High energy and high voltage integrated photo-electrochemical double layer capacitor

Alberto Scaliaabc, Alberto Varzi*ab, Andrea Lambertic, Elena Tressoc, Sangsik Jeongab, Timo Jacobad and Stefano Passerini*ab
aHelmholtz Institute Ulm (HIU), Helmholtzstrasse 11, 89081 Ulm, Germany. E-mail: alberto.varzi@kit.edu; stefano.passerini@kit.edu
bKarlsruhe Institute of Technology (KIT), P.O. Box 3640, 76021 Karlsruhe, Germany
cDepartment of Applied Science and Technology – DISAT, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy
dInstitute of Electrochemistry, Ulm University, Albert-Einstein-Allee 47, 89081 Ulm, Germany

Received 3rd January 2018 , Accepted 18th February 2018

First published on 19th February 2018

A novel, monolithic harvesting–storage (HS) device composed of a dye-sensitized solar cell (DSSC)-based module and a high voltage supercapacitor with impressive discharge capacity after photocharging is herein proposed. Both the harvesting and the storage sections are fabricated onto conductive glass substrates, paving the way to a smart and easy integration in window facades for energy self-sustainable buildings. In addition, the HS device can also be integrated in portable electronics or drive remote, off-grid sensor networks requiring high power intermittent electrical energy. The harvesting photovoltaic section is constituted by a series of four DSSCs integrated in a single W-type module while the storage section consists of an activated carbon-based supercapacitor (SC) utilizing Pyr14TFSI ionic liquid as the electrolyte. The testing of the two separated sections as well as of the integrated system is reported here. In particular, the integration is evaluated through photo-charge and subsequent discharge protocols performed at different galvanostatic currents, showing that the SC handles photo-charges up to 2.45 V while delivering discharge capacities exceeding 1.8 mA h (0.1 mA h cm−2) upon 1 mA discharge current. To the best of our knowledge this is a never reported before, absolute record value, for stable and reliable integrated HS devices.


In the last decade our society has been increasing its efforts to disengage its energy needs from fossil fuels since pollution and global warming may irremediably change the environment quality in the near future. In this perspective, power production policies to exploit renewable energy sources have already been established by many governments and international associations. Among the renewable energy sources, photovoltaic (PV) power generation is probably the most investigated in terms of research activity.1–4 This is mainly due to the possibility of exploiting cheap5 and environmentally friendly materials6 appealing for the third-generation PVs, in particular for applications where slightly lower efficiencies than those of silicon-based PVs do not negatively affect working operation. However, due to the intermittent and non-uniform nature of the sun's radiation, coupling of PVs with a storage element is of primary importance to effectively utilize the power when required. In addition, a storage section may facilitate the power transfer from the renewable energy sources to the electric grid, avoiding the problems related to transient periods in which the grid is highly loaded and, eventually, reducing the total power losses.7

Recently, the idea to integrate a PV solar cell with an electrical storage unit in a monolithic structure has evoked profuse interest in the research community.8 The majority of reported systems so far make use of a supercapacitor (SC) for the energy storage.9–11 This is obviously related to the simple configuration and less restrictive technical specifications (voltage, current) required by SCs compared to batteries. Furthermore, SCs are more appealing than batteries for harvesting storage (HS) devices because they can provide exceptionally long cycle life (1[thin space (1/6-em)]000[thin space (1/6-em)]000 cycles).12 This feature is particularly important for a PV-based integrated device since, obviously, a large number of cycles are expected in a single day. Additionally, SCs offer substantially high power densities,13 making them functional in different emerging and topical applications, such as consumer electronics, voltage stabilizers, grid power buffers and automotive applications. HS devices can also be smartly used in the majority of these fields making the integration in a monolithic arrangement very proficient, e.g., consumer electronics, such as mobile phones or cameras. Finally, SCs are substantially safer than batteries (at least when operated under moderate current loads), since no redox reactions involving the formation of dangerous compounds (e.g., the growth of Li metal dendrites or O2 evolution might occur during operation or when overcharging a lithium ion battery) take place. It should be considered that, in fact, the temperature of PVs increases under sunlight exposure and their proximity to the storage sections might trigger hazardous side-reactions.

Among the third generation PV technologies a good compromise between cost, efficiency and long life stability is represented by the so-called dye-sensitized solar cells (DSSCs). With the 14.3% (ref. 14) energy conversion efficiency record, DSSCs are amongst the most reliable alternatives to silicon-based PV technologies. This is mainly due to their remarkable capability to work well also with diffuse radiation and under low illumination conditions. Actually, the efficiency improvement of DSSCs under these particular environmental conditions is continuously progressing.15 Last but not least, the ease of fabrication is another factor explaining why DSSCs are the most studied PV technology for HS devices.

Recently, Lu et al.16 reported a stack-integrated HS device composed of one DSSC and one SC assembled on bi-polar anodic titanium oxide nanotube arrays, which showed an overall photoelectric conversion and storage efficiency (OPECSE) of 1.64%, with a maximum storage potential of 0.61 V. In the same work, four of these HS devices were also externally connected in series so that the storage section could be charged up to 2.5 V. However, the limited capacitance, i.e., stored charge, of the SC led to a discharge time of few seconds even for the lowest current applied (0.1 mA cm−2). An important result regarding the overall efficiency of DSSC-based HS devices was obtained by Yang et al.,17 achieving an OPECSE value of 5.12% with a maximum SC voltage during photo-charge of 0.72 V. Even higher OPECSE values are reported in the literature regarding perovskite solar cell (PSC)-based HS devices. Kim et al.18 reported an OPECSE of 10.97%, reaching a voltage plateau of 0.91 V during photo-charge, for the HS device employing a PVA/H3PO4-based supercapacitor. However, even if PSCs had achieved photo-conversion efficiencies (PCEs) higher than 20%, their stability in time appears to be critical due to rapid degradation, particularly at high temperatures and/or in high humidity environments.19 Therefore, DSSCs are presently a more reliable solution regarding the energy-harvester section.

The large majority of HS devices reported in the literature present a maximum voltage below 1 V. This is due to the presence of a single solar cell as the energy harvester. A value slightly below 2.3 V was reported only by Chien et al.20 They fabricated an HS device with a series of eight organic solar cell modules as the energy harvester and a graphene-based SC using 1 M Et4NBF4 in propylene carbonate (PC) as the electrolyte. Therefore, to the best of our knowledge, this is the only work in which a value higher than 1 V is obtained for a monolithic HS device. Nevertheless, as stated in their article, the limiting factor regarding the voltage obtained during photo-charge is the SC electrolyte (1 M Et4NBF4 in propylene carbonate). In fact, as they affirm, the maximum voltage is limited by the voltage beyond which the electrolyte reacts with the electrode and breaks down, irremediably reducing device lifetime. They also tried to put two SCs in series to increase the voltage output, but again evident stability problems were revealed from electrochemical measurements of the storage section and photocharge of the integrated device. Here we introduce an innovative highly stable HS device, which is completely fabricated onto a glass substrate (see Fig. 1). The harvesting section is composed of a W-type dye-sensitized solar module (DSSM) consisting of four cells connected in series, coupled with a SC employing activated carbon (AC) as the active material and the ionic liquid Pyr14TFSI (i.e., N-butyl-N-methylpyrrolidinium bis(trifluoromethanesulfonyl)imide) as the electrolyte. Compared to organic electrolytes (e.g., PC or acetonitrile), the use of an ionic liquid guarantees high thermal stability during the relatively high temperature sealing as well as operation. In addition, as already reported in the literature by Chien et al.,20 the use of organic electrolytes (e.g., PC) in photo-capacitors can lead to voltage saturation for a SC potential higher than approximately 2.2 V even with a much higher PV open circuit voltage (VOC) of the module (4.91 V, 8 series connected cells), thus demonstrating electrolyte instability and an overall irreversible behaviour.

image file: c8se00003d-f1.tif
Fig. 1 Schematic drawing and pictures of the assembly process for the integrated photo-electrochemical double layer capacitor.

In this work, the characteristics of the harvesting and storage sections were first individually tested. Subsequently, a compact monolithic device was developed by their integration. Finally, the energy harvesting and storage performance was tested acquiring the photocharge–discharge curves under 1 sun (100 mW cm−2) radiation intensity. By examining the feasibility of high voltage photo-capacitors, this research work is of primary importance because only such an integrated device can address the energy needs of real applications, requiring higher operating voltages than the VOC of a single solar cell.



Sodium iodide (NaI) and iodine (I2) were purchased from Sigma-Aldrich (Milan, Italy). Methoxypropionitrile (CH3OCH2CH2CN) and 4-tert-butylpyridine (TBP) were purchased from Merck. TiO2 Paste DSL 18NR-AO was purchased from Dyesol. The sensitizing dye, cis-bis(isothiocyanato)(2,2′-bipyridyl-4,4′-dicarboxylato)(4,4′-di-nonyl-2′ bipyridyl)ruthenium(II) (Z907, Ruthenizer 520-DN) was purchased from Solaronix.

Activated carbon (AC, P4) was obtained from SGL Carbon (Germany),21,22 while conductive carbon (CC, Super C65) and sodium carboxymethyl cellulose (CMC, Walocell CRT 2000) binder were obtained from Imerys Graphite & Carbon and Dow Wolff Cellulosics, respectively. The ionic liquid electrolyte was synthesized in-house as previously described in the literature.23,24

For the device manufacturing, fluorine-doped tin oxide (FTO)-coated glass (sheet resistance 7 Ω sq−1) and Meltonix 1170-60 films (60 μm) were purchased from Solaronix (Aubonne, Switzerland).

Dye-sensitized solar module fabrication

Conducting glass plates were used as substrates for the DSSM. The forward plate was 4 × 4 cm2 while the backward plate was 4 × 5 cm2, in order to have additional space for contacting the electrodes. Four inlet holes (1 mm diameter) were drilled into the backside substrate for electrolyte filling. The FTO was properly patterned with a diamond tip in order to electrically disconnect the counter electrode from the photoanode of the contiguous DSSC (see Fig. 1S in the ESI).

First, the glass plates were rinsed in acetone and ethanol in an ultrasonic bath for 10 min. Afterwards, 3 and 10 nm-thick Pt films were deposited onto the counter electrode regions of the front and back side glass substrates, respectively, by means of sputtering (EM ACE 600, LEICA); using a 30 mA current and 0.04 nm s−1 deposition rate. To deposit the semiconductor paste layer, adhesive tapes (80 μm thick) with circular holes (two 10 mm for the front illuminated side and two 12 mm for the back illuminated side) were placed on the glass plates. The paste was then deposited in the holes by means of the doctor blade technique. The two substrates were then calcined at 500 °C for 30 min in order to remove the solvent and binders and to electrically interconnect the TiO2 nanoparticles. The sintered electrodes were then soaked for 15 h in 0.3 mM Z907 solution in ethanol. The devices were sealed utilizing 2 × 60 μm-thick Meltonix films with a hot press set at 120 °C. The liquid electrolyte, 0.45 M sodium iodide, 0.056 M iodine and 0.55 M 4-tert-butylpyridine dissolved in methoxypropionitrile (MPN), was injected by vacuum backfilling through the 4 holes that were subsequently sealed.

Supercapacitor fabrication

FTO glass was cut into 7.5 × 6 cm2 plates, which were used as substrates for casting the SC electrodes. Prior to use, the FTO glasses were rinsed with acetone and ethanol in an ultrasonic bath for 10 min.

A water based slurry was prepared by dispersing the electrode components (in total 1 g of dry mass) in 50 mL ultrapure (milliQ) H2O with the appropriate ratio. The chosen electrode composition was 90 wt% of activated carbon (AC), 5 wt% of conductive carbon and 5 wt% binder (sodium carboxymethylcelluose, CMC). The obtained slurry was then deposited onto the FTO substrate (deposition area: 4 × 4.5 cm2) via the doctor blade technique. Different active material loadings were deposited on the positive and negative electrodes by adjusting the coating thickness. In particular, 100 mg and 60 mg were deposited at the positive and the negative electrodes, respectively, to account for the slightly asymmetric capacitive response of the chosen AC in Pyr14TFSI electrolyte as determined via investigation in 3-electrode cells making use of an Ag wire quasi-reference electrode (results not reported). The resulting electrodes were firstly pre-dried overnight at 80 °C, and then dried in vacuum for 12 h at 130 °C.

Glass fiber separators (GF/D, 260 μm thickness, Whatman) were cut into 4.5 × 5 cm2 sheets and dried in a glass oven (Büchi) under vacuum for 12 h at 120 °C. The fabrication and sealing of the SC were performed in a dry room to reduce contamination by moisture. A separator sheet onto which some IL droplets were uniformly spread was placed between the SC electrodes. Meltonix films were used to thermally seal the device by means of the same hot press used for the DSSM. Upon pressing the SC electrodes, the liquid electrolyte uniformly impregnated the glass fiber separator and the AC layer.

Integration of the dye-sensitized solar module with the supercapacitor

DSSM and SC were integrated by means of a hot press placing two Meltonix films (each 60 μm thick) between the DSSM and the SC counter electrode, leaving the DSSM contact region outside the SC surface in order to easily connect harvesting and storage contacts during photocharging. In this way also a proper sealing of the backward DSSM holes was ensured.

Characterization of the devices

The DSSM performance was tested with a HelioSim-CL60 (Voss Electronics) solar simulator and a PG510/590 (Heka) potentiostat in order to measure the photo-generated current under simulated solar radiation with an AM1.5 spectrum. In this phase, the terminals of the potentiostat were connected to the DSSM electrodes of the HS device.

The SC performance was evaluated by means of constant current (CC) charge–discharge cycling and cyclic voltammetry (CV) performed on a programmable multi-channel potentiostat–galvanostat (VMP3, Biologic Science Instruments, France). All tests were performed at 20 °C in a climatic chamber (Binder, KBF-115). Capacitance values were calculated from the CC discharge curve (see Fig. 2A) according to the following equation:

image file: c8se00003d-t1.tif(1)
where i is the galvanostatic current (2 mA), s the slope in the linear voltage region (0–2.3 V), and A the active electrode area (18 cm2). Electrochemical impedance spectra (EIS) were recorded on the discharged SC at VOC, by means of an Impedance/Gain-Phase Analyzer 1260 (Solartron Analytical) between 0.1 MHz and 1 mHz, with an AC amplitude of 10 mV.

image file: c8se00003d-f2.tif
Fig. 2 Electrochemical characterization of the IL-based SC. (A) Voltage profile during constant current charge–discharge (CCCD) experiment recorded at 2 mA, (B) cyclic voltammetry performed at 1 mV s−1, (C) cycling stability CCCD test performed at 20 mA for 10[thin space (1/6-em)]000 cycles, (D) electrochemical impedance spectroscopy measurement (Nyquist plot).

Upon photo-charging, the voltage of the SC was measured with a potentiostat PG 510/590 (Heka), while the HS device was irradiated by the solar simulator.

Results and discussion

Characterization of the Pyr14TFSI based-SC

Fig. 2A displays the electrochemical behaviour of the SC during the constant current charge–discharge cycling tests with an imposed current of 2 mA (0.11 mA cm−2). The voltage profile shows the triangular shape typical of double layer capacitor charge/discharge. A slight deviation from the ideal capacitive response is noticed, however, due to the moderate ionic conductivity of the IL electrolyte, which is in full agreement with previous reports.25 From the discharge curve, a capacitance value of 233 mF cm−2 is found for the SC. This is an impressive value in the photo-capacitor literature and is also comparable to high quality ionic liquid-based SC devices.22 Moreover, if compared to the only other high voltage photocapacitor reported in the literature so far,20 we obtained a two order of magnitude higher value of capacitance (233 mF cm−2 with respect to 2.5 mF cm−2). Most importantly, the voltage profiles upon charge and discharge do not show plateaus associated with faradaic reactions within the maximum cut-off voltage set at 2.8 V, confirming the stability of the storage section upon photo-charging of the HS device. Again, with respect to Chien et al.,20 a much more triangular shape, characteristic of double layer capacitors, and an evidently much higher coulombic efficiency were obtained with nearly the same normalized imposed constant current. In addition, their CC measurement shows instability even below 2 V, demonstrating that our work is the first high voltage reliable photocapacitor, capable of consistently fulfilling a long cycling life.

In Fig. 2B the cyclic voltammetry (CV) measurement of the SC, performed at 1 mV s−1, is shown. Again, the theoretical rectangular shape results slightly distorted. Here the deviation from the ideal behaviour appears more pronounced since the used scan rate is “equivalent” to charging/discharging the SC in half the time compared to the CC experiment. Under such conditions, however, the somewhat resistive FTO glass substrate could be responsible for the limited power performance, as also suggested by the decrease in specific capacitance and increase in resistive behaviour when increasing the CV scan rate (see Fig. 3S).

In order to test the reliability of the storage section, a long-term cycling test, rarely performed in photocapacitor literature and never done previously for high voltage photocapacitors, was performed by CC cycling with an imposed current of 20 mA, limiting the charging voltage to 2.5 V (value of interest for this specific HS device, since the final voltage during photo-charge is expected to be ca. 2.45 V). Capacitance and energy density retentions are found to be 100% of the initial values after 10[thin space (1/6-em)]000 cycles. For the sake of completeness, energy retention as a function of the cycle number is also reported in the ESI (Fig. 4S). In Fig. 2C the coulombic efficiency as a function of the cycle number is displayed. After an initial adjustment, probably associated with the slow wetting of the porous activated carbon by the viscous IL electrolyte, the coulombic efficiency remained higher than 99% for 10[thin space (1/6-em)]000 cycles, demonstrating impressive long-term stability. EIS measurements were also performed in order to further characterize the storage section (see the Nyquist plot in Fig. 2D, and the related equivalent circuit). As suggested by Zhang et al.,26 the equivalent series resistance (ESR) was evaluated using the real part of the impedance at 1 kHz and by the prosecution of the 45° region with the x axis. In both cases an ESR value of 26.8 Ω was found. Although for conventional EDLCs (with electrodes coated on an Al current collector) the major contribution to the ESR arises from the electrolyte bulk resistance, the high ESR value detected in the cell herein reported is primarily due to the high resistivity of the FTO glass support. At high frequencies, the intercept with the real axis can be associated with the bulk cell resistance (R1). Then, a semicircle associated with the carbon–FTO glass interface is observed (modelled by a R2|Q2 element, where Q stands for a constant phase element). The intermediate frequency region shows a linear behaviour with a slope of approximately 45°, which can be associated with the limited ion diffusion in the pores of the electrodes, and therefore described by a Warburg element (ZW).27 At very low frequencies, the EIS spectrum is not entirely vertical, meaning that the response of the SC deviates from that of an ideal capacitor, as already observed in the CC and CV tests. For this reason, a constant phase element (Q3) more accurately describes this behaviour. Among other factors, the limited ionic conductivity of the electrolyte and electronic conductivity of the glass conductive substrate play the major role in such a deviation.

Characterization of the DSSM

The harvesting part of the HS device is constituted by a W-type, four cell series DSSC module. In this type of module half of the DSSCs are front-illuminated, while the others are back-illuminated, which results in different incident power densities. To ensure that all cells provide comparable current values, the front- and back-illuminated cells had different active surfaces, as already reported in the literature.28 Specifically, front-illuminated cells were fabricated with a diameter of 10 mm while the back-illuminated cells had a diameter of 12 mm, resulting in a back-/front- areal ratio of 1.44. Also, the area of the cells was rather small in order to limit the photo-charging current of the SC. This choice was dictated by the rather low conductivity of the FTO current collector.

Given a fixed number of cells in series, the maximum achievable voltage is fixed since VOC is not a function of the active surface or illumination intensity. However, capacitance and capacity decrease for higher current density values.22 In addition, due to the Joule-effect a lower heating rate is obtained. Nonetheless, DSSM active surface equalling that of the SC would be feasible employing more conductive SC current collectors. Accordingly, the charge time would be proportionally decreased.

MPN was selected as the solvent for the electrolyte because of the higher boiling point compared to the most commonly reported electrolyte acetonitrile. As for the storage section, this choice was carried out to reduce/prevent electrolyte evaporation, during both the high temperature sealing process and the device operation. In fact, part of the electromagnetic radiation not converted into electrical energy is absorbed as heat, eventually leading to the PV module temperature rise. To date, the long-life limiting factor of DSSCs is the leakage of the liquid electrolyte.19 Therefore, the less volatile solvent appeared to be a worthy compromise between high reliability, stability and efficiency.

Fig. 3A shows the IV characteristics of the DSSM, obtained under 1 sun (100 mW cm−2) illumination conditions, while the most relevant photovoltaic parameters are reported in Table 1. The total efficiency was evaluated according to the following equation:

image file: c8se00003d-t2.tif(2)
where VOC is the open circuit voltage of the W-type module, ISC is the short circuit current, FF is the fill factor, and G is the incident power density radiation (100 mW cm−2). VMP, IMP and PMP are the voltage, the current and the power provided by the W-type module in the point of maximum power, respectively.

image file: c8se00003d-f3.tif
Fig. 3 (A) Current–voltage measurement of the DSSM performed under 1 sun illumination condition, (B) power-voltage curve of the DSSM.
Table 1 Peculiar photovoltaic parameters of the DSSM collected under 1 sun illumination condition
VOC [V] ISC [mA] FF IMP [mA] VMP [V] η [%]
2.57 6.91 0.49 5.35 1.61 2.25

With respect to the only other work reported in the literature regarding high voltage photo-capacitors, the choice to use a DSSM brought about a substantial increase of the photovoltaic efficiency (2.25% instead of 1.57% obtained by Chien et al.20 with an organic photovoltaic module).

The DSSM reproducibility was investigated by fabricating several identical devices. All showed stable and reliable performance in terms of power output and efficiency. To this purpose the IV curves of three DSSMs are reported in the ESI (Fig. 5S).

Fig. 3B shows the power–voltage characteristics of the module. Here particular attention has to be paid to the shape of the curve and to VMP and IMP, since they can provide precious information about the photo-charging process. In fact, as the power has a maximum at 1.61 V, the corresponding maximum OPECSE has to be presumably investigated around this voltage value. The IV measurement test was repeated after storing the DSSM under dark conditions for 3 months. Noteworthy, the profile and efficiency remained nearly the same as those of the initial test.

Characterization of the HS device

The HS device was tested using a specific photocharging–discharging cycling protocol. Upon photo-charging 1 sun radiation intensity (100 mW cm−2) was used to facilitate the comparison with the literature, since almost every report uses this illumination condition. The SC electrodes were connected to the terminals of the DSSM through the galvanostat/potentiostat in order to follow the voltage during photo-charge and apply the galvanostatic discharge current (see Fig. 2S in the ESI). In Fig. 4A are reported a few of such cycles, photocharging curves followed by galvanostatic discharges, collected at increasing discharge current (ranging from 1 mA (0.0.056 mA cm−2) to 10 mA (0.56 mA cm−2)). The voltage of the SC approaches a plateau close to the VOC of the DSSM. The photocharging was stopped at 2.45 V, which is, to the best of our knowledge, the highest charge voltage ever reported for a photo-capacitor. In Fig. 4B the measurement was repeated stopping the photocharging process at 2 V, i.e., avoiding the aforementioned plateau and minimizing the charging time.
image file: c8se00003d-f4.tif
Fig. 4 Performance of the HS device. (A) Photocharging to 2.45 V and subsequent discharge curves at increasing current. (B) Photocharging to 2 V and subsequent discharge curves at increasing current. (C) OPECSE as a function of photo-charging time. (D) OPECSE as a function of the changing potential. (E) Photocharging to 2.45 V and discharge curves at increasing current after 1 h rest time. (F) Discharge capacity as a function of the discharge current for photocharging to 2.45 V, photocharging to 2 V, and photocharging to 2.45 V with 1 h rest time.

To fully understand the photo-charging efficiency process, the OPECSE value was calculated as a function of time according to the following formula:18

image file: c8se00003d-t3.tif(3)
where C is the capacitance of the SC evaluated from the galvanostatic (CC) discharge step (233 mF cm−2), ΔV is the difference between SC potential and its open circuit voltage (∼0.3 V, initial point of the photo-charging curve), G is the incident electromagnetic power density (100 mW cm−2), t is the photocharging time and S is the total active surface of the DSSM.

As reported in the literature,16,18,29 OPECSE increases quickly in the very first part of the photocharge (see Fig. 4C), since the energy stored increases with the square of the SC voltage which, in turn, increases almost linearly in this region. However, the increasing SC voltage results in a lower current provided by the DSSM, since the photovoltaic module works under higher voltage conditions, which is in accordance with its IV characteristics (Fig. 3A). Thus, the SC voltage increased more slowly. On the other hand, the power provided by the sun simulator is constant and the OPECSE denominator value increased linearly as a function of time. Thus, the maximum OPECSE (1.83%) is obtained after 7.9 min, i.e., before the SC voltage plateau. This is the highest efficiency ever attained for a monolithic photocapacitor with a high voltage storage section. In fact, the only other report dealing with a high voltage (ca. 2.3 V) photocapacitor20 reported the maximum efficiency of the PV section to be 1.57%, thus resulting in a lower OPECSE according to the equation:

OPECSE = ηPVηstorage (4)

From the energy point of view, having a maximum OPECSE at 1.63 V in such a relatively short time is obviously a positive aspect. However, it would be even more appealing to reach it at a higher voltage since, as already mentioned, real applications require voltages higher than the VOC of a single solar cell. Nonetheless, the HS storage capacity and OPECSE remain quite interesting upon short photocharges as shown in Fig. 4B. In fact, photo-charging up to 2 V corresponds to an OPECSE of 1.65%, which is 90% of the maximum efficiency value. In addition it corresponds to almost 80% of the final achievable voltage (VOC of the DSSM). These values are extremely important under real environment conditions considering that, according to the data so far reported in the literature, DSSCs usually improve their PV efficiency under low illumination conditions or diffuse radiation.30

The most important characteristics of the HS device, i.e., the SC voltage, time of photocharge and relative OPECSE, are summarized in Table 2. As it can be clearly noticed in Fig. 4D, where the OPECSE is plotted as a function of the voltage, the same efficiency value can be obtained at two distinct voltages (parallel line with respect to the x-axis). Consequently, in Fig. 4C two corresponding time values needed to obtain that particular efficiency can be found. In terms of real applications, the points at higher voltages (after the maximum OPECSE) are certainly more appealing.

Table 2 Peculiar parameters regarding the photocharging phase: ratio between potential and maximum achievable potential, photocharging time, OPECSE, ratio between OPECSE and maximum obtained OPECSE
Voltage [V] V/VOC [%] Time [m] OPECSE [%] OPECSE/max(OPECSE) [%]
1.63 63.4 7.9 1.83 100
2 77.8 13.8 1.65 90.1
2.45 93.4 91 0.41 22.2

Since often in real applications the energy stored is not needed immediately after the photocharge, a different test was performed in which the device was discharged at various constant currents 1 hour apart from the photocharge. This is also, to the best of our knowledge, a test never reported before for photocapacitor devices. The corresponding curves are presented in Fig. 4E. Overall, all the discharge curves subsequent to photocharge have a linear behaviour in all the three different measurements performed. This was not experienced by Chien et al.,20 where a fast decrease in voltage was measured. The authors attribute this behaviour to the porosity saturation effect. However, in this case the discharge phase lasts just for few seconds at a voltage higher than 1 V, and below 1 V proceeds with a completely different slope. In this perspective, we can affirm that our proposed structure is certainly more suitable for high voltage applications, and the only one to date fabricated that can be utilized in applications that must be driven at a moderate voltage for more than 1 m. The capacity of the HS recorded under various charge and discharge conditions is compared in Fig. 4F. In detail, the blue, green and red markers refer to the discharge capacity at different currents after, respectively, photocharging up to 2.45 V, photocharging up to 2.45 V followed by 1 h waiting period (in the dark), and photocharging up to 2 V. As expected, the delivered capacity decreased with increasing discharge currents due to increasing voltage drop. However, appreciably, it was found that the difference between discharge capacity with and without rest time is quite limited.

For the lowest currents applied (1–2 mA, i.e., 0.056 mA cm−2 − 0.11 mA cm−2) a discharge capacity of 0.1 mA h cm−2 was obtained after photo-charging up to 2.45 V and immediate discharge. This value is about twenty times higher than the best result (0.005 mA h cm−2) reported in the literature,20 especially considering the higher current used in this work (0.11 mA cm−2 versus 0.075 mA cm−2 in ref. 20), outlining the extraordinary performance of the HS device herein reported. In addition, it was found that increasing the value of the imposed discharging constant current of 1 order of magnitude a capacity retention of at least 50% was found. In the case in which the photocharging was stopped at 2 V the capacity retention approached 60%. This is again an interesting result, considering also that a higher voltage drop is obtained changing the discharge current from 1 to 10 mA (from 0.056 mA cm−2 to 0.56 mA cm−2) and proving that the energy recovery is also experienced for the highest current applied. In order to fully understand the energetic performance of the HS device, the voltage drop under different discharging conditions was evaluated, which is reported in the ESI (Fig. 6S). For all the three experiments, the voltage drop linearly increases with increasing the discharge current. Also, as expected, a higher voltage drop is observed when the photo-charge is limited to only 2 V, with respect to the case in which the photo-charge reaches 2.45 V. This is due to the larger total current variation (positive photo-current minus negative imposed discharge current) for the “2 V experiment” compared to the “2.45 V experiment”. In fact, as previously discussed, a pseudo-plateau was obtained when the charging was stopped at 2.45 V (see Fig. 4A). Thus, in the latter case the photo-charging current was almost null when switching from charge to discharge. When the device was discharged for 1 h after the photo-charge, the voltage drop appears to be even higher with respect to the other experiments. However, this increase is certainly related to the SC self-discharge that, even if limited (as testified by the small difference in recorded discharge capacity values, especially for the higher discharge current applied of 0.56 mA cm−2, see Fig. 4F), resulted in a decrease of the initial discharge voltage.


In this work we present an innovative, integrated energy harvesting and storage system obtained by coupling a DSSM constituted of four serially connected cells, and an electrical double layer supercapacitor with an ionic liquid-based electrolyte. The system achieved an outstanding photo-charging potential of 2.45 V, which so far has not been attained. Glass conductive substrates were employed for both the harvesting and storage sections. Different photoanode active surface areas and counter electrode Pt film thicknesses were employed for the serially connected front- and back-illuminated solar cells in order to achieve similar photo-generated currents. The storage section consisted of a supercapacitor employing Pyr14TFSI as the electrolyte, enabling high efficiency operation under the charging conditions provided by the DSSM. The operation of both the SC and DSSM units was studied through constant current charge–discharge, cyclic voltammetry, electrochemical impedance spectroscopy (SC), and IV measurement (DSSM), respectively. Subsequently, the integration was tested via photocharge (under 1 sun illumination conditions) and discharge curves of the combined device. An impressive discharge capacity value slightly over 1.8 mA h (0.1 mA h cm−2) after photo-charge was attained under 1 and 2 mA (0.056 and 0.11 mA cm−2) current discharges, demonstrating the remarkable energy density of the HS system. In addition, discharge curves were also recorded after 1 h SC rest time, subsequent to photo-charging, proving only a slight decrease in discharge capacity. Besides the excellent performance, the reasonably easy fabrication of the here proposed integrated HS device provides very promising scenarios for a possible future industrial scale-up.

Conflicts of interest

There are no conflicts to declare.


A. S., A. V. and S. P. acknowledge the financial support of the Helmholtz Association. Layal Daccache and Julius Gröne are acknowledged for allowing the use of the solar simulator. Last but not least, the authors would like to thank SGL Carbon, Imerys Graphite & Carbon and Dow Wolff Cellulosics for kindly providing their materials.

Notes and references

  1. L. M. Gonçalves, V. de Zea Bermudez, H. A. Ribeiro and A. M. Mendes, Energy Environ. Sci., 2008, 1, 655 Search PubMed.
  2. H. Cheema, R. R. Rodrigues and J. H. Delcamp, Energy Environ. Sci., 2017, 10, 1764 CAS.
  3. M. A. Green, Prog. Photovoltaics Res. Appl., 2001, 9, 123–135 CrossRef CAS.
  4. H. S. Jung and J. K. Lee, J. Phys. Chem. Lett., 2013, 4, 1682–1693 CrossRef CAS PubMed.
  5. M. Gerosa, A. Sacco, A. Scalia, F. Bella, A. Chiodoni, M. Quaglio, E. Tresso and S. Bianco, IEEE Journal of Photovoltaics, 2016, 2514702,  DOI:10.1109/jphotov.
  6. F. Bella, C. Gerbaldi, C. Barolo and M. Grätzel, Chem. Soc. Rev., 2015, 44, 3431–3473 RSC.
  7. M. Farhadi and O. Mohammed, IEEE Transactions on Smart Grid, 2015, 6, 54–62 CrossRef.
  8. D. Schmidt, M. D. Hager and U. S. Schubert, Adv. Energy Mater., 2016, 6, 1–11 Search PubMed.
  9. M. Skunik-Nuckowska, K. Grzejszczyk, P. J. Kulesza, L. Yang, N. Vlachopoulos, L. Häggman, E. Johansson and A. Hagfeldt, J. Power Sources, 2013, 234, 91–99 CrossRef CAS.
  10. C. H. Ng, H. N. Lim, S. Hayase, I. Harrison, A. Pandikumar and N. M. Huang, J. Power Sources, 2015, 296, 169–185 CrossRef CAS.
  11. T. Chen, L. Qiu, Z. Yang, Z. Cai, J. Ren, H. Li, H. Lin, X. Sun and H. Peng, Angew. Chem., Int. Ed., 2012, 51, 11977–11980 CrossRef CAS PubMed.
  12. S. Zhang and N. Pan, Adv. Energy Mater., 2015, 5, 1–19 Search PubMed.
  13. P. Simon and Y. Gogotsi, Nat. Mater., 2008, 7, 845–854 CrossRef CAS PubMed.
  14. K. Kakiage, Y. Aoyama, T. Yano, K. Oya, J. Fujisawa and M. Hanaya, Chem. Commun., 2015, 51, 15894–15897 RSC.
  15. F. Bella, A. Sacco, D. Pugliese, M. Laurenti and S. Bianco, J. Power Sources, 2014, 264, 333–343 CrossRef CAS.
  16. J. Xu, H. Wu, L. Lu, S.-F. Leung, D. Chen, X. Chen, Z. Fan, G. Shen and D. Li, Adv. Funct. Mater., 2014, 24, 1814 CrossRef.
  17. Z. Yang, L. Li, Y. Luo, R. He, L. Qiu, H. Lin and H. Peng, J. Mater. Chem. A, 2013, 1, 954–958 CAS.
  18. J. Kim, S. M. Lee, Y.-H. Hwang, S. Lee, B. Park, J.-H. Jang and K. Lee, J. Mater. Chem. A, 2017, 1–7 Search PubMed.
  19. J. Gong, K. Sumathy, Q. Qiao and Z. Zhou, Renew. Sustain. Energy Rev., 2017, 68, 234–246 CrossRef CAS.
  20. C. T. Chien, P. Hiralal, D. Y. Wang, I. S. Huang, C. C. Chen, C. W. Chen and G. A. J. Amaratunga, Small, 2015, 11, 2929–2937 CrossRef CAS PubMed.
  21. C. Ramirez-Castro, C. Schütter, S. Passerini and A. Balducci, J. Power Sources, 2014, 270, 379–385 CrossRef CAS.
  22. A. Varzi, C. Schütter, J. Krummacher, R. Raccichini, C. Wolff, G. T. Kim, S. Rösler, B. Blumenröder, T. Schubert, S. Passerini and A. Balducci, J. Power Sources, 2016, 326, 162–169 CrossRef CAS.
  23. G. B. Appetecchi, S. Scaccia, C. Tizzani, F. Alessandrini and S. Passerini, J. Electrochem. Soc., 2006, 153, A1685 CrossRef CAS.
  24. M. Montanino, F. Alessandrini, S. Passerini and G. B. Appetecchi, Electrochim. Acta, 2013, 96, 124–133 CrossRef CAS.
  25. A. Varzi, A. Balducci and S. Passerini, J. Electrochem. Soc., 2014, 161, 368–375 CrossRef.
  26. S. Zhang and N. Pan, Adv. Energy Mater., 2015, 5, 1–19 Search PubMed.
  27. R. Kötz, M. Hahn and R. Gallay, J. Power Sources, 2006, 154, 550–555 CrossRef.
  28. F. Giordano, E. Petrolati, T. M. Brown, A. Reale and A. Di Carlo, IEEE Trans. Electron Devices, 2011, 58, 2759–2764 CrossRef CAS.
  29. J. Xu, Z. Ku, Y. Zhang, D. Chao and H. J. Fan, Adv. Mater. Technol., 2016, 1–5 CrossRef.
  30. A. Scalia, F. Bella, A. Lamberti, S. Bianco, C. Gerbaldi, E. Tresso and C. Fabrizio, J. Power Sources, 2017, 359, 311–321 CrossRef CAS.


Electronic supplementary information (ESI) available. See DOI: 10.1039/c8se00003d

This journal is © The Royal Society of Chemistry 2018