You are on page 1of 13

Energy Conversion and Management 276 (2023) 116567

Contents lists available at ScienceDirect

Energy Conversion and Management


journal homepage: www.elsevier.com/locate/enconman

Spectral-splitting concentrator agrivoltaics for higher hybrid solar energy


conversion efficiency
Zhisen Zhang a, Fangxin Zhang a, Wei Zhang b, Ming Li a, Wenjun Liu b, Altyeb Ali Abaker Omer a,
Jianan Zheng a, Xinyu Zhang a, b, 1, Wen Liu a, b, 1, *
a
Department of Optics and Optical Engineering, University of Science and Technology of China, Hefei 230026, China
b
Xiong’an Institute of Innovation, Xiong’an 071799, China

A R T I C L E I N F O A B S T R A C T

Keywords: By integrating solar cells into agricultural lands, agrivoltaics is a promising route to widely deploy photovoltaics,
Spectral splitting and it can reduce land competition for food and energy production. Although various agrivoltaics have been
Solar energy conversion proposed, their optical properties and effects on plant growth remain disputed, which largely limits the wide­
Agrivoltaics
spread deployment of agrivoltaics. Here, we present spectral-splitting concentrator agrivoltaics (SCAPV), which
Semitransparent solar cells
can effectively harvest photosynthetically excess light energy for photovoltaic power without compromising crop
Tunable photovoltaics
productivity. This system transmits a selected light spectrum for plant growth while reflecting the remaining
spectrum for electricity generation. Field trials show that SCAPV can increase plant biomass by 13% and decrease
plant heat dissipation by approximately 50%. Meanwhile, SCAPV has a maximum photovoltaic power conversion
efficiency (PCE) of 9.9%. The overall solar energy conversion efficiency is higher than the theoretical limitation
of photosynthesis. Economic analysis shows that SCAPV has a levelized cost of energy (LCE) of $0.033/kWh. Our
results show that with proper spectral management, only a portion of sunlight is sufficient to support or even
enhance plant growth. Compared to other spectral-splitting agrivoltaics, SCAPV is a scalable technology with
flexible spectrum and high transparency and can be implemented using low-cost components.

light than it can use and has different protective mechanisms to avoid
1. Introduction damage from overexcitation [9,10]; a major path to dissipate the excess
energy is through dissipation in the form of heat [11], which accounts
Solar energy is known as the most abundant and reliable energy for over 50 % of the light energy absorbed by photosystem II (PSII)
source. Photosynthetic organisms and photovoltaic devices are the most chlorophylls [12].
scalable and mature ways to utilize solar energy. For the present silicon As photosynthesis is inefficient, various agrivoltaics can be seen as a
solar cells, the highest confirmed efficiency is 26.7 ± 0.5 % [1]. The game-changing strategy for hybrid light use systems on agricultural
theoretical limit of photosynthetic efficiency (εc, from solar energy into lands [8,13–15]. Agrivoltaics is also important because the rising de­
biomass) is 4.6 % for C3 plants and 6 % for C4 plants. Under common mand for PV will lead to heightened land use competition between en­
field conditions, crops typically show an εc of less than 1 % [2–5], which ergy and food, especially in densely populated areas [16–18]. Although
is considerably lower than that of solar cells. This inefficiency of large-scale PV power stations in remote areas are feasible and
photosynthesis is largely due to the inefficient use of photons. For plants, economical, long-distance energy transmission is expensive, risky and
sunlight is too bright and encompasses the broadband spectrum. Plant associated with substantial losses [19]. Previous studies have estimated
pigments selectively absorb blue, red and far-red (FR) light (Fig. 1a) that the potential of deploying PV is greatest on agricultural lands
[6–8]. In addition, even for photons within the radiation band [20,21]. In addition, agrivoltaics have been proven to improve the
comprising photosynthetically active irradiation (PAR), a large pro­ microclimate [16,22].
portion of oversaturated light energy is mostly dissipated as heat However, the opaque agrivoltaics currently used in China and all
(Fig. 1b), resulting in a low photosynthetic εc [2,9,10]. In fact, the plant over the world usually have limitations, as they cause spatial and tem­
photosynthetic apparatus usually encounters and absorbs much more poral shading, which usually decreases agricultural productivity

* Corresponding author at: Department of Optics and Optical Engineering, University of Science and Technology of China, Hefei 230026, China.
E-mail address: wenliu@ustc.edu.cn (W. Liu).
1
The two authors have the same contribution to this study.

https://doi.org/10.1016/j.enconman.2022.116567
Received 22 September 2022; Received in revised form 17 November 2022; Accepted 10 December 2022
Available online 20 December 2022
0196-8904/© 2022 Elsevier Ltd. All rights reserved.
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

Nomenclature PTC parabolic trough concentrator


Fv/Fm the maximum photosynthetic efficiency of PSII
PV photovoltaic NPQ nonphotochemical quenching
CPV concentrating photovoltaic PPFD photosynthetic photon flux density
CPVT concentrating photovoltaic/thermal G solar radiation flux (W/m2)
SCAPV spectral-splitting concentrator agrivoltaics Τ temperature (K)
MPF multilayer polymer films Pm maximum power point (W)
εc photosynthetic efficiency Vm the voltage at the maximum power point (V)
PAR photosynthetically active irradiation Im the current at the maximum power point (A)
PSII photosystem II Isc the short-circuit current (A)
OPV organic photovoltaics Voc the open-circuit voltage (V)
DSSC dye-sensitized solar cells Id the dark saturation current (A)
LSC luminescent solar concentrators Eg bandgap energy (eV)
PVSC perovskite solar cells RS spectral response of silicon cells (A/W)
a-Si amorphous silicon solar cells Wp cost per Watt of PVs ($)
PCE power conversion efficiency (of solar cells) LCE levelized cost of energy ($/kWh)

[23–25]. Due to the spectral selective absorption properties of plants absorptance; the spectral property of PVSCs comes from tuning the band
(Fig. 1a), strategy of designing various spectral-splitting agrivoltaics has gap through element control; to make it more adjustable, structural
also been actively explored, such as semitransparent organic photovol­ management, such as micro-structured islands [34], or light manage­
taics (OPVs) [26–30], dye-sensitized solar cells (DSSCs) [24,31–33], ment, such as optical filters, must be engaged [42].
perovskite cells (PVSCs) [34,35], luminescent solar concentrators (LSCs) Meanwhile, according to our literature survey on the knowledge gap,
[36,37], amorphous silicon solar cells (a-Si) [38,39], and other solutions some studies did not conduct any test to evaluate the effects of these
[40,41]. These solar cells under this strategy usually have wavelength- technologies on plants [28,33,34,43,44]. Although several other studies
selective transparency and have been of particular interest recently. conducted plant tests, they were usually qualitative without statistical
However, the average transmittance of these solar cells is usually low, repeat and evaluated in a short-term period [30,45,46], using artificial
the sizes of these laboratory-stage PV panels are usually small (hence light sources [47], or cultured in pots in greenhouses [45,48]. In general,
unscalable for massive plant trials), and the spectral tuning capabilities for currently available semitransparent solar cells, there remains dispute
are inflexible. In fact, the optical properties of a-Si are inherently diffi­ about plant growth [26,38,47,49]. As maintaining crop production is of
cult to regulate, including fixed transmitting spectra and low trans­ high priority for agrivoltaics, the prerequisite of balancing agricultural
parency; among the few reports thus far, the effects on plants have been and electrical yields brings critical challenges to designing spectral-
negative [38]. OPVs have the property of narrow-band absorption, splitting agrivoltaics.
which is not conducive to high efficiency but brings the chance to In previous publications, the design concept of spectral-splitting
flexibly adjust spectra; by adjusting the photoactive layer, OPVs can concentrator agrivoltaics (SCAPV) was also proposed, which holds po­
allow PAR to pass through and use only UV or IR for power generation tential for high transparency and flexible spectra [19,50–52]. This
[29]; new donors and acceptors can also allow a green absorption peak concept usually contains a parabolic trough concentrator/compound
[27], which makes the transmitting properties closer to the spectra in parabolic concentrator (PTC/CPC) and solar spectral-splitting dichroic
Fig. 1a. However, the transparencies of OPVs are mostly limited to low mirrors and supports photosynthesis with a selected portion of the
percentages of less than 10–20 %, which needs further improvement. incident sunlight while generating electricity with the excess light.
DSSCs are electrochemical devices that use light-absorbing dye mole­ However, these were usually notional or laboratory-stage prototypes,
cules to generate electricity. Although DSSCs have high transparency and the in situ efficiencies and productivities are still unclear, which
[14], the absorption property is adjusted by dye sensitizers, which is less largely limits the utilization of these systems. In this study, the optical
flexible. PVSCs also have high transparency due to their low properties and photosynthetic and photovoltaic efficiencies of SCAPV

Fig. 1. Feasibility and schematic of the spectral-splitting strategy in agrivoltaics. (a) Absorption spectra of photosynthetic pigments [6–8]. (b) Light response
curve of photosynthesis. An represents net photosynthetic rates. (c) The transmission spectrum of multilayer polymer films (MPF) and the response spectrum of
crystalline silicon.

2
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

are experimentally evaluated on a scaled-up SCAPV field. In the field makes SCAPV scalable. We have thoroughly studied the characteristics
trial, the SCAPV shows a short-term PCE of 9.9 % and gains an in situ and design of MPF [56], and the overall cost of MPF could be as low as
electrical power generation of ~ 10.86 kWh/m2 within the lettuce $1.0 USD per m2 per polymer film-processing study. As a proof of
growth period of 50 days; meanwhile, the SCAPV increases the crop concept, commercially available MPF (3 M Chill, 3 M PR90, USA) was
productivity of multiple plant species. Most notably, the overall effi­ used herein. The MPF has high transmittance (~87 %) at 397–493 nm
ciency breaks the theoretical limitation of photosynthesis. We also and 604–852 nm, and the average reflectance of other bands is over 90
observe an improved microclimate. Furthermore, based on chlorophyll % (Fig. 1c), which is generally in accordance with the absorption bands
fluorescence measurements, we observe a decreased plant heat dissi­ of plant photosynthesis, except that it transmits more infrared light than
pation and photoinhibition; we conclude that the essence of maintaining can be used (750–852 nm). In the future, customized rather than
production relies on the fact that the spectral-splitting strategy can commercially available MPF can better match the absorption spectra of
release the plant photoprotection and cleverly turn the photosyntheti­ plants.
cally excess radiation into a significant amount of electricity. Some semitransparent PV technologies have optical performances
independent of the incidence angle of the light, such as OPV and DSSC.
2. Materials and methods Multilayer interference films are, however, known to show angular
dependence (20, 23). To reduce this effect and achieve satisfactory solar
2.1. Spectral-splitting concentrator agrivoltaics (SCAPV) splitting and tracking performance in the field with diurnal changes in
solar angles, the bent paraboloidal glass panel of a PTC is used as a
2.1.1. System description substrate for the flexible MPF and as a reflector to direct the reflected
The SCAPV described herein was developed based on multilayer bands toward focal-point PV cells, thus creating the SCAPV. Further­
polymer film (MPF), PTC, and silicon cell (Fig. 2). The flexibility in size, more, silicon cells are used in the SCAPV; although multijunction III-V
transparency and spectrum of SCAPV relies on MPF, which has cells hold potential for higher efficiency, silicon cells are cost competi­
wavelength-selective interference effects through scaling layer thick­ tive for a scalable solution, and can maintain performance under high
ness; light of particular wavelengths (pass bands) can be transmitted by incident radiation within a concentration ratio of 3–100 [57].
adjusting the composition and thickness of the polymer materials
[53,54]. Two polymer materials with different optical refractive indices 2.1.2. Installation details
can be alternately superposed layer by layer with prescribed gradient, In this study, a large-scale project (128 × 81 W) was constructed to
leading to a broad pass band (from blue to far red) stack and a green- test both PV and plant productivity. The experimental site in Fuyang
reflective stack; multiwavelength selective MPF for photosynthesis can (32.93◦ N, 115.80◦ E) has a warm temperate subhumid monsoon climate,
be generated by coextruding these stacks. Compared to expensive characterized by significant monsoons, four distinct seasons, a mild
vacuum-depositional dielectric films, MPF fabricated by polymer coex­ climate, and moderate rainfall (annual average rainfall of 820–950 mm).
trusion and stretching is characterized by low cost and production in The PTC of the SCAPV has a 45◦ rim angle and geometrical concentra­
large coverage areas [55,56], which is the essential component that tion ratio of 6.7 suns. The extreme tracking angle is ± 45◦ from east to
the west and 10◦ north and 35◦ south from north to south. The me­
chanical structure of the SCAPV is composed of welded angle steel and
steel plates and then assembled with screws (Fig. 3a). The SCAPV is also
equipped with a dual-axis solar tracking system to generate more power
throughout the day. According to the nature of the solar angle variation,
the tracking mode is 1.25◦ every 5 min, and the movement of reflected
light on the surface of the PV module is within 23 mm, which is
acceptable.
The height of the column is 1.5 m, while the ground pile penetrates
into the earth at approximately 2 m, and the land is leveled on a large
scale before system deployment. The pile spacing is 6.1 m between the
north and south and 4 m between the east and west. The number of PTCs
installed on each row is 32 sets. The total land area used is 422.4 m2. The
length from north to south is 26.4 m, and the east to west length is 16 m
(Fig. 3b, c).
A double-sided monocrystalline PV from Sunpower (C60) with a
physical size of 1090 mm × 140 mm (24 pcs of 1/3 sliced solar cells in
series) is the electrical power generating component in the SCAPV. The
PV module comprises a front cover made of low-iron glass, the PV
laminate (EVA, solar cells and EVA) and a back aluminum substrate. The
cells were thermally bonded to the substrate using thermally conductive
and electrically insulating silica gel. The details and electrical properties
of the PV module under the AM1.5 spectrum (T = 298 K, G = 1000 W/
m2) are listed in Table 1.

2.2. Evaluation of electrical performance

In our SCAPV design, the PCE is defined as the generated Pm divided


by the solar radiation flux G reaching the effective area of the reflective
panel A:
Fig. 2. Spectral-splitting concentrator agrivoltaics (SCAPV). Multilayer
Isc ⋅Voc ⋅FF
polymer films (MPF) transmit a portion of the incident solar spectrum for plant PCE = (1)
G×A
growth, while other portions of the incident solar spectrum are reflected by the
PTC and focused on solar cells for electricity generation. The essence is to mathematically and experimentally evaluate Pm

3
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

Fig. 3. Schematic of the large-scale SCAPV field. (a) 3D structure of the proposed SCAPV system. (b) Schematic drawing of the SCAPV field. (c) Photographs of the
SCAPV field in Fuyang (32.93◦ N, 115.80◦ E), China.

2.2.1.1. Calculation of ultimate efficiency limits. The ultimate PV effi­


Table 1
ciency limits were calculated based on the S-Q limit theory. For the
Properties of the PV module.
standard diode, the current in a single-junction solar cell is given by
Electrical properties Pm Vm Im Voc Isc [58,59]:
25 Wp 11.9 V 2.1A 15.9 V 2.4A ( )
Temperature coefficient V:-0.27 %/℃; P:-3.2 %/℃
qV
I = Isc − Id exp( )− 1 (2)
PV module size 1090 mm × 140 mm × 48 mm nkT
PV module structure 5 stacks: glass (3.2 mm) - EVA (0.5 mm) - solar cells
(0.165 mm) - EVA (0.5 mm) - backplane (0.35 mm); where Isc is the short-circuit current, which is:
aluminum heat sink and silica gel ∫
G(λ)
Isc = q × Q = q dλ (3)
qE
under the selective bands of the spectral-splitting strategy.
where q is the elementary charge and E is the energy of a photon.
Id is the dark saturation current, which is:
2.2.1. Mathematical modelling of PV efficiency.
The seminal research of Shockley and Quisser calculated the theo­ Id = I0 exp(−
qEg
) (4)
retical efficiency limit of solar cells (S-Q limit) [58]. However, this was nkT
under the assumption of broadband sunlight and ideally square-shaped where the thermodynamic limit of I0 is
external quantum efficiency (EQE). To estimate the ultimate efficiency
limits of the split sunlight and the ideal efficiency of the SCAPV, the I0 =
2πkTq3 Eg 2
(5)
following assumptions are made: h3 c2
The electrical efficiency can be determined from:
• AM1.5 solar irradiation is concentrated by an ideal concentrator with
negligible optical losses. max(V × I)
PCE = ∫ ∞ (6)
• The spectral management is ideal so that the transmittance at 0
G(λ)dλ
397–493 nm and 604–852 nm is 100 %, and the other bands are The original S-Q limit assumed that the sun had a 6000 K blackbody
completely retained for PV, without optical losses. spectrum. Here, the incident sunlight G(λ) is assumed to be the AM1.5G
• The calculation of the ultimate efficiency limits is based on an ideally spectrum. The solar cell temperature is assumed to be 300 K. An in-
square-shaped EQE, corresponding to 2.2.1.1. house Python code was developed to calculate the ultimate efficiency
• The calculation of the ideal PV efficiency of the SCAPV is based on limits as a function of bandgap energy by numerically maximizing the
silicon cells with a measured spectral response curve, corresponding (V × I) for each bandgap (100 sampling points in the range of 0.4 ~ 3.0
to 2.2.1.2. eV).

2.2.1.2. Calculation of the ideal PV efficiency of the SCAPV. Then,

4
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

another model was performed to evaluate the ideal PV efficiency of the the aboveground dry biomass of the 12 representative plants under
SCAPV by calculating the efficiency loss based on the measured spectral SCAPV or in the open air was determined at 7 weeks after planting. The
response curve (RS(λ)) of the silicon cells. biomass was harvested and dried at 70 ◦ C in an oven for 3 days to attain
According to Eq. (2) a constant weight.
( )
Voc qVoc
= Isc − Id exp( )− 1 (7) 2.3.2. Chlorophyll fluorescence measurements
R nkT
Chlorophyll fluorescence was measured on the youngest fully func­
Voc is the open-circuit voltage, and when R = ∞, we can calculate tional leaf of lettuce plants using a Mini-PAM II (Walz, Effeltrich, Ger­
that [50]: many). Briefly, the maximum photosynthetic efficiency of PSII was
determined as Fv/Fm = (Fm – F0)/Fm, where Fm is the maximal fluores­
kT Isc + Id kT Isc
Voc = ln ≈ ln (8) cence level under a saturating pulse of 5000 μmol m− 2 s− 1 and F0 is the
q Id q Id
ground fluorescence in dark-adapted leaves. Fv/Fm was measured both
in the early morning (4:00 a.m.) when plants were fully dark-adapted
kT a − ln(a + 0.72)
Pm = Isc ⋅Voc ⋅FF = Isc a (9) overnight and at daytime (7:00 and 13:00) when plants were manu­
q a+1
ally dark-adapted for 30 min.
where a is approximate to a constant After measuring the instantaneous fluorescence level F under
Isc ambient light, a saturating pulse of 5000 μmol m− 2 s− 1 was applied to
a = ln ≈ 16.1 (10) measure Fm′ . The effective PSII quantum yield Y(II) was calculated using
Id
the following formula: Y(II) = (Fm′ - F)/Fm′ . Furthermore, the non­
For the short-circuit current, the voltage V is zero and photochemical quenching NPQ was calculated as: NPQ = (Fm - Fm′ )/Fm′ .

Isc = G(λ)⋅RS(λ)dλ (11) 2.4. Measurements of microclimate

An in-house code was developed to integrate G(λ)⋅RS(λ) and The microclimatic conditions under SCAPV or in the open air were
accordingly calculate the ideal loss of Isc (%) and hence the ideal loss of recorded by weather stations (Shandong Renke Control Technology,
Pm (%) due to solar splitting. The ideal PV efficiency of the SCAPV was China), including the ambient temperature and humidity, water evap­
calculated based on the nominal PV efficiency value from the supplier oration, soil temperature and moisture, and CO2 concentration. Soil
(22.5 %) and the ideal loss of Pm (%). temperature and moisture sensors were placed at different depths (5 cm,
10 cm) under SCAPV or in the open air, respectively. Evaporation con­
2.2.2. Experimental evaluation of the PV efficiency tainers were used to evaluate water evaporation without plant transpi­
To experimentally evaluate the electrical properties of the SCAPV, ration. The evaporation container (white plastic buckets) has a top
multiple I-V curves under concentrated sunlight in the outdoor SCAPV diameter of 26.0 cm, a bottom diameter of 22.0 cm, and a height of 28.5
field were tested between 10:00 and 16:00 on a typical sunny day using cm, with holes of 1 cm diameter at the bottom. Each microclimatic
a portable PV power meter (PROVA-1011, Taishi, Taiwan, China). The parameter was logged at a 10-minute interval. Solar radiation was
solar irradiation, ambient temperature and wind velocity varied within recorded by a PAR sensor (Li-190R, Li-cor, USA), and spectra were
the ranges of 725–900 W/m2, 303–309 K and 0–1.5 m/s, respectively. recorded by a handheld spectrometer (PLA-30, Everfine, China).
To derive the I-V curves in stable operating conditions, a single array
was stripped from the grid-connected SCAPV field and measured inde­ 2.5. Statistical analysis
pendently; the value of the external load was gradually varied between
its minimum and maximum values to cover the entire operating range of Datasets were processed using GraphPad Prism 9.3.1 (3 5 0) for
the PV module. Data for each testing sequence were obtained within 20 macOS, GraphPad Software, San Diego, California USA, https://www.
s, constituting a constant irradiation environment. Furthermore, the in graphpad.com.
situ power generation of the SCAPV field was also recorded, and the
output power (W/m2) and electricity generation (kWh/m2) were 3. Results and discussion
recorded by the inverter every 5 mins and transmitted using a 4G
transmission module. It is important to note that the in situ power gen­ 3.1. Spectral properties under SCAPV
eration results were assessed based on the vertical projected area of
SCAPV, which was 1.08 m2 for a single array; due to the semitransparent The spectral irradiance at different locations under SCAPV will vary
property, the assumption here is that SCAPV does not require spacing to according to their specific positions and the time of day. We measured
avoid shading and that farmland can be densely covered with the ver­ the spectral irradiance close to the plant canopy at 12:00, and spectra
tical projected area of SCAPV. under SCAPV and in the open air are depicted in Fig. 4a. Compared to
the transmission property of the MPF in Fig. 1c, the in situ transmission
2.3. Plant experiments spectrum under SCAPV had smooth transition edges due to angular
dependence and light scattering. The transparency of 400–800 nm
2.3.1. Plant materials and cultivation sunlight reached 65 % (Fig. 4b), which is significantly higher than other
Potato (Jinshu No. 1) tubers were germinated in the field in six rows spectral-splitting agrivoltaics, such as OPV (10–20 %). More impor­
and three repeats (320 cm × 180 cm block, 80 cm row spacing, and 14 tantly, the SCAPV reduced the amount of non-useful photons trans­
cm plant spacing) under SCAPV or in open air (control). Lettuce mitted to the plant canopy, including most green and infrared bands. In
(Jonction RZ 81–45) seeds were germinated in a greenhouse and contrast, most other solutions did not accurately split the green band for
transplanted to the field in three rows and three repeats (180 cm × 56 PV [45,46,51]. In general, the spectrum under SCAPV predominantly
cm block, 18 cm row spacing and 14 cm plant spacing) under SCAPV or matches the absorption spectra of plant pigments. These results indicate
in open air (control). The field was prepared two weeks prior to trans­ that SCPAV, as an optimal spectra management, can split and utilize
plant by rototilling. For potato plants, at 12 weeks after planting, the broadband spectra accurately and efficiently.
aboveground fresh weight of two rows from each repeat under SCAPV or
in the open air was harvested, and then 12 representative plants under
SCAPV or in the open air were used to determine biomass. For lettuce,

5
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

Fig. 4. Spectral properties of SCAPV. (a) Transmission spectrum under SCAPV and in the open air (control), measured at 12:00 a.m. (b) Transparency of the
spectral components of 400–800 nm sunlight.

3.2. Electrical performance of the system Meanwhile, this efficiency is higher than the reported results of 8.84 %
[50] and 3 % [51] in previous prototypes of similar designs.
3.2.1. Numerical simulation of PV efficiency As a field productivity trial, we also tested the in situ power gener­
Based on the S-Q limit theory and ideally square-shaped EQE, we ation of the field (all grid-connected arrays), including the diurnal
evaluated the ultimate efficiency limits with the ideally reflective bands profile of PV output power (W/m2) on a typical sunny day and electricity
of sunlight (detailed methods are introduced in Section 2.2.1.1). The generation (kWh/m2) during the full life cycle of crop production. On a
thermodynamic limiting efficiency for a single-junction PV as a function typical sunny day, we measured an in situ maximum output power of
of bandgap energy (Eg) is presented in Fig. 5a. With the consideration of 61.8 W/m2, and the power was maintained at ~ 50 W/m2 during most
silicon cells (Eg = 1.12 eV), the ultimate efficiency limit is 17 %. This daytime, except for the downfall at 13:00 due to temporary overcast
represents the upper limit for harvesting the photosynthetically excess (Fig. 7a). The peak output power occurs at approximately 9:00, which
radiation for PV conversion. Furthermore, even if too much infrared may suggest that the tracking system can be further optimized to gain
light is transmitted (750–852 nm), this ultimate efficiency limit value is more power during most of this day. We also calculated an in situ elec­
still similar to the previous prediction of 17 % in Ref.[27] which tricity generation of 10.86 kWh/m2 during the lettuce growth (see
assumed narrower transmission bands, showing that with further spec­ Section 3.3) period of 50 days, among which the maximum daily elec­
tral management, there is room for balancing the energy between PV tricity generation was 0.43 kWh/m2 (Fig. 7b). The SCAPV and the PTC
and photosynthesis. systems contained therein mainly work with direct beam radiation, so
The above ultimate efficiency limit is based on an ideal EQE, and the electrical performance decreased on some cloudy days, and the
there are multiple losses in actual PV devices, such as incomplete light decrease was not linear to the decrement in solar radiation; thus, the
absorption or collection of carriers, recombination, parasitic resistance, electricity generation was extremely low on June 16 and 28, July 11 and
etc. We subsequently calculated the ideal PV efficiency of the SCPAV 17, etc. Despite this, the in situ electricity generation, which was
based on the measured spectral response and the nominal efficiency of equivalent to ~ 80 kWh/m2-year, still represents quite a high level. As a
silicon cells (detailed methods are introduced in Section 2.2.1.2). The reference, the reported semitransparent LSC agrivoltaics showed a
calculated efficiency loss of the split sunlight compared to broadband power generation of ~ 47 kWh/m2-year [37].
sunlight is 54 %. As the nominal efficiency value of the silicon PV from In the future, higher PV efficiencies can be achieved by replacing the
our supplier is 22.5 %, we report that the ideal PV efficiency of the existing monocrystalline solar cells with multijunction concentrator
SCPAV is 10.35 % (Fig. 5b). Compared to the aforementioned ultimate cells, adjusting the optical properties of MPF, and continuously opti­
efficiency limit of 17 %, this ideal efficiency of 10.35 % is more prag­ mizing the system structure and solar tracking performance.
matic, as the ideal EQE is replaced with the actual spectral response
curve, and the nominal efficiency value of 22.5 % has already consid­ 3.3. Plant performance under SCAPV
ered multiple losses. Furthermore, as we have introduced, the spectral
property of the MPF used herein is not perfect, as it allocated too much 3.3.1. Plants under agrivoltaics showed uncompromised biomass
infrared light for plants rather than PV, which limited the PV efficiency. To demonstrate agricultural biomass production under this system,
field productivity trials were conducted in 2020. Lettuce (Lactuca sativa)
3.2.2. Experimental evaluation of the PV efficiency and potato (Solanum tuberosum) plants were grown under SCAPV with
To verify the PV efficiency, we experimentally evaluated the elec­ local famers’ agronomic practices (Fig. 8a, b, d, e). Compared with those
trical properties of the SCAPV. We first tested multiple I-V curves of a grown in the open field, lettuce and potato plants grown under SCAPV
single array under concentrated sunlight. Vm and Im (the voltage and both showed an increased aboveground biomass of 13 % (Fig. 8c, f). The
current when operating at the maximum power point Pm) of all the I-V results imply that a large portion of incident solar energy can be used to
curves are listed in Supplementary Data S1, and the multiple values of generate photovoltaic electricity without negatively influencing plant
Pm and PCE are depicted in Fig. 6a. The maximum power conversion productivity or even with potential growth benefit.
efficiency (PCE) was 9.9 % when Vm was 13.1 V, Im was 6.2 A, and the As introduced, due to limited spectral properties and sizes of
solar irradiation was 825 W/m2. The I-V curve corresponding to the laboratory-stage semitransparent PVs, in most previous spectral-
maximum PCE is presented in Fig. 6b. This peak efficiency is close to the splitting agrivoltaics studies, plant performances were merely evalu­
simulated ideal efficiency mentioned in Section 3.2.1 (10.35 %). ated qualitatively or under protected crop cultivation environments of

6
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

Fig. 5. Numerical simulation of PV efficiency. (a) The ultimate efficiency limits of split sunlight (black line) and full sunlight (red line) as a function of bandgap
energy. The red line agrees well with the original S-Q curve [58]. (b) A schematic of the ideal efficiency of the SCAPV.

Fig. 6. Photovoltaic efficiency of the system. (a) Pm, PCE and solar irradiation from multiple I-V curves measured under concentrated sunlight. (b) Representative
I-V curve with the maximum PCE.

7
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

Fig. 7. The in situ power generation of the SCAPV field. (a) Profile of the output power (W/m2) of the SCAPV field on a typical sunny day. (b) Daily electrical
power generation and solar irradiation per sqm land during 50 days of the lettuce growth period.

Fig. 8. Crop productivity trials for SCAPV. (a, b) Photographs of potato plants under SCAPV and control conditions. (d, e) Photographs of lettuce plants under
SCAPV and control conditions. (c, f) Dry biomass of plants in SCAPV field trials. Bars represent the average SD of 10 independent measurements. P values indicate
that the differences are not significant.

8
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

greenhouses. Our results were evaluated in open-air croplands and with encountered too much sunlight and had to dissipate the overexcitation
local farmers’ agronomic practices and thus have practical significance. light energy by heat as characterized by NPQ. This is partially why
In addition, although the plants were evenly watered during the trial, plants show uncompromised biomass using only selected bands of the
the rainwater distribution was altered by the system. As water is an solar spectrum; most importantly, this decreased heat dissipation under
important variable for plant growth [60,61] and was reduced under SCAPV was funnelled to PV for electrical generation. We also observed
SCAPV, we anticipate that SCAPV could even have the potential to higher Y(II) of plants grown under SCAPV than the control throughout
enhance plant biomass under the assumption of equal water availability the day (Fig. 9b), especially when the PPFD of natural sunlight is high.
or in hot arid climates. This result partially reflects that the photosynthetic efficiency under
Maintaining or enhancing plant productivity with partial sunlight is SCAPV is improved due to this optimized light environment, especially
of great significance to solar energy polygenerations. In previous opin­ under high light. The brief midday rise of Y(II) is due to the overcast at
ions, a 1 % radiation decrement usually results in a 0.5 to 1 % decrease noon, and the quick recovery of Y(II) at approximately 16:00 is also
in biomass for most crops on a large statistical scale [62]. The SCAPV because of the overcast afternoon (Fig. 9c).
might alter the linear relationship as a game changer. These results also The maximum photochemical efficiency of photosystem II is
provide insights into the design and application of other spectral- measured as Fv/Fm. For lettuce plants grown in the open air, Fv/Fm at
splitting agrivoltaics, such as OPV, which also holds potential for 13:00 was much lower than that at 4:00 a.m. and 7:00 a.m. (Fig. 9d).
maintaining or improving crop production if adjusted to proper spectral This indicates severe photoinhibition and partial inactivation of the
properties. photosynthetic apparatus at noon, and plants must involve a variety of
repair processes to prevent the accumulation of photodamage [9,12].
3.3.2. Plants showed decreased heat dissipation and increased photosystem However, under SCAPV, this photoinhibition is largely relieved.
II operating efficiency Chlorophyll fluorescence measurements can only partly reflect the
There are several photoprotective mechanisms to cope with excess photosynthetic efficiency, and plant photosynthesis and physiological
sunlight, and heat dissipation in the photosystem II antenna is mechanisms under SCAPV still require further research. However, we
commonly evaluated as nonphotochemical quenching (NPQ) [63]. This have preliminarily revealed that the uncompromised biomass under
heat dissipation results in a decrease in the quantum efficiency of SCAPV is partly due to relieved photoinhibition and reduced photo­
photosystem II (Y(II)) [9,64]. In situ chlorophyll fluorescence measure­ protection. In essence, the photons that were originally wasted by plants
ments were conducted to evaluate NPQ, Y(II) and other fluorescence are cleverly applied to generate electricity. This may not only be the
parameters in lettuce plants, in order to support the biomass produc­ process under SCAPV but also the physiological basis for various agri­
tivity results in Section 3.3.1. voltaics; in other words, photosynthesis is so inefficient, thus PV and
Compared to the control group, the NPQ of lettuce plants under photosynthesis can be tandemly utilized on agricultural lands.
SCAPV was significantly lower throughout the day (Fig. 9a). During 8:00
and 14:00, the NPQ values under SCAPV decreased by approximately 50 3.4. Improved microclimate under SCAPV
% compared to the control. At 16:00, the difference in NPQ was nar­
rowed due to the overcast afternoon, as indicated by the photosynthetic The microclimate conditions under SCAPV were evaluated in terms
photon flux density (PPFD) in Fig. 9c. Plants in the control group of solar radiation, air temperature and humidity, soil–water status, etc.,

Fig. 9. Chlorophyll fluorescence measurements. (a) NPQ of lettuce plants under SCAPV and control conditions. (b) Y(II) of lettuce plants under SCAPV and the
control. (c) PPFD under SCAPV and control conditions. (d) Fv/Fm measured at 7:00, 13:00, and 4:00. Bars represent the average SD of 6–11 independent mea­
surements. P values are indicated, * represents a significant difference at P less than 0.05 compared to the control, ** P ≤ 0.01, *** P ≤ 0.001.

9
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

Fig. 10. Measurements of microclimate. (a-d) PAR, CO2 concentration, air temperature and humidity over 2 typical sunny days under the SCAPV and control.

and measured from June 4 to June 8 (typical growing season). The covered when it rained; then they were constantly weighed using an
average atmospheric pressure was 99.99 kPa, and the average wind electronic scale over three days, and the results showed that evaporation
speed was 1.59 m/s. The PAR was diurnally lower under SCAPV under SCAPV (700 g) was slightly lower than that under the control (762
(Fig. 10a). The CO2 concentration in the air showed irregular fluctua­ g) (Fig. 11c). In general, the differences in evaporation were not sig­
tions (Fig. 10b). The air temperature under SCAPV was generally lower nificant, and plant transpiration may be a more important factor in the
(Fig. 10c), as was the air humidity (Fig. 10d). These results represent an soil moisture difference in Table 2, which can be clearly concluded from
improved microclimate under SCAPV due to optimal spectra Fig. 11b and Fig. 11c. In summary, SCAPV placed plants in a new
management. microclimate where the soil–water status is improved, which usually
We monitored the soil–water status as characterized by soil tem­ relieves drought stress and stomatal closure.
perature and soil humidity (relative amount of soil moisture content) at
5 cm and 10 cm depths of the SCAPV and control sites. Data were 3.5. Combined efficiency of SCAPV
averaged to a single representative value and are listed in Table 2, and
profiles are depicted in Fig. 11a, b. Under SCAPV, the average soil At present, there is no official standard available for the character­
temperature at 5 cm and 10 cm was generally lower, and the soil hu­ ization of hybrid efficiencies of agrivoltaics. Based on our literature
midity was generally higher than the natural control. survey, the combined efficiency of SCAPV should be evaluated by
To distinguish between soil evaporation and plant transpiration, the comparably cumulating photosynthetic and PV efficiencies. First, short-
evaporation of soil water measured using evaporation containers term (rapid growth phase) photosynthetic efficiency εc was calculated
(without plant cultivation) was also evaluated. After being fully watered based on crop biomass and local solar radiation. One mol of 1C carbo­
and discharged with gravity water, the evaporation containers were at hydrate unit (CH2O, 30 g), which is one-sixth of a mole of glucose,
the initial soil water content of the field capacity and then placed under contains 477 kJ of energy [3]; thus, 1 kg of biomass contains 15900 kJ of
either SCAPV or in the open air. The evaporation containers were energy. Photosynthetic light use efficiency εc can be determined from
the following equations: (total biomass (kg/m2) × 15900 (kJ/kg))/local
radiation (kJ/m2). For lettuce plants under SCAPV, this value is 1.13 %,
Table 2 while that under natural sunlight is 1.0 %. As illustrated in the Intro­
Average soil temperature and soil humidity under SCAPV and the control.
duction Section, the εc is extremely low, and when considering the net
Average T(℃) Average T (day)(℃) Average T (night)(℃) annual productivity and efficiency, the value would be even lower.
control SCAPV Control SCAPV control SCAPV PV and photosynthetic efficiencies must be compared with a process
5CM 28.77 26.81 30.83 28.00 26.21 25.33
in which PV must store energy in chemical bonds; based on this, the
10CM 28.27 26.74 28.58 27.05 27.89 26.37 generated electrical power can be used to electrolyze water [65]. Ac­
cording to this method, the overall short-term light use efficiency is
Soil moisture Soil moisture (day) Soil moisture (night)
(%) (%) (%) determined from the following equation: εc + photovoltaic efficiency ×
modern commercial electrolyser efficiency = 1.13 %+ 9.9 %×80 %=
control SCAPV control SCAPV control SCAPV
9.05 %[65]. Although this efficiency value falls far below the bench­
5CM 16.16 21.58 16.39 21.70 15.88 21.42 mark of most opaque Si-PV modules or other hybrid systems, such as
10CM 21.71 26.14 21.77 26.15 21.64 26.13
concentrating photovoltaic/thermal (CPVT) [66], which is due to the

10
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

Fig. 11. Measurements of soil temperature and moisture. (a) Soil temperature over 3 days at 10 cm underground. (b) Soil moisture over 3 days at 10 cm
underground. (c) Water evaporation over 3 days under SCAPV and the control.

inefficiency of photosynthesis, the SCAPV represents a hybrid system 3.6. Comparative evaluation of various spectral-splitting agrivoltaics
with an overall solar energy conversion efficiency higher than that of
any existing biological photosynthetic system, i.e., the theoretical limi­ As we have introduced, there are several other candidates for
tation of εc of 4.6 % for C3 plants and 6 % for C4 plants. spectral-splitting agrivoltaics, which have been of tremendous interest
Another way to comparably calculate the combined efficiency of the in recent years [13,14,29]. Here, we compare different technologies in
SCAPV is by completely converting the energy to plant biomass. The terms of technology maturity, spectral flexibility, transparency, cost,
generated electricity can be used to further enhance plant productivity, weathering performance (durability), and reported impacts on plants.
such as illumination in artificial light plant factories, supplemental The available technologies considering these key parameters are pre­
lighting in the understory or supplemental light at night. The electricity sented in Table 3, among which mono-Si and poly-Si are for comparison.
cost per unit dry mass in the current LED-based plant factory is As emerging technologies, OPVs and PVSCs are still in the learning
approximately 75 kWh/kg for leafy microgreens [67]. The SCAPV re­ stage and are technically immature, especially in large-size
ported here generated a photovoltaic electricity power of 10.86 kWh/m2 manufacturing and durability. The technology maturities of the DSSC
during the lettuce growth period, which is equivalent to a biomass and LSC are relatively high compared to those of the OPV and PVSC, as
increment of 0.14 kg/m2 and an εc increment of 0.26 %. For plants under well as durability. In general, most technologies are not mature enough
SCAPV, the combined efficiency of converting solar energy into biomass to be commercialized yet, with only a few prototypes having demon­
(1.13 %+0.26 %=1.39 %) is 39 % higher than that grown under natural strated reliable performance until now . In contrast, all the components
sunlight (1.0 %). of SCAPV, including MPF, PTC and c-Si, are already commercially
By combining biological light conversion with PV electricity gener­ available.
ation, the SCAPV reported here represents a novel efficient strategy of Evaluation of cost (Wp ($)) shows that the SCAPV is cost-effective
utilizing broadband solar energy. Most importantly, for SCAPV, the light compared to most other spectral-splitting agrivoltaics at current stage.
spectrum can be fine-tuned to balance the demand for plant growth and It’s worth noting that, the estimated low costs for OPVs and PVSCs are
power generation, i.e., the transmitted spectrum of MPF can be custom- based on assumptions of future mass production. In fact, more conser­
designed and fabricated. For example, it may be designed to decrease vative data suggest that the costs would be 0.23–0.34$ or higher for OPV
the transmission of blue light [68], decrease the reflection of green light [13] and 0.5–2.9$ for PVSC [73]. Fabrication methods of mass pro­
[69], increase the reflection of infrared light in Fig. 1c for more PV duction are still required to reduce the current high cost. Costs for DSSC
power generation, etc. Thus, photosynthetic and photovoltaic effi­ and LSC are estimated as approximately 0.5–0.94$ and 0.65$, respec­
ciencies can be simultaneously optimized by proper choice of MPF tively. Compared to Si-PV, SCAPV consumes fewer Si materials, which
spectra. In addition, SCAPV extends the earlier concept of concentrator have a high carbon footprint and cost. It reduces PV module cost per
photovoltaic systems (CPVs) by integrating biomass production Watt by up to 50 % by replacing relatively expensive Si-PV with low cost
concurrently with electricity generation. CPV has been well studied in MPF. Based on our research and interviews with suppliers, the estimated
past decades, among which parabolic trough concentrators (PTCs) are material costs of the SCAPV are approximately $0.1/W for PV cells +
widely deployed and account for approximately 85 % of all active CPV $0.02/W for MPF + $0.45/W for hardware (concentrator and tracking
installations [57]. Since 2012, CPV’s market share continues to fall [70], mechanism) = $0.57/W. In addition, note that for DSSC, LSC and
largely due to the rapid decline in the cost of Si-PV. Novel multi­ SCAPV, there could be a great margin in cost reduction if a more
generation systems such as SCAPV may increase the interest in PTC and extended production is considered.
CPV among researchers and the market. As introduced, plant performances under these solar cells were
merely qualitatively tested. Despite this, the effects were generally
neutral (several studies show that plants grown under semitransparent

Table 3
Comparisons for various spectral-splitting agrivoltaics.
PV Technology maturity Spectral flexibility Transparency Cost/Wp ($) Durability Effects on plants

mono-Si ***** * * 0.19–0.38 [70] ***** –


poly-Si ***** * * 0.16–0.29 [70] ***** –
a-Si ***** ** *** – ***** Negative[38]
OPV ** **** *** 0.05 [8] ** Neutral [28,30,45–48]
DSSC *** *** ***** 0.5–0.94[71] *** Unknown [33]
PVSC * *** **** 0.12[72] * Unknown [34,43]
LSC *** *** **** 0.65[37] *** Neutral [37]
SCAPV ***** ***** ***** 0.57 **** Positive

***** represents high technology maturity, spectral flexibility, transparency and durability; the reduction of * means relatively low maturity, spectral flexibility,
transparency and durability. The effects of mono-Si and poly-Si on plants are not within the scope of our discussion.

11
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

solar cells showed similar or improved yields, see Table 3), which also the SCAPV. We calculated that this system can achieve a hybrid light use
partly supports our conclusion. Combined with our results on plant efficiency of 9.05 % (from solar light to chemical bonds), which is higher
performance, we assume that, on the premise of relatively high trans­ than the maximal theoretical photosynthetic efficiency, i.e., 6 %. Thus,
parency and proper spectra, all these technologies under the spectral- SCAPV provides a physical means to increase energy use efficiency
splitting strategy are promising. However, systematic theories and ex­ compared to contemporary agricultural systems based purely on crops.
periments are still needed to establish the full impact of split sunlight on Furthermore, the economic analysis shows that the LCE of SCAPV is
plants in future studies. 0.033 $/kWh, indicating the economic feasibility of SCAPV.
We compared the SCAPV with various spectral-splitting agrivoltaics,
showing the different technologies with this similar strategy and their
3.7. Feasibilities and challenges of SCAPV
pros and cons. Our results show that SCAPV is a proper technology with
scalable size, flexible spectrum, high transparency and low cost; mean­
To assess the economic viability of SCAPV, a levelized cost of energy
while, the progress and potential of other spectral-splitting agrivoltaics
(LCE) analysis was performed. LCE, defined as the ratio of the total life-
are also noteworthy. Although these tunable PVs are usually not cost
cycle cost (TLCC) to the equivalent electrical production during the
competitive (compared to conventional Si-PVs) for now, agrivoltaics
entire life of the project (EP ), can be calculated by [74]:
provide a niche market where they can provide advantages over con­
TLCC C0 + ΣLi=1
t
COM (1 + fR )i− 1 (1 + iR )− i ventional Si-PVs and hence have the opportunity to be commercialized
LCE = = (12) and improved. Using spectral-splitting agrivoltaics to harvest the
EP ΣLi=1
t
QA (1 + iR )− i
damaging excess light energy for electricity generation suggests that it is
where C0 is the initial capital cost of the SCAPV, which is 81 W/m2 × a feasible strategy that may be used to support the realization of the UN
0.57$/W = 47$/m2 according to Fig. 6 and Table 3; Lt is the estimated millennium developmental goal of agricultural and water sustainability,
lifetime of the SCAPV, which is assumed to be 25 years [75]; COM is the energy structure transformation, and the global effort to develop a
annual operating and maintenance cost, which is assumed to be 2 % of carbon–neutral society.
C0 ; iR and fR are the rates of interest and inflation, which are assumed to
be 10 % and 5 %, respectively [74]; QA is the equivalent yearly electrical CRediT authorship contribution statement
production. Regarding the evaluation of QA , we cumulated the in situ
electricity generation of 80 kWh/m2-year (Section 3.2.2) and the po­ Zhisen Zhang: Conceptualization, Methodology, Investigation,
tential electrical cost equivalent to the dry biomass production (Section Writing – original draft, Writing – review & editing. Fangxin Zhang:
3.3.1). Specifically, the potato plants in Section 3.3.1 gained a biomass Resources, Investigation. Wei Zhang: Data curation, Visualization.
production of 0.31 kg/m2 (an average dry weight of 0.049 kg per plant, Ming Li: Resources, Validation. Wenjun Liu: Data curation. Altyeb Ali
with a cultivating density of 6.25 plants/m2), which is equivalent to a Abaker Omer: Validation. Jianan Zheng: Investigation. Xinyu Zhang:
biomass production of 0.62 kg/m2-year (a conservative assumption of Investigation, Visualization. Wen Liu: Conceptualization, Supervision,
two cultivations in a year). Considering the method in Section 3.5, this is Funding acquisition, Writing – review & editing.
equivalent to a potential electrical production of 121 kWh/m2-year. In
total, a comprehensive QA of 201 kWh/m2-year is calculated. Based on Declaration of Competing Interest
these results and assumptions, the LCE of the SCAPV is 0.033 $/kWh,
which is lower than that of Refs.[74,75]. The authors declare that they have no known competing financial
There are still limitations and challenges for SCAPV. The first chal­ interests or personal relationships that could have appeared to influence
lenge is the potential cost disadvantage compared to novel semi­ the work reported in this paper.
transparent solar cells in the future. For the current situation, the SCAPV
is cost-effective compared to most spectral splitting agrivoltaics; how­ Data availability
ever, when considering future cost, OPVs and PVSCs may have cost
advantages due to the abundance of related elements on Earth and the Data will be made available on request.
expected low processing costs. Another limitation the SCAPV is faced is
the durability and stability of the complex mechanical structure Acknowledgements
compared to a conventional silicon module; according to Eq. (12), the
economic feasibility is highly related to the operating and maintenance We thank Professor Xin-guang Zhu for the guidance of photosyn­
cost and the estimated lifetime. This is also a common challenge for thesis theory and experimental experience. We thank Xin-liang Liu for
various novel energy polygeneration systems. The policy of protecting plant care and management. We thank Tian-shu Chu for language pol­
cultivated land in many countries is also a major limitation, and this ishing. This work was financially supported by the Plan for Anhui Major
concern calls for more investigations on photosynthetic yields and Provincial Science & Technology Project (Grant No. 202203a06020002)
economic feasibilities of SCAPV and other agrivoltaics. and Science & Technology Program of Hebei (Grant No. 22327215D).

4. Conclusions References

The SCAPV reported in this study represents a hybrid solar energy [1] Green MA, et al. Solar cell efficiency tables (version 59). Prog Photovolt Res Appl
2022;30(1):3–12.
conversion system that selectively splits solar light into two portions, [2] Melis A. Solar energy conversion efficiencies in photosynthesis: minimizing the
with one portion supporting photosynthesis and the other supporting chlorophyll antennae to maximize efficiency. Plant Sci 2009;177(4):272–80.
photovoltaic electricity generation. The optimal light management of [3] Zhu X-G, Long SP, Ort DR. What is the maximum efficiency with which
photosynthesis can convert solar energy into biomass? Curr Opin Biotechnol 2008;
SCAPV resulted in substantially decreased photoprotective heat loss and 19(2):153–9.
improved quantum efficiency of plant photosystem II. Hence, one [4] Bailey-Serres J, et al. Genetic strategies for improving crop yields. Nature 2019;575
counterintuitive result is that even without considering the electrical (7781):109–18.
[5] Long SP, Marshall-Colon A, Zhu X-G. Meeting the global food demand of the future
power, plants grown under partial sunlight did not show any decrease in by engineering crop photosynthesis and yield potential. Cell 2015;161(1):56–66.
biomass; in fact, multiple plants grown under the SCAPV showed the [6] Chen M, Blankenship RE. Expanding the solar spectrum used by photosynthesis.
potential for increased biomass. The experimentally measured PV effi­ Trends Plant Sci 2011;16(8):427–31.
ciency of the SCAPV was 9.9 %, which was in line with our simulation of
the ultimate efficiency limit (17 %) and the ideal efficiency (10.35 %) of

12
Z. Zhang et al. Energy Conversion and Management 276 (2023) 116567

[7] Delegido J, et al. Retrieval of chlorophyll content and LAI of crops using [42] Xia R, et al. High-Throughput Optical Screening for Efficient Semitransparent
hyperspectral techniques: application to PROBA/CHRIS data. Int J Remote Sens Organic Solar Cells. Joule 2019;3(9):2241–54.
2008;29(24):7107–27. [43] Subhani WS, et al. Anti-solvent engineering for efficient semitransparent
[8] La Notte L, et al. Hybrid and organic photovoltaics for greenhouse applications. CH3NH3PbBr 3 perovskite solar cells for greenhouse applications. Journal of
Appl Energy 2020;278:115582. Energy Chemistry 2019;34:12–9.
[9] Murchie EH, Niyogi KK. Manipulation of photoprotection to improve plant [44] Yildirim N, Bilir L. Evaluation of a hybrid system for a nearly zero energy
photosynthesis. Plant Physiol 2011;155(1):86–92. greenhouse. Energ Conver Manage 2017;148:1278–90.
[10] Niyogi KK, Björkman O, Grossman AR. The roles of specific xanthophylls in [45] Liu Y, et al. Unraveling Sunlight by Transparent Organic Semiconductors toward
photoprotection. Proc Natl Acad Sci 1997;94(25):14162–7. Photovoltaic and Photosynthesis. ACS Nano 2019;13(2):1071–7.
[11] Zhu X-G, Long SP, Ort DR. Improving photosynthetic efficiency for greater yield. [46] Song W, et al. Foldable Semitransparent Organic Solar Cells for Photovoltaic and
Annu Rev Plant Biol 2010;61:235–61. Photosynthesis. Adv Energy Mater 2020;10(15):2000136.
[12] Ort DR. When There Is Too Much Light. Plant Physiol 2001;125(1):29–32. [47] Ravishankar E, et al. Balancing crop production and energy harvesting in organic solar-
[13] Lu L, et al. Comprehensive review on the application of inorganic and organic powered greenhouses. Cell Reports Physical. Science 2021:100381.
photovoltaics as greenhouse shading materials. Sustainable Energy Technol Assess [48] Song W, et al. Ultra-flexible light-permeable organic solar cells for the herbal
2022;52:102077. photosynthetic growth. Nano Energy 2021;86:106044.
[14] Gorjian S, et al. Progress and challenges of crop production and electricity [49] Zisis C, et al. Organic Photovoltaics on Greenhouse Rooftops: Effects on Plant
generation in agrivoltaic systems using semi-transparent photovoltaic technology. Growth. Mater Today: Proc 2019;19:65–72.
Renew Sustain Energy Rev 2022;158:112126. [50] Liu W, et al. A novel agricultural photovoltaic system based on solar spectrum
[15] Mamun MAA, et al. A review of research on agrivoltaic systems. Renew Sustain separation. Sol Energy 2018;162:84–94.
Energy Rev 2022;161:112351. [51] Sonneveld PJ, et al. Feasibility study for combining cooling and high grade energy
[16] Barron-Gafford GA, et al. Agrivoltaics provide mutual benefits across the production in a solar greenhouse. Biosyst Eng 2010;105(1):51–8.
food–energy–water nexus in drylands. Nat Sustainability 2019;2(9):848–55. [52] Ma Q, et al. Photovoltaic/spectrum performance analysis of a multifunctional solid
[17] Valle B, et al. Increasing the total productivity of a land by combining mobile spectral splitting covering for passive solar greenhouse roof. Energ Conver Manage
photovoltaic panels and food crops. Appl Energy 2017;206:1495–507. 2022;251:114955.
[18] Ravi S, et al. Colocation opportunities for large solar infrastructures and [53] Weber MF, et al. Giant birefringent optics in multilayer polymer mirrors. Science
agriculture in drylands. Appl Energy 2016;165:383–92. 2000;287(5462):2451–6.
[19] Miskin CK, et al. Sustainable co-production of food and solar power to relax land- [54] Macleod HA. Turning value monitoring of narrow-band all-dielectric thin-film
use constraints. Nat Sustainability 2019;2(10):972–80. optical filters. Optica Acta: International Journal of Optics 1972;19(1):1–28.
[20] Adeh EH, et al. Solar PV Power Potential is Greatest Over Croplands. Sci Rep 2019; [55] Alfrey Jr T, Gurnee E, Schrenk W. Physical optics of iridescent multilayered plastic
9(1):11442. films. Polym Eng Sci 1969;9(6):400–4.
[21] Fernández EF, et al. Global energy assessment of the potential of photovoltaics for [56] Li M, et al. Polymer multilayer film with excellent UV-resistance & high
greenhouse farming. Appl Energy 2022;309:118474. transmittance and its application for glass-free photovoltaic modules. Sol Energy
[22] Marrou H, et al. Microclimate under agrivoltaic systems: Is crop growth rate Mater Sol Cells 2021;229:111103.
affected in the partial shade of solar panels? Agric For Meteorol 2013;177:117–32. [57] Tagle-Salazar PD, Nigam KDP, Rivera-Solorio CI. Parabolic trough solar collectors:
[23] Armstrong A, Ostle NJ, Whitaker J. Solar park microclimate and vegetation A general overview of technology, industrial applications, energy market,
management effects on grassland carbon cycling. Environ Res Lett 2016;11(7): modeling, and standards. Green Process Synth, 2020;9(1):595–649.
074016. [58] Shockley W, Queisser HJ. Detailed Balance Limit of Efficiency of p-n Junction Solar
[24] Allardyce CS, et al. The influence of greenhouse-integrated photovoltaics on crop Cells. J Appl Phys 1961;32(3):510–9.
production. Sol Energy 2017;155:517–22. [59] Huang G, Wang K, Markides CN. Efficiency limits of concentrating spectral-
[25] Cossu M, et al. Solar radiation distribution inside a greenhouse with south-oriented splitting hybrid photovoltaic-thermal (PV-T) solar collectors and systems. Light Sci
photovoltaic roofs and effects on crop productivity. Appl Energy 2014;133:89–100. Appl 2021;10(1):28.
[26] Emmott CJM, et al. Organic photovoltaic greenhouses: a unique application for [60] Ai Z, et al. Global bioenergy with carbon capture and storage potential is largely
semi-transparent PV? Energ Environ Sci 2015;8(4):1317–28. constrained by sustainable irrigation. Nat Sustainability 2021;4(10):884–91.
[27] Meitzner R, Schubert US, Hoppe H. Agrivoltaics—The Perfect Fit for the Future of [61] Ort DR, Long SP. Limits on Yields in the Corn Belt. Science 2014;344(6183):484–5.
Organic Photovoltaics. Adv Energy Mater 2021;11(1):2002551. [62] Marcelis LFM, et al. Quantification of the growth response of light quantity of
[28] Shi H, et al. Spectral Engineering of Semitransparent Polymer Solar Cells for greenhouse grown crops. Acta Hortic 2006;711(2006):711.
Greenhouse Applications. Adv Energy Mater 2019;9(5):1803438. [63] Kromdijk J, et al. Improving photosynthesis and crop productivity by accelerating
[29] Zhao Y, et al. A review on semitransparent solar cells for agricultural application. recovery from photoprotection. Science 2016;354(6314):857–61.
Mater Today Energy 2021;22:100852. [64] Genty B, Briantais J-M, Baker NR. The relationship between the quantum yield of
[30] Wang D, et al. High-performance and eco-friendly semitransparent organic solar photosynthetic electron transport and quenching of chlorophyll fluorescence.
cells for greenhouse applications. Joule 2021;5(4):945–57. Biochimica et Biophysica Acta (BBA)-General Subjects 1989;990(1):87–92.
[31] Dessì A, et al. D-A–π–A organic dyes with tailored green light absorption for [65] Blankenship RE, et al. Comparing photosynthetic and photovoltaic efficiencies and
potential application in greenhouse-integrated dye-sensitized solar cells. recognizing the potential for improvement. Science 2011;332(6031):805–9.
Sustainable Energy Fuels 2021;5(4):1171–83. [66] Khodadadi M, Sheikholeslami M. Review on poly-generation application of
[32] Barichello J, et al. Stable Semi-Transparent Dye-Sensitized Solar Modules and photovoltaic/thermal systems. Sustainable Energy Technol Assess 2022;52:
Panels for Greenhouse Application. Energies 2021;14(19):6393. 102172.
[33] Dessì A, et al. Thiazolo[5,4-d]thiazole-based organic sensitizers with improved [67] Pattison P, et al. LEDs for photons, physiology and food. Nature 2018;563(7732):
spectral properties for application in greenhouse-integrated dye-sensitized solar 493–500.
cells. Sustainable Energy Fuels 2020;4(5):2309–21. [68] Hamdani S, et al. Changes in the photosynthesis properties and photoprotection
[34] Weng S, et al. Band-Gap-Engineered Transparent Perovskite Solar Modules to capacity in rice (Oryza sativa) grown under red, blue, or white light. Photosynth
Combine Photovoltaics with Photosynthesis. ACS Appl Mater Interfaces 2021;13 Res 2019;139(1–3):107–21.
(33):39230–8. [69] Terashima I, et al. Green Light Drives Leaf Photosynthesis More Efficiently than
[35] Xue Q, et al. Recent advances in semi-transparent polymer and perovskite solar Red Light in Strong White Light: Revisiting the Enigmatic Question of Why Leaves
cells for power generating window applications. Energ Environ Sci 2018;11(7): are Green. Plant Cell Physiol 2009;50(4):684–97.
1688–709. [70] Benda V, Černá L. PV cells and modules – State of the art, limits and trends.
[36] Corrado C, et al. Power generation study of luminescent solar concentrator Heliyon 2020;6(12).
greenhouse. J Renewable Sustainable Energy 2016;8(4). [71] Roslan N, et al. Dye Sensitized Solar Cell (DSSC) greenhouse shading: New insights
[37] Loik ME, et al. Wavelength-Selective Solar Photovoltaic Systems: Powering for solar radiation manipulation. Renew Sustain Energy Rev 2018;92:171–86.
Greenhouses for Plant Growth at the Food-Energy-Water Nexus. Earth’s Future [72] Faheem MB, et al. Insights from scalable fabrication to operational stability and
2017;5(10):1044–53. industrial opportunities for perovskite solar cells and modules. Cell Reports
[38] Thompson EP, et al. Tinted Semi-Transparent Solar Panels Allow Concurrent Physical Science 2022;3(4):100827.
Production of Crops and Electricity on the Same Cropland. Adv Energy Mater 2020; [73] Chang NL, et al. Manufacturing cost and market potential analysis of demonstrated
10(35):2001189. roll-to-roll perovskite photovoltaic cell processes. Sol Energy Mater Sol Cells 2018;
[39] Hassanien RHE, Li M, Yin F. The integration of semi-transparent photovoltaics on 174:314–24.
greenhouse roof for energy and plant production. Renew Energy 2018;121:377–88. [74] Yuan Y, et al. Experimental investigation of full solar spectrum utilization based on
[40] Tamang A, et al. Combining Photosynthesis and Photovoltaics: A Hybrid Energy- nanofluid spectral splitter for greenhouse applications. Energ Conver Manage
Harvesting System Using Optical Antennas. ACS Appl Mater Interfaces 2020;12 2022;254:115215.
(36):40261–8. [75] Schindele S, et al. Implementation of agrophotovoltaics: Techno-economic analysis
[41] Osterthun N, et al. Spectral engineering of ultrathin germanium solar cells for of the price-performance ratio and its policy implications. Appl Energy 2020;265:
combined photovoltaic and photosynthesis. Opt Express 2021;29(2):938–50. 114737.

13

You might also like