Hexakis(2,3,6-tri-O-methyl)-α-cyclodextrin–I5– complex in aqueous I–/I3– thermocells and enhancement in the Seebeck coefficient

Supramolecular thermocell composed of I– (yellow balls), I3– (trio of red balls), I5– (five connected dark red balls) and Me18-α-CD (gray cone-shaped cylinder).


Preparation of aqueous solution
Various kinds of stock solutions were prepared as summarized in Table S1 and   Table S2. CD were dissolved into 40 ml water so that initial concentration of KI and I 2 to be 12.5 and 2.5 mM, respectively. The concentration of Me 18 -α-CD was listed in Table S1. The solutions were called as S1-x where x corresponds to the concentration of Me 18 -α-CD (mM).

Seebeck Coefficient Measurements
40 ml of Solution 1 (S1-0.0 to S1-8.0) were poured into the H-shaped glass cell consists of two half-cells soaked into a couple of water baths at different temperature (Fig. S2). Each half-cell has a diameter of ca. 2 cm and the distance between them is ca. 10 cm for creating a stable temperature difference. The temperatures inside the half-cells were monitored with thermometers (TM201, AS ONE, Japan) and the cold side was kept at approximately 10 ℃. Meanwhile, the electrolyte was stirred during the measurements to accelerate the equilibrium in the cell. Platinum wires (φ = 1 mm) were washed with concentrated sulfuric acid and used as electrodes for measuring a potential difference generated by the temperature difference. The potential difference was recorded by a source meter 2401 (KEITHLEY).

Figure S2
. Schematic illustration of the H-shaped cell. The two half-cells were immersed in cold and hot water baths. A stirrer was put into each side to accelerate the thermal equilibrium. Two platinum wire electrodes were used to measure the open circuit voltage (V oc ) and the temperature difference was evaluated by thermocouples.

Isothermal Titration Calorimetry (ITC) measurements
The The fitting model is not the general single set of identical sites (SSIS) or two sets of independent sites (TSIS), as discussed in the manuscript. Thus, the experimental results were analyzed by a novel fitting method based on the single set of identical sites (SSIS).
In the novel fitting model, the concentration of the substances in Eq. 2 is dependent on the initial binding reaction (Eq. 1). Me 18 -α-CD + I 3 − ⇌ Me 18 -α-CD-I 3 − (Eq. 1) The total heat generated can be regarded as a sum of Eq. 1 and Eq. 2 and generated heat at each injection can be written as Eq. S1, where ΔQ(i) is the heat generated of the i th injection step, ΔQ(i) 1 and ΔQ(i) 2 are heat generated by Eq. 1 and Eq. 2, respectively.
The reaction of each step can be fitted by SSIS model.

SSIS model
The SSIS model is introduced as follows: 3 The binding constant K and the relationship of the total and free guest are described as: Combing Eq. S1 and Eq. S2 gives The total heat content Q of the solution contained in V o at fractional saturation Θ is Where ΔH is the molar heat of guest binding. Solving the quadratic equation (Eq. S4) for Θ and then substituting this into Eq. S5 gives The value of Q above can be calculated at the end of the i th injection and designated Q (i). The parameter of interest for comparison with experiment, however, is the change in heat content from the completion of the (i-1) th injection to completion of the i th injection. The expression for Q in Eq. S5 only applies to the liquid contained in volume V o . Therefore, after completing an injection, it is obvious that a correction must be made for displaced volume (i.e., ΔV i = injection volume) since some of the liquid in V o after the i-1 th injection will no longer be in V o after the i th injection, even though it will contribute By dividing with moles in the i th injected volume, the normalized heat, ΔQ(i) is Δ ( ) obtained, which is described by three parameters n, ΔH, and K.

The fitting of the present research
In the current fitting model, the total generated heat can be separated by two kinds of heat Q 1 (i) and Q 2 (i), which are generated according to Eq. 1, and Eq. 2, respectively.
Where, ΔH 1 = Enthalpy change at the initial binding stage; Then, the experimental ITC curve can be fitted by ΔQ = ΔQ(i) 1 + ΔQ(i) 2 , where ΔQ(i) 1 and ΔQ(i) 2 are the released heat for i th titration in the initial and second stages, respectively.
The result in the thermodynamic parameters ΔH 1 , K 1 , ΔH 2 , and K 2 . Are shown in Table   S3. ΔS 1 and ΔS 2 were obtained by van't Hoff equation. Table S3. Parameters for initial and second binding stages.

UV-vis spectroscopy
UV-vis spectra were recorded by UV-vis spectrometer (V670, JASCO, Japan) in the wavelength range between 200 to 600 nm, with a resolution of 0.5 nm and a fixed slit width of 0.5 nm. 0.2 mm path length quartz cuvettes were used.    Table S1) at varied Me 18 -α-CD concentrations.
The color of solutions changes from yellow to red black then to deep red-black. When the concentration of Me 18 -α-CD is increased beyond 3 mM, the color returns to red black

Raman spectroscopy
Raman spectra were recorded on a micro-Raman spectrometer (Jasco NRS 3100KK) equipped with YAG laser (power 1.5 mW, Wavelength 532 nm) and a thermoelectrically cooled CCD detector (DU401-BV-120, Andor). All the measurements were performed at ambient temperature.
UV-vis spectra reveal that I 3 − /I − aqueous solution in 3 mM α-CD and 2.1 mM Me 18 -α-CD possess the maximum I 3 − and I 5 − respectively, thus the Raman spectra were recorded at such conditions and the corresponding fittings are shown in Fig. S10. A triiodide cation shows a strong peak at ca. 110 cm −1 and a weak peak at ca. 140cm −1 , 6-10 , which are attributed to the symmetric (ν 1 ) and asymmetric (ν 2 ) stretch vibration respectively.
Another weak peak at 160 cm −1 is assigned to the Fermi resonance between ν 1 and the overtone of bending mode (ν 3 ) of triiodide, though the ν 3 is proposed to be at 50-70 cm −1 while it is very weak and not observed in the spectra. 7,9,11 When 3 mM α-CD was added to the I 3 − /I − solution, the increment of the peak intensity of ν 1 indicates that the host-guest interaction. A new peak appeared at ca. 170 cm −1 , which was assigned to the I-I stretch mode and a fraction of the captured I 3 − in α-CD should be described as an adduct of I − ·I 2 (ν I-I = 170 cm −1 ). 7,9,11 In the case of Me 18 -α-CD as a host, the intensity of ν 1 peak decreased and the of a peak at 170 cm −1 significantly enhanced. I 5 − in solid state exists as an adduct of I − ·2I 2 or I 3 − ·I 2 , 7-9,12,13 and a peak at 170 cm −1 was enhanced in the case of I − ·2I 2 . Thus, Me 18 -α-CD stabilizes the I − ·2I 2 adduct in aqueous solution. 3 mM α-CD in addition of (a) and (c) 2.1 mM Me 18 -α-CD in (a). Each spectrum was fitted with Pseudo Voigt functions. All experimental data was overlapped in (d).