Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2026 Mar 26;16:15154. doi: 10.1038/s41598-026-44966-8

Novel strategy for boosting thermoelectric performance of organic materials with low electrical conductivity

Masato Nakaya 1,✉, Shun Yamamoto 1, Satoshi Ogawa 2, Toshiaki Nishii 1,3, Jun Onoe 1,✉
PMCID: PMC13171998  PMID: 41888372

Abstract

To realize a wearable power generator, organic thermoelectric materials are desired to exhibit a dimensionless figure-of-merit [zT = α2σT/(κe + κl)] of more than unity, where T, α, and σ denote the temperature, thermopower, and electrical conductivity, whereas κe and κl are the thermal conductivities mediated by carriers and phonons, respectively. Although conventional materials exhibit a maximum power factor (PFmax = α2σ) in the high-σ region of more than 10 Scm− 1, zT is not necessarily maximized at the σ giving PFmax because of a marked increase in κe in the high-σ region, which has been a barrier for boosting the zT of organic materials up to a practical level. Here, we report that an exceedingly large PFmax of 1.1 × 10− 3 Wm−1K− 2 was realized for the composite film consisting of fullerene molecules and molybdenum trioxide nanoclusters by keeping the giant Seebeck effect at a low σ of less than 10− 2 Scm− 1. Since κe is negligibly small in the low-σ region, zT of 0.81 is expected to be achieved at room temperature, which is the highest value among organic materials to the best of our knowledge.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-44966-8.

Keywords: Energy harvesting, Thermoelectric conversion, Seebeck effect, Nanocomposites, Fullerene, Metal oxides

Subject terms: Energy science and technology, Materials science, Physics

Introduction

The thermoelectric (TE) conversion from body temperature via the Seebeck effect is attractive for a power generator built in various wearable devices such as healthcare sensors and smart glasses used in the advanced information-oriented society. To harvest widespread waste heat from the human body, flexible, large-area, and high-performance TE materials have been required. The energy conversion efficiency of a TE device is determined by the dimensionless figure-of-merit [zT = α2σT/(κe + κl)] of TE materials, where T is the absolute temperature, α is the thermopower, σ is the electrical conductivity, and κe and κl are the thermal conductivities (κ = κl + κe) mediated by carriers and phonons of TE materials, respectively. Here, zT > 1 has been preferred for practical use1. Organic materials such as conductive polymers and polymer/nanocarbon composites have been widely developed as novel TE materials owing to not only their ease for constructing wide-area and flexible devices by printing technology but also their low κl on the order of 10− 1 Wm−1K− 12−4. In conventional organic materials, a numerator of zT, the so-called power factor (PF = α2σ), must be maximized at a high σ of more than ca. 10 Scm− 1 by controlling the concentration of carriers via chemical doping5, because there is a trade-off relationship between σ and |α| as functions of carrier density (expressed as ne for electrons and nh for holes)5–10. However, the maximum zT is not necessarily achieved at the σ giving PFmax because of a marked increase in κe in the high-σ region with increasing ne and/or nh (Wiedemann−Franz law),1 which has been a large barrier for boosting the zT of organic materials up to more than 0.5 so far6–10. However, this issue can be resolved by realizing PFmax at a low σ of less than ca. 1 Scm− 1, because the effect of κe on the thermal conductivity of organic materials is negligibly small (κ ≃ κl)11.

The giant Seebeck effect (GSE), which is the emergence of large |α| values within the 10 − 250 mVK− 1 range for organic semiconductor films consisting of small molecules such as pentacene, fullerene (C60), and thiophene derivatives12–15, has been expected to provide novel strategies toward the achievement of a high zT in the low-σ region for nondoped and/or low-density-carrier-doped organic materials, because zT and PF are proportional to the square of α. However, despite the giant |α| and low κ, the zT values of the nondoped organic semiconductor films were estimated to be less than 10− 3 owing to their extremely small σ of less than 10− 6 Scm− 1 at room temperature (RT)12–15. Although the σ of organic semiconductor films can be markedly improved to be more than 10− 1 Scm− 1 by the chemical doping of carriers, |α| values of less than 1 mVK− 1 were reported for the organic semiconductor films doped with impurity dopants such as metal atoms and metal complexes12,14,16–20. Thus, it has been difficult so far, to control the σ of organic semiconductor films at a high-σ level while keeping the GSE.

Recently, we have succeeded in improving the σ of organic semiconductor films while maintaining the GSE by forming nanocomposite films consisting of C60 molecules and molybdenum trioxide (MoO3) nanoclusters [(MoO3)n]20. By investigating the TE properties of a C60(MoO3)x film with a wide range of composition ratios (x) of 0.00 − 11.8, we found that the C60(MoO3)0.034 film exhibits PFmax of 1.8 × 10− 4 Wm−1K− 2, which is two orders of magnitude larger than that of pristine C60 films20. This large PF was realized by increasing σ to be three orders of magnitude larger than that of pristine C60 films with a giant |α| value of 36 mVK–1maintained. The scanning tunneling microscopy/spectroscopy (STM/STS) and X-ray absorption fine structure (XAFS) analyses of the C60(MoO3)x film indicate that the (MoO3)n nanoclusters behaved as hole dopants for C6020,21. Thus, C60(MoO3)x films exhibited P-type (α > 0) TE properties, although pristine C60 films (x = 0) exhibit N-type (α < 0) ones12,20.

To develop practically useful C60-based TE materials exhibiting zT ≃ 1 at RT, PFmax is preferred to be more than ca. 10− 3 Wm−1K− 2 at a low σ of less than 1 Scm− 1, by assuming the thermal conductivity of κl ≃ 0.4 Wm1K–1for a pristine solid C60 at RT22. Although PFmax of 1.8 × 10− 4 Wm−1K− 2 has been achieved in C60(MoO3)x composite films, it is still one order of magnitude smaller than the practical value of 10− 3 Wm− 1 K− 2. C60(MoO3)x films are categorized into binary nanocluster solids whose properties and functionalities can be tuned via the coupling of physical properties among nanoclusters23–26. This implies that a wide controllability of the TE properties of C60(MoO3)x films is expected to be realized by tuning the size (the number of MoO3 units per cluster) and valence state of MoO3 nanoclusters.

Here, we demonstrate that the TE performance of C60(MoO3)x films can be boosted up to a practical level by controlling the giant α value at a low σ of less than 10− 2 Scm− 1 by the thermally-induced chemical reduction of (MoO3)n nanoclusters. C60(MoO3)x films showed an excellent stability in thermal annealing temperature (Ta) ranges of 22–100 °C for x of 0.03 − 0.09 and 22–250 °C for x of 11.8, which provide a sufficient thermal stability for practical use at RT as a component of wearable devices. The PF values of C60(MoO3)0.05 and C60(MoO3)0.09 films were markedly increased by thermal annealing at slightly higher Ta values within the 100–150 °C range. Note that an exceedingly high PFmax of 1.1 × 10− 3 Wm−1K− 2 was achieved at 22℃ for the C60(MoO3)0.05 film by realizing a giant α of 110 mVK− 1 at σ of 9 × 10− 4 Scm− 1. In this case, since κe is negligibly small in the low-σ region, zT is estimated to be 0.81 at RT.

Results and Discussion

Figure 1 shows changes in the α (blue circles) and σ (red squares) of C60(MoO3)x films with (a) x = 0.00, (b) x = 0.05, (c) x = 0.09, and (d) x = 11.8 as functions of Ta. The α and σ of as-grown C60(MoO3)x films were measured at Ta = 22 ℃. The nondoped (pristine) C60 (x = 0.00) and C60(MoO3)x (x = 0.05, 0.09, 11.8) films respectively exhibited negative and positive α values before thermal annealing. This is consistent with the previous reports that the carriers of pristine C60 and C60(MoO3)x films are electrons (N-type) and holes (P-type), respectively12,20. When the pristine C60 film was annealed at Ta within the 22 − 150℃ range, σ gradually increased from 1.6 × 10− 6 to 2.6 × 10− 5 Scm− 1 with Ta. This is because the small σ of the C60 film originates not only from a small ne but also from a low mobility of electrons (µe = 1 cm2 V–1 s–1) owing to the scattering of carriers at high-density geometrical defects such as domain boundaries and stacking faults in the C60 film formed at RT27. The increase in σ (Fig. 1a) was obviously caused by an increase in µe, because the density of domain boundaries remarkably decreases via the thermal rearrangement of C60 domains at Ta of 20 − 200℃.28 Correspondingly, the giant |α| value of the pristine C60 film monotonically decreased from 121 mV K− 1 to 43 mV K–1 with increasing Ta. This strongly suggests that the GSE of the C60 film is affected by the scattering of carriers at geometrical defects, which is consistent with the previous results that the multiple-trapping-and-release mechanism of carriers at the traps due to the defects and/or fluctuations at the band edge of the molecular film induces the emergence of the GSE of nondoped and low-density-carrier-doped organic semiconductors14,20. Thus, a pristine C60 film shows a trade-off relationship between σ (increase) and |α| (decrease) when thermally annealed in association with changes in µe.

Fig. 1.

Fig. 1

Open in a new tab

Changes in the σ and α of C60(MoO3)x films as functions of thermal annealing temperature, which were measured for composite films with (a) x = 0.00 (pristine C60 film), (b) x = 0.05, (c) x = 0.09, and (d) x = 11.8. For the σ and α measurements, 300-nm-thick C60(MoO3)x films were formed on natural mica substrates. Each result of σ and α at Ta = 22℃ in (a)−(d) corresponds to the value measured for the as-grown film.

On the other hand, the σ and α of C60(MoO3)x films show different behaviors against Ta from those of the pristine C60 film. Before the annealing (red squares and blue circles at Ta = 22 ℃ in Figs. 1a−1c), C60(MoO3)x (x = 0.05 and 0.09) films show three-to-four-order magnitude larger σ values and one-to-two-order magnitude smaller |α| values than the pristine C60 films (x = 0.00), which depends on x. This is due to the hole donation from (MoO3)n nanoclusters to C60 molecules, as reported previously20,21. The σ and α values of C60(MoO3)x films remained unchanged even after thermal annealing at Ta in the range of 22 − 100℃, indicating that the mobility (µh) and concentration (nh) of holes in C60(MoO3)x films were maintained before and after thermal annealing. Supplementary Fig. S1 shows atomic force microscopy (AFM) topography images (top) and energy-dispersive X-ray spectroscopy (EDX) elemental imaging maps of C (middle) and Mo (bottom) for C60(MoO3)0.16 films measured (b) before and (c) after thermal annealing at 150℃. In the AFM images and Mo elemental imaging map, granular-shape domain structures with a size of 50–100 nm and a uniform spatial distribution of Mo in the composite film remained unchanged before and after thermal annealing. This indicates that the rearrangement of C60 domains and the aggregation among MoO3 nanoclusters did not occur at the micrometer and/or sub-micrometer scale during thermal annealing, which is consistent with the thermal stabilities of σ and α of C60(MoO3)x films in the Ta range of 22 − 100℃. Moreover, the σ and α of the C60(MoO3)11.8 film were still maintained even at Ta = 250℃ (Fig. 1d), which is comparable to the desorption temperature of C60 molecules from the film28,29. Therefore, C60(MoO3)x films exhibit an excellent thermal stability for practical use at RT.

The high thermal durability of C60(MoO3)x films with structural and thermoelectric properties was realized by covalent bonding between C60 molecules and (MoO3)n nanoclusters. Supplementary Fig. S2a shows Fourier-transform infrared (FT-IR) spectra of pristine C60 (green) and C60(MoO3)2.85 (blue) films before thermal annealing. Four intense IR peaks originated from the radial motion (526 cm− 1), in-plane motion (1428 cm− 1), and a mixture of the two motions (575 and 1182 cm− 1) of C60 molecules30. These four peaks were also observed in the C60(MoO3)2.85 film in addition to a wide and strong band appearing in the range of 490–1030 cm− 1, which was consistently ascribed to the IR spectra of (MoO3)n nanoclusters with various sizes (Supplementary Fig. S3a). Supplementary Fig. S2c shows quantitative changes in the area (circles) and full-width-at-half-maximum (FWHM, triangles) of the IR peak originating from C60 (1182 cm–1, Supplementary Fig. S2b), which were measured in the pristine C60 (green) and C60(MoO3)2.85 (blue) films as a function of Ta. Note that the number of C60 molecules contained in both pristine C60 and C60(MoO3)2.85 films was the same. Before thermal annealing (left side in Supplementary Fig. S2c), the area of the IR peak in the C60(MoO3)2.85 film (blue circle) was obviously smaller by almost one-third that in the pristine C60 film (green circle) whereas the FWHM of the C60(MoO3)2.85 film (blue triangle) was certainly larger than that of the pristine C60 film (green triangle). Supplementary Fig. S3b shows theoretical IR peaks (1182 cm− 1) of pristine C60 (green line) and C60−(MoO3)n (n = 1, 2, 4, 6, 8, 9, blue line). In the spectra of C60−(MoO3)n, the main peak split into two components at higher and lower wavenumbers because the degenerated IR mode of C60 with Ih symmetry was resolved by the formation of covalent bonds with (MoO3)n nanoclusters. Furthermore, the total area of the theoretical IR peak (1182 cm− 1) of C60 (green circle) was obviously reduced by covalently bonding with (MoO3)n nanoclusters (n = 1, 2, 4, 6, 8, 9, blue circles), as shown in Supplementary Fig. S3c. This can be explained by a decrease in the IR activity of C60 molecules by the formation of covalent bonds with (MoO3)n nanoclusters.

The peak area of the C60(MoO3)2.85 film remained unchanged after thermal annealing at Ta < 100 ℃ (right side of Supplementary Fig. S2c). Although this peak area gradually increased again after thermal annealing at Ta > 100 ℃, it saturated below ca. 50% even after the annealing at Ta = 250℃ compared with that of the pristine C60 film. This indicates that covalent bonds between C60 molecules and (MoO3)n nanoclusters remained. These experimental results are consistent with previous theoretical results showing that (MoO3)n nanoclusters (n = 1 − 9) are bound to adjacent C60 molecules via two strong C − O and C − Mo bonds with covalent and ionic characters, respectively21 , which is expected to disturb the thermal diffusion and aggregation of (MoO3)n nanoclusters in the C60(MoO3)x film during thermal annealing.

On the other hand, the σ of C60(MoO3)0.05 and C60(MoO3)0.09 films markedly decreased by more than one order magnitude after thermal annealing at Ta in the range of 110 − 180℃, as shown in Figs. 1b and c (red squares), respectively. Since the flatness, continuity, and MoO3 dispersibility of the C60(MoO3)x film did not deteriorate even after thermal annealing at Ta = 150℃ (Supplementary Figs. S1b and S1c), we consider that the decrease of σ was caused by decreasing nh of holes injected from (MoO3)n nanoclusters. The capability of (MoO3)n nanoclusters to function as a hole dopant was deactivated by the thermally-induced chemical reduction of the (MoO3)n nanoclusters, which will be discussed later in detail. Conversely, α was markedly increased by almost five (x = 0.05) and seven (x = 0.09) times after thermal annealing at Ta in the range of 110 − 150℃.

The thermally induced trade-off relationship between σ (decrease) and α (increase) for C60(MoO3)x films improves their PF, because PF is proportional to the square of α. Figure 2a shows changes in the PF of C60(MoO3)x films with x values of 0.00 (black triangles), 0.05 (red squares), 0.09 (blue circles), and 0.16 (green triangles) as a function of Ta. Although the PF values of C60(MoO3)0.05, C60(MoO3)0.09, and C60(MoO3)0.16 films are larger by 20 − 50 times than that of pristine C60 films (x = 0.00) before thermal annealing (data obtained at Ta = 22℃ in Fig. 2a), the PF values of the composite films were further increased by thermal annealing at Ta in the range of 100 − 150℃. As a result, extremely large PFmax values of 1.1 × 10− 3 and 0.74 × 10− 3 Wm−1K− 2 were achieved for the C60(MoO3)0.05 and C60(MoO3)0.09 films, respectively. In contrast, the PF value of the C60(MoO3)11.8 film was maintained to be small (ca. 4 × 10− 6 Wm−1K− 2) before and after thermal annealing in a range of 22–400 °C, indicating that the precise control of x at less than ca. 0.16 is required to markedly improve the PF of the C60(MoO3)x film by thermal annealing.

Fig. 2.

Fig. 2

Open in a new tab

Changes in the PF (left axis) and/or expected zT (right axis) of C60(MoO3)x films with x = 0.00 (black triangles), x = 0.05 (red squares), x = 0.09 (blue circles), and x = 0.16 (green triangles) as functions of (a)Ta and (b)σ. In (b), data of purple stars show changes in the PF and expected zT of C60(MoO3)x films with x of 0.00–11.8 as a function of σ. PF (= α2σ) was estimated using σ and α obtained experimentally, and then the expected zT was evaluated using zT = PFT/κ at T = 22℃ and κ = 0.4 Wm–1 K–1 (value of C60 solid). Each result of PF at Ta = 22℃ in (a) corresponds to the value measured for the as-grown film. (c) Changes in κ of C60-based materials as a function of σ, which were estimated using Eq. (1).

Figure 2b shows the PF − σ curves obtained from the results in Fig. 2a, together with that (purple stars) of C60(MoO3)x (x = 0.00–11.8) before thermal annealing. Note that the remarkable improvement in the PF of C60(MoO3)x films by the control of x and Ta was realized in the σ region of less than 10–2 Scm–1, as shown in Fig. 2b. In this low-σ region, the thermal conductivity (κ) of C60-based materials is constant (Fig. 2c) given that the κ of C60-based materials is expressed as the follows1,11

graphic file with name d33e2110.gif 1

Here, L0 is the Lorentz number (1.5 × 10–8 V2K− 2) estimated for nondegenerate semiconductors using a single parabolic band with an acoustic phonon scattering model31,T is Celsius temperature (22 ℃), and κl is the phonon-mediated thermal conductivity of pristine solid C60 (0.4 W m–1 K–1)22 . In this case, the maximum zT values of C60(MoO3)0.05 and C60(MoO3)0.09 films at RT were expected to be 0.81 and 0.55, respectively, as shown in the right axis of Fig. 2b. To the best of our knowledge, the estimated zT (0.81 at 22 ℃) of the C60(MoO3)0.05 film obtained in this work is the highest value when compared with those of high-performance organic TE materials recently reported so far, for example, molecule-doped polymers [zT = 0.2 (127 °C), 0.42 (24 °C), and 0.5 (27 °C)]6–8, assemblies of n-doped fullerene derivatives [zT = 0.34 (120 °C)]9, and zinc-based charge transfer complexes [zT = 0.23 (25 °C) and 0.45 (200 °C)]10.

To obtain zT more accurately, it is preferable to be calculated using the κ value obtained experimentally for C60(MoO3)x films before and after thermal annealing rather than the referred value of C60 solid crystal. On the other hand, we found that the high zT is expected to be realized for C60(MoO3)x films with low x such as C60(MoO3)0.05 and C60(MoO3)0.09, which were composed of C60 domains doped with a low density of (MoO3)n nanoclusters. Thus, we consider that there was a small difference in κl among pristine C60 and C60(MoO3)x films before and after thermal annealing. A low κl (0.4 Wm–1 K–1) of single C60 crystal was reported to originate mainly from the strong phonon scattering induced by the rotation and vibration of C60 molecules that interacted with each other via the van der Walls force at RT22 . The κl of C60 films further decreases as a result of phonon scattering at geometrical defects such as grain boundaries. Conversely, it was theoretically determined that the κl of C60 films markedly increases to be on the order of 10 Wm–1K–1 at RT owing to the formation of two-dimensional networks of covalently bound C60 molecules32, in which the rotation of C60 is strongly disturbed. In the present FT-IR measurements of the C60(MoO3)x films before and after thermal annealing at 250 °C (Supplementary Fig. S2a), we observed no spectral features arising from the formation of covalent bonds between C60 molecules by processes such as [2 + 2] cycloaddition reaction33 and/or generalized Stone–Wales transformations.34,35 This indicates that the rotation and van der Waals intermolecular interaction of C60 molecules were maintained even in the C60(MoO3)x film before and after thermal annealing at 250 °C. In addition, AFM measurements (Supplementary Figs. S1b and S1c) also demonstrated that the size and spatial distribution of grains in C60(MoO3)0.16 films remain unchanged before and after thermal annealing at 250 °C. These results indicate that the strong phonon scattering in C60 film is maintained even in the C60(MoO3)x film before and after thermal annealing.

Furthermore, note that the conduction type (N/P) of C60(MoO3)x films was also controlled by thermal annealing. Although C60(MoO3)x (0.03 ≤ x ≤ 11.8) exhibited the P-type property (α > 0) before thermal annealing20, the N-type property (α < 0) emerged in C60(MoO3)0.03, C60(MoO3)0.09, and C60(MoO3)0.16 films thermally annealed at Ta in the range of 130 − 180℃ (Fig. 1c and Supplementary Fig. S4). This indicates that the carriers changed from holes to electrons. Since the change in carrier type and the decrease in σ (Fig. 1b and c) were simultaneously induced in the similar Ta range, these might originate from the change in the chemical state of (MoO3)n nanoclusters after thermal annealing. To confirm the change in the chemical state before and after thermal annealing, we next examined the (MoO3)n nanoclusters in the composite films by X-ray photoelectron spectroscopy (XPS) and XAFS analysis.

We first measured high-resolution XPS spectra of Mo 3d for pristine MoO3 [Fig. 3a], C60(MoO3)0.16 [Fig. 3b], and C60(MoO3)11.8 [Fig. 3c] films before (top panels) and after (center panels) thermal annealing at 200℃ or 400℃. The horizontal bars in the inset of Fig. 3 denote the binding energy range of each Mo 3d5/2 peak of Mo6+ (red), Mo4+ (green), and Mo (metal) (black), which were taken from the XPS database as references36. For the as-grown films shown in the top panels of Figs. 3a and c, two strong peaks (dashed red lines) were attributed to the Mo 3d3/2 (235.2 eV) and Mo 3d5/2 (232.2 eV) of Mo6+. In addition, a small shoulder (dashed blue lines) appears in a low-binding-energy region for the 3d3/2 and 3d5/2 peaks of Mo6+. Since this minor component was present in larger amounts for the C60(MoO3)0.16 and C60(MoO3)11.8 films rather than for the pristine MoO3 film, it originated from Mo with a slightly smaller valence l (4+ ≤ l ≤ 6+) that was generated by the hole injection from (MoO3)n nanoclusters to C60 molecules in the composite films.

Fig. 3.

Fig. 3

Open in a new tab

XPS spectra of (a) MoO3, (b) C60(MoO3)0.16, and (c) C60(MoO3)11.8 films measured before (top panels) and after (middle panels) thermal annealing in the binding energy range of 227.5–238.4 eV. The bottom panels in (a), (b), and (c) show changes in the content ratios of Mo+ 6, Mo (l) (+4 ≤ l ≤ +6), and Mo (m) (0 ≤ m ≤ +4) in each film as a function of Ta, which were estimated using each of the XPS peak areas of Mo6+ (dashed red line), Mo (l) (dashed blue line), and Mo (m) (dashed green line). For all XPS measurements, we used 30-nm-thick C60(MoO3)x films formed on Ta substrates.

The Mo (l) of C60(MoO3)0.16 and C60(MoO3)11.8 films markedly increased after thermal annealing at 200℃ (center panel of Fig. 3b) and 400℃ (center panel of Fig. 3c), respectively. Furthermore, the 3d3/2 and 3d5/2 peaks of Mo with a smaller valence m (0 ≤ m ≤ 4+) appeared in the C60(MoO3)11.8 film after thermal annealing at 400℃ (dashed green lines). As shown in the bottom panels of Figs. 3a and b, we examined changes in the content ratios of Mo6+ and Mo (l) in the MoO3 and C60(MoO3)0.16 films using the peak area ratios of Mo6+ and Mo (l) as a function of Ta. The ratio of Mo (l) in the C60(MoO3)0.16 film remarkably increased at Ta of more than ca. 110℃, whereas no increase was observed in the pristine MoO3 film. This is similar to that observed for the α and σ of C60(MoO3)0.05 (Fig. 1b) and C60(MoO3)0.09 (Fig. 1c) films. In a similar manner, marked increases in the content ratios of Mo (l) and Mo (m) in the C60(MoO3)11.8 film were obtained in a temperature range (Ta > 300℃) similar to that inducing the trade-off relationship between α and σ (Fig. 1d).

We next examined the changes in the coordination structures of Mo in the C60(MoO3)x films using XAFS. The blue solid curves in Fig. 4 show the radial distribution functions obtained by the Fourier transform of the extended X-ray absorption fine structure (EXAFS) of Mo K-edge spectra for the α-MoO3 crystal (a), and the C60(MoO3)0.16 (b) and C60(MoO3)11.8 (c) films before thermal annealing. Four intense peaks of the α-MoO3 crystal originate from X-ray scattering paths at Mo–O (OI and OII) and Mo–Mo (MoI and MoII). By comparing with the theoretical radial distribution functions of Mo–O dimers (dashed pink and light blue lines), we found that the peaks OI and OII mainly originate from the Mo–O coordination with interatomic distances (dMo−O) of 0.166 and 0.217 nm, respectively, which agrees well with two of six dMo−O in the MoO3 crystal37. Peaks OI and OII were observed in both C60(MoO3)0.16 and C60(MoO3)11.8 films without any shift in radial distance. Although the peak positions of OI and OII remained constant before (blue) and after (red) thermal annealing at 200℃ for the C60(MoO3)0.16 film (Fig. 4b) and at 300℃ for the C60(MoO3)11.8 film (Fig. 4c), the peak intensity of OI decreased after thermal annealing. This indicates that the number of O atoms coordinating with Mo atoms decreased after thermal annealing, which is consistent with the decrease in the Mo valence observed by XPS (Figs. 3b and c). There are three possible reasons behind the chemical reduction in Mo valence in association with the decrease in the coordination number of O atom: (MoO3)n bound to C60 is thermally dissociated to MonO3n−2 and O2 [reaction (a)], (MoO3)n oxidizes C60 to form C60Ok and MonO3n−k [reaction (b)], or CO2 is emitted from C60(MoO3)n to form C59–(MonO3n−2) [reaction (c)].

Fig. 4.

Fig. 4

Open in a new tab

Effect of thermal annealing on the coordination structure of Mo atoms in C60(MoO3)x films. Radial distribution functions obtained by the Fourier transform of EXAFS (wavenumber range: 0.3–1.2 nm− 1) for the (a) MoO3 crystal, (b) C60(MoO3)0.16, and (c) C60(MoO3)11.8, which were measured before (blue lines) and after (red lines) thermal annealing. Dashed lines in (a) show the theoretical radial distribution functions of Mo–Mo dimers with dMo−Mo = 0.340 nm (black) and dMo−Mo = 0.375 nm (purple), and those of Mo–O dimers with dMo−O = 0.166 nm (pink) and dMo−O = 0.217 nm (light blue) calculated using the ARTEMIS package.58 For XAFS measurements, we used 10-µm-thick C60(MoO3)0.16 and 1-µm-thick C60(MoO3)11.8 films formed on glass substrates.

Inline graphic (a).

Inline graphic (b).

Inline graphic (c).

In the case of reaction (a), the activation energy for the decomposition of (MoO3)n itself, (MoO3)n → MonO3n−2 (n = 1 − 6 ) + O2, is more than ca. 2.0 eV38. This indicates that it was difficult for reaction (a) to proceed at Ta in the range of 110 − 200℃, which is consistent with the XPS results that Mo with a small valence l (4 + ≤ l ≤ 6+) is a minor component of the pristine MoO3 film even after thermal annealing at 170℃ (bottom panel of Fig. 3a). In addition, we observed no peaks corresponding to O2 emitted from the C60(MoO3)x films during thermal annealing by in situ quadrupole mass spectroscopy (QMS) (see Fig. 5 as described in detail later). Accordingly, reaction (a) can be excluded in the present system.

Fig. 5.

Fig. 5

Open in a new tab

In situ QMS measurements of (a) 10-µm-thick C60(MoO3)0.16 and (b) 1-µm-thick C60(MoO3)11.8 films, and (c) bare Ta substrate during thermal annealing within the 25–405℃ range in UHV (base pressure: 2 × 10–7 Pa). The Ta substrate was cleaned by thermal annealing at 450℃ for 6 h before the formation of C60(MoO3)x films. The vertical axis corresponds to an increase in the ion current (ΔIi) of M/z = 16 (O, green squares), M/z = 28 (CO, blue triangles), M/z = 32 (O2, orange squares), and M/z = 44 (CO2, red circles) determined by subtracting Ii before thermal annealing.

On the other hand, in the case of reaction (b), the chemical reduction of (MoO3)n possibly proceeds via O atoms reacting with surrounding C60 molecules as a reductant in C60(MoO3)x films. Then, we examined the FT-IR spectra of C60(MoO3)2.85 films after thermal annealing at Ta of 100, 150, 200, 250, and 300℃ (Supplementary Fig. S2d). However, we observed no IR peaks originating from C60 oxides39,40, such as C60O, C60O2, C120O3, C120O, and C120O2, in the range of 1600–2200 cm–1, where the IR peaks corresponding to C60 oxides appear.41–44 Hence, reaction (b) can also be excluded.

Finally, in the case of reaction (c), since CO2 should be emitted from C60(MoO3)x films during thermal annealing, we also performed in situ QMS measurements of C60(MoO3)x films during thermal annealing. Figure 5 shows changes in the ion current (ΔIi) originating from O (M/z = 16), CO (M/z = 28), O2 (M/z = 32), and CO2 (M/z = 44), individually, which were obtained by the QMS of the C60(MoO3)0.16 (a) and C60(MoO3)11.8 (b) films, and a bare Ta substrate (c) during thermal annealing at Ta in the range of 25–405℃. As shown in Figs. 5a and c, the ΔIi of CO2, CO, and O for the C60(MoO3)x films markedly increased at Ta > 110℃, whereas it could be negligible for the Ta substrate even at Ta of up to 400℃. Here, CO and O were fragments from CO2, because the ΔIi behavior of both CO and O shows a similar Ta dependence to that of CO2. Thus, CO2 was assigned to the main product emitted from the C60(MoO3)x films during thermal annealing. This finding clearly demonstrates that reaction (c) occured in the C60(MoO3)x films during thermal annealing.

We next discuss the impacts of reaction (c) on the TE characteristics of C60(MoO3)x films (see Fig. 1b and c) using theoretical calculations. A previous study using density functional theory (DFT) showed that (MoO3)n (n = 1 − 6) nanoclusters have a large HOMO−LUMO gap of more than ca. 2.5 eV, whereas MonO3n−1 and MonO3n−2 have a smaller HOMO−LUMO gap of less than one-third via their HOMO shift to a higher energy region38. This implies that (MoO3)n nanoclusters weaken their activity as a hole dopant owing to a decrease in their ionization potential by chemical reduction. To clarify the mechanisms behind the emergence of N-type TE characteristics of C60(MoO3)x films (x = 0.03, 0.09, and 0.16) after thermal annealing (see Supplementary Fig. S4), we theoretically evaluated the amount of effective charges on both C60 and (MoO3)n of C60−(MoO3)n and on both C59 and MonO3n−2 of C59−MonO3n−2 for n = 1, 2, 4, 6, 8, and 9.

Figure 6 shows (a) schematics of the optimized structure of C60−(MoO3)n (upper panel) and C59−MonO3n−2 (lower panel), and (b) plots of the effective charges on C60 of C60−(MoO3)n [blue circles] and C59 of C59−MonO3n−2 [red circles] with respect to the number of Mo atoms (n). Before the chemical reduction of C60−(MoO3)n, some amount of holes are doped from (MoO3)n to C60 except at n = 1 and increase with n basically. This can well explain the emergence of the P-type TE characteristics of the C60(MoO3)x films, as shown in Figs. 1b−1d. On the other hand, it is noted that a some amount of electrons are doped into C59 from MonO3n−2. This well corresponds to changes in the N-type TE characteristics of the C60(MoO3)x films after thermal annealing. To examine the bonding interactions between C60 and (MoO3)n in detail, we focused on the bonding area in C60−(MoO3)6 and C59−Mo6O16 models. For bonding area in C60−(MoO3)6 (Supplementary Fig. S5a), although the C atom directly bound to Mo has large negative charges, since almost all C atoms in the vicinity of the bonding area have positive charges, the total charge of C60 is positive. On the other hand, for bonding area in C59−Mo6O16 (Supplementary Fig. S5b), although some C atoms around the bonding area have positive charges, the total charge of C60 is negative.

Fig. 6.

Fig. 6

Open in a new tab

(a) Schematics of theoretically optimized structure of C60−(MoO3)n (upper panel) and C59−MonO3n-1 (lower panel) heterodimers at n = 1, 2, 4, 6, 8, and 9. Red and blue dashed arrows in (a) indicate C − Mo and C − O bonds, respectively. (b) Changes in the amount of effective charges on C60 in the C60−(MoO3)n heterodimers and C59 in the C59−MonO3n−1 heterodimers as a function of n. Since the formation of C60–(MoO3)n (n = 3, 5, 7) heterodimers hardly proceeds energetically because of their large positive ΔG values,21 they were excluded in this work.

Interestingly, the number of electrons doped into C59 from MonO3n−2 decreases with increasing n, particularly at n = 8 and 9, which is the alternative tendency of hole doping into C60 before reaction (c). These results reasonably explain the experimental results that the emergence of the N-type TE properties and the marked decrease in σ were simultaneously induced in the C60(MoO3)x films (x = 0.09) by thermal annealing at Ta > 130 ℃ (Fig. 1c). Conversely, the results in Fig. 6 also suggest that the concentration of N-type carriers (electrons) in C60(MoO3)x films is increased by the chemical reduction of (MoO3)n nanoclusters with sizes less than n = 6. Therefore, the control of both the size of (MoO3)n nanoclusters and the Mo valence plays a key role in realizing practical high-performance N-type C60-based TE materials.

In summary, we have formed novel organic TE materials exhibiting high PFmax at a low σ in the range of 10− 5−10− 2 Scm− 1, using C60(MoO3)x films. Among them, it is noted that the C60(MoO3)0.05 film exhibits an exceedingly high PFmax of 1.1 × 10− 3 Wm−1K− 2 at σ = 9 × 10− 4 Scm− 1 after thermal annealing, whose zT (T = 22℃) is expected to be 0.81. To the best of our knowledge, this is the highest value among organic TE materials to date. The key factors leading to this result are the emergence and control of the GSE of C60 films at σ in the range of 10− 5−10− 2 Scm− 1 by doping (MoO3)n nanoclusters with a precise composition and controlling the Mo valence via the thermally induced chemical reduction. This has never been realized in various organic semiconductors doped with metal atoms and metal−molecule complexes so far14,20. To date, the zT of TE materials has been improved on the basis of several concepts, such as utilizing a low-dimensional electron system45, a design guideline of phonon−glass electron−crystal systems46, and the formation of organic−inorganic composites2–4. The findings obtained in the present work provide a new strategy and concept for realizing high-zT materials through the emergence and control of GSE in the low-σ region, which overcomes the barrier caused by the strong positive correlation between κ and σ in the high-σ region of more than 10 Scm− 1. The doping of metal−oxide nanoclusters will be a promising method to achieve a high zT for not only C60 films but also other organic films with GSE. In other words, this strategy can be applied to realizing novel high-performance flexible TE materials.

Methods

Formation of C60(MoO3)x films with control of x

C60(MoO3)x films were formed and thermally annealed in the same ultrahigh vacuum (UHV, a base pressure of less than 1 × 10–6 Pa) system consisting of multiple chambers equipped with measurement systems for σ and α, and a QMS system. Natural mica, tantalum (Ta, 99.95% in purity), Si(111), cesium iodides (CsI), quartz, and glass substrates were thermally cleaned at 280 °C under UHV conditions before the formation of C60(MoO3)x films. C60(MoO3)x films were formed by codepositing C60 and MoO3 at a substrate temperature of 22℃ by the thermal evaporation of C60 (MTR Ltd., 99.98% in purity) and MoO3 (Rare Metallic, 99.9% in purity) powders. The mole ratio x (MoO3/C60) was precisely controlled by changing the deposition rates of C60 (0.2–7.35 nm/min) and MoO3 (4.4–4.9 × 10–3 nm/min). The x of C60(MoO3)x was measured by XPS (ESCALAB250) using C60(MoO3)x on a Ta substrate20. In all XPS measurements, the Au 4f7/2 orbital (binding energy of 84.00 eV) was used as a reference for the correction of chemical shift in binding energy via charge-up effects. Supplementary Fig. S1a shows photographs of C60(MoO3)x films with x of 0.00, 0.09, 0.90, and 11.8. The color of the films changed from yellow to dark brown with increasing x. This is due to an increase in their optical absorbance in a wavelength range of 670–1050 nm, which originates from the hole injection from (MoO3)n nanoclusters to C60 molecules via the covalent bonding between C60 and (MoO3)n21.

Characterization of C60(MoO3)x films

The surface morphology of the C60(MoO3)x films thus formed was observed by AFM (Hitachi High-Tech. Science SPA400) with a dynamic force mode using Si cantilevers in air. The elemental mappings of Mo and C in the C60(MoO3)x films were carried out by detecting characteristic X-rays (Mo Lα and C Kα, respectively) using an EDX system (Horiba EMAX) equipped with a scanning electron microscope (SEM, Hitachi SU-1500, 15 keV for an incident energy of an electron beam). To evaluate the coordination structure of MoO3 in the C60(MoO3)x films, XAFS analysis at the Mo K-edge was conducted at BL5S1 in Aichi Synchrotron Radiation Center using a Si(111) double-crystal monochromator in the fluorescence mode at RT (Experiment No: 202306041). Photon energy was compensated for before and after each measurement using Mo foil as a reference. Changes in structural properties of C60(MoO3)x films were examined before and after thermal annealing by in situ FT-IR spectroscopy (JASCO FT/IR-6600, resolution: 0.5 cm–1) in a UHV chamber (base pressure: < 3 × 10–6 Pa). Furthermore, we also analyzed by QMS (GME220) the gas species emitted from C60(MoO3)x films during thermal annealing to determine chemical reactions between C60 and (MoO3)n nanoclusters during thermal annealing in the films. By QMS, we evaluated an increase in ion current (ΔIi) on the basis of Ii before the annealing for various species within the M/z range of 1–200 as a function of Ta. The σ and α of C60(MoO3)x composite films were measured without air exposure after the formation of the films. For the evaluation of σ, we obtained the average σ by repeating the four-probe measurements more than ten times at different positions using four phosphor bronze probers (50 μm in tip diameter) connected to a source meter (Keithley 2450). α was measured using a laboratory-made system consisting of two detection probes (50 μm in tip diameter), one heater probe (50 μm in tip diameter), and one thermocouple probe (TC probe, 85 μm in tip diameter)20. Before α measurements, an array of microsize square electrodes (150 μm × 150 μm in area and 100 nm in thickness with 90 μm intervals between adjacent electrodes) was formed on the sample surface by depositing titanium (Ti) atoms through a mesh-shaped shadow mask. Thereafter, the heater probe was carefully placed in contact with one Ti electrode, and ΔT of 5–6℃ was generated between the adjacent electrodes at a substrate temperature of 22℃. Then, the thermoelectromotive force between the electrodes (ΔV) was measured using two detection probes connected to a nanovolt meter (Keithley 2182 A), and α was estimated as α = –ΔV/ΔT. Using these schemes, we obtained the average α by repeating the measurements more than five times at different positions.

Theoretical calculation of C60–(MoO3)n heterodimers

First-principles calculations were performed using Gaussian1647 based on DFT48,49. Geometrical structures of C60–(MoO3)n and C59–MonO3n–2 heterodimers were optimized using the LanL2DZ basis set50 and B3LYP exchange-correlation functional51. Here, the core electrons of Mo were considered an effective core potential in all the calculations52, and the basis set used in this study was taken from the “Basis Set Exchange Library”53. In addition, the neutral singlet state was calculated for the entire system. The same results were obtained for both the spin-polarized and non-spin-polarized states. The amount of effective charges on C60, C59, (MoO3)n, and MonO3n–2 in the heterodimers was obtained by the summation of Mulliken effective charges54–57.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (2.3MB, pdf)

Acknowledgements

The computation in this work was carried out using the supercomputer “Flow” in Nagoya University. One of the authors (M.N.) is thankful to Prof. K. Tanaka (Graduate School of Science, Nagoya University) and Prof. T. Torimoto (Graduate School of Engineering, Nagoya University) for AFM and SEM-EDX measurements, respectively.

Author contributions

M.N. conceived and designed the overall project. J.O. supervised the project. M.N. and S.Y. prepared all samples and measured TE properties, FT-IR spectra, and XPS spectra, and conducted EDX mapping and AFM imaging. S.O. performed XAFS measurements. T.N. carried out theoretical calculations. M.N., S.Y., S.O., T.N., and J.O. discussed the results. M.N. and J.O. cowrote the manuscript. All authors have read and commented on the manuscript.

Funding

This work was financially supported partly by Grants-in-Aid for Scientific Research (KAKENHI) (Grant Nos. 21200032, 17H03384, 20H02463, and 23H01708) of Japan Society for the Promotion of Science (JSPS), partly by “The R&D program for advanced nuclear power systems” from MEXT (Ministry of Education, Culture, Sports, Science and Technology), and partly by the joint research fund of J-Power.

Data availability

The datasets used in the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Contributor Information

Masato Nakaya, Email: m-nakaya@energy.nagoya-u.ac.jp.

Jun Onoe, Email: j-onoe@energy.nagoya-u.ac.jp.

References

  • 1.Snyder, G. J. & Toberer, E. S. Complex thermoelectric materials. Nat. Mater.7, 105–114. 10.1038/nmat2090 (2008). [DOI] [PubMed] [Google Scholar]
  • 2.Du, Y., Shen, S. Z., Cai, K. & Casey, P. S. Research progress on polymer–inorganic thermoelectric nanocomposite materials. Prog Polym. Sci.37, 820–841. 10.1016/j.progpolymsci.2011.11.003 (2012). [Google Scholar]
  • 3.Russ, B., Glaudell, A., Urban, J. J., Chabinyc, M. L. & Segalman, R. A. Organic thermoelectric materials for energy harvesting and temperature control. Nat. Rev. Mater.1, 16050. 10.1038/natrevmats.2016.50 (2016). [Google Scholar]
  • 4.Jin, H. et al. Hybrid Organic–Inorganic Thermoelectric Materials and Devices. Angew Chem. Int. Ed.58, 15206–15226. 10.1002/anie.201901106 (2019). [DOI] [PubMed] [Google Scholar]
  • 5.Wang, H., Yu, C. & Thermoelectrics, O. Materials Preparation, Performance Optimization, and Device Integration. Joule3, 53–80 (2019). [Google Scholar]
  • 6.Sun, Y. et al. Organic Thermoelectric Materials and Devices Based on p- and n-Type Poly(metal 1,1,2,2-ethenetetrathiolate)s. Adv. Mater.24, 932–937. 10.1002/adma.201104305 (2012). [DOI] [PubMed] [Google Scholar]
  • 7.Kim, G-H., Shao, L., Zhang, K. & Pipe, K. P. Engineered doping of organic semiconductors for enhanced thermoelectric efficiency. Nat. Mater.12, 719–723. 10.1038/nmat3635 (2013). [DOI] [PubMed] [Google Scholar]
  • 8.Wang, H. et al. Thermally Driven Large N-Type Voltage Responses from Hybrids of Carbon Nanotubes and Poly(3,4-ethylenedioxythiophene) with Tetrakis(dimethylamino) ethylene. Adv. Mater.27, 6855–6861. 10.1002/adma.201502950 (2015). [DOI] [PubMed] [Google Scholar]
  • 9.Liu, J. et al. A., N-type organic thermoelectrics: demonstration of ZT > 0.3. Nat. Commun.11, 5694. 10.1038/s41467-020-19537-8 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Li, X. et al. Transparent charge transfer complex with high thermoelectric performance. Joule7, 2814–2829. 10.1016/j.joule.2023.10.008 (2023). [Google Scholar]
  • 11.Zapata-Arteaga, O. et al. Reduction of the Lattice Thermal Conductivity of Polymer Semiconductors by Molecular Doping. ACS Energy Lett.5, 2972–2978. 10.1021/acsenergylett.0c01410 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Kojima, H. et al. Giant Seebeck effect in pure fullerene thin films. Appl. Phys. Exppress. 8, 121301. 10.7567/APEX.8.121301 (2015). [Google Scholar]
  • 13.Kojima, H. et al. Universality of the giant Seebeck effect in organic small molecules. Mater. Chem. Front.2, 1276–1283. 10.1039/c7qm00596b (2018). [Google Scholar]
  • 14.Nakamura, M. et al. Giant Seebeck effect over 0.1 V K– 1 – is this an intrinsic phenomenon in organic semiconductors? Faraday Discuss.250, 361–376. 10.1039/d3fd00127j (2024). [DOI] [PubMed] [Google Scholar]
  • 15.Okayama, K. et al. Giant seebeck effect in an undoped single crystal of 2,5,8-triphenylbenzo[1,2-b:3,4-b’:5,6-b’’]trifuran. Chem. Lett.53, upad054. 10.1093/chemle/upad054 (2024). [Google Scholar]
  • 16.Sumino, M. et al. Thermoelectric properties of n-type C60 thin films and their application in organic thermovoltaic devices. Appl. Phys. Lett.99, 093308. 10.1063/1.3631633 (2011). [Google Scholar]
  • 17.Menke, T. et al. A comparison of two air-stable molecular n-dopants for C60. Org. Electron.13, 3319–3325. 10.1016/j.orgel.2012.09.024 (2012). [Google Scholar]
  • 18.Menke, T., Ray, D., Meiss, J., Leo, K. & Riede, M. In-situ conductivity and Seebeck measurements of highly efficient n-dopants in fullerene C60. Appl. Phys. Lett.100, 093304. 10.1063/1.3689778 (2012). [Google Scholar]
  • 19.Barbot, A., Bin, C. D., Lucas, B., Ratier, B. & Aldissi, M. N-type doping and thermoelectric properties of co-sublimed cesium-carbonate-doped fullerene. J. Mater. Sci.48, 2785–2789. 10.1007/s10853-012-6824-1 (2013). [Google Scholar]
  • 20.Nakaya, M., Yamamoto, S., Kawai, T. & Onoe, J. High-performance and anomalous thermoelectric properties of organic–inorganic nanocomposites consisting of fullerene molecules and molybdenum trioxide nanoclusters. Carbon241, 120374. 10.1016/j.carbon.2025.120374 (2025). [Google Scholar]
  • 21.Nakaya, M., Kawai, T., Watanabe, S. & Onoe, J. Micro-/macroscopic and density functional studies of the interactions between molybdenum trioxide and C60 molecule. J. Chem. Phys.158, 054701. 10.1063/5.0134207 (2023). [DOI] [PubMed] [Google Scholar]
  • 22.Yu, R. C., Tea, N., Salamon, M. B., Lorents, D. & Malhotra, R. Thermal Conductivity of Single Crystal C60. Phys. Rev. Lett.68, 2050–2053. 10.1103/PhysRevLett.68.2050 (1992). [DOI] [PubMed] [Google Scholar]
  • 23.Claridge, S. A. et al. Cluster-Assembled Materials. ACS Nano. 3, 244–255. 10.1021/nn800820e (2009). [DOI] [PubMed] [Google Scholar]
  • 24.Özdoğan, C. et al. The Unusually Stable B100 Fullerene, Structural Transitions in Boron Nanostructures, and a Comparative Study of α- and γ-Boron and Sheets. J. Phys. Chem. C. 114, 4362–4375. 10.1021/jp911641u (2010). [Google Scholar]
  • 25.Gautam, R. et al. Creation of flexible spin-caloritronic material with giant transverse thermoelectric conversion by nanostructure engineering. Nat. Commun.15, 2184. 10.1038/s41467-024-46475-6 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Okada, J. T. et al. Improvement of thermoelectric properties of icosahedral AlPdRe quasicrystals by Fe substitution for Re. J. Appl. Phys.101, 103702. 10.1063/1.2716212 (2007). [Google Scholar]
  • 27.Könenkamp, R., Priebe, G. & Pietzak, B. Carrier mobilities and influence of oxygen in C60 films. Phys. Rev. B. 60, 11804–11808. 10.1103/PhysRevB.60.11804 (1999). [Google Scholar]
  • 28.Nakayama, T., Onoe, J., Takeuchi, K. & Aono, M. Weakly bound and strained C60 monolayer on the Si(111)√3×√3R30°-Ag substrate surface. Phys. Rev. B. 59, 12627–12631. 10.1103/PhysRevB.59.12627 (1999). [Google Scholar]
  • 29.Hamza, A. V. & Balooch, M. The chemisorption of C60 on Si (100)-(2×1). Chem. Phys. Lett.201, 404–408. 10.1016/0009-2614(93)85092-3 (1993). [Google Scholar]
  • 30.Stanton, R. E. & Newton, M. D. Normal Vibrational Modes of Buckminsterfullerene. J. Phys. Chem.92, 2141–2145. 10.1021/j100319a012 (1988). [Google Scholar]
  • 31.Kim, H. S., Gibbs, Z. M., Tang, Y., Wang, H. & Snyder, G. J. Characterization of Lorenz number with Seebeck coefficient measurement. APL Mater.3, 041506. 10.1063/1.4908244 (2015). [Google Scholar]
  • 32.Mortazavi, B. Structural, electronic, thermal and mechanical properties of C60-based fullerene two-dimensional networks explored by first-principles and machine learning. Carbon213, 118293. 10.1016/j.carbon.2023.118293 (2023). [Google Scholar]
  • 33.Onoe, J. et al. Electron-beam irradiation of photopolymerized C60 film studied using in situ scanning tunneling microscope, in situ Fourier-transform infrared spectroscopy, and first-principles calculations. AIP Adv.10, 085212. 10.1063/5.0018985 (2020). [Google Scholar]
  • 34.Takashima, A., Nishii, T. & Onoe, J. Formation process and electron-beam incident energy dependence of one-dimensional uneven peanut-shaped C60 polymer studied using in situ high-resolution infrared spectroscopy and density-functional calculations. J. Phys. D: Appl. Phys.45, 485302. 10.1088/0022-3727/45/48/485302 (2012). [Google Scholar]
  • 35.Nakaya, M. et al. Immobilization of CO2 at Room Temperature Using the Specific Sub-NM Space of 1D Uneven-Structured C60 Polymer Film. Adv. Sustainable Syst.5, 2000156. 10.1002/adsu.202000156 (2021). [Google Scholar]
  • 36.Lee, A. Y. et al. Development of the NIST X-ray Photoelectron Spectroscopy (XPS) Database, Version 5. Data Sci. J.23, 1–9. 10.5334/dsj-2024-045 (2024). [Google Scholar]
  • 37.Scanlon, D. O. et al. Theoretical and Experimental Study of the Electronic Structures of MoO3 and MoO2. J. Phys. Chem. C. 114, 4636–4645. 10.1021/jp9093172 (2010). [Google Scholar]
  • 38.Sumer, A. Molybdenum Oxide Clusters: Structure, Stability, and Electronic Properties. J. Phys. Chem. A. 125, 5201–5211. 10.1021/acs.jpca.1c03052 (2021). [DOI] [PubMed] [Google Scholar]
  • 39.Heymann, D. & Weism, R. B. Fullerene oxides and ozonides. C. R. Chim.9, 1107–1116. 10.1016/j.crci.2006.02.003 (2006). [Google Scholar]
  • 40.Tabari, L. & Farmanzadeh, D. Interesting adsorption behavior of C60O fullerene oxide isomers toward O3 and CO molecules: A DFT study. Appl. Surf. Sci.479, 569–575. 10.1016/j.apsusc.2019.02.085 (2019). [Google Scholar]
  • 41.Datta, A., Kelkar, R. Y., Boroojerdian, P., Kulkarni, S. K. & Datta, M. An FT-IR and TGA Study of the Thermal Oxidation of C60. Bull. Chem. Soc. Jpn. 67, 1517–1521 (1994). [Google Scholar]
  • 42.Nisha, J. A. et al. Studies of C60 Oxidation and Products. J. Phys. Chem.100, 4503–4506. 10.1021/jp9532576 (1996). [Google Scholar]
  • 43.Krause, M. et al. Vibrational signatures of fullerene oxides. J. Chem. Soc. Faraday Trans.94, 2287–2294. 10.1039/A802560F (1998). [Google Scholar]
  • 44.Cataldo, F. Ozone Reaction with Carbon Nanostructures 1: Reaction Between Solid C60 and C70 Fullerenes and Ozone. J. Nanosci. Nanotechnol. 7, 1–7. 10.1166/jnn.2007.326 (2007). [DOI] [PubMed] [Google Scholar]
  • 45.Dresselhaus, M. S. et al. New Directions for Low-Dimensional Thermoelectric Materials. Adv. Mater.19, 1043–1053. 10.1002/adma.200600527 (2007). [Google Scholar]
  • 46.Nolas, G. S., Morelli, D. T. & Tritt, T. M. A Phonon-Glass-Electron Crystal Approach to Advanced Thermoelectric Energy Conversion Applications. Annu. Rev. Mater. Sci.29, 89–115. 10.1146/annurev.matsci.29.1.89 (1999). [Google Scholar]
  • 47.Frisch, M. J. et al. Gaussian 16, Revision C.01. Gaussian (Inc., 2019).
  • 48.Hohenberg, P. & Kohn, W. Inhomogeneous Electron Gas. Phys. Rev.136, B864–B871. 10.1103/PhysRev.136.B864 (1964). [Google Scholar]
  • 49.Kohn, W. & Sham, L. J. Self-Consistent Equations Including Exchange and Correlation Effects. Phys. Rev.140 A1133 – A1138. 10.1103/PhysRev.140.A1133 (1965). [Google Scholar]
  • 50.Weigend, F. & Ahlrichs, R. Balanced basis sets of split valence, triple zeta valence and quadruple zeta valence quality for H to Rn: Design and assessment of accuracy. Phys. Chem. Chem. Phys.7, 3297–3305. 10.1039/B508541A (2005). [DOI] [PubMed] [Google Scholar]
  • 51.Yanai, T., Tew, D. P. & Handy, N. C. A new hybrid exchange–correlation functional using the Coulomb-attenuating method (CAM-B3LYP). Chem. Phys. Lett.393, 51–57. 10.1016/j.cplett.2004.06.011 (2004). [Google Scholar]
  • 52.Andrae, D., Häuẞermann, U., Dolg, M., Stoll, H. & Preuẞ, H. Energy-adjusted ab initio pseudopotentials for the second and third row transition elements. Theor. Chem. Acta. 77, 123–141. 10.1007/BF01114537 (1990). [Google Scholar]
  • 53.Pritchard, B. P., Altarawy, D., Didier, B., Gibsom, T. D. & Windus, T. L. New Basis Set Exchange: An Open, Up-to-Date Resource for the Molecular Sciences Community. J. Chem. Inf. Model.59, 4814–4820. 10.1021/acs.jcim.9b00725 (2019). [DOI] [PubMed] [Google Scholar]
  • 54.Mulliken, R. S. Electronic Population Analysis on LCAO–MO Molecular Wave Functions. I. J. Chem. Phys.23, 1833–1840. 10.1063/1.1740588 (1955). [Google Scholar]
  • 55.Mulliken, R. S. Electronic Population Analysis on LCAO–MO Molecular Wave Functions. II. Overlap Populations, Bond Orders, and Covalent Bond Energies. J. Chem. Phys.23, 1841–1846. 10.1063/1.1740589 (1955). [Google Scholar]
  • 56.Mulliken, R. S. Electronic Population Analysis on LCAO–MO Molecular Wave Functions. III. Effects of Hybridization on Overlap and Gross AO Populations. J. Chem. Phys.23, 2338–2342. 10.1063/1.1741876 (1955). [Google Scholar]
  • 57.Mulliken, R. S. Electronic Population Analysis on LCAO–MO Molecular Wave Functions. IV. Bonding and Antibonding in LCAO and Valence–Bond Theories. J. Chem. Phys.23, 2343–2346. 10.1063/1.1741877 (1955). [Google Scholar]
  • 58.Ravel, B. & Newville, M. ATHENA and ARTEMIS: Interactive Graphical Data Analysis using IFEFFIT. Phys. Scr.T115, 1007–1010. 10.1238/Physica.Topical.115a01007 (2005). [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (2.3MB, pdf)

Data Availability Statement

The datasets used in the current study are available from the corresponding author on reasonable request.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES