CHAPTER 5 Results and discussion
5.2 Future strategy
As described in Section 1.2.3., the efficiency of thermoelectric energy converters is limited by the material thermoelectric figure of merit (𝒁𝑻̅). The recent advances in 𝒁𝑻̅ depends on the fundamental properties of thermoelectric material, and further improvement of efficiency of 𝒁𝑻̅ cannot be expected unless innovative new classes of materials are explored.
In the case of inorganic materials, further improvement of 𝒁𝑻̅ has been achieved mainly based on two strategies. One is the use of nanostructures to limit the phonon heat conduction and decrease the lattice thermal conductivity (𝜿𝐩𝐡). The other is band structure engineering to enhance the local density of states in conduction or valence bands and increase 𝑺 while maintaining the 𝝈.
For organic thermoelectric materials, which suffer from insufficient 𝑺 and 𝝈 , nanostructuring is not attractive because the thermal conductivity (𝜿) cannot be reduced below the amorphous limit and complicated structures are required as shown in Fig. 5.1.
Figure 5.1: Super lattice structure used for Bi2Te3/Sb2Te3 to obtain high 𝒁𝑻̅ an lower 𝜿𝐩𝐡. In contrast, enhancing 𝑺 through a distortion of the electronic density of states should be critical; thus, research on organic thermoelectric materials has been focus on this route.
The basis for the enhancement of 𝑺 can be explained based on the Mahan-Sofo theory,125 which suggests the study of systems in which there is a local increase in the density of states (DOS) 𝑵(𝑬) over a narrow energy range, as shown schematically in Fig. 5.2.
Such a situation can occur when the valence or conduction band of the host semiconductor resonates with one energy level of a localized material in a semiconductor matrix.125 The effect of this local increase in DOS on 𝑺 is given by the Mott expression (Eq.
5.1). Here, 𝑺 depends on the energy derivative of the energy-dependent electrical conductivity 𝝈(𝑬) = 𝒒𝝁(𝑬)𝒏(𝑬) taken at the Fermi energy (𝑬𝐅), where 𝒏(𝑬) is the carrier density defined by 𝒏(𝑬) = 𝑵(𝑬)𝑭(𝑬), 𝑭(𝑬) is the Fermi function, 𝒒 is the carrier charge, and 𝝁(𝑬) is the mobility as shown in Eq. 5.2.
Bi2Te3/Sb2Te3 Multilayers
κ ＝ κ ⊥
2D Superlattice
83
Figure 5.2: Schematic representation of the density of electron states of the valence band of pure PbTe (dashed line) contrasted to that of Tl-PbTe, which has a Tl-related level increase the density of states. The figure of merit 𝒁𝑻̅ increases when the slope of 𝑵(𝑬) at the Fermi energy (𝑬𝐅) of the holes increases because of an enhancement of 𝑺 while maintaining 𝝈.
𝑺
𝑻= 𝝅𝟐𝒌𝑩𝟐
𝟑𝒒 × [𝝏𝐥𝐧𝝈(𝑬)
𝝏𝑬 ]
𝑬=𝑬𝐅
Mott expression (Equation 5.1) =𝝅𝟐𝒌𝑩𝟐
𝟑𝒒 × [ 𝟏
𝒏(𝑬)
𝝏𝐥𝐧𝒏(𝑬)
𝝏𝑬 + 𝟏
𝝁(𝑬)
𝝏𝐥𝐧𝝁(𝑬)
𝝏𝑬 ]
𝑬=𝑬𝐅
𝝈(𝑬) = 𝒒𝝁(𝑬)𝒏(𝑬) = 𝒒𝝁(𝑬)𝑵(𝑬)𝑭(𝑬) (Equation 5.2)
Equation 5.1 shows that there are two mechanisms that can increase 𝑺. One is an increased energy dependence of 𝝁(𝑬), for instance by a scattering mechanism that strongly depends on the energy of the charge carriers. The other one is an increased energy dependence of 𝒏(𝑬), for instance by a local increase in 𝑵(𝑬).
The latter is the basis of the Mahan-Sofo theory, provided that 𝑬𝐅 of the semiconductor aligns properly in the range of the excess DOS in the band (Fig. 5.2). The concept can also be expressed in terms of effective mass 𝒎∗, as shown for degenerate semiconductors:147
𝑺
𝑻=𝟖𝝅𝟐𝒌𝑩𝟐𝒎∗
𝟑𝒒𝒉𝟐 (𝝅
𝟑𝒏)𝟐/𝟑 (Equation 5.3) with
DOS of EV, N(E)
Energy,E
EF
Small S Large S
Pure PbTe
Tl-PbTe
DOS of valence band, N(E)
84
𝑵(𝑬) =(𝒎∗)𝟑/𝟐
ℏ𝟑𝝅𝟐 √𝟐𝑬. (Equation 5.4)
Because 𝒁𝑻̅ also depends on 𝒏 via 𝝈, the value of 𝑬𝐅 that maximizes 𝒁𝑻̅ is somewhat different from the value that maximizes 𝑺 and 𝒎∗.147 Such band structure engineering results in a remarkable increase of 𝒁𝑻̅ in inorganic thermoelectric material systems. Use of this new physical principle could further enhance 𝒁𝑻̅ for organic thermoelectric materials and enable more widespread use of organic thermoelectric generators.
To use this new physical principle, precise band structure engineering is necessary.
However, organic solids having various complex structures and poor crystallinity; thus, calculating the band structure of organic solids by computational science is much more difficult than for inorganic materials. Furthermore, the band structure becomes more complicated if dopants are inserted into the organic solid and mutual interaction occurs between the organic solid and dopants as shown in Fig. 5.3.
Figure 5.3: Schematic representation of the density of electronic states of (a) a single molecule having a single energy state, (b) an amorphous organic solid having a widely dispersed DOS, (c) an organic crystal having a relatively narrowly dispersed DOS, and (d) an amorphous organic solid with a dopant exhibiting an overlap of the DOS of the amorphous organic solid and dopant.
Thus, a method that enables direct observation of electronic structure is necessary for band structure engineering, and ultraviolet photoelectron spectroscopy (UPS) and inverse
(c)
(a) (b) (d)
85
photoelectron spectroscopy (IPES) are the most effective methods, as shown in Fig. 5.4. UPS provides information on the occupied DOS near and at 𝑬𝐅 as discussed in Section 3.3.5. In contrast, IPES provides information on the unoccupied-DOS near and at 𝑬𝐅. Generally, the unoccupied DOS is indirectly estimated from UPS results and optical absorption spectra, but IPES can observe the unoccupied DOS directly.
The energy of photons emitted when electrons with a constant energy (𝑬𝐢) from an electron beam incident on a substance relax to a lower energy unoccupied state (𝑬𝐟) is given by the conservation of energy as 𝑬𝐢 = 𝑬𝐟+ 𝒉𝝂. By measuring 𝑬𝐢 and 𝒉𝝂, information about the unoccupied states of a solid can be found through the inverse process of UPS as shown in Fig. 5.4a. Electronic states of the substance can be express by combining UPS and IPES as described in Fig. 5.4b. Furthermore, transitions from the occupied DOS to the unoccupied DOS have to match with the optical absorption spectra.
Figure 5.4: (a) The principles of UPS and IPES. (b) Example of actual spectra from UPS and IPES. EA is electron affinity, WF is work function, IE is ionization energy.
(b)
(a)
86
After investigating the band structure of organic solids with dopants using UPS-IPES, band structure engineering becomes possible by searching for and applying various candidates to enhance the local DOS. Therefore, future research on organic thermoelectric materials should strategically use UPS-IPES and focus not only on organic materials and dopants but also on DOS enhancers.
87
Appendix A
A-1 Fabrication of P3HT nano- and micro-films
Figure A-1: Fabrication process of P3HT nano- and micro-films by (a) spin coating and (b) drop casting along with pictures ((c) and (d), respectively) of the fabricated films.
Figure A-2: AFM topography images of the P3HT (a) nano- and (b) micro-films. (c) Microscope image of micro-film.
Uniformity of the active layer thickness over the entire substrate is very important when fabricating a module in order to make uniform contact with the heat source. Furthermore, etching speed is dependent on film thickness, so non-uniform thickness will affect the patterning of the active layer. While spin coating is a familiar and very easy method for fabricating smooth nano-films whereas, make smooth micro-films with methods such as screen printing and drop-casting can be very difficult. Drying in a solvent vapor environment, introduced in this study, allows smooth micro-film fabrication without the formation of a highly convex surface.
Checking the surface uniformity of the samples, the P3HT nano- and micro-films appear flat when observed by the naked eye as shown in Fig. S1, and atomic force microscopy (AFM, JSPM-5400, JEOL) imaging below the 3 × 3 μm2 scale (see Figs. A-2a and b) shows that the surfaces of the films are amorphous and smooth (root mean square roughness was below 2 nm). However, optical microscopy images (see Fig. A-2c) show rough grains of several hundred micrometers for the micro-films at scales between the unmagnified optical images and the AFM images.
(a) (c)
(b) (d)
(a) (b)
topo500 nm
0.00 nm
13.2 nm
0.00 nm 12.5 nm
topo
500 nm
(c)
100 μm
88
A-2 Patterning by photo-etching process
Figure A-3: Patterning steps for individual thermoelectric devices using the photo-etching process. (a) Four-bottom electrodes, each 23 mm × 3 mm and 3 mm apart, are vacuum deposited onto the substrate (2.5 × 2.5 cm2). (b) P3HT films on the patterned electrodes/PET substrate are fabricated using spin coating or drop casting. (c) Dry film-type photoresist is laminated onto the organic layer. (d) The active area (2 mm × 23 mm) is protected by patterning the photoresist. (e) The exposed parts of the P3HT are removed by immersion in an etching solution. (f) The photoresist is removed to expose the patterned active area.
After the fabrication of the P3HT nano- and micro-films, unnecessary parts were removed by photo-etching. Because the adhesion and wettability between most conductive polymers and solution-type photoresists are very poor, dry film-type photoresist (TMMF-S2045 TOKYO OHKA KOGYO Co., LTD) was used in this study. The photo-etching process is described in Fig.
A-3, and the photolithography process (lamination, pre-baking, masking, UV exposure, developing, post-baking, and removal) was done following the process described in the manual for TMMF.
The etching process for the P3HT layer, which is new in this study, was developed based on the solubility of the polymer, which depends strongly on the temperature and the polarity of the solvent used. An etching-solution composed of a main etchant (oDCB, Wako) and a polar coordinator (ethylene glycol monobutyl ether, TCI), in a volume ratio of 5:1 was used.
When etching was performed, the patterned photoresist was used to protect the active layers while the unneeded parts were removed by etching, as shown in Fig. A-3d. Samples were immersed in the etching solution at 55-60°C for 10-60 min.
(a) (b) (c)
(e)
(d) (f)
Electrodes
Substrate
Photoresist P3HT
Active layer
Protect layer
89
A-3 Changes of doped state of P3HT and stability of electrical conductivity in air condition
Figure A-4: Route for the de-doping process.
Though the ferric ion showed good doping properties, heavily doped conductive polymers can be de-doped by side reactions with moisture in ambient conditions. The de-doping is mainly thought to be caused by the instability of the counter anion and can be a source of degradation of the electrical characteristics of devices. In particular, the intermediate ferrous ion produced by the side reaction has high reactivity in air and is thought to be critical to the de-doping process described in Fig. A-4. This may explain the poor lifetime in the nano-film P3HT device with Fe3+-tos3·6H2O doping.
Bipolaron
Unstable counter anion De-doping
Polaron
90
A-4 Thermoelectric performance of recycled P3HT
Figure A-5: Voltage and power output as a function of temperature gradient along with performance parameters for single micro-film thermoelectric devices with active layers of pristine and recycled P3HT doped with Fe3+-tos3·6H2O (50 mM in acetonitrile solution).
Table A-1: Key characteristics of the single thermoelectric devices with active layers of pristine and recycled P3HT doped with Fe3+-tos3·6H2O.
Condition Film thickness (μm)
Fe3+-tos·6H2O (mM)
Rp
(Ω)
σ (S cm−1)
S (μV K−1)
σS2 (μW m−1K−2)
New 42.0 51.0 5.94 61.2 79.3 38.5
Recycled 41.5 52.5 5.92 64.9 76.2 37.7
After the etching process, the removed P3HT was collected and purified by filtering the etching solution and washing the residue with several solvents (pure water, methanol, acetone and ethyl acetate) to remove the insoluble residual photoresist and impurities. After drying in a vacuum oven, the recycled P3HT was used to fabricate devices using the same procedures as pristine P3HT. Device fabrication for both devices was the same as for the other devices in Chapter 2 except that annealing was at 230 °C for 1 h and the metal electrodes were 100-nm-thick gold deposited on top of the organic layer. Figure A-5 shows the thermoelectric voltage and power output of the single micro-film OTEGs fabricated with pristine and recycled P3HT, and the performance of both devices was found to be similar.
However, the power factors of the two single OTEGs using top-electrodes were slightly higher than for the single OTEGs using bottom-electrodes described in Fig. 2.3a. Improved electrode contact along with coverage of the polymer edge by the top-deposited gold is expected to contribute to the improved power factors by improving the electrical conductivity and temperature gradient.
0 3 6 9 12 15 18
0.0 0.3 0.6 0.9 1.2
TCOLD= 25 °C New
Recycled
Thermoelectric voltage, ΔV TE (mV)
Temperature gradiant, ΔT (K)
0 10 20 30 40 50 60 70 80
NEW Recycled
Maximum power output, P Max (nW) Tos・6H2O (mM)
91
A-5 Thermoelectric measurement system for flexible thermoelectric modules
Figure A-6: Diagram of the measurement scheme for the organic thermoelectric modules.
The heater and cooler provide a temperature gradient across the module. The temperature is measured by Pt-RTSs. Electrical conductivity is measured by two terminal sensing.
Figure A-6 shows the measurement scheme for evaluating the thermoelectric properties of the flexible organic thermoelectric modules. Contact between the substrate and cold-side copper block is made using an insulating diamond paste (12.56 W m−1 K−1, OC7, JOUJYE). In contrast, because the top surface of the module is covered with conductive gold electrodes, a conductive copper block cannot be used as a heat source in direct contact with the electrodes. Instead, a 1-mm-thick aluminum nitride (AlN) block, which has a high thermal conductivity (285 W m−1 K−1) and is a good electrical insulator, was used between the hot-side copper block and device. Diamond paste was also included between the AlN and copper blocks, and contact between the top of the module and the hot-side AlN block was made by mechanical pressing. Gold wires (25 μm diameter, Nilaco) were connected to the modules by a small piece of Kapton tape (thickness 50 μm) for measuring the thermoelectric properties.
The steady-state temperature was measured by thin-film (200-μm thick) platinum resistance temperature sensors (Pt-RTS, NFR-CF-PT100, NETSUSHIN). Two Pt-RTSs were used:
one inserted in the diamond paste on the cold side and the other in the diamond paste on the hot side (See Figure A-6). Considering the thickness of the AlN block (2 mm) and its much higher thermal conductivity compared to the PET substrate and usual organic materials (< 1 W m−1 K−1), the temperature gradient that is generated across the AlN block should be negligible, and the temperature gradient should be mainly generated across the active layer and substrate of the module.
92
Appendix B B-1 Synthesis
2,2'-(2,5-dibromo-1,4-phenylene)diethanol (1)19,148: With ice bath cooling and under nitrogen flow, sulfuric acid (90 g) was cautiously added to water (45 g) in a 300 mL flask. Once the mixture had cooled back down to room temperature, 1,4-bis(2-hydroxyethyl)benzene (30 g, 166.22 g/mol, 180 mmol, TCI) was added with stirring, followed by N-bromosuccinimide (77 g, 178.0 g/mol, 433 mol, Wako), and the resulting suspension was stirred at room temperature for 24 h. The suspension was added with stirring to an ice/water mixture (200 mL). After neutralizing with KOH aqueous solution, the mixture was extracted with EtOAc. The combined organic extracts were washed with brine and dried (Na2SO4), and the solvent was evaporated in vacuo to give compound 1 as a white solid (15.1 g, 24%). M.p. 135-136 °C. 1H NMR (500 MHz, DMSO-d6): δ 2.80 (t, J = 6.8 Hz, 4H), 3.58 (q, J = 6.8 Hz, 4H), 4.77 (t, J = 5.3 Hz, 2H), 7.55 (s, 2H). 13C NMR (75 MHz, DMSO-d6): δ 38.2, 60.0, 122.8, 134.6, 138.5. HR-MS (ESI):
calcd for C10H12Br2O2 344.9096 [M+Na]+; found 344.9094.
(2,2'-(2,5-dibromo-1,4-phenylene)bis(ethane-2,1-diyl))bis(oxy)bis(tert-butyldimethylsilane) (2): Compound 1 (16.8 g, 48.7 mmol) was dissolved in 200 mL of DMF and treated with imidazole (14 g, 205.6 mmol, 4.2 equiv., Wako) and TBDMSCl (18.9 g, 125.4 mmol, 2.6 equiv., TCI) at 0 °C. The reaction solution was warmed to room temperature and stirred for 24 h and then was extracted with hexane. The combined organic extracts were washed with brine and dried (Na2SO4), and the solvent was evaporated in vacuo to give compound 2 as a white solid (27.1 g, 95%). M.p. 50-51 °C. 1H NMR (500 MHz, CDCl3): δ -0.03 (s, 12H), 0.87 (s, 18H), 2.89 (t, J = 6.8 Hz, 4H), 3.79 (t, J = 6.8 Hz, 4H), 7.43 (s, 2H). 13C NMR (75 MHz, CDCl3): δ -5.3, 18.4, 26.0, 38.9, 62.2, 123.1, 135.3, 138.3. HR-MS (ESI): calcd for C22H40Br2O2Si2 573.0826 [M+Na]+; found 573.0835.
(2,2'-(2,5-bis(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)-1,4-phenylene)bis(ethane-2,1-diyl))bis(oxy)bis(tert-butyldimethylsilane) (3): Compound 2 (5.32 g, 9.6 mmol), pinacoldiboron ester (7.34 g, 28.9 mmol), KOAc (5.69 g, 57.9 mmol), and PdCl2(dppf) (472 mg, 0.58 mmol) were dissolved in DMF (20 mL) under nitrogen. The mixture was heated at 80 °C for 24 h, cooled, and then diluted with EtOAc. The organic layer was washed with water and brine and then dried (Na2SO4), and the solvent was evaporated. The crude product was purified through column chromatography (silica gel, hexane/EtOAc=8/1). Recrystallization from EtOAc gave compound 3 as a white solid (1.8 g, 28%). M.p. 133-134 °C. 1H NMR (500 MHz, CDCl3): δ -0.02 (s, 12H), 0.87 (s, 18H), 1.33 (s, 24H), 3.08 (t, J = 7.5 Hz, 4H), 3.73 (t, J= 7.5 Hz, 4H), 7.62 (s, 2H). 13C NMR (75 MHz, CDCl3): δ -5.1, 18.5, 25.0, 26.2, 38.9, 65.7, 83.6, 137.9, 138.1, 142.1. HR-MS (ESI): calcd for C34H64B2O6Si2 669.4346 [M+Na]+; found 669.4351.
93
6,6'-(2,5-bis(2-(tert-butyldimethylsilyloxy)ethyl)-1,4-phenylene)bis(3-bromopyridine) (4):
Compound 3 (1.35 g, 2.1 mmol), 2,5-dibromopyridine (1.6 g, 6.8 mmol), CsCO3 (2.7 g, 8.3 mmol), and Pd(PPh3)4 (200 mg, 0.17 mmol) were dissolved in THF/H2O (16/12 mL) under nitrogen. The mixture was heated at 100 °C for 24 h, cooled, and then diluted with EtOAc. The organic layer was washed with water and brine and then dried (Na2SO4), and the solvent was evaporated. The crude product was purified through column chromatography (silica gel, hexane/EtOAc=20/1) and further purified using vacuum thermal gradient sublimation (under 10-3 Pa). Finally, recrystallization from EtOAc gave compound 4 as a white crystal (540 mg, 36.5%). M.p. 140-142 °C. 1H NMR (500 MHz, CDCl3): δ -0.08 (s, 12H), 0.80 (s, 18H), 2.94 (t, J
= 7.5 Hz, 4H), 3.71 (t, J = 7.5 Hz, 4H), 7.31 (s, 2H), 7.38 (d, J = 8.3 Hz, 2H), 7.89 (dd, J = 8.3 Hz, J = 2.5 Hz, 2H), 8.75 (d, J = 1.8 Hz, 2H). 13C NMR (75 MHz, CDCl3): δ -5.3, 18.4, 26.0, 26.1, 36.2, 64.4, 119.4, 125.6, 132.5, 135.1, 139.0, 139.8, 150.4, 158.1. HR-MS (ESI): calcd for C32H46Br2N2O2Si2 705.1537 [M+H]+; found 705.1505.
Siloxyethyl-substituted poly(pyridine phenylene) (5): A solution of 4 (1 g, 1.42 mmol), bis(1,5-cyclooctadiene)nickel (0) (1 g, 3.65 mmol), 1,5-cyclooctadiene (200 μL, 1.6 mmol), and 2,2’-bipyridine (250 mg, 1.6 mmol) in 200 mL of anhydrous DMSO was stirred for 24 h at 80 °C under nitrogen. The reaction mixture was cooled to room temperature and then re-precipitated into ice-cooled methanol and filtered by vacuo twice. After washing with EDTA 2Na dehydrate aqueous solution and pure water, the precipitate was collected, dried, dissolved in chloroform, and then re-precipitated from methanol twice. Finally, the precipitate was purified by the Soxhlet extraction with methanol, hexane, THF and EtOAc. The precipitate was re-dissolved in chloroform, and the solvent was evaporated in vacuo to give compound 5 as a pale yellowish white film (428mg). 1H NMR (500 MHz, CDCl3): δ -0.04 (s, 12H), 0.83 (s, 18H), 3.10 (br, 4H), 3.81 (br, 4H), 7.47 (br, 2H), 7.66 (br, 2H), 8.07 (br, 2H), 9.03 (br, 2H).
Ethylene-bridged poly(pyridinium phenylene) (6): To a solution of 5 (428 mg) in 100 mL of anhydrous acetonitrile, SOCl2 (18 mL) was added, and the mixture was stirred at room temperature for 24 h under nitrogen flow. The solution was evaporated, and then the resulting solid was washed with DMF, THF and methanol to give polymer 6 as a yellow film (265.5mg).
1H NMR (500 MHz, D2O): δ 3.57 (br, 4H), 5.04 (br, 4H), 8.37 (br, 2H), 8.84 (br, 2H), 9.06 (br, 2H), 9.50 (br, 2H).
94
B-2 Measurement of Seebeck coefficient of single OTEGs
Figure B-1: Schematic of the experimental setup for in-plane Seebeck coefficient (𝑺) measurement: (a) temperature measurement mode and (b) thermoelectric voltage (∆𝑽𝐓𝐄) measurement mode. (c) Typical ∆𝑽𝐓𝐄 curves and (d) 𝑺 obtained from Eqs. B-3 and 4. (e) Calculated temperature gradient (∆𝑻𝐂𝐚𝐥) obtained from Eq. B-6 versus temperature gradient (∆𝑻) for alumel as a reference material.
With every experiment, accurate measurement is important. Seebeck coefficient (𝑺) is particularly easy to overestimate due to difficulty of measuring precise temperature gradients (∆𝑻) and thermoelectric (Seebeck) voltages. These difficulties arise from the low electrical conductivity of organic materials and the regular use of small-scale devices, i.e., thin films and narrow channel lengths.109–111
Low electrical conductivity results in a high resistance between the two-electrodes (output impedance) that are used to measure the thermoelectric voltage (∆𝑽𝐓𝐄). This high output impedance disturbs accurate measurement of ∆𝑽𝐓𝐄 due to floating electromotive force. To suppress the floating electromotive force and accurately measure ∆𝑽𝐓𝐄, the ratio between input impedance of the voltage meter and output impedance of the device should be over 103.112
A digital multi meter (Keithley DMM2000 61/2, input impedance > 1010 Ω, resolution 0.1 μV) was used in the measurement system here; thus, the output impedance of the sample must be < 1MΩ. Considering the sample configuration (𝑾 = 23 mm, 𝑳 = 3 mm, 𝒕 = 100-300 nm), 𝝈 of the sample must be > 10-2 S cm-1. This is why 𝑺 of intrinsic organic
95
semiconductors is so difficult to measure. For these reasons, people using thin-film materials prefer a narrow channel length, e.g., several tens of micrometers, in order to decrease output impedance. However, an extremely narrow channel length makes applying a controlled and steady ∆𝑻 difficult.
In this study, 𝑺 of single OTEGs were measured with a homemade setup as described in Fig. B-1a. To determine 𝑺, it is necessary to apply a ∆𝑻 across the sample and measure
∆𝑽𝐓𝐄 and ∆𝑻. Fine chromel and alumel wires (φ 50 μm, Nilaco) were used in order to resolve these problems. Chromel and alumel are alloys with known 𝑺 and well known as materials for forming K-type thermocouples.
Even though chromel and alumel have high thermal conductivity compared with organic materials, a thermocouple of the two wires has high thermal resistance due to their fine and long geometry. Thus, heat loss through these thermocouple while measuring thermoelectric voltage can be prevented, and this ultrafine thermocouple can provide fast, precise, and accurate monitoring of local temperature changes of the sample.113
In addition, thermoelectric voltages can be measured through the inner two electrode strips using the chromel-chromel or alumel-alumel wire pairs after the temperature stabilized (steady-state). These ∆𝑽𝐓𝐄 can be expressed as
∆𝑽𝐂𝐡𝐫‐𝐂𝐡𝐫 = ∆𝑻(𝐒𝛘− 𝑺𝐂𝐡𝐫), Equation B-1
∆𝑽𝐀𝐥𝐮‐𝐀𝐥𝐮 = ∆𝑻(𝑺𝐀𝐥𝐮− 𝑺𝛘). Equation B-2
Here ΔVChr-Chr and ΔVAlu-Alu are the thermoelectric voltages measured across the inner two chromel-chromel and alumel-alumel wire pairs, respectively. ΔT is the temperature gradient applied to the sample. 𝑺𝐂𝐡𝐫 and 𝑺𝐀𝐥𝐮 are the Seebeck coefficients of chromel (+22 μV K−1) and alumel (−19 μV K−1) respectively.114 Finally, 𝐒𝛘 is the Seebeck coefficient of the sample being measured and can be expressed using Eqs. B-1 and B-2 as follows:
𝑺𝛘,𝐂𝐡𝐫‐𝐂𝐡𝐫 = ∆𝑽𝐂𝐡𝐫‐𝐂𝐡𝐫
∆𝑻 + 𝑺𝐂𝐡𝐫, Equation B-3 𝑺𝛘,𝐀𝐥𝐮‐𝐀𝐥𝐮 = −∆𝑽𝐀𝐥𝐮‐𝐀𝐥𝐮
∆𝑻 + 𝑺𝐀𝐥𝐮, Equation B-4 𝑺𝛘 = 𝑺𝛘,𝐂𝐡𝐫‐𝐂𝐡𝐫 = 𝑺𝛘,𝐀𝐥𝐮‐𝐀𝐥𝐮 =∆𝑽𝐂𝐡𝐫‐𝐂𝐡𝐫−∆𝑽𝐀𝐥𝐮‐𝐀𝐥𝐮
𝟐∆𝑻 +𝑺𝐂𝐡𝐫+𝑺𝐀𝐥𝐮
𝟐 . Equation SII-5
From Eqs. B-3-5, 𝐒𝛘 can be evaluated using either the chromel-chromel or alumel-alumel wire pair or both. Here, 𝑺𝛘,𝐂𝐡𝐫‐𝐂𝐡𝐫 and 𝑺𝛘,𝐀𝐥𝐮‐𝐀𝐥𝐮 are the 𝐒𝛘 estimated through the inner two chromel-chromel and alumel-alumel wire pairs respectively.
96
The reason why two equations and wire pairs were used when estimating 𝑺𝛘 is that the temperature gradient (∆𝑻𝐂𝐚𝐥) also can be calculated from Eqs. B-3 and B-4 as
∆𝑻𝐂𝐚𝐥 = − (∆𝑽𝐂𝐡𝐫‐𝐂𝐡𝐫+∆𝑽𝐀𝐥𝐮‐𝐀𝐥𝐮
𝑺𝐂𝐡𝐫−𝑺𝐀𝐥𝐮 ). Equation B-6
Thus, the accuracy of 𝑺𝛘and ∆𝑻 can be confirmed by comparing 𝑺𝛘,𝐂𝐡𝐫‐𝐂𝐡𝐫 with 𝑺𝛘,𝐀𝐥𝐮‐𝐀𝐥𝐮 and comparing ∆𝑻 measured by the two K-type thermocouples at the hot and cold sides with ∆𝑻𝐂𝐚𝐥.115
The in-plane 𝑺 of alumel was measured as a reference using the homemade setup described in the experimental section to ensure the accuracy of the measurement system (see Figs. SII-1c-d). While varying the temperature gradient through control of the current of the Peltier devices, ∆𝑽𝐓𝐄 versus ∆𝑻 was automatically recorded by computer. The applied ∆𝑻 increased from zero up to +2.5 K (forward), decreased from +2.5 K down to −2.5 K (reverse), and then increased back to zero. If the measured material was pyroelectric, exhibited ion conductivity, or did not achieve thermal steady state, this cycle would show hysteresis because these phenomena are related to the temperature rise rate.116 A typical ∆𝑽𝐓𝐄 versus ∆𝑻 curve when measuring alumel as a reference material is shown in Figure SII-1c. The ∆𝑽𝐀𝐥𝐮‐𝐀𝐥𝐮 is practically zero over the entire region because the measured alumel sample was measured with alumel wires, so ∆𝑽𝐓𝐄 should cancel each other regardless of the applied temperature gradient. In contrast, ∆𝑽𝐂𝐡𝐫‐𝐂𝐡𝐫 should change at a rate of 𝑺𝐀𝐥𝐮− 𝑺𝐂𝐡𝐫. Thus, the slopes measured through the inner alumel-alumel and chromel-chromel wire pairs were 0 µV K−1 and -41 µV K−1 respectively. A Seebeck coefficient of −19 ± 0.1 µV K−1 was obtained from the linear slopes of lines fit to each ∆𝑽𝐓𝐄-∆𝑻 curve over all the data and Eqs. SII-3-5, which matches well with literature.
Figure B-1d shows instantaneous 𝑺𝛘,𝐂𝐡𝐫‐𝐂𝐡𝐫 and 𝑺𝛘,𝐀𝐥𝐮‐𝐀𝐥𝐮 at each 𝚫𝑻 obtained using Eqs. B-3 and 4 and the individual data points in Fig. B-1c. 𝑺𝛘,𝐂𝐡𝐫‐𝐂𝐡𝐫 and 𝑺𝛘,𝐀𝐥𝐮‐𝐀𝐥𝐮 obtained from Eqs. B-3 and 4 converge to 𝑺𝐀𝐥𝐮 at about |𝚫𝑻|> 1 K. For |𝚫𝑻|< 1 K, the error is large because of high output impedance and floating electromotive force.109–111 Furthermore Fig.
B-1e shows that the calculated temperature gradient ∆𝑻𝐂𝐚𝐥 obtained from Eq. B-6 matches well with the measured temperature gradient 𝚫𝑻.
The 𝑺 values of several other reference samples were also measured to further confirm the accuracy of measurement system. The measured S of chromel, gold, nickel, and platinum were 21.9 ± 0.2 µV K−1, 2.0 ± 0.4 µV K−1, −19.3 ± 0.4 µV K−1, and −0.9 ± 0.65 µV K−1 respectively.
97
B-3 Comparison between DMBI and NaNap as electron donors
Figure B-2: Optical absorption spectra of a pristine P(PymPh) film, a film annealed at 150 °C for 30 min, and films doped with DMBI or NaNap. Inset: The molecular structures of P(PymPh), DMBI, and NaNap.
Another n-type dopant that has been used in OTEGs is 4-(1,3-dimethyl-2,3-dihydro-1H- benzoimidazol-2-yl)phenyl)dimethylamine (DMBI). To compare the effectiveness of doping with DMBI, the optical absorption spectra of a pristine P(PymPh) film, a film annealed at 150 °C for 30 min, and films doped with DMBI or NaNap are shown in Fig. B-2. The NaNap doping was done the same way as described in the experimental section. For DMBI doping, DMF was used as solvent and heated to 150 °C to activate DMBI.117,118 All processing was done in nitrogen. The higher ratio of the absorption peaks attributed to charged polymer species (592 nm, 1100-2500 nm) to those of the neutral polymer (428 nm) for the film doped with NaNap than the annealed film and film doped with DMBI suggests that NaNap is the stronger reductant.
98
B-4 XPS
Figure B-3: XPS (a) Na 1s, (b) Cl 2p, and (c) C 1s spectra of pristine, annealed, and NaNap-doped films of P(PymPh). The concentration of NaNap in the doping solution ranged from 1.0-6.0 mM as indicated on the plots. As references, Ag and NaNap were also measured.
Comparing the XPS Na 1s and Cl 2p signals in Figs. B-3a and b, peaks derived from the formation of sodium chloride, 1072.6 eV at Na 1s and 199.6 eV at Cl 2p, can be observed.123 Thus, the sodium introduced by NaNap likely reduces the pyridinium ions and reacts with the chloride anion to form sodium chloride. In contrast, although peaks consisting of C-N were obtained in Fig. B-3c, a large chemical shift following the doping was not obtained.
B-5 Thermally stimulated doping of P(PymPh)
Figure B-4: Conductivity (𝝈) of annealed P(PymPh) films at room temperature.
To investigate the effect of annealing treatment on electrical conductivities (𝝈), a device with bottom silver electrodes on a glass substrate was used. The bottom silver electrodes and a P(PymPh) thin film were prepared the same way as described in the experimental section and Fig. 3.3a. The film was measured, annealed for 30 min in nitrogen atmosphere, cooled to
99
room temperature, and re-measured, and this process was repeated for consecutively high temperatures. The 𝝈 measured by 4T-sensing method at room temperature are shown in Fig.
B-2b for the different annealing temperatures. The film and metal electrode fabrication and annealing treatments were carried out under nitrogen without any air exposure.
From Fig. B-4, 𝝈 can be seen to start to increase at an annealing temperature of about 125 °C. One reason for the increase could be the degassing of oxygen and water. Another one would be self-doping due to the cleaving of N+-Cl− chemical bonds in accordance with thermal stimulation.
B-6 TG-DTA measurement
Figure B-5: TG-DTA curves of P(PymPh) in nitrogen.
A commercial TG-DTA system (TG-DTA2400SA, Bruker) was used for thermogravimetric (TG)/differential thermal analysis measurements (DTA) measurements. After loading the sample of P(PymPh), nitrogen gas was flowed at a rate of 200 mL min−1 during the measurement. The resulting TG-DTA curves are shown in Fig. B-5.
Below 100-125 °C, the TG continuously decreases with increasing temperature. Because P(PymPh) is very polar, water and oxygen are expected to be absorbed during the sample loading, so this may correspond to degassing of the sample. Thus, an annealing temperature of 150 °C was chosen to degas water and oxygen without the decomposition of P(PymPh), which begins at about 200 °C.
100
B-7 XRD
Figure B-6: (a) Out-of-plane XRD patterns of pristine and annealed P(PymPh) on Si substrates. (b) Main crystallographic structures of P(PymPh).
The effects of doping on the films were further studied using X-ray diffraction (XRD) analysis because the crystallinity of the polymer matrix can be changed by annealing treatment and dopants physically penetrating into or moving out of the polymer during doping. Thus, the polymer films can undergo swelling, shrinking, or other morphological changes that can also impact the electrical properties.107,135
The changes in the crystal structure of P(PymPh) films on silicon substrates were investigated using a Rigaku Ultima IV XRD system. The out-of-plane XRD patterns of pristine and annealed P(PymPh) films are shown in Fig. B-6a. Results for doped films are not shown in Fig. B-6a because the doped films were too unstable in ambient air to reliably measure in our XRD system.
In all cases, only one small peak was found at 𝟐𝜽 = 23-24°, which indicates that the matrix of P(PymPh) should be almost amorphous compared with P3HT, which is a general conducting polymer that can exhibit a large number of diffraction peaks such as (100) to (300) and (010) to (020).107 The peak for P(PymPh) corresponds to the (010) π-π stacking distance, and lattice constants of 𝒅 = 3.5-3.7 Å were estimated according to Bragg's law (𝟐𝒅𝐬𝐢𝐧𝜽 = 𝒏𝝀). Thus, the P(PymPh) polymer crystallites have preferential face-on orientation and are predominantly oriented horizontally to the substrate surface as shown in Fig. SII-6b.
Comparing the full widths at half maximum (FWHM) of the (010) peaks for the pristine and annealed films, the annealed film (0.94 Å) is much narrower than the pristine film (1.12 Å), which may be advantageous for charge transport (i.e., electrical conduction) due to an increase in crystallinity and also play a role in the results in B-5.107
9 7 8 6 4
3 7 5
98 6 4
10 20
10 20 30
5 3
Counts (a.u.)
Si sub.
Pristine P(PymPh) / Si sub.
Anealed P(PymPh) at 150 °C / Si sub.
40
(010)
d (Å)
2Θ (°)
(a) (b)
Face-on orientation (a00)
Substrate (0b0)
π-π stacking
(00c)
2Θ = 24.6 ° d = 3.62 Å
101
B-8 Hall measurement
When a current-carrying semiconductor is kept in a magnetic field, the charge carriers of the semiconductor experience a Lorentz force in a direction perpendicular to both the magnetic field and the current, which is called the Hall effect. Thus, Hall measurements make discrimination of carrier type and estimation of carrier density and mobility possible.149
The relationships among the Hall coefficient, mobility, and carrier density can be expressed as described in Eqs. B-7 and 8. Here, 𝑹𝐇 is the Hall coefficient, 𝑽𝐇 is the Hall voltage, 𝑩 is the magnetic field, 𝑰 is the current, 𝒕 is the thickness of the sample, and 𝒒 is the charge of an electron. The values 𝒏𝐡 and 𝒏𝐞 are the charge carrier densities of holes and electrons, respectively, and 𝝁𝐡 and 𝝁𝐞 are the hole and electron mobilities. Finally, 𝝈 is electrical conductivity, which can be broken into components from holes (𝝈𝐡) and electrons (𝝈𝐞).
𝑹𝐇 =𝑽𝐇
𝑩𝑰𝒕 = 𝒏𝐡𝝁𝐡𝟐−𝒏𝐞𝝁𝐞𝟐
|𝒒|(𝒏𝐡𝝁𝐡+𝒏𝐞𝝁𝐞)𝟐 Equation B-7
𝝈 = 𝝈𝐡+ 𝝈𝐞 = |𝒒|(𝒏𝐡𝝁𝐡+ 𝒏𝐞𝝁𝐞) Equation B-8
When a single carrier species, i.e., electrons or holes, dominates the current flow in the semiconductor, the equations can be simplified as Eqs. B-9 and 10. Here, 𝒏𝐇 and 𝝁𝐇 correspond to the charge carrier density and carrier mobility of the dominate carrier species.
𝒏𝐇 = 𝟏
|𝒒𝑹𝐇| Equation B-9
𝝁𝐇 = |𝑹𝐇|𝝈 Equation B-10
Hall measurements of P(PymPh) doped with NaNap were performed to obtain further information on carrier type, carrier density, and mobility. Device fabrication was almost the same as described in the experimental section, but a different device configuration was adopted based on the Van der Pauw method (Fig. B-7a). All processing and measurements were carried out under nitrogen without any exposure to air.
For the Hall measurements, a commercial magnet system (ResiTest 8340DC/CSE, TOYO Corporation) was used. A constant magnetic field of −0.5 T or +0.5 T and a constant DC current (Source Meter 2400, Keithley) were applied to the sample. To eliminate the drift of background voltage and thermoelectric voltage in accordance with the Peltier effect in the sample, delta-mode measurement was adopted for measuring the precise Hall voltage (Nanovolt Meter 2182A, Keithley).