Solomon Demiss
Negedu‡
ab,
Raphael
Tromer‡
c,
Chinmayee Chowde
Gowda
d,
Cristiano F.
Woellner
e,
Femi Emmanuel
Olu
b,
Ajit K.
Roy
f,
Prafull
Pandey
g,
Douglas S.
Galvao
*ch,
Pulickel M.
Ajayan
*i,
Partha
Kumbhakar
*a and
Chandra S.
Tiwary
*a
aDepartment of Metallurgical and Materials Engineering, Indian Institute of Technology Kharagpur, West Bengal 721302, India. E-mail: parthakumbhakar2@gmail.com; chandra.tiwary@metal.iitkgp.ac.in
bMaterials Science and Engineering, Jimma Institute of Technology, Jimma University, Jimma, Ethiopia
cApplied Physics Department, University of Campinas, Brazil. E-mail: galvao@ifi.unicamp.br
dSchool of Nano Science and Technology, Indian Institute of Technology Kharagpur, West Bengal 721302, India
ePhysics Department, Federal University of Parana, UFPR, Curitiba, PR 81531-980, Brazil
fMaterials and Manufacturing Directorate, Air Force Research Laboratory, Wright Patterson AFB, OH 45433-7718, USA
gDepartment of Materials Engineering, Indian Institute of Science, Bangalore 560012, India
hCenter for Computational Engineering and Sciences, State University of Campinas, Campinas, SP 13083-970, Brazil
iDepartment of Materials Science and Nano Engineering, Rice University, Houston, TX 77005, USA. E-mail: ajayan@rice.edu
First published on 30th March 2022
Two-dimensional (2D) materials have been shown to be efficient in energy harvesting. Here, we report the use of waste heat to generate electricity via the combined piezoelectric and triboelectric properties of 2D cobalt telluride (CoTe2). The piezo-triboelectric nanogenerator (PTNG) produced an open-circuit voltage of ∼5 V under 1 N force and the effect of temperature in the range of 305–363 K shows a four-fold energy conversion efficiency improvement. The 2D piezo-tribogenerator shows excellent characteristics with a maximum voltage of ∼10 V, fast response time, and high responsivity. Density functional theory was used to gain further insights and validation of the experimental results. Our results could lead to energy harvesting approaches using 2D materials from various thermal sources and dissipating waste heat from electronic devices.
Here, we demonstrate a high-performance and flexible piezo-triboelectric nanogenerator using 2D CoTe2 as an active material. The device can function as a generator to harvest waste heat and collect mechanical energy from deformations at high temperatures. 2D CoTe2 was synthesized using a simple sonication-assisted liquid exfoliation method from its bulk crystal. A detailed materials characterization using spectroscopy and microscopy has been performed. The energy conversion and energy harvesting properties of 2D CoTe2 have been carefully investigated by carrying out experimental and supporting density functional theory (DFT) calculations. The PTNG fabrication and working mechanism based on temperature dependence and mechanical strain harvesting are shown in Fig. 1. The piezoelectric and triboelectric effects are demonstrated by harvesting mechanical strain (tapping and bending) and heat energy dissipated from various sources (electronic devices). DFT and ab initio molecular dynamics simulations have been used to analyze the temperature-dependent behavior of our PTNG cell under strain conditions. Therefore, utilization of these PTNG devices to effectively harvest energy from such overlooked sources can be realized.
Fig. 1 Schematic representation of the utilization of waste heat by generating electricity and the energy harvesting concept of piezo-tribo generators using 2D CoTe2. |
To clearly explore the semiconducting nature of the exfoliated sample, we measured the UV-vis absorption spectra with different size variations and it shows a clear absorption band in the UV (∼360 nm) and infrared regions (∼900–1050 nm) (Fig. 3a). The optical absorption results show a clear change in the absorption spectra as the size is decreased to a few nanometers, and the IR absorption decreased systematically and became flat. Fig. 3b shows the intensity ratio between the IR and UV bands as a function of particle size. It reveals the opening of the bandgap in the exfoliated samples when the dimensions of the material are reduced (d ≤ 4 nm). As the effective thickness approaches a few nanometers, more high-energy photons are absorbed than low-energy ones. This has also been reported in many studies where the material shows metallic behavior in bulk and a semiconductor one at nanosize.33 In Fig. 3c, we present the calculations for the infinite system to confirm that it is metallic but changes to a semiconductor one when the size of its corresponding finite flakes is decreased. We verified this by calculating the molecular orbitals of the systems with three different sizes: 36, 144, and, 324 atoms, respectively, as shown in Fig. 3c. We can see that the systems with 144 and 324 atoms are already metallic within DFT precision, while the one with 36 atoms exhibits a LUMO–HOMO difference (which can be associated with the bandgap) of about ∼0.13 eV. In Fig. S4a and b,† we present the optimized structure of the 2D CoTe2 unit cell, which contains 6 atoms (4Te + 2Co). The corresponding lattice parameters were obtained with the SIESTA (and Quantum Espresso) code corresponding to an orthorhombic unit cell with lattice vectors given by 3.03 and 8.07 Å along the X and Y directions, respectively. We observed five different bond length values: 2.45 and 2.66 Å for Co–Te, 2.98 and 3.17 Å for Te–Te and, 2.90 Å for Co–Co. Thus, there are two different bond types for Co–Te and Te–Te. In Fig. S4c and d,† we present the electronic band structure of layered CoTe2 and the corresponding projected density of states (PDOS) for the optimized structure shown in Fig. S4a and b.† It shows the metallic characteristics where the weight of the 3d atomic orbital is higher than those of others. We also investigated the effects of Co and Te vacancies and the system maintains its metallic behavior. Our simulations included spin polarization effects, and we verified that the system has an antiferromagnetic characteristic because the energy difference between the ferro and anti-ferro configurations is positive,
ΔE = Eferromagnetic = Eanti-ferromagnetic = 0.4 eV | (1) |
Electrochemical impedance spectroscopy (EIS) was performed in order to observe the capacitive behavior of the material at different frequencies. In Fig. 3d, we plotted the negative real impedance (−Z′) vs. positive imaginary impedance (Z′′), as the Nyquist plot of CoTe2. The diagram indicates a very good capacitive nature. The absence of semi-circles and a 45° line reveals a near-zero interfacial impedance and excellent charge transfer capability. The internal resistance obtained from the intercept on the x-axis is only ∼3.21 Ω, manifesting the ultrafast charge conduction inside the electrode. Another important analysis derived from the EIS data is shown in the inset of Fig. 3d. Here we have plotted the negative phase angle vs. frequency. At a certain frequency, the larger the corresponding negative phase angle, the stronger the capacitance characteristic. To distinguish the capacitance and resistance of the device, 45° is the key boundary. For CoTe2, the corresponding negative phase angle appears at ∼120 Hz. Therefore, the exfoliated sample shows capacitive behavior which helps store the generated charge during applied strain. Most significantly, the material shows the permittivity as a function of frequency and it was found that at low frequency, the value of permittivity is ∼7 (Fig. 3e). Therefore, our material shows dielectric behavior, which helps in the generation of charge as a function of strain. Furthermore, we have plotted the specific capacitance as a function of frequency (inset of Fig. 3e). At low frequency, CoTe2 shows a high value of capacitance and it confirms the charge storage capacity of the sample. These results support that the synthesized ultrathin sample can be fruitfully used in energy harvesting applications.
The fabrication of a PTNG using ultrathin CoTe2 has been illustrated in the Experimental section and ESI.†Fig. 4a shows the output voltage as a function of a periodic force (F) of 1 N. The maximum output voltage reaches up to ∼5 V. We calculated the response time of the fabricated PTNG, and it was found to be ∼4 ms (Fig. 4b). The generated output voltage from the PTNG by the repetitive finger (index and middle fingers) press and release motion is shown in the inset of Fig. 4c. By increasing the applied pressure (different fingers), the device produced an output voltage of ∼0.56 V and it increased up to ∼1.6 V (Fig. 4c). This also shows the great flexibility and voltage produced for bending measurements (Fig. 4d). A linear increase in the output voltage was observed which was directly proportional to the bending angle. Additionally, a maximum voltage of ∼1.5 V for the cell was achieved when it was connected over a bridge rectifier to obtain a positive voltage cycle (Fig. S5a and b†). We also compared the output performance to those of bulk CoTe2 samples, and a ten-fold increment of the output voltage under a periodic force of 1 N (Fig. 4e) was observed. These results confirmed that the generated output voltage originated from the ultrathin CoTe2. To know the internal resistance (RL) of the PTNG, the output voltage was recorded as a function of external load resistance (KΩ to MΩ), as shown in Fig. 4f. It is observed that the PTNG has an internal resistance of ∼9 MΩ. The maximum output power density (Pmax), output electrical energy (Eelec), total input mechanical energy (Emech), and energy-conversion efficiency (η) were calculated to know the efficiency of the fabricated PTNG34 and the detailed calculations are given in the ESI.† A Pmax value of ∼2.56 mW m−2 was obtained for the PTNG at an RL of 9 MΩ. The Eelec value was calculated to be ∼4.3 × 10−8 J at a maximum output voltage of ∼5 V, whereas the Emech value for an applied force of 1 N was 5.24 × 10−7 J. Therefore, the maximum conversion efficiency of the PTNG was found to be ∼8.2%. The current density was measured under the application of 1 N load on the PTNG cell at RT and found to be ∼60 nA m−2 (as seen in Fig. S6†). The proposed flexible PTNG will also provide effective electrical energy for real-world applications such as powering a blue LED in a sustainable manner (Fig. S7†). The voltage outputs of the LED with minimum glow were due to the low current output as mentioned above. The fabricated PTNG can provide a sustainable voltage with a current to power the connected LEDs for a longer duration when connected in series.
In order to investigate the elastic properties of ultrathin CoTe2, we replicated the unit cell of 2 × 2 × 1 along with the X and Y directions, building one orthorhombic supercell composed of 36 atoms. In Fig. 5a, we present the stress–strain curve for ultrathin CoTe2. We observed a significant anisotropy for the X and Y directions (see also Fig. 5b). The Young's modulus was calculated considering the linear regime for the strain value until 1%. The estimated Young's modulus values (from the linear regime of the stress/strain curve) were 163.73 and 72.13 GPa for the X and Y directions, respectively. We also analyzed the charge distortion effects in the elastic regime to obtain the piezoelectric coefficients using the expression:
dijl = Pi/σjl, | (2) |
The charge generation mechanism of the PTNG under external stress is shown in Fig. S8 (a and b†), which is based on the piezoelectricity generated by ultrathin CoTe2 and the triboelectricity produced from the contact electrification process between Kapton and electrode materials (as discussed above, the dielectric properties of 2D CoTe2).We have schematically presented the working principle of the as-fabricated PTNG. In order to explain the voltage generation mechanism for the exfoliated CoTe2 based PTNG, we used a similar theoretical model to that reported by Yousry et al.35 Fig. S8a (aI and II)† shows the piezoelectric model, wherein mechanical stress is applied on the material with a non-centrosymmetric nature, and the centers of the positive and negative charges shift, which is also confirmed from DFT calculations, as discussed above. The charges generally move from one electrode to another until a point is reached where the potential difference between the electrodes is in a balanced (equilibrium) condition (Fig. S8a (I and II)†). The pressing action leads to stretching in the device, creating a piezoelectric polarization. The piezoelectric effect decreases when the external force is released, and the electrons flow back to the top and neutralize the positive potential. When the two surfaces are not in full contact, the triboelectric effect becomes prominent and produces a substantial electrical output (Fig. S8a (III)†). Simultaneously, as a result of this, polarization-free charges flow through the external circuit, maintaining the balance in the potential. Therefore, when an external pressure is applied, the dipole momentum and overall polarization in the pressed state are changed from the initial state; thus an electric current (IP) is generated as a response to the induced applied pressure (Fig. S8 (IV and V)†). After that, when the pressure is released, the opposite electrical potential change is induced, and the structure returns to its original state due to the triboelectric effect. We further analyzed the contribution of piezo and triboelectric modes in our PTNG. In Fig. S8(b),† we present the output performance of the triboelectric, piezoelectric, and hybrid energy harvesters under the same experimental conditions. The fabricated nanogenerator shows a maximum open-circuit voltage output of ∼2.4 V in triboelectric mode while it exhibits a voltage output of ∼0.6 V in piezoelectric mode, both under finger tapping conditions. However, for the hybrid PTNG device, the maximum open-circuit voltage output can reach up to ∼4 V (finger tapping), taking advantage of the synergistic combination of the two effects, which was much higher than the individual piezoelectric and triboelectric outputs. The output electrical potential depends on the deformation of the lattice structure of CoTe2, and it is due to the applied stress, as charges appear on the opposite faces in the material. The generated charges flow through the contact process between electrodes. The exfoliated CoTe2 suffices as a dielectric material in the contact process, which helps in charge transfer. Appropriate connections of the external electric circuit can generate an enhanced and reversible output signal with the electrons moving in the same direction. We also verified that the layer-dependent charge generation effect based on the Mulliken population is negligible, as shown in Fig. S9 in the ESI.† The durability demonstrated for the fabricated device under constant repetitive pressure showed stable behavior over a prolonged period (Fig. S10†). A fatigue test was carried out for the fabricated PTNG. The cell response is continuous and was studied for up to 200 cycles as seen in Fig. S11.† We observed that there is a drop in voltage from ∼5 V to ∼4.4 V, which indicates the good stability of the device. Using first-principles calculations, the piezo-coefficient of 2D CoTe2 is found to be 12.19 pm V−1. Compared to previous reports (see Fig. S12†), 2D CoTe2 produces a significant output response.
Telluride-based structures are widely researched for their thermoelectric behavior. Therefore, ultrathin CoTe2 can also be used as an energy harvesting device from different thermal sources. The utilization of waste heat dissipated for the generation of energy can account for waste heat management. Here we have measured the temperature-dependent behavior of 2D CoTe2. The temperature of the fabricated device was monitored using a laser thermometer. Meanwhile, the device output voltage at each changing temperature was recorded using a digital oscilloscope. With the slightest increase in temperature (in the range from 304 K to 373 K), the output voltage linearly increased (Fig. 6a) and then reached saturation. After the removal of the heat source, the maximum voltage attained is reduced to a minimum gradually. Fig. 6b shows the average values of the output voltage at increasing and decreasing temperatures. It was observed that the output voltage increased almost linearly and is directly proportional to the temperature change (Fig. 6c). The variation of the polarization of surface charges on the application of heat tends to generate significant thermionic vibrations; these vibrations are responsible for producing a drastic variation in the voltage. These produced charges are, in turn, responsible for the temperature-dependent piezoelectric effect that is observed in 2D CoTe2. An open-circuit voltage of ∼10 V was obtained from the PTNG, under a temperature variation from 305 K up to 363 K. We calculated the variation of output electrical energy for increased temperature values at a constant input mechanical energy of ∼5.24 × 10−7 J. It was observed that for the temperature increase from 305 K to 320 K, there was a gradual increase in efficiency (Fig. 6d and Table S1†). After that, the rate of efficiency increase tempered down. The CoTe2 based PTNG device shows a four-fold increment in conversion efficiency at the maximum temperature compared to RT. To confirm the temperature-induced strain in the atomically thin CoTe2, laser power-dependent Raman measurements were performed. It shows gradual shifting with increasing temperature (Fig. S13†). The temperature induced strain changes the lattice structure by absorbing the heat that creates surface charge and increases the output voltage. We would like to stress that, although the XPS results indicate the presence of Te–O bonds formed on the sample surfaces, these bonds are not very stable and break and/or dissociate under thermal treatments; thus the resulting surfaces are oxide-free. Theoretical studies were also performed in the same temperature region to analyze the origin of the higher energy conversion efficiency of the PTNG cell, discussed later. Therefore, this temperature range was feasible for low-grade-temperature device operation under non-ambient conditions, indicating internal overheating issues.
The electronic devices used in our day-to-day activities radiate temperatures in the range of ∼300 K–325 K on constant usage. One such example is a laptop; we demonstrate that energy can be harvested from overheating of these devices. The average operating temperature of a laptop is below 313 K and above this temperature, internal problems of the device appear. In order to harness this energy, the fabricated PTNG was attached under a laptop (keyboard and touchpad), where most of the heat flow (overheating) is obtained during the working condition. To evaluate the performance of the fabricated PTNG cell, the generated output voltage was measured directly during the working condition of the laptop, as shown in the inset of Fig. 6f. The possibility to harvest waste-heat energy dissipated from the laptop by installing the described system on the source device was demonstrated successfully. The PTNG shows increased output voltage as a function of laptop temperature. A linear change is observed in the generated relative output voltage (ΔV/V0), with respect to increasing temperature (Fig. 6e). This indicates the generation of voltage in the PTNG cell, which is due to the temperature-dependent piezoelectric effect under the given experimental conditions. The responsivity (RS) (measurement of the input–output gain of a detector system) of a device is also a key parameter, and it was calculated using, Rs = (ΔV/V0)/T, where V0 is the initial output voltage at RT, and ΔV is the voltage change at temperature T. It is clearly noted that the device was responsive to temperature variations and exhibited a positive linear relationship in the temperature range of RT to 321 K with a response of 2.013 K−1. Therefore, the results confirm the application of our PTNG cell in scavenging waste heat dissipated during overheating of a laptop. Bulk materials generally harness large amounts of heat and temperatures; usually, small amounts of heat dissipated go unnoticed. Furthermore, the fabricated device was also sensitive as a function of laptop weight despite being a temperature sensing unit. Here, we have taken a different laptop and monitored their output voltage as a function of weight and heat generation (Fig. 6f). The calculated responsivity as a function of applied load (F = mg and m = weight of the laptop) also shows a linear behavior. Therefore, the fabricated lightweight and highly-sensitive PTNG cell can be used for monitoring the temperature of electronic gadgets.
The coupling of the piezoelectric and triboelectric properties of CoTe2 creates a polarized electric field with charge separations observed in a few layers of 2D CoTe2, which is a result of the time-dependent change in temperature.6,36 To explore the possible origin of this output voltage, we have used theoretical calculations on the application of strain and temperature. In Fig. 6g and h, we present the 3D electron density plot (upper) from a top view and the electron density map (lower) from a side view. For a small strain of 1% within the linear regime and a temperature of 320 K, we observed the charge distortion responsible for the changes in the electrical dipole value. In Table 1 we present the piezoelectric coefficients considering the linear elastic regime for a strain value of 1%. The results were obtained using optimized geometries from DFT calculations at T = 0 K and ab initio molecular dynamics for finite temperatures. We observed the same trend for the piezoelectric coefficient discussed in the present manuscript only for stress along the X-direction, as seen in Table 1. These cases occur as the coefficient values change when the temperature increases. With these above coefficients, we were further able to calculate the current density for various temperature conditions (details in the ESI†).37 Therefore, the current density (Isc) was calculated, and maximum current densities of ∼51 nA m−2 for 300 K and ∼67 nA m−2 for 320 K were obtained. The above theoretical simulation is consistent with experimental observation.
Coefficients (pc N−1) | T = 0 K | T = 300 K | T = 320 K |
---|---|---|---|
d 11 (x) | −0.62 | 7.32 | 8.64 |
d 12 | −4.75 | 57.8 | 41.3 |
d 21 | −7.12 | 82.3 | 31.5 |
d 22 (y) | 3.31 | 17.05 | 19.99 |
We would like to remark that while our theoretical results for infinite 2D CoTe2 shows metallic behavior, the experimental data show a semiconductor one after size reduction. The experimental samples considered in the manuscript are very small, because of the quantum confinement effects, and we can expect a transition from a metal to a semiconductor when the system size is decreased, and this is observed in our calculations (see Fig. 3c). The structural models in Fig. 3c did not consider defects and/or strain effects, and both contribute towards increasing the bandgap opening. Eqn (2) that we used to obtain the piezoelectric coefficients is valid only for semiconductors. The piezoelectric coefficients calculated along the X and Y directions for the system composed of 36 atoms are presented in Table 1. Interestingly, some coefficients are negative, but no clear pattern as a function of temperature is observed. This behavior has been also observed for other cobalt telluride families. We presented the results for the infinite system only to obtain some insights into the charge variation induced by the strain regime.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2nr00132b |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2022 |