Qiang
He‡
ac,
Teck Lip Dexter
Tam‡
*b,
Xue Qi
Koh
c,
Nguk Neng
Tham
c,
Hong
Meng
*de,
Wei
Huang
*a and
Jianwei
Xu
*bcf
aFrontiers Science Center for Flexible Electronics (FSCFE), Institute of Flexible Electronics (IFE), Northwestern Polytechnical University (NPU), 127 West Youyi Road, Xian 710072, P. R. China. E-mail: iamwhuang@nwpu.edu.cn
bInstitute of Sustainability for Chemical, Engineering and Environment (ISCE2), Agency of Science, Technology and Research (A*STAR), 1 Pesek Road, Singapore 627833, Singapore. E-mail: dexter-tam@isce2.a-star.edu.sg; jw-xu@imre.a-star.edu.sg
cInstitute of Materials Research and Engineering (IMRE), Agency of Science, Technology and Research (A*STAR), 2 Fusionopolis Way, Innovis, Singapore 138634, Singapore
dSchool of Electronics and Information, Northwestern Polytechnical University, 127 West Youyi Road, Xi'an, 710072, P. R. China. E-mail: menghong@nwpu.edu.cn
eSchool of Advanced Materials, Peking University Shenzhen Graduate School, 2199 Lishui Road, Shenzhen, 518055, P. R. China
fDepartment of Chemistry, National University of Singapore, 3 Science Drive 3, Singapore 117543, Singapore
First published on 25th November 2022
Ambipolar bulk heterojunction polymer thermoelectrics were demonstrated for the first time using a classic n-type poly(benzimidazobenzophenanthrolinedione) ladder polymer (BBL) and a new p-type 6H-pyrrolo[3,2-b:4,5-b′]bis[1,4]benzothiazine ladder polymer (PBBTL). The blend films were solution-processed and can be p-/n-doped using a FeCl3/benzyl viologen radical cation (BV˙+). Well-balanced p- and n-doped conductivities can be achieved at a 1:1 blend ratio, and a full thermoelectric device consisting of a pair of p- and n-channel legs from a single layer blend film was fabricated. The full device exhibited a power density of 0.27 μW cm−2 at ΔT = 3 K, operating at a voltage of 0.93 mV and a current density of 0.29 mA cm−2. The simplification of the device fabrication process using such ambipolar bulk heterojunction blend films may help the realization of polymer micro-thermoelectric devices in the future.
In thermoelectric devices, the use of both p- and n-channel legs (instead of using metals to replace one of the elements) will result in more efficient device performance due to the larger differences in the thermal voltage (both p- and n-channel contributions) and larger thermal resistance (metals have much higher thermal conductivity). Micro-thermoelectric devices, with their smaller dimensions than their macroscopic counterparts, can provide a faster response, higher temperature resolution and greater power density.14 These properties, together with their inherent smaller size, make micro-thermoelectric devices very suitable for wearable devices. While both p- and n-channel legs of large devices can be screen-printed with ease, micro-thermoelectric devices with dimensions of a few tens of microns will face technical difficulties even with inkjet printing.15 The use of both p- and n-channel legs in micro-thermoelectric devices is even more crucial due to the need to maintain a thermal gradient over small distances. One way to overcome the need to separately print the micron-size p- and n-channel legs using p- and n-type polymers is to use an ambipolar polymer that can be either p- or n-doped to form the respective p- and n-channel legs in selected areas using fabrication techniques such as thermal evaporation of p- and n-dopants via shadow-masks. Such an approach allows the fabrication of small devices with dimensions down to 10 μm,16 without the need of micropatterning the p- and n-channel legs separately, while still taking advantage of the solution-processability of the polymer.
Easier said than done, there is only one very recent report on an ambipolar conjugated polymer doped using p- and n-dopants for OTEs, achieving maximum p- and n-doped conductivities of 1.6 and 2.75 × 10−5 S cm−1, respectively, and maximum power factors of ∼1 and ∼0.003 μW m−1 K−2 for p- and n-doping, respectively.20 The lack of ambipolar conjugated polymers that can be both p- and n-doped is due to the frontier molecular orbital energy levels, i.e., the ambipolar conjugated polymer must have the suitable highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels that can be oxidized and reduced by the p- and n-dopants, respectively. Such a criterion will require the careful design of conjugated polymers, and the resulting polymers should also have balanced hole and electron transport properties. Alternatively, a blend of p- and n-type conjugated polymers may also accomplish the same result. Such a blend of p- and n-type conjugated polymers is analogous to the all-polymer bulk heterojunction (BHJ) OPVs that have shown a dramatic power efficiency improvement over the past decade,21 and more recently in high-performance BHJ OECTs.19 Interestingly, the use of p- and n-type conjugated polymer BHJ has not been demonstrated in OTEs, albeit the stimulating work on density of states engineering using pure p- or n-type blends reported by the groups of Katz, Campoy-Quiles and Kemerink.22–26 Here, we describe for the first time the use of p- and n-type ladder conjugated polymer (PBBTL as p-type and BBL as n-type) blends that can be either p- or n-doped for OTEs (Fig. 1). The thermoelectric properties of individual p-doped PBBTL and n-doped BBL have been investigated by us and other research groups previously.18,27–31 The thermoelectric performance of the BHJ film can be tuned not only by varying the doping time, but also by varying the p- and n-type polymer ratios. A full thermoelectric device based on a pair of p- and n-channel legs was fabricated using a 1:1 BHJ blend film doped at 1 s immersion for both p- and n-doping, respectively, demonstrating the simplicity of such a device architecture. Under such conditions, power factors of 2.18 ± 0.17 and 1.12 ± 0.19 μW m−1 K−2 for p- and n-doped BHJ films can be achieved, respectively, and the full device exhibited a power density of 0.27 μW cm−2 at ΔT = 3 K, operating at a voltage of 0.93 mV and a current density of 0.29 mA cm−2. While the devices fabricated here are not in the micron dimension, we foresee that the simplified processing arising from using a single ambipolar blend film will enable the realization of polymer micro-thermoelectrics in the future.
Fig. 1 (a) Chemical structures of the p- and n-type ladder conjugated polymers and (b) p- and n-dopants used in this work. (c) Energy levels of the individual components from ref. 17–19. (d) Schematic of a thin film thermoelectric module. |
The Seebeck coefficient decreased with longer immersion times for both p- and n-doping for the most part (Fig. 2c and d). The positive and negative Seebeck coefficients obtained after FeCl3- and BV˙+-doping prove their p- and n-type behaviours, making this the first demonstration of an ambipolar BHJ OTE device. For all blend ratios and p-/n-doping, the power factor, a product of conductivity and the square of the Seebeck coefficient, generally reached a maximum at around 5–10 s of immersion (Fig. 2e and f). A maximum power factor of 3.03 ± 0.20 μW m−1 K−2 was achieved for a p-doped PBBTL:BBL 2:1 ratio at 10 s immersion and 3.42 ± 0.43 μW m−1 K−2 for a n-doped PBBTL:BBL 1:2 ratio at 5 s immersion. Not surprisingly, these values are again lower as compared to their p- and n-doped neat films (PBBTL, 7.73 ± 0.24 μW m−1 K−2; BBL, 4.85 ± 0.46 μW m−1 K−2), respectively. It is worth noting that the p- and n-doped conductivities achieved at a 1:1 blend ratio were quite well-balanced, with considerably higher power factors as compared to the recently reported ambipolar conjugated polymer.20
Taking advantage of the ambipolarity of the blend PBBTL:BBL film, we fabricate a full thermoelectric device consisting of single p- and n-channel legs using the PBBTL:BBL 1:1 ratio blend film as a proof of concept (Fig. 3a). The p- and n-channels were selectively p- and n-doped via immersion in their respective dopant solutions. To maximize an output power in thermoelectric devices, resistance matching of the p- and n-channel legs must be accounted. Taking this into consideration, the p- and n-channel legs were subjected to 1 s immersion each in the FeCl3 and BV˙+ solutions. After thermal annealing, the total resistance of the full device was about 120 kΩ. The current density and the power density plot of the full device subjected to a ΔT of 3 K were recorded (Fig. 3b). The open-circuit voltage of the full device could reach a value greater than 1.8 mV and the short-circuit current could reach almost 0.6 mA cm−2. The resulting maximum output power density of the full device reached 0.27 μW cm−2, operating at a voltage of 0.93 mV and a current density of 0.29 mA cm−2.
To understand the microstructure of the blend films, we first measured their GIWAXS (Fig. 4a and b). The pure PBBTL has a predominant face-on stacking with lamellar and π–π stacking distances of ∼5.5 Å (q = ∼1.15 Å−1) and ∼3.4 Å (q = ∼1.85 Å−1), respectively. BBL, on the other hand, has an edge-on stacking with lamellar and π–π stacking distances of ∼7.9 Å (q = ∼0.80 Å−1) and ∼3.4 Å (q = ∼1.82 Å−1), respectively. By considering the full width at half maximum of the π–π stacking peak for both pure polymers, one can tell that BBL is a more crystalline polymer than PBBTL. Going from the pure PBBTL to the PBBTL:BBL blend and to the pure BBL, we observed a gradual decrease of the PBBTL π–π stacking signal in the out-of-plane direction, with an increase of the BBL lamellar stacking signal. In the in-plane direction, a gradual increase of the BBL π–π stacking signal was also observed with an increasing BBL content. It also appears that BBL can better disrupt the microstructure of PBBTL but not vice versa, probably due to the higher crystallinity of the former. In UV-Vis absorbance and FTIR transmittance spectra, similar observations in terms of signal contribution can be made when the PBBTL:BBL ratio is changed (Fig. 4c and d). In these spectra, no significant shift could be observed, indicating no ground state electron transfer between the two polymers.
By using an atomic force microscopy-infrared (AFM-IR) technique, we mapped the morphology of the PBBTL:BBL blend films by scanning the samples at 1700 cm−1, which is characterized by CO stretching that is present only in BBL (Fig. 5). First, all the blend ratios gave smooth films (Fig. 5a–c) with a low RMS value (∼10 nm) and this was supported by the scanning electron microscope (SEM) images (Fig. S1, ESI†). Smooth films are critical to provide reliable results for such characterization. Second, with a higher BBL content, i.e., a PBBTL:BBL ratio from 2:1 to 1:2, the blend films showed a stronger CO stretching signal (Fig. 5d–f). Third, the AFM height images of the blend films revealed porous nanostructures with dimensions of ∼50 nm, similar to those of the pure PBBTL and BBL films, as well as other ladder-type conjugated polymers.18,27,32 We attribute this to the film formation process, where the wet polymer films consisting of methanesulfonic acid (MSA)-swelled polymer networks shrink to form the porous nanostructures when the MSA leaches out during immersion in the anti-solvent. Thus, we believe that this should be universal to all (ladder) polymers processed in this manner. Fourth, the AFM-IR images also showed features with dimensions of ∼50 nm. While the positions of these features do not seem to match that of the AFM height images, they suggest that both PBBTL and BBL can be well-mixed on the nanoscale. In addition, the observed conductivities over a large channel length (0.25 cm) after p- and n-doping suggest the formation of percolation pathways that transverse well beyond the micron scale even though both PBBTL and BBL are well-mixed in the nanoscale. Lastly, to confirm the results, we performed selected area IR scans with different CO stretching intensities for the 1:1 film (Fig. 5g). We indeed observed higher CO stretching intensity ratios with respect to the CN stretching in areas with stronger CO signals, where the CN stretching is mainly contributed by PBBTL.
The UV-Vis-NIR absorption spectra of the p- and n-doped polymer blend films are shown in Fig. 6. The spectra of the p-doped blends are normalized at the λmax for BBL at around 585 nm since BBL cannot be p-doped using FeCl3. Likewise, the spectra of the n-doped blends are normalized at λmax for PBBTL at around 747 nm since PBBTL cannot be n-doped using BV˙+. For these films, blends with higher p- or n-type polymer ratios resulted in a larger change in the UV-Vis-NIR absorption upon p- or n-doping, respectively. A significant change in the UV-Vis-NIR absorption was observed for all immersion times for p-doping in FeCl3 solution. In comparison with the pure PBBTL film, it appears that the addition of BBL in the blend films enables more efficient FeCl3 p-doping. However, there was no significant change in the UV-Vis-NIR absorption of the blend films after 1 s immersion in the BV˙+ n-dopant solution. And in comparison with the pure BBL film, it appears that the addition of PBBTL in the blend films impedes the BV˙+ n-doping process.
While the polymers and dopants used in this work are not new and the device performances of the BHJ polymer thermoelectrics are still relatively low, our work here demonstrates the possibility of using the BHJ donor–acceptor blend that can be either p- or n-doped. This is akin to the early stages of BHJ OPV research where known materials were used and power conversion efficiencies were low. However, since then, new science, new materials, new concepts, and research areas have sprung out from OPV research, with their power conversion efficiencies reaching almost 20% now, almost on par with silicon and perovskite solar cell technologies. Moreover, the science of BHJ polymer blend thermoelectrics is fundamentally different from that of BHJ OPV as the former involves chemical doping, thermoelectric effect, and conducting materials while the latter deals with the photovoltaic effect and semiconducting materials. There is much more to understand in this new system, and can benefit or give rise to new concepts/applications in the future.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2tc03684c |
‡ Both authors contribute equally to the manuscript. |
This journal is © The Royal Society of Chemistry 2023 |