Mixed-Valence Compounds as a New Route for Electrochromic Devices with High Coloration Efficiency in the Whole Vis-NIR Region

Electrochromic devices (ECDs) that allow the modulation of light transmission are very attractive in the research field of energy saving. Here all-in-one gel switchable ECDs based on mixed-valence electroactive compounds were developed. The use of the thienoviologen/ferrocene couple as cathode and anode, respectively, leads to a significant electrochromic band in the visible range (550–800 nm), with a color change from yellow to green, and to a lower band in the NIR region (1000–1700 nm), due to the presence of one electroactive-chromic species. Replacement of the electroactive ferrocene with a fluorene-diarylamine electroactive-chromic species, allows to extend and intensify the absorption in the NIR region, thus affording modulation of the solar radiation from 500 up to 2200 nm. High optical contrast, fast coloration and bleaching times and outstanding coloration efficiencies were measured for all observed absorption bands upon the application of small potential differences (1.4 V < DV < 2 V).


Introduction
Limiting climate change and fostering sustainable growth are two of the main diktats of modern life. The last report on global warming, called "GLOBAL WARMING OF 1.5 • C", reported by the ONU Climate Committee, predicted that, already in 2030, the global warming could exceed the threshold of 1.5 degrees from the pre-industrial levels [1]. The scientists' proposal provides different possible paths with a mix of different tools such as: (i) cutting emissions that can be actuated by switching to renewable energies and electric vehicles, energy efficiency and recycling of waste and (ii) "removal" of CO 2 in terms of reforestation, carbon capture and storage. The way indicated by the ONU is the "greenest" and focuses on energy saving and reforestation. Environmental challenges on energy saving provide the driving force for many studies in the field of electronic devices for energy, including the electrochromic (EC) one that can be exploited for smart windows. The latter technology can afford, simultaneously, energy efficiency in buildings and good indoor comfort [2][3][4][5][6], since it is able to cut and/or modulate a variable transmission percentage of solar radiation both in the near-infrared region, which is responsible of thermal contribution (heating and cooling) and in the visible one, which is responsible of lighting [4,7]. In fact, the electrochromism is defined as the reversible variation of chemical species between two redox states with distinguishable absorption spectra as effect of a stimulus of electrical nature [8,9].
It is particularly interesting for building application since a significant portion of consumed energy is for lighting, heating and cooling. Switchable windows and mirrors did not represent the only area of interest, indeed, many others can be added such as: controllable light-reflective or light-transmissive devices for optical information and storage, electromagnetic shutter, controllable aircraft canopies, protective eyewear for the military, glare reduction system for automotive and transport [10][11][12]. To date, open issues for successful EC products are low-cost manufacturing and long-term durability and these pave the way for the research of the most suitable materials. Vis-NIR EC materials include several transition metal oxides [13][14][15], inorganic coordination complexes [16,17], conjugated polymers [18][19][20] and small organic molecules [21][22][23][24][25][26]. Viologens and arylamine-based compounds are among the most studied cathodic and anodic materials in the class of small organic molecules. These compounds allow an easy structural functionalization, thus offering a wide modulation of the absorption properties as a function of the inserted functional units, all over the Vis-NIR spectrum. Viologens are generally transparent in the dicationic state but, upon reduction, a highly colored and exceptionally stable radical cation is formed [8,9,[27][28][29][30]. The radical-cationic form of viologens exhibits different coloration depending on the nature of the substituents on the nitrogen atoms as well as of the π-system that can be inserted between the pyridinium rings [30][31][32][33]. Arylamine derivatives are well known for their photoactive and electroactive properties, which usually have many advantageous such as good solubility, thermal stability, strong electro-donating and excellent hole-transmission capability [34][35][36][37]. During the oxidation-reduction process, arylamines always switch their colors and thus were developed as a new family of electrochromic materials [38][39][40]. Both, viologens and arylamines radical cations belong to the class of mixed valence compounds (MVs), resulting particularly interesting since they may offer NIR Electrochromism [41][42][43][44][45]. Actually, the arylamines NIR EC is extremely widespread rather than that of viologens. Organic MVs are donor-acceptor compounds that possess an open shell ground state [41]. The donor and acceptor moieties, i.e., the redox centres, are chemically identical units but in different oxidation states. They are connected by a π-conjugated bridging unit, which plays a crucial role for exploiting intramolecular electron transfer (ET) processes [41]. Indeed, when the redox centers are electronically coupled, an electron transfer process from the reduced to the oxidized center may be thermally or optically induced. In the latter case, the process gives rise to the so-called Intervalence Charge Transfer Transition (IVCT), which generally has an energy falling in the NIR region.
Herein we assembled highly transmissive all-in-one gel devices based on the thienoviologens/ferrocene (TV/Fc) and thienoviologens/fluorene-diarylamine (TV/FD) systems, that were characterized by extensive electrochemical and spectroelectrochemical studies in order to disclose their potential applicability. Particularly, we studied the devices at different operating voltage highlighting an interesting tuning of the electroptical properties of device components. The voltage-induced switching studies highlight that the reduction and oxidation processes occurring in the devices, generate high optical contrast (∆%T up to 82%) within fast response time (τ ON ≈ 2 s; τ OFF ≈ 5 s) and with low surface charge injection, as revealed by the measurements of the coloration efficiency (CE) both in the visible and near-infrared region.
The TV/Fc EC polymer gels were prepared with 2% (w/w) of the ferrocene, the 2% (w/w) of thienoviologens, the 44% (w/w) of PVF, and 52% (w/w) of NMP. The TV/FD EC polymer gels were prepared with 10% (w/w) of the anodic, electroactive MV compounds, the 2.5% (w/w) of thienoviologens, the 35% (w/w) of PVF and 52.5% (w/w) of NMP. The two electrochromic active species were first dissolved in NMP at room temperature and then mixed with the polymer under continuous stirring at 120 • C for 30 min.
ITO (indium tin oxide)/EC gel/ITO devices were assembled by drop casting of the hot EC mixture onto a ITO-coated glass support (Visiontek Systems Ltd. (Upton Chester Cheshire CH2 1JL United Kingdom) with a sheet resistance of 25 Ω sq −1 and a thickness of 1 mm) with a second ITO electrode used to create a sandwich with active area of about 1.5 × 3 cm 2 . The cell gap was controlled by inserting cylindrical spacers with mean base diameter of 30 × 10 −6 m. The final cell gap was calculated by the interference fringes in its absorption spectrum in the NIR spectroscopy range (900−1500 nm).
Standard cyclic voltammetry of the electrochromic polymer gel (EC gel) was performed with an Amel Instruments 7050 model potentiostat on the ITO/EC gel/ITO cell in which the two ITO plates worked as working and counter electrodes. Spectroelectrochemical experiments were performed, on the two-electrode device, with the Vertex 80 (Bruker) spectrophotometer. The potential was supplied by means of an Amel 2049 model potentiostat. Measurements were performed at 25 • C.
Coloration efficiency measurements were performed by coupling to the above describe spectrophotometer a current source measure, Keithley 6517A Electrometer [48].

Spectroelectrochemistry of the Devices
The electrochemical properties of the thienoviologen as well as that of the fluorene-diarylamine derivative in solution have been previously studied and reported by our group [33,[47][48][49]. Cyclic voltammetry (CV) of the polymer gel (EC gel) was performed on a ITO/EC gel/ITO sandwich cell without a reference electrode, in which two ITO coated glass plates served as working and counter electrodes ( Figure S1). The CV of the device with the TV/Fc complementary components shows two redox processes in the negative scan, at potential biases (the potential difference between the two ITO electrodes) of 1.10 V and 1.30 V. The first one is relative to the reduction of the thienoviologen dication (TV ++ ) into the radical-cation (TV •+ ) while the second is due to the formation of the neutral species (TV). The CV of the device containing the TV/FD system is more complex and this is mainly due to: (i) the higher numbers of processes involved and (ii) the overlap of the redox processes of the TV •+ and FD •+ species within the same range of potential bias (0.75 V < ∆V < 2.0 V). In both devices therefore, the redox processes start at potential biases higher than about 0.75 V, above which huge spectral changes (electrochromism) should be expected.
The electro-optical properties of these devices were investigated by spectroelectrochemical measurement (Figure 1). The spectroelectrochemistry of the TV/Fc EC gel shows that in absence of applied potential a strong absorption band is found at wavelength values < 500 nm, due to the π-π* transition of the TV ++ species [50]. Reduction to the radical cation species leads to the growing of a new broad band in the region 530-880 nm, and another in the range 1150-1700 nm (NIR region). Both these bands increase with the increase of the applied potential (Figure 1a). At about 1.7 V a partial bleaching of the NIR band at~1457 nm occurs (see the isosbestic point at 1370 nm) with the increase of a new absorption between~900-1300 nm. The application of a very small potential bias (≥1.1 V) enables the device to absorb almost all the visible region being accompanied by a color change of the gel from yellow to black-green ( Figure 1b). Replacement of the Fc + /Fc redox couple with the anodic FD, allows to broad the absorption range fully covering the near infrared region (NIR) up to 2000 nm ( Figure 1c). This NIR band, starts to grow up for potential bias of 1.3 V, where oxidation to the MV radical cation (FD •+ ) and reduction to the thienoviologen radical cation (TV •+ ) take place (Scheme 1). Therefore, the NIR absorption band extending from about 800 nm up to 2000 nm, is the result of the superposition of the NIR bands of the two radical cations above, which are attributed to inter-valence charge-transfer (IVCT) transitions typical for MV compounds [47]. At this potential, there is the concomitant rising up of another absorption band (confirmed by the absence of the isosbestic point) in the range 550-800 nm, due to the formation of the TV radical cation (TV •+ ) on the ITO slide opposite to that where the FD oxidation occurs. As the potential applied to the device is increased up to 2 V, a new intense band occurs in the range 840-980 nm. This band is typical of the dication FD ++ species. Actually, at this potential bias, the fluorene derivative should be fully oxidized to the dication species (FD ++ ) and the thienoviologen should be fully reduced to the neutral species (TV). Therefore, the corresponding radical cation bands should be bleached. In contrast, they are still present in the spectrum, though their intensity begins to decrease. The persistence of the radical cation bands at such potential value, can be explained by the invoking the following comproportionation reactions in the bulk of the EC film: Reactions (1) and (2) have rather high comproportionation constants (KCO ≈ 10 3 , 10 4 ) [33,47]. In addition, the reaction between the fully oxidized FD ++ and the fully reduced neutral TV, which is a strong reducing agent [30], easily occurs: This device switches from a transmissive off state (pale yellow) to different opaque on states, depending on the applied voltage pulse sequence, leading to an almost black colored device ( Figure  1d).

Device Performance Metrics
The electrochromic performances of these all-in-one gel based devices were further evaluated by switching them between alternating ON and OFF pulses with different voltages and times ( Figure  2, Table S1). The transmittance contrast (Δ%T), defined as the difference between the transmittance in the ON and OFF states, and the switching times, defined as the time required for reaching 90% of the full change in transmittance in a single switching time, are reported in Table S1. It is important to note that the voltage windows used correspond to the ranges where the monocations species are Scheme 1. Schematization of the formation of the radical cation species in the FD (a) and TV (b) mixed valence species.
As the potential applied to the device is increased up to 2 V, a new intense band occurs in the range 840-980 nm. This band is typical of the dication FD ++ species. Actually, at this potential bias, the fluorene derivative should be fully oxidized to the dication species (FD ++ ) and the thienoviologen should be fully reduced to the neutral species (TV). Therefore, the corresponding radical cation bands should be bleached. In contrast, they are still present in the spectrum, though their intensity begins to decrease. The persistence of the radical cation bands at such potential value, can be explained by the invoking the following comproportionation reactions in the bulk of the EC film: Reactions (1) and (2) have rather high comproportionation constants (K CO ≈ 10 3 , 10 4 ) [33,47]. In addition, the reaction between the fully oxidized FD ++ and the fully reduced neutral TV, which is a strong reducing agent [30], easily occurs: This device switches from a transmissive off state (pale yellow) to different opaque on states, depending on the applied voltage pulse sequence, leading to an almost black colored device (Figure 1d).

Device Performance Metrics
The electrochromic performances of these all-in-one gel based devices were further evaluated by switching them between alternating ON and OFF pulses with different voltages and times ( Figure 2, Table S1). The transmittance contrast (∆%T), defined as the difference between the transmittance in the ON and OFF states, and the switching times, defined as the time required for reaching 90% of the full change in transmittance in a single switching time, are reported in Table S1. It is important to note that the voltage windows used correspond to the ranges where the monocations species are generated and reach their maximum intensity (∼1430 and ∼670 nm, respectively), according to the spectroelectrochemistry of the devices. Figure 2 and Table S1 show that the TV/Fc system can reach, at λ = 670 nm, transmittance changes ranging from 74% to 82%. Analogous values, for the same absorption band, were found switching the TV/FD device. In this last, ∆%T values ranging from 58% to 80% were found for the absorption band at λ = 1430 nm, depending on the V ON and its time of application ( Figure 2, Table S1). For all the devices and for all absorption bands coloration times of few seconds (from 1.5 s to 5 s) and a little bit slower bleaching times (from 5 s to 8.5 s) were recorded.
In order to compare the performance of the EC devices it is of noteworthy importance the correlation between the measurement of the charge passed during the oxidation and reduction processes and the transmittance change at a specific wavelength. This correlation parameter is known as coloration efficiency (CE, η). The η is given by the equation: This also determines the amount of optical density change (∆OD) as a function of the injected/ejected electronic charge (Qi) per unit area. T b and T c are, respectively, the transmittance in the bleached and colored states. Figure 3 shows the change in optical density at three wavelengths (670, 890 and 1430 nm) as a function of the surface charge density. The CE value, extracted as the slope of the linear fitting of the reported curves, for both systems (TV/Fc and TV/FD), at 670 nm, is as high as 500 cm 2 /C, which is an outstanding result compared with analogous systems [51][52][53]. The CEs registered for the NIR bands in the TV/FD system, due to the electrogeneration of the TV •+ /FD •+ and TV ++ /FD ++ species, respectively, are 180 cm 2 /C and 170 cm 2 /C, thus, approximately three times smaller than that of the calculated at 670 nm and due to the formation of TV •+ . Furthermore, the coloration efficiency of the band at 670 nm calculated for the TV/FD devices, is only slightly smaller than that calculated for the TV/Fc devices (490 cm 2 /C vs. 500 cm 2 /C), thus suggesting that the replacement of the ferrocene anodic component with the FD one, does not affect significantly the EC performances of the device in the visible wavelength range. Importantly, the coloration efficiency calculated at 1430 nm in the TV/Fc device is almost half that found in the TV/FD one (105 cm 2 /C vs. 180 cm 2 /C). In order to compare the performance of the EC devices it is of noteworthy importance the correlation between the measurement of the charge passed during the oxidation and reduction processes and the transmittance change at a specific wavelength. This correlation parameter is known as coloration efficiency (CE, η). The η is given by the equation: This also determines the amount of optical density change (ΔOD) as a function of the injected/ejected electronic charge (Qi) per unit area. Tb and Tc are, respectively, the transmittance in the bleached and colored states. Figure 3 shows the change in optical density at three wavelengths (670, 890 and 1430 nm) as a function of the surface charge density. The CE value, extracted as the slope of the linear fitting of the reported curves, for both systems (TV/Fc and TV/FD), at 670 nm, is as high as 500 cm 2 /C, which is an outstanding result compared with analogous systems [51][52][53]. The CEs registered for the NIR bands in the TV/FD system, due to the electrogeneration of the TV •+ /FD •+ and TV ++ /FD ++ species, respectively, are 180 cm 2 /C and 170 cm 2 /C, thus, approximately three times smaller than that of the calculated at 670 nm and due to the formation of TV •+ . Furthermore, the coloration efficiency of the band at 670 nm calculated for the TV/FD devices, is only slightly smaller than that calculated for the TV/Fc devices (490 cm 2 /C vs. 500 cm 2 /C), thus suggesting that the replacement of the ferrocene anodic component with the FD one, does not affect significantly the EC performances of the device in the visible wavelength range. Importantly, the coloration efficiency calculated at 1430 nm in the TV/Fc device is almost half that found in the TV/FD one (105 cm 2 /C vs. 180 cm 2 /C). Let's try to explain the above differences in the CE of the three bands. The infinitesimal change of absorbance at the wavelength of maximum absorption, at time t, is given by the Equation (5): where, , is the sample optical depth (cm), , and are the molar extinction coefficient (dm 3 mol −1 cm −1 ) and the infinitesimal change in the concentration of the radical cation species. Integration of Equation (2) gives the change in the optical density of the sample at time t: being 0 = 0. It follows that: Equation (7) tells us that the coloration efficiency is a function of the molar extinction coefficient of the active electrochromic species (the radical cation in our case) and of its concentration at time t. The molar extinction coefficient of the two species are 1.9 × 10 5 mol −1 dm 3 cm −1 and 1.2 × 10 4 mol −1 dm 3 cm −1 respectively for the TV •+ and FD •+ . In order to estimate the concentration of the radical cation at time t in the device, we can assume that the heterogeneous electron transfer reactions at the ITO electrode are of first order with rate coefficients . The rate of change of is therefore expressed as Let's try to explain the above differences in the CE of the three bands. The infinitesimal change of absorbance dA at the wavelength of maximum absorption, at time t, is given by the Equation (5): where, l, is the sample optical depth (cm), ε + , and dc + (t) are the molar extinction coefficient (dm 3 mol −1 cm −1 ) and the infinitesimal change in the concentration of the radical cation species. Integration of Equation (2) gives the change in the optical density of the sample at time t: being c + (0) = 0. It follows that: Equation (7) tells us that the coloration efficiency is a function of the molar extinction coefficient of the active electrochromic species (the radical cation in our case) and of its concentration at time t. The molar extinction coefficient of the two species are 1.9 × 10 5 mol −1 dm 3 cm −1 and 1.2 × 10 4 mol −1 dm 3 cm −1 respectively for the TV •+ and FD •+ . In order to estimate the concentration of the radical cation at time t in the device, we can assume that the heterogeneous electron transfer reactions at the ITO electrode are of first order with rate coefficients k e . The rate of change of c + (t) is therefore expressed as where, c p (t) is the concentration of the precursor species (the dication state for TV and the neutral state for FD) at time t. Integration of the differential Equation (8) leads to the desired equation that expresses the concentration of the radical cation species as a function of time: Equation (9) shows that the concentration of the monocation species is an exponential growing function of time with characteristic time k −1 e (s) and depends on the initial concentration of the precursor species c p (0). The higher is the kinetic constant of the heterogeneous electron transfer process, the faster is the coloration kinetics. Figure 4 displays the optical density change of the devices at the two wavelengths investigated, as a function of time. It clearly shows the exponential growth of the OD with time that can be very well fitted by the Equations (6) and (9) (R 2 > 0.99). The fitting results are summarized in Table 1.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 12 Figure 4 displays the optical density change of the devices at the two wavelengths investigated, as a function of time. It clearly shows the exponential growth of the OD with time that can be very well fitted by the Equations (6) and (9) (R 2 > 0.99). The fitting results are summarized in Table 1.    It is interesting to note that the rate coefficient k e for the reduction of the thienoviologen dication is almost 8 times larger than that for the oxidation of the neutral diarylamine species at the ITO electrode surface. This should be expected due to the charged nature of the first species which can migrate more rapidly toward the cathode surface upon application of the DC bias.
The difference in the CE observed for the two bands can then be rationalized on the basis of the Equation (4) rearranged as follows and the coloration kinetics data: It is clear from the Equation (10) that the faster is the kinetics of coloration and the higher is the extinction coefficient, the larger is the coloration efficiency, being the same the thickness of the absorbing layer (l) and the injected/ejected surface charge, as is the case for the thienoviologen compared to the arylamine.

Discussion
Herein, we presented all-in-one thermoplastic gel-based EC devices whose components are organic mixed valence compounds belonging to the class of small molecules. In the TV/Fc system we used the thienoviologen as the only electro-active/-chromic species while, in the case of TV/FD system we incorporated two electro-active/-chromic species as complementary cathodic (thienoviologen) and anodic (the fluorene-diarylamine derivative) components. Replacement of ferrocene with the MV fluorene-diarylamine allows a more efficient control of the solar radiation over a large part of the vis-NIR spectrum. Cyclic voltammetry and spectroelectrochemistry demonstrate the good electrochemical behavior of the anodic and cathodic compounds, highlighting that low operating voltages (potential difference as 1.4 V < ∆V < 2 V) are enough to reach high optical contrasts both in the visible range and in the NIR region, low response times and outstanding coloration efficiencies. The dual EC response of the Vis-NIR absorbing system coupled with a fine tuning of the coloration at low voltages indicate a great potential for practical use of these systems in many technological fields such as smart windows, electronic displays, optical communication, information storage, and thermal control.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-3417/10/23/8372/s1, materials: cyclic voltammetry of the ITO/EC/ITO devices containing the Fc/TV and the TV/FD systems ( Figure S1), transmittance changes and switching times for the TV/FD and TV/Fc devices (Table S1). Funding: The authors are grateful to the Ministero dell'Istruzione dell'Universita e della Ricerca Italiano (MIUR) and the University of Calabria for supporting this project in the framework of the ex 60% budget grant.

Acknowledgments:
The authors are grateful to the Ministero dell'Istruzione dell'Università e della Ricerca Italiano (MIUR) and the University of Calabria for supporting this project in the framework of the ex 60% budget grant.

Conflicts of Interest:
The authors declare no conflict of interest.