates how well a material converts heat into electricity, including how easily it conducts heat, is the dimensionless figure of merit \(zT\). With absolute temperature \(T\) and total thermal conductivity \(\kappa\), the following relationship holds:
\[ zT = \frac{S^2 \sigma T}{\kappa} = \frac{PF \cdot T}{\kappa} \]
The thermal conductivity \(\kappa\) here includes heat transport by electrons and by lattice vibrations. Even if PF is high, \(zT\) will not necessarily improve by the same factor if a large amount of heat escapes through the material. Evaluating the conversion efficiency of an actual generator also requires considering the temperature difference between the hot and cold sides, contact resistance with the electrodes, and heat loss to the surroundings. In other words, the 250-fold figure reported here is the improvement in power factor, an electrical measure of a material's power-generating performance; it does not mean that generation efficiency rose 250 times.
The team focused on chromium nitride, a nitride with excellent wear and corrosion resistance that is also expected to serve as a thermoelectric material. However, achieving high thermoelectric performance can require high-temperature deposition or epitaxial growth, and films made at room temperature tend to have high electrical resistivity. One cause is a defect known as a "nitrogen vacancy," where a nitrogen atom is missing from its normal position in the crystal. Nitrogen vacancies act as donors that supply electrons, but as charged defects they also act as scattering sources that impede electron movement. In this study, suppressing these nitrogen vacancies without high-temperature processing was the key to improving performance.
Room-Temperature Sputtering and Control of Titanium Composition
The team used reactive sputtering, which can produce CrN thin films without heating the substrate. They used thermally oxidized silicon substrates (SiO₂/Si) and alkali-free glass substrates, forming the films at room temperature with no substrate heating during deposition. The experimental design aimed to improve performance through composition control during deposition rather than through property improvement by high-temperature processing.
The films were made by reactive co-sputtering in a radio-frequency magnetron sputtering system. High-purity metallic chromium (Cr) and titanium (Ti) were placed as targets, and plasma was generated in a mixed gas of argon (Ar) and nitrogen (N₂). The deposition chamber was evacuated to a base pressure of \(5.0 \times 10^{-5}\) Pa or lower, and during deposition the gas flow rates were set to 5 sccm for argon and 2 sccm for nitrogen, with an operating pressure of 0.14 Pa.
In the experiments, the power applied to the chromium target was fixed at 50 W, while the power applied to the titanium target was varied in four steps: 0 W, 40 W, 50 W, and 60 W. All the films were about 100 nm thick. Compositional analysis by energy-dispersive X-ray spectroscopy (EDX) showed titanium atomic concentrations, as a share of all metal elements, of 0 at.%, 4.6 at.%, 6.9 at.%, and 9.3 at.%, respectively. Using the sample without titanium as a reference, the team examined how properties changed with the amount added.
The compositional analysis also confirmed that adding titanium suppressed nitrogen deficiency. Expressing the film composition as \(\mathrm{Cr_{1-x}Ti_xN_y}\), the undoped CrN film had a nitrogen ratio \(y\) of 0.89, below the stoichiometric value of 1.00. In the Ti-doped films, \(y\) was 0.92 to 0.96, indicating a higher nitrogen content. Titanium has a higher affinity for nitrogen than chromium does and holds nitrogen more stably in the crystal, which is thought to have suppressed nitrogen vacancy formation.
The nitrogen ratio did not, however, increase perfectly monotonically with titanium concentration. Although nitrogen deficiency decreased overall, changes in crystal orientation and lattice constant are also thought to affect the material's properties. The team examined in detail how these changes in composition and crystal structure relate to electrical transport properties.
Mobility Rises About 47-Fold as Suppressing Nitrogen Vacancies Improves Electron Transport
The changes in electrical properties from titanium doping were confirmed through Hall effect and electrical resistivity measurements at room temperature (300 K). The undoped CrN film had an electrical resistivity of 7.44 mΩ cm, equivalent to a conductivity of about 134 S cm⁻¹. Resistivity fell as the amount of titanium increased, reaching 1.21 mΩ cm, or a conductivity of about 826 S cm⁻¹, in the 9.3 at.% film. Compared with the undoped film, the resistivity was reduced to about one-sixth.
The main reason for this improved conductivity was not an increase in the number of electrons but the fact that electrons could move through the material more easily. The team measured carrier density and Hall mobility through Hall effect measurements and quantitatively evaluated the changes caused by titanium doping. Comparing the undoped film with the 9.3 at.% Ti-doped film, both Hall mobility and power factor improved substantially.
{
"type": "bar",
"title": "室温(300K)における無添加と9.3 at% Ti添加薄膜の物性比較",
"unit": "各指標の実測値",
"categories": ["無添加 (0 at% Ti)", "9.3 at% Ti添加"],
"series": [
{ "name": "ホール移動度 (cm² V⁻¹ s⁻¹)", "values": [0.0123, 0.578] },
{ "name": "出力因子 PF (µW cm⁻¹ K⁻²)", "values": [0.021, 5.26] }
],
"caption": "無添加対照群と最大PF達成膜の実測値比較。移動度は約47倍、PFは約250倍へ向上",
"source": "Journal of Materials Chemistry A (2026)"
}The Hall mobility of the undoped CrN film was 0.0123 cm² V⁻¹ s⁻¹, rising about 47-fold to 0.578 cm² V⁻¹ s⁻¹ in the 9.3 at.% Ti-doped film. According to the paper, the maximum Hall mobility was actually obtained in the sample with 6.9 at.% titanium; it declined slightly at 9.3 at.% but remained high.
Meanwhile, the electron carrier density fell by about an order of magnitude, from roughly \(6.8 \times 10^{22}\) cm⁻³ in the undoped film to roughly \(8.9 \times 10^{21}\) cm⁻³ in the 9.3 at.% Ti-doped film. This is thought to be because fewer nitrogen vacancies meant fewer electrons supplied by defects. Conductivity improved despite this because mobility rose by more than enough to offset the drop in carrier density. Even with fewer electrons, each one was less likely to be scattered by defects and could move through the material more easily, which lowered the overall electrical resistance.
The Seebeck coefficient also changed significantly. It was negative in all samples, showing n-type behavior in which electrons are the main charge carriers. The undoped film had a Seebeck coefficient of −12.5 µV K⁻¹, while the 4.6 at.% Ti-doped film reached −106 µV K⁻¹, the largest absolute value. The decrease in carrier density is thought to have increased the contribution of entropy carried per carrier, raising the Seebeck coefficient. With higher titanium concentrations the absolute value decreased somewhat, reaching −79.8 µV K⁻¹ in the 9.3 at.% film.
As a result of these changes, the power factor PF, determined by the square of the Seebeck coefficient times the electrical conductivity, rose from 0.021 µW cm⁻¹ K⁻² in the undoped film to 5.26 µW cm⁻¹ K⁻² in the 9.3 at.% Ti-doped film, about 250 times higher. The composition with the largest absolute Seebeck coefficient did not coincide with the one with the highest PF, which shows that both the Seebeck coefficient and the electrical conductivity must be considered to improve thermoelectric performance.
The team cites the reduction of nitrogen vacancies acting as donors, through strong titanium–nitrogen bonding, and the resulting suppression of electron scattering by ionized impurities as the main reason for the mobility improvement. It also suggests that ordering of the local crystal structure may have improved electron transport. In its announcement, Tohoku University explains that, in addition to the compositional analysis, X-ray photoelectron spectroscopy confirmed a decrease in components attributable to nitrogen defects. However, the contribution of electron scattering by individual defects was not directly isolated and measured; the mechanism of mobility improvement through defect suppression is an interpretation drawn from the combined results of compositional analysis and electrical transport measurements.
Power Generation Performance and Power Density of the Prototype Devices
Beyond confirming improvements in the properties of the material itself, the team built prototype thermoelectric generators that draw power from an applied temperature difference. Three types were evaluated: an n-type single-leg device using titanium-doped CrN (Ti-CrN), a p-type single-leg device using oxygen-doped CrN (O-CrN) for comparison, and an in-plane composite device in which the two were connected on the same substrate. A single-leg device uses one type of thermoelectric material, while the in-plane composite device places p-type and n-type materials on a substrate and uses the electromotive force of both.
The main evaluation results for each material and device are as follows.
| Subject | Metric and result | Comparison | Measurement conditions | Notes on interpretation |
|---|---|---|---|---|
| Ti-CrN film (Ti 9.3 at.%) | Power factor PF: 5.26 µW cm⁻¹ K⁻² | Undoped CrN: 0.021 µW cm⁻¹ K⁻² | Room temperature (300 K), film thickness about 100 nm | A material property, not the thermoelectric conversion efficiency itself |
| Ti-CrN film (Ti 9.3 at.%) | Hall mobility: 0.578 cm² V⁻¹ s⁻¹ | Undoped CrN: 0.0123 cm² V⁻¹ s⁻¹ | Room temperature (300 K), Hall effect measurement | The maximum mobility was obtained in the 6.9 at.% Ti sample |
| Ti-CrN film (Ti 4.6 at.%) | Seebeck coefficient: −106 µV K⁻¹ | Undoped CrN: −12.5 µV K⁻¹ | Room temperature (300 K), temperature-gradient method | The composition with the largest absolute Seebeck coefficient differs from the one with the highest PF |
| n-type single-leg device (Ti-CrN) | Maximum output: 455.8 nW; power density: 38.0 mW cm⁻² | p-type single-leg device: 12.4 nW | Temperature difference 96 K | Power density is normalized by the film's cross-sectional area |
| In-plane composite device (n-type + p-type) | Open-circuit voltage: 180 mV; maximum output: 122.7 nW | n-type single-leg device: 455.8 nW | Temperature difference 97 K | High electromotive force was obtained, but internal resistance limited the output |
The single-leg device using the n-type Ti-CrN film with 9.3 at.% titanium produced a maximum of 455.8 nW (nanowatts) when a 96 K temperature difference was applied across its ends. The power density at that point is reported as 38.0 mW cm⁻². The team says this power density is comparable to that of existing high-performance thermoelectric thin films made with high-temperature processes.
The 38.0 mW cm⁻² figure needs careful interpretation, however. The power density is normalized by the cross-sectional area through which current flows in the film, that is, the product of the roughly 100 nm thickness and the channel width, and not by the planar projected area of the substrate that the device occupies. In thermoelectric research, performance is sometimes compared on the basis of effective cross-sectional area, but this differs from the power per installed area commonly used for solar cells and similar devices. Because the film is extremely thin, even a large cross-section-normalized power density yields only 455.8 nW from the whole device. The team also reports a power density normalized by the square of the temperature difference: 4.12 µW cm⁻² K⁻².
By contrast, the comparison p-type O-CrN single-leg device produced a maximum of only 12.4 nW, even though a larger temperature difference of 103 K was applied. Oxygen-doped p-type CrN has low electrical conductivity and high internal resistance, which is thought to have prevented sufficient current from being drawn. The result shows that even if the n-type material improves substantially, a poorly performing p-type partner limits the output of the entire generator.
The same problem appeared in the in-plane composite device, in which n-type and p-type films were connected in series. With a 97 K temperature difference, the composite device generated a high open-circuit voltage of 180 mV, because the electromotive forces from the Seebeck effect in the n-type and p-type materials add together. The maximum power actually obtained was 122.7 nW, however, well below the 455.8 nW of the n-type single-leg device. Raising the voltage does not increase the available power if current is limited by the large internal resistance of the p-type region or the contact resistance at the junctions.
From these results, the team points out that practical generator modules will require improved electrical transport in the p-type material and lower resistance at electrodes and junction interfaces. Achieving a high power factor in a single material and drawing sufficient power from a combination of several materials are different technical challenges.
Room-Temperature Deposition Widens Possible Applications, but Hurdles Remain Before Practical Use
The key achievement of this study is that room-temperature deposition without substrate heating reached a power factor comparable to that of thermoelectric thin films made with conventional high-temperature processes. Manufacturing methods that require high-temperature treatment or special single-crystal substrates for good thermoelectric performance are heavily constrained and hard to integrate with existing electronic devices. Ti-CrN films that can be made by room-temperature sputtering, by contrast, are materials that can more readily be considered for use on heat-sensitive substrates or devices with complex fine structures.
For example, some thermoelectric materials, such as bismuth–tellurium and silicon–germanium systems, are made by processes that need high-temperature crystallization or sintering to bring out high performance. Such steps make direct integration difficult on low-heat-resistance resin substrates, organic electronics, and semiconductor chips with fine wiring. Ti-CrN films could be expected to find use in back-end semiconductor processes and flexible circuits. However, being deposited at room temperature does not by itself guarantee full compatibility with existing production lines; adhesion to substrates and effects on the overall manufacturing process will need separate verification.
Nor can these results be taken as the immediate realization of practical waste-heat power generation. The paper demonstrates improved material properties in a laboratory setting and power generation in small prototype devices. It has not shown that the material provides sufficient output, conversion efficiency, and reliability for a commercial waste-heat recovery system.
The first issue is the amount of power that can actually be drawn. The prototype n-type single-leg device produced a maximum of 455.8 nW, short of the microwatt-to-milliwatt range generally needed to run an IoT microcontroller or wireless communication continuously. Since the power consumption of electronic circuits varies greatly with operating conditions and the frequency of intermittent operation, nanowatt-level generation could still have uses. For broader use, though, techniques will be needed to integrate many thin-film elements to raise output and to reduce losses in wiring and junctions.
The second issue is that thermoelectric conversion efficiency remains unknown. In this study the power factor was calculated from measured electrical conductivity and Seebeck coefficient, but the in-plane thermal conductivity \(\kappa\) of the films was not measured. As a result, neither the dimensionless figure of merit \(zT\) nor the net conversion efficiency, which shows what percentage of the heat input to the device is converted into electricity, has been determined. If titanium doping also increased thermal conductivity, \(zT\) would not necessarily have improved by the same factor as the 250-fold gain in power factor. Evaluating thermal transport properties is essential for judging the material's overall thermoelectric performance.
The third issue is long-term reliability in real operating environments. The paper's supplementary materials describe thermal cycling tests in which electrical resistance was measured from room temperature to 400 °C in an argon atmosphere, confirming reversible changes in electrical properties. These are results from an inert gas, however, and do not guarantee oxidation resistance under prolonged high-temperature exposure in air or durability in environments with repeated temperature differences. Element diffusion from long-term heating and mechanical fatigue from thermal expansion and contraction also need future verification.
Practical thermoelectric technology requires combining a range of techniques beyond improving the material itself, including bonding to electrodes, structural design to control heat flow, and integration of multiple elements. The significance of the Ti-CrN film is that it shows that controlling nitrogen vacancies can deliver a high power factor even when the film is deposited at room temperature. Developing practical generators, though, will require pairing it with a high-performance p-type material, lowering interface resistance, evaluating thermal conductivity, and testing durability in real environments. The focus going forward is how far material-level performance gains can be turned into more usable electric power.