Photovoltaic Partner Selection for High-Efficiency Photovoltaic-Electrolytic Water Splitting Systems: Brief Review and Perspective

Photovoltaic-electrolysis water splitting (PV-EWS) is the most promising approach for high solar-to-hydrogen (STH) efficiency. The present PV-EWS systems achieve the highest STH performance by using a III-V triple-junction configuration, which, however, involves a complex and expensive manufacture process. Therefore, in this work, we demonstrate a III–V double junction device that can be used as an alternative to the III–V triple-junction device for high STH conversion efficiency of the PV-EWS systems. We estimate the STH performance via coupling world-recorded multi-junction photovoltaic (PV) and our experimented cell configurations with an EWS system. The results show that the III–V double junction, owing to the good trade-off between the efficiency loss and compensation, exhibits a higher STH efficiency than the III–V triple-junction. Furthermore, strategies for improving the efficiency of the III–V double junction device for low-cost PV-EWS system are discussed.


Introduction
Solar-to-hydrogen (STH) production via electrolysis water splitting (EWS) holds promise as a strategy for utilizing thoroughly renewable and sustainable solar energy sources, and convert these into storable and transportable energy sources without releasing harmful byproducts such as the CO 2 gas [1][2][3]. Diverse configurations of the EWS systems, which are based on photocatalysts and/or solar cells have been popu l a r l y s u g g e s t e d , s u c h a s p h o t o c a t a l y t i c , photoelectrochemical (PEC), and photovoltaic-electrolysis water splitting (PV-EWS) configurations [4]. However, the main challenge for STH production continues to be the significantly higher manufacture cost compared with the H 2 gas produced from the traditional fossil fuel: for example, the H 2 threshold cost according to the calculation by the United States Department of Energy has been $2.00 to $4.00 per gallon of gasoline equivalent [5], whereas the up-to-date production cost of H 2 from electrolysis is up to $3.26 to $6.62 per gallon of gasoline equivalent [6]. Therefore, numerous efforts have been focused on further enhancing the STH conversion performance using simply established EWS configurations for a cost-effective manufacture process. For example, some efforts that focused on improving the Si-based photocathodes achieved an STH conversion efficiency of 11.5% [7][8][9][10][11][12][13]. Table 1 summarizes briefly recent results related to these efforts. Others, which were devoted to wireless monolithic solar water splitting, exhibited simplicity and hence, costeffectiveness [3,14,15]. Further efforts were dedicated to improving the power conversion efficiency (PCE) of the PV devices [16][17][18]. Some impressive results have been summarized in Table 2 [16,[18][19][20][21][22][23][24]. To date, the highest recorded efficiency of the state-of-the-art photocatalysts and PEC systems is 5% [25] and 12.4% [26,27], respectively. However, the most recent impressive efficiencies of 30% [19] and 24.4% [20] STH conversion were achieved by the PV-EWS systems, which were based on concentrator PV modules using III-V triple-junction solar cells directly connected with polymer electrolyte electrochemical cells. Although electricity-conversed PV devices and PV-EWS systems based on a III-V triple-junction configuration have achieved a world-record efficiency of 37.9% [28] and 30% [19], respectively, their complex manufacture process and high material cost limits the industrial applications of this configuration to specific fields such as aerospace.
Owing to its relatively simpler manufacture technology, the III-V double junction device is more commercially costeffective than the III-V triple-junction device. Moreover, the former can generate a high open-circuit voltage in the range of 2-2.5 V, which can sufficiently fulfil the practical minimum photo-voltage requirement of the EWS systems (which is normally ≥1.5 V) [23]. This V oc threshold of the III-V double junction device is as insignificant as that of the III-V triplejunction cells (which is normally in the range of 2.5-3.5 V) compared with the practical minimum required voltage (1.5 V) of the EWS systems; this limits the energy losses in the EWS process [23,29]. However, despite the apparent costeffectiveness, the capability of the III-V double junction cells to produce a high PV-EWS efficiency as that currently achieved by the III-V triple-junction cells remains unestablished. Table 2 presents some recent reports, which show that an efficiency as high as 18% was achieved by the PV-EWS system using a III-V double junction device. Owing to the recent persistent advantages of the light concentrator and manufacture technologies, the III-V double junction device is predicted to be competitive with the silicon PV in terms of efficiency and cost in certain applications [23].
Coupling the III-V triple-junction and double junction PV configurations separately with an EWS system, we demonstrated that the III-V double junction devices can achieve higher STH conversion performance, which indicates that despite their currently lower PCE, they hold promise as an alternative to the III-V triple-junction devices for high PV-EWS efficiency. Furthermore, we estimated and compared the STH conversion performance of a PV-EWS system based on the recently world-recorded cells with those of the laboratoryexperimented cells (i.e., the III-V triple-and double-junction devices). We also compared the STH conversion performance of the III-V materials with that of the Si-based materials: the results indicated that the former are appropriate for highefficiency and low-cost PV-EWS systems, owing to the good trade-off between energy loss and compensation. Moreover, the recent and persistent developments in the III-V device technology allow considerable room for further enhancement in the PV-EWS system performance.

Experiment
Si-based multijunction devices including hydrogenated amorphous Si (a-Si:H) / hydrogenated nanocrystalline Si (nc-Si:H) double junction, and a-Si:H / Si heterojunction (SHJ) double  4 ) and H 2 gases were used as precursors for the intrinsic absorption layers, and diborane (B 2 H 6 ) and phosphine (PH 3 ) gases were used as doping precursors for the ptype and n-type layers, respectively. In the a-Si:H / SHJ double junction, the a-Si:H (300 nm) top cell was directly deposited on the SHJ (180 μm) bottom cell, which functions as a substrate. For PECVD deposition, a radio frequency power source (13.56 MHz) was used for the p-type or n-type doped layers whereas a very high radio frequency power source (40 MHz) was utilized for intrinsic a-Si:H layers. In the III-V double junction and triple devices, the cells were implemented by the metal organic chemical vapor deposition (MOCVD) method: for the III-V double junction configuration, the p-type GaAs materials with (001) orientation were used as a substrate for the like-epitaxy MOCVD growth process of sequential layers including 3.0-μm thick p-GaAs bottom and 0.4-μm p-GaInP top cells, while for the III-V triplejunction configuration, p-type Ge material was used as a substrate. The cross-sections of all multijunction cell configurations are illustrated in Fig

Result and Discussion
The STH conversion efficiency (η) based on EWS is generally defined by the thermodynamic potential of 1.23 V, current density of the EWS (J WS ), the Faradaic efficiency (η FE ), and the incident irradiance power or input power (P in ), as expressed in the following relation: Generally, in a PV-EWS system, the P in is provided by the PV device system and therefore, P in nearly corresponds to the power input of the system. The relation between P in and PCE of the PV device system is well-known and defined as follows: where, J sc is the short-circuit current density, V oc is the opencircuit voltage, and FF is the fill factor.
Using Eq. (2), the value of η in Eq. (1) can be rewritten as a function of PCE and the cell parameters, as follows: Under ideal operation conditions, assuming maximum electrolysis efficiency, the PV-EWS system can operate under short-circuit conditions. This means that J WS is nearly equal to J sc and η FE is 100%. Therefore, Eq. (3) can be simplified and rewritten as follows: Herein, if we define E loss = 1.23 / V oc and E com = PCE / FF, and Eq. (4a) can be expressed as: Thus, Eq. (4a) shows that the efficiency of the PV-EWS system can be estimated using the basic cell parameters, i.e., V oc , FF, and PCE, of the PV device. Consequently, we estimated the STH conversion performance of some standard III-V and Si-based multijunction configurations based on their confirmed cell parameters [28,[30][31][32][33][34][35], as shown in Table 3. While the Si-based triple-junction configuration exhibited a low STH conversion efficiency of 12.2%, the Si double junction structure exhibited a higher STH efficiency of 16.58%. However, despite gaining higher STH efficiency, the V oc of the Si double structure was as low as 1.342 V, which is below the practical photo-voltage requirement (≥1.5 V) for a commercial PV-electrolysis device [16,29] and hence, insufficient for practical PV-ECP systems. To overcome this issue, some authors proposed the use of a DC/DC converter with a maximum power point tracker [36][37][38][39]. However, this complicated the design, increased power loss, and as a result, involved extra cost [36]. Nevertheless, it is possible to obtain a high STH conversion of over 17% via coupling with III-V multijunction configurations, owing to their high V oc of over 1.5 V. Moreover, the III-V double junction devices seemed to generate a slightly higher STH efficiency than that of the III-V triple configurations, although the latter exhibit a higher PCE than that of the former. In Eq. (4b), η depends on two parameters: E loss and E com , where E loss and E com represent the efficiency loss and compensation, respectively. While E loss can result from a high applied voltage, E com results from the high PCE of a PV device. Therefore, a high η value mainly depends on the tradeoff between E loss and E com . Moreover, the lower V oc in the III-V double junction PV devices results in lower E loss than the III-V triple-junction devices: The good tradeoff between E loss and E com results in a slightly higher η value for the III-V double junction devices than the III-V triple devices. This indicates that the manufacture of the III-V double junction configuration is more cost-effective than the III-V triple structure. This is because the former manufacture process involves a larger amount of, and more complex material than that of the latter.
Under practical operation conditions, it is difficult for a PV-EWS system to operate at the V oc of the PV device, and therefore, it should be controlled to operate at a maximum power point (P MPP ) of the PV devices [40]. Considering the same ideal operation conditions as previously mentioned, the value of J WS in this case is nearly equal to J MPP , and hence, Eq. (5) can be simplified as follows: Thus, Eq. (6) shows that the essential efficiency loss of a PV-EWS system is due to a V MPP value that is significantly higher than the thermodynamic potential (1.23 V) of the system, while the PCE can compensate for this loss. Therefore, we implemented multijunction PV devices and used Eq. (6) to estimate the PV-EWS performance. Figure 1 shows the cross-  section schematics of the multijunction devices including a-Si/ nc-Si double junction, a-Si/SHJ double junction, GaInP/GaAs double junction, and GaInP/GaAs/Ge triple-junction devices. The J-V curve characteristics of these devices along with their maximum power points (P MPP ) are illustrated in Fig. 2, and the device parameters and STH efficiency are shown in Table 4. In the a-Si/nc-Si double junction device, a V MMP as low as 1.1 V is apparently insufficient for the EWS voltage requirement although its estimated STH efficiency is 12.6%. While the a-Si top cell exhibits a high V oc of approximately 1 V, the low V oc (normally <0.6 V) of the nc-Si bottom cells remains a drawback for the total V oc in this device. Therefore, a total V oc enhancement can be affected by the SHJ bottom cell, which can currently reach a high V oc of over 0.7 V, resulting in a total V oc of 1.61 V. However, the a-Si/SHJ device exhibits a low J sc of 9.6 mA/cm 2 , resulting in a low PCE of 11.2%; such low PCE can insignificantly compensate the E loss and result in a low STH performance of 7.37%. It is noteworthy that the high efficiency of 14.3% derive from III-V double junction device. In addition, the III-V double junction device exhibits higher STH performance than that of III-V triple junction device. Furthermore, considering the commercial manufacture cost, the III-V double junction device is apparently more cost effective than the III-V triple device for a PV-EWS system. The above results emphasize the need for strategies to improve the efficiency of PV-EWS systems using a more costeffective III-V double junction technology. Besides, Eq. (6) shows that the STH efficiency of the PV-EWS system can be significantly enhanced by improving the PCE of a III-V double junction device while maintaining a nearly constant voltage. This can be fulfilled by a current state-of-the-art concentrator photovoltaics (CPV) technology, which illuminates a small area of the PV device with very highly concentrated suns; this is one of the most promising solar technologies that is low-cost and highly efficient [41]. The Fraunhofer Institute for Solar Energy Systems has developed the FLATCON® concentrator system [42], which is expected to be costcompetitive to the silicon PV technology in certain applications [43]. Although the FLATCON® concentrator tandem solar cells can produce a sufficient V oc > 2 V and high PCE efficiency for the PV-EWS systems [23], it is also important to consider a simple and inexpensive PV-EWS system for a real solution to STH production. The III-V multijunction devices are conventionally grown like-epitaxy on GaAs or Ge bulk substrates: the current state-of-the-art epitaxial lift-off (ELO) technology can release III-V multijunction thin film solar cells out of the underlying GaAs or Ge expensive bulk substrates and transfer the cells to diverse inexpensive substrate types such as a possible thin film, flexible plastic, glass, or metal substrate. For example, Sharp Corporation reported a 29.4% efficiency with a III-V double junction on a flexible metal film using the ELO technology [44,45]. Furthermore, an ultralight flexible III-V double junction device on plastic with high efficiency (≥28%) using a novel layer transfer technique has been reported [46]. These reports demonstrate that the ELO technology has the potential to reduce material cost and achieve high efficiency without significantly affecting the cell operation. In addition, the use of the ELO technique can directly attract the monolithic integration of the III-V double junction cell on PEC system to yield a simple, inexpensive, and high-efficiency STH production system [14,23]. The overall cost of a PV-EWS system for STH production depends on the cost of many different factors such as the material cost, manufacture process, installations, work effort, transportation, and maintenance. Many reports have predicted the targeted H 2 cost for STH production: for example, the European  Commission anticipates a cost of 3 € kg −1 by 2030, while the US Department of Energy anticipates 4 $ kg −1 by 2020 [47]. The value in range of 2-4 $ kg −1 is attributed to the possibly competitive to traditional H 2 production from fossil fuels [47]. However, a thorough estimation of the final prices for STH production of a persistently developing PV-EWS system is both difficult and unreliable. Nakamura et al. [20], who reported the 24.4% efficiency of the PV-EWS system, suggested that the PV-EWS system is one of the most realistic methods for future renewable STH production. Moreover, although the practical efficiency of a PV-EWS system remains debatable, PV-EWS systems using III-V double junction devices hold promise as a simple, inexpensive, and highly efficient system. In other aspects, the next PV generations such as perovskite, dye-sensitized and organic PV devices are great potential for low-cost PV technologies and thus PV-EWS system. However, these PV generations have suffered from considerably low performance, compared with Si-and III-V based PV devices, and are still under development for commercial-scale production. Consequently, few report has focused these PV kinds for a PV-EWS system. Their performance and stability need to further improvement for diverse applications in the future.

Conclusion
We estimated and compared STH the conversion efficiency of a PV-EWS system by coupling various multijunction PV devices with an EWS system: The multijunction PV devices included a III-V double, III-V triple-junction, and Si-based multijunction solar cells. The results showed that the a-Si/nc-Si and a-Si/SHJ double junction PV devices exhibited lower efficiency, while high efficiency was achieved using the III-V double and triple-junction PV devices. Especially, we demonstrated that the PV-EWS system using the III-V double junction solar cells exhibited higher STH conversion performance than that of III-V triple-junction devices. Furthermore, the III-V double junction PV devices required a simpler and inexpensive manufacture technique than the triple junction devices. These results demonstrate that III-V double junction can be used as an alternative for the III-V triple-junction PV devices for high-efficiency and low-cost PV-EWS systems. Practically, with the persistent development of the state-ofthe-art concentrate PV and ELO technologies, there is great potential for further improving the efficiency and costeffectiveness of the PV-EWS systems for STH production, which can be competitive to the traditional H 2 production from fossil fuels in the future.