A Comprehensive Review of Agrivoltaics: Multifaceted Developments and the Potential of Luminescent Solar Concentrators and Semi-Transparent Photovoltaics

Abstract

In the context of rapid decarbonization, photovoltaics (PV) has played a key role. Traditionally, PV installations require large land areas, leading to competition between PV and agriculture for land use. This conflict must be addressed as the demand for both energy and food continues to rise. Additionally, it poses broader challenges, potentially leading local communities to perceive PV energy production as a threat to their economic activities and food security. An emerging and promising solution is agrivoltaics (AV), a combination of agriculture and PV. AV comes in many different forms, ranging from the simple coexistence of crops and PV installations on the same patch of land to a full synergy of the two, producing better crops while also harvesting energy from the sun. This paper paints a complete picture of the scientific work produced so far throughout the field, with special attention to the use of third-generation PV and luminescent solar concentrators (LSCs). Both technologies minimize shading and enable wavelength selection and enrichment (when functionalized with fluorescent materials) to better align with the photosynthetic needs of plants. The viability of AV has also been evaluated from an economic standpoint. This work aims to assess the current landscape of AV research and to point out possible future developments. It also seeks to evaluate whether the advantages of semi-transparent devices are substantial enough to justify their development and employment on a scale comparable to traditional PV.

1. Introduction

The entire area of the Earth’s surface is 510 × 10⁶ km², 71% of which is covered by seawater. That means that all human activities take place on the remaining 29%. Out of the entire Earth’s landmass, the area deemed to be habitable is 112 × 10⁶ km², 45% of which (i.e., 50 × 10⁶ km²) is entirely devoted to agricultural purposes. In particular, 16% of it is used to grow crops for human consumption, 80% is dedicated to livestock grazing and crop production for livestock feeding, and the remaining 4% is devoted to crop production for different uses [1]. From 2001 to 2022, cropland grew by 5%, which is roughly 0.8 × 10⁶ km² [2].

At the same time, energy demand has been steadily increasing for the last 50 years at least, reaching 642 EJ in early 2024 [3], and this growth is projected to persist. Between 2012 and 2022, energy consumption per hectare of utilized agricultural area increased in 17 EU countries [4]. Shifting attention to agriculture, in 2022, 3% of total direct energy consumption in the EU was needed for agriculture and forestry. Fifty-eight percent of the energy used by the agriculture sector is produced using oil and other petroleum products. The necessity to meet global energy demand and, perhaps more importantly, to achieve decarbonization goals, has led to the rapid expansion of solar and photovoltaic power generation.

As of early 2024, the total global installed capacity of photovoltaic (PV) systems was 1600 GWp [5]. While figures regarding the total land area covered by these installations are not available, a reasonable estimation of land use per unit of power generated can be obtained. Considering the average extension of the largest PV parks in the world, the result is 0.3 km² MW⁻¹. At present, with 1.6 TW of cumulative PV power installed worldwide [6], the total land area devoted to PV energy generation should be around 0.5 × 10⁶ km², with roughly 1% of this land being devoted to agriculture. This situation is creating distrust in some local communities, as reported in [7,8], as it is giving rise to competition over land usage priorities between energy and food production. Clearly, it is necessary to resolve this conflict in order to ensure sustainable development in both sectors.

A pivotal concept that has emerged in recent years is the water-energy-food-ecosystem (WEFE) nexus [9]. This approach takes into account the intertwined nature of water, energy, and food security, as well as water, soil, and land resources. The WEFE nexus is a response to the increasing global population, resource exploitation, and the exacerbation effects caused by climate change. Its aim is to develop mutually beneficial strategies for each resource, ensuring water, energy, and food security for populations while preserving and benefiting ecosystems and environments. Agrivoltaics (AV), also called agro-photovoltaics, is the integration of PV technologies into agricultural fields in order to optimize land utilization. This practice aligns with the WEFE nexus objectives and aims to ease the competition between agriculture and clean energy production. AV is a logical step forward for PV implementation and energy production, as PV technologies have been at the forefront of sustainable development in-mixed use applications for a long time. A prime example of this is building-integrated photovoltaics (BIPV), which involves the integration of PV technologies within the urbanized landscape to improve buildings’ energy efficiency [10,11,12]. PV devices have also found innovative applications in photocatalytic water decontamination [13], providing the possibility of purifying clear waters from a wide range of contaminants without the need for substance-specific filters or chemical processes.

The goal of this review is to outline the many forms of AV, presenting different implementations published in the scientific literature and the opportunities it offers in terms of environmental benefits. First, the manuscript’s methodology will be introduced. This section is necessary for the comparison of different research results, defining the different classifications for AV systems, as well as some quantitative parameters and concepts. Secondly, a comparison of actual results, divided according to the generation (I, II, and III) of the PV devices used in the experiments, will be performed. Greater emphasis is placed on luminescent solar concentrators (LSCs), which are discussed in a dedicated sub-section. In addition, a distinct section is reserved for technologies related to spectral separation and management without energy production. Lastly, further discussion and conclusions will be provided, commenting on the previously presented material and proposing possible future developments.

2. Materials and Methods

To effectively review and compare the different research works, the definition of a common reference frame to categorize the various types of AV systems is as crucial as the actual examination of the results.

2.1. Classification of AV Systems

The AV systems reviewed in this work have been categorized according to the parameters reported in

Table 1. Definition of categories for the classification of AV systems. The superscripts ^a,b,c,d,e link the definitions to their respective acronyms.

Name	Definition	Acronym(s)
Openness	A field is open if it is exposed to the surrounding environment and closed when it is encased in a structure that completely separates it from the environment.	-
Structure	Denotes the geometry of how the PV modules are mounted in relation to the crops. - Interspersed: arrays mounted on the ground with rows of PV panels and crops alternating between each other; - Overhead/roof mounted: arrays mounted above the crops or acting as/covering the roof of the structure enclosing the crops; - Side mounted: PV modules cover the walls or the sides of a structure. The structure can include additional parameters such as module elevation, tilt angle, facing direction, employment of solar tracking, etc.	-
Transparency	Optical transparency of the devices: - Opaque: PV modules made of contiguous traditional opaque PV cells; - See-Through Opaque a: the module is made of opaque cells arranged with gaps between them; - Semi-transparent b: PV technologies that are innately semi-transparent to radiation; - Transparent: high-transparency devices such as LSCs.	a STO b STPV
Coverage Ratio c	Fraction of ground area covered by the considered AV device;	c CR
PV Generation	Generation of the employed PV devices: - I: thick crystalline cells (e.g., monocrystalline or polycrystalline Si); - II: thin-film cells (CIGS, CdTe, amorphous Si, etc.); - III: newer technologies (organic cells, perovskites, quantum dots cells, LSCs, multijunction cells, and concentrated photovoltaics).	-
Degree of Synergy e	Relationship between the PV system and the crops. - Coexisting: simple installation with the purpose of growing crops and producing energy at the same time; - Cooperative: PV systems are used not only to produce energy but also to create more favorable conditions for the agricultural process. Cooperation can be in terms of microclimate control (lower temperatures on the ground, reduced evaporation of water, more efficient water usage), wavelength selection or enhancement (e.g., using LSCs with a luminophore to absorb some wavelengths that are less useful to the plants and possibly re-emitting light at different wavelengths that are more suited to the needs of the plants). An AV system can employ different types of cooperation at the same time.	e DoS

.

Figure 1 summarizes the possible combinations of characteristics and applications of AV, while Figure 2 illustrates different types of mounting structures for PV modules in AV and Figure 3 shows the transparency of different PV devices.

2.2. Quantitative Parameters and Key Concepts

Power conversion efficiency (PCE) is the total radiation impinging on the PV panel to electrical power produced by the module. Calling P [W] the generated power, S the surface area of the device [m²], and I [W m⁻²] the irradiance incident on the device, the PCE is defined as follows:

$$\mathrm{P}\mathrm{C}\mathrm{E}={\displaystyle \frac{\mathrm{P}}{\mathrm{S} \mathrm{I}}}.$$

(1)

When LSC panels are considered, the PCE is strictly related to the waveguide optical efficiency (OE). The OE ranges between 0 and 1; the higher the OE, the greater the amount fluorescent light delivered to the solar cells. For this reason, traditional LSC modules are designed to have as high an OE as possible. However, if these panels are used in agricultural contexts, a lower OE may still be associated with the optimal configuration. The OE is defined as the ratio between the light optically guided by an LSC to its edge and the total light received by the LSC. It can be calculated as the product of all the efficiencies of the physical processes involved [15]:

$$\mathrm{O}\mathrm{E}={\displaystyle \frac{{\mathrm{I}}_{\mathrm{g}}}{{\mathrm{I}}_{\mathrm{r}}}}=\left(1-\mathrm{R}\right){\mathrm{P}}_{\mathrm{T}\mathrm{I}\mathrm{R}}{\mathsf{\eta }}_{\mathrm{a}\mathrm{b}\mathrm{s}}{\mathsf{\eta }}_{\mathrm{P}\mathrm{L}\mathrm{Q}\mathrm{Y}}{\mathsf{\eta }}_{\mathrm{S}\mathrm{t}\mathrm{o}\mathrm{k}\mathrm{e}\mathrm{s}}{\mathsf{\eta }}_{\mathrm{h}\mathrm{o}\mathrm{s}\mathrm{t}}{\mathsf{\eta }}_{\mathrm{T}\mathrm{I}\mathrm{R}}{\mathsf{\eta }}_{\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{f}},$$

(2)

where I_g is the amount of light guided to the edges, I_r the amount of radiation incident on the LSC, R is the reflectance of the waveguide surface, P_TIR is the total internal reflection efficiency (determined by the refraction index of the material), η_abs is the fraction of solar light absorbed by the fluorophore, η_PLQY is the photoluminescence yield of the fluorophore, η_Stokes is the energy dissipated as heat by absorption and emission processes, η_host is the transport efficiency of photons through the waveguide, η_TIR is the reflection efficiency of the waveguide related to the smoothness and defects of the waveguide surface, and η_self is the transport efficiency of the waveguided photons related to the reabsorption of fluorescence photons by the luminophore. Lowering P_TIR and η_TIR increases the amount of light received by crops. Depending on whether the PV cells are coupled to the edges of the LSC or positioned front facing, lowering P_TIR and η_TIR can result in a severe or relatively low PCE decrease, respectively.

The interactions between radiation and plants are the triggers for photosynthesis and are determined by the absorption spectra of the pigments present in the plants. The main pigments found in most photosynthetic organisms are chlorophyll a and chlorophyll b (Chl a and Chl b), but all plants can have other specific pigments that play a part in the photosynthesis process. Generally, the absorption curves of pigments range between 400–700 nm wavelengths, even though the shape of the curves depends on the specific pigment molecule. The 400–700 nm wavelength range falls under the name of photosynthetically active radiation (PAR). The amount of PAR received by the plant is measured as photosynthetic photon flux density (PPFD) in mol m⁻² s⁻¹ [16].

Another key parameter in plants is water use efficiency (WUE), which is defined as the ratio between the dry biomass (D) produced by a plant and the mass of water (W) used by it [17].

$$\mathrm{W}\mathrm{U}\mathrm{E}={\displaystyle \frac{\mathrm{D}}{\mathrm{W}}}.$$

(3)

WUE can be defined and calculated at leaf level, plant level, or field level. In this study, the WUE will be reported as it is presented in the original research paper. WUE links the rate of transpiration and evapotranspiration of a plant to its biomass yield and thus gives an indication of how water availability affects its growth. This also means that WUE is not an efficiency in the strict sense, and so it can assume values greater than 1. WUE is dependent on a number of plant characteristics and environmental factors including, but not limited to, temperature, precipitation, CO₂ amounts, soil water evaporation, leaf transpiration, etc. [18].

Considering water, another essential concept is evapotranspiration, which refers to water loss through soil evaporation and plant leaf transpiration. Transpiration occurs when stomas in the leaf open, allowing for gas exchange (such as CO₂ uptake) and water transpiration from the leaf to the atmosphere. The rate of gas exchange and transpiration is given by the stomatal conductance [mmol m⁻² s⁻¹]. The opening of the stomas and stomatal conductance are tied to various parameters, including the amount and the type of spectrum of light received by the plant.

Finally, when the land available for installation is considered, the effectiveness of multiple-use systems on shared land is calculated using the land equivalent ratio (LER), which is defined as the sum of the ratios between the yields in dual-use systems and single use systems over a fixed area. In the case of land shared between crops and PV modules, the LER is calculated as follows:

$$\mathrm{L}\mathrm{E}\mathrm{R}={\displaystyle \frac{{\mathrm{Y}}_{\mathrm{d}\mathrm{u}\mathrm{a}\mathrm{l}}}{{\mathrm{Y}}_{\mathrm{m}\mathrm{o}\mathrm{n}\mathrm{o}}}}+{\displaystyle \frac{{\mathrm{E}}_{\mathrm{d}\mathrm{u}\mathrm{a}\mathrm{l}}}{{\mathrm{E}}_{\mathrm{m}\mathrm{o}\mathrm{n}\mathrm{o}}}},$$

(4)

where Y_dual/E_dual and Y_mono/E_mono are the crop/PV energy yields in dual-use and single-use systems, respectively. Considering this definition, LER > 1 means that implementing a dual-use system is advantageous, LER 1 denotes no advantages, and LER < 1 indicates that a dual-use system is disadvantageous and that employing two separate single-use systems would be more effective [19].

3. Evolution of Agrivoltaics Research and Interest

It is useful to briefly look at how the field of agrivoltaics has evolved over the years. A keyword search for “agrivoltaic” on Scopus [20] reveals that its first appearance in the scientific literature dates back to 2011, with only one paper mentioning it. As reported in Figure 4, by 2017, only eight papers containing the word “agrivoltaic” had been published. The field nonetheless gained popularity from 2018 onward and Scopus now yields a total of 432 documents containing the term “agrivoltaic”, 133 of which were published in 2024 alone. Searching “agrivoltaics” with an “s” at the end of the word shows a similar history, with a total of 402 documents, 144 of which were published in 2024. Naturally, some documents appear under both keywords and must be accounted for to avoid duplication. The number of overlapping results can be found by inputting “agrivoltaic AND agrivoltaics” (“AND” serves as a Boolean operator, showing documents containing both the words) in the Scopus search, which produced 184 results. This means that, as of December 2024, a total of 650 distinct papers have been published in the field of AV, in a timespan of just a little over a decade, with the majority appearing in the last 4 years (from 2020 to 2024).

Searching “photovoltaic AND agriculture” shows around 20 results from the 1970s to the year 2000 and roughly 50 results from 2001 to 2010, suggesting that an embryo of agrivoltaic studies was already present long before the concept solidified and its name came into existence. In particular, the first real proposal for the simultaneous use of land for both photovoltaic energy production and agricultural activities can be attributed to Goetzberger and Zastrow [21] in 1982, while the term “agrivoltaic” seems to have been coined by Dupraz et al. [22] in 2011.

The Web of Science database also confirms the number obtained from Scopus, with negligible differences in the number of publications.

The countries most involved in the AV field so far seem to be China, Germany, India, Italy, Spain, and the United States, according to data from Scopus regarding the number of publications from each country. France, Japan, the United Kingdom, and Canada are also big contributors [20].

4. AV Literature Results

This section discusses scientific research papers from the last 10 years on agrivoltaics. They are divided in terms of the generation of PV devices employed and then differentiated based on their various setup structures. The comparison of results is performed using some common criteria while still highlighting key findings unique to each specific study. The greatest focus will be placed on the effects of AV systems on plant growth.

4.1. I Generation

The most studied configuration of AV, especially when considering first-generation PV, is the overhead/roof-mounted one, as it maximizes both the solar exposure and the shielding effect of PV panels without any precautions.

Apriani et al. [23] conducted a four-stage study on the effectiveness of a bok choy cultivation with overhead PV devices in Indonesia under a tropical climate. The devices considered were a traditional opaque crystalline Si PV panel and an STO panel. First, a simulation computed the average illuminance E_av and the average PPFD_av at ground level under both overhead coverings at different heights (from 1 m up to 3 m) and inclinations (up to 20° with respect to the ground). The results showed that the STO module yielded the highest E_av and PPFD_av at all heights and inclinations when compared to the opaque one, with differences ranging from 12% to 2%. The STO also exhibited the highest uniformity in lighting at ground level. The highest E_av, PPFD_av, and uniformity were all achieved with a tilt angle of 20° for both modules. However, at this angle, the modules reached the lowest PCE compared to any other configuration. In general, the opaque module has a slightly higher PCE than the STO one (which leads to an energy density of 115 kWh/m² compared to 110 kWh/m²). The second part of the study focused on the growth of bok choy plants in two batches: one under STO, and one with no cover as a control group. The plants grown under STO had a mean crop yield (based on leaf width, plant height, and weight) 1.4–1.8% greater than that of the plants grown with no cover.

Blando et al. [24] characterized the growth of berries (red raspberry, wild strawberry, and blackberry) inside three commercial greenhouses: one covered with transparent polycarbonate (T), one with 25% of its roof area covered with STO PV panels (called S-T in this particular study), and one fully covered with PV panels (non-T). The effects of the various illumination conditions were evaluated in terms of the bioactive compound content of the various cultivars. Anthocyanin content (an important dye in plant biology for protection against UV and antioxidant effect) was highest in the S-T and lowest in the T configuration for raspberries. It was similar for strawberries between the T and S-T configurations but higher in the non-T configuration. In blackberries, anthocyanin content was highest in the non-T and lowest in the T configuration. Glucose and fructose were higher in S-T strawberries, while in blackberries, glucose was lower in the S-T than in the T and non-T configurations. Similar results were also reported for phenols and TEAC antioxidant activity.

Still considering strawberries, Tang et al. [25] set up a greenhouse with opaque Si panels covering roughly 50% of the roof area. This type of system showed beneficial results, as the strawberry plants cultivated under this shading reported a higher yield per plant (+17%), higher average fruit mass (+10%) and size, and higher solid soluble content per fruit (16.4% vs. 13.1%).

Buttaro et al. [26] conducted a study on three greenhouses (100% coverage with poly-Si PV panels, full coverage with transparent polycarbonate modules, and 25% coverage with semi-transparent poly-Si PV modules) to assess the effects on wild rocket growth. The crop yield and dry weight were lowest for the 100% PV panel covering, the yield was comparable under semi-transparent and polycarbonate cover, and the dry weight was only slightly higher under polycarbonate than under the semi-transparent PV modules.

AV cultivations also yielded similar results to traditional ones in a study by Colantoni et al. [27]. Iberis, petunia, and cyclamen were cultivated in a greenhouse with STOPV panels on the roof (shown in Figure 5). Some PV modules were fixed on the roof above the floriculture, and they contributed to a 33% shading inside the greenhouse. The other PV modules were movable; they were put next to the fixed ones on sunny days, creating a 66% shading in the greenhouse, and moved out of the way on cloudy days when the irradiance was lower. This study highlights that Iberis showed no significant difference in plant height, diameter, and number of flowers under PV modules and control conditions. Petunia had a slightly higher plant diameter under PV, while cyclamen had a slightly higher plant height and an almost 4 cm bigger plant diameter under PV. On the other hand, control cyclamen yielded slightly more flowers per plant.

In 2023 alone, 192 × 10⁶ tons of tomatoes were produced globally [28], making them the most cultivated species among fruit and vegetables (categories that exclude tubers and cereals, the actual most cultivated crops worldwide). The popularity of tomatoes can be attributed, among various factors, to the possibility to grow them all year round and the convenience of growing them in greenhouses. This is the reason for which many AV studies focus on tomatoes. Cossu et al. (2014) [29] studied a tomato greenhouse (orientation E-W) with 50% of its roof area covered in m-Si opaque PV modules (14.2% PCE and 235 Wp). The greenhouse was built for horticultural purposes only and later used for AV scopes, and it was equipped with supplementary lighting and an air heating system. The roof configuration proved to be inconvenient for tomatoes’ needs, resulting in strong shading and uneven lighting conditions inside the greenhouse. The supplementary lighting system, although used only in winter, proved incapable of balancing the low tomato productivity, and consumed more energy than the PV modules were able to provide. Nonetheless, in winter, the greenhouse helped maintain a higher average temperature on the crops when compared to the outside, a feature deemed positive for the growth of tomatoes. Different geometrical configurations for greenhouses and different PV devices might yield more favorable results, both in electrical and agricultural terms.

Ezzaeri et al. [30,31] also assessed the effects of opaque PV modules on a greenhouse roof on tomato cultivation. The cover ratio of the PV modules on the greenhouse was 10%. The temperature of the AV system remained slightly lower than the temperature of the control greenhouse, with up to a 1.47 °C difference between the two. The same was true for relative humidity, with a maximum difference of 5% between the greenhouses. The plant height, stem diameter, number of fruits per plant, and total yield were measured for each cultivation in the two greenhouses and no statistically significant differences were observed in either case.

The effects of PV modules on the greenhouse microclimate and the growth of tomatoes and lettuce were studied by Hassanien et al. [32,33]. The PV modules were made of opaque Si PV cells with some spacing between each other, rendering the modules themselves semi-transparent. In both studies, it was observed that the temperature in the AV greenhouse was lower than that in the polyethylene greenhouse only on sunny days between noon and 2:00 p.m., with a difference of 1–3 °C. Overall, the irradiance in the polyethylene control greenhouse was 75–80% of the outside irradiance, while in the AV greenhouse, the irradiance was 30–35% of the outside one. For both lettuce and tomatoes, the fresh and dry weight of the crops were not significantly affected by the light reduction. Both lettuce and tomatoes produced bigger leaves, although the number of leaves per plant was lower for lettuce when compared to controls.

Other results for lettuce were obtained by Tani et al. [34], who cultivated lettuce in a greenhouse covered with STO modules made of interspaced opaque Si cells. Besides the control treatment, a lettuce crop was grown under PV modules with non-diffuse light (PV-T treatment) and some other lettuce was grown under PV modules with a diffusing film (PV-D treatment). Lettuce growth was reduced in the PV-T treatment due to the fluctuating light conditions, whereas diffused light at the same PPFD yielded better results in terms of biomass and leaf growth compared to PV-T, being only slightly lower than the control results. Ascorbic acid content (a key nutritional value), however, was lowest in the PV-D treatment.

Marrou et al. [35] designed an AV system with the goal of studying how shading affects water flows in a cultivation. The system consisted of PV panels placed at a 4 m elevation over an open field under three conditions: full density (50% of total radiation available for crops), half density (70% of radiation available for crops), and full sun (control plot, no shading, 100% radiation for crops). The crops chosen for the study were two varieties of crisphead lettuce, two varieties of cutting lettuce, and one variety of cucumber. The study revealed that all the studied crops, except the “Bassoon” lettuce, exhibited a higher WUE under the full-density treatment compared to under the half-density one. In addition, the relative dry matter accumulation and actual evapotranspiration were higher in the full-density treatment. Overall, the PV modules allowed for a 14–29% saving of evapotranspired water depending on the shading level.

Gonocruz et al. [36] cultivated rice in four open fields with stilt-mounted opaque PV modules placed at a 3 m elevation. The cover ratios were 29% and 14% for two separate parts of Farm A, 30% for Farm B, 39% for Farm C, and 34% for Farm D. The four farms were located in different parts of Japan and different rice varieties were cultivated in each. Crops were cultivated between 2014 and 2018. Shading affected various properties of the crops, such as SPAD (soil–plant analysis development), panicle number, and grain quality. However, the study established that up to around 36% cover ratio, the overall crop yield of the cultivations was at least 80% of that of the control cultivations.

Another AV study, conducted in Japan by Sekiyama and Nagashima [37], studied an experimental farm of 100 m² divided into three configurations: control, low PV module density, and high PV module density. The PV modules were stilt-mounted overhead opaque PV panels with an inclination of 30° with respect to the ground at an elevation of 2.7 m. The panels also had a self-cleaning glass surface. The crop used in this study was corn, and the results showed that, when compared to control, the low-density configuration yielded corn stoves with slightly higher average fresh weight, average biomass, and corn yield (roughly +5–6% each), while the high-density configuration showed a decrease of roughly 3% in each parameter. These results are consistent with the concept of photosynthetic saturation and that too much exposure to sunlight can induce other types of damage in the crops, thus making the small shading provided by the low-density configuration beneficial, while the high-density configuration causes too much shading, making light the limiting factor in PE.

Kavga et al. [38] cultivated pepper in a greenhouse with 20% of its roof covered with opaque poly-Si PV modules. The pepper fruits’ weight, size, and thickness were roughly the same between the AV greenhouse and the control one. Moreover, there were no statistically significant differences in the phenolic content or antioxidant and antiradical activities of fruits from the two greenhouses, showing that slight shading does not affect the growth of peppers. Temperatures were basically the same between the AV and the control greenhouse (covered in glass), and overall, the total crop yield was slightly higher in the AV greenhouse.

Gadhiya et al. [39] studied an insect net semi-open house with overhead opaque Si PV modules arranged with gaps between them. The experiment took place in India, in a tropical environment characterized by a rainy (June to September) and a dry season (October to May). Capsicum grown in the net house yielded 15.60 t/ha (tons/hectares), while the one grown in the open field had a yield of 10.20 t/ha. This improvement was attributed to the better microclimate created by the partial shading on the crops, resulting in an average temperature decrease of 1–2 °C, which reduced evapotranspiration. The LER of the system was calculated to be 1.97, basically double that of single-use land for the cultivation of only one crop.

In another case study based in India, Malu et al. [40] modeled cultivation fields in India with rows of grape crop interspaced at 1.8 m from each other with opaque PV modules in between. The crops went along the E-W axis and the modules were considered at various elevations and tilt angles. The simulation assumed a ground coverage ratio of ~26%, and the results highlighted that a large-scale implementation in grape farms in India might amount to roughly 3 GW.

The implementation of Java tea cultivation in AV systems was investigated from different points of view by Othman et al. [41,42,43] considering the possible economic benefits, the thermal effects on the crops, and the effects on attractiveness to pests and bugs. A Java tea cultivation under opaque PV systems experienced significantly lower heat stress, a condition in which the plant is at a temperature >10–15 °C higher than the surrounding area. This condition mostly occurs during peak sun hours at a ~1.2 m elevation from the ground, close to the back surface of PV modules. AV systems lead to lower temperatures at the crop level, and thus tend to attract fewer bugs. However, Java tea AV cultivations also require careful water drainage, since high-humidity conditions can attract specific pests, such as lace bugs, despite the lower temperatures.

Trommsdorff et al. [44] studied an optimization of overhead opaque PV panels on a cultivated field by calculating the plants’ biomass yield and the electrical yield as functions of various parameters, such as received PAR and PV row distance, among others, while also trying to minimize the surface devoted to the mounting structures of the PV panels so as not to reduce the farmable area. A case study was then conducted in Germany with bifacial PV panels over an agricultural field. The AV system was 25.3 m wide and 136.3 m long, with 3.4 m wide panels interspaced by 9.5 m and oriented 52.5° SW. The PV modules’ tilt angle was 20° and the overall PV surface area was 1206 m². In terms of electricity, the system generated 246 MWh, 17% lower than what a standard PV park (with monofacial panels) in the same place and with the same size would have generated (295.4 MWh). Over the years, results varied for the crops’ yield compared to the control crops. In 2017, a −5% yield was recorded for clover grass, while celeriac, potatoes, and winter wheat saw an 18–19% yield decrease. The following year, 2018, an 8% yield decrease was recorded for clover grass while +12%, +11%, and +3% yields were observed for celeriac, potatoes, and winter wheat, respectively. The LER of the system over the years ranged from 1.56 and 1.87.

Sforza et al. [45] employed a PV panel on a bioreactor for microalgae cultivation. The PV panel covered ⅓ of the front surface of the reactor. The research highlighted how the primary challenges of an outdoor microalgae cultivation’s photosynthetic efficiency are photoinhibition and saturation, which can occur even in winter at middle latitudes. The use of traditional opaque PV panels can mitigate this issue by reducing the amount of light entering the culture while simultaneously producing electricity.

AV is not only a synergy of PV and agriculture but also a possible tool for ecosystem and biodiversity preservation. For this purpose, Nakatani and Osawa [46] evaluated the suitability of seminatural grasslands for AV purposes. These grasslands support a rich biodiversity of flora and fauna that have adapted to some level of human activity. As a result, it seems that the disuse of this land threatens this biodiversity. Stilt-mounted PV modules, designed to minimize excessive shading, might be an effective solution to use this land without disrupting the local biodiversity, allowing plants to grow and producing clean energy at the same time. The study also concludes that the installation and maintenance costs of an AV system in seminatural grassland would not be much different from the cost of a traditional PV park, and that this kind of implementation could lead to higher social acceptance given the benefits to the local ecosystem.

Bambara and Athienitis [47] performed an energetic and economic analysis of a 923.6 m² greenhouse covered with STO modules made of opaque Si-cells. The mathematical model comprises various parameters, encompassing the thermal, electrical, economic, and optical characteristics of the greenhouse. The weather and irradiance data inputted into the calculation refer to data collected in Ottawa, Canada. The study found that, as of 2018, a system of this type built in Ottawa was not self-sufficient and might not yield significant economic benefits.

General results on the proportionality between CR and actual shading for I generation PV were found by Cossu et al. (2018) [48]. They calculated the distribution of solar radiation inside four different geometries and configurations of AV greenhouses with opaque Si panels: 100% cover ratio at a 2.5 m elevation and 20° tilt; 60% cover ratio with a venlo-type roof at a 4.5 m elevation and 26° tilt; 50% cover ratio with a gable roof at a 2.5 m elevation and 22° tilt; 25% cover ratio with a gable roof at a 2.5 m elevation and 22° tilt. The distribution of solar radiation was calculated twice for each greenhouse, once for the E-W alignment and once for the N-S alignment. For a cover ratio < 100%, the distribution was calculated considering all the panels to be adjacent to each other and arranged in a checkerboard pattern. The general findings show that total radiation inside the greenhouses decreased by 0.8% per each additional 1% of cover ratio, and that the N-S orientation allowed for more radiation inside the greenhouse–up to +24% more with respect to the E-W orientation.

Gao et al. [49] modeled an AV greenhouse with opaque PV modules based on different sun-tracking mechanisms (closed, quasi-perpendicular, no-shading, open) with various combinations of PV module densities, inclinations, and roof coverage. Among the different results, the simulation showed that no-shading sun tracking produced 6.91% more electricity. At the same time, it yielded a more uniform irradiance than traditional methods, such as quasi-perpendicular tracking, in high PV module density configurations inside the greenhouse. In low-density configurations, the best results were obtained with quasi-perpendicular tracking.

A different method for avoiding excessive shading was proposed by Vadiee et al. [50], who developed a solar blind system for AV greenhouses in which PV panels are rotated to cover the greenhouse roof once the temperature inside the greenhouse reaches the chosen threshold. Once the temperature inside drops back below this threshold, the panels then rotate back to their original position, which does not create any shading in the greenhouse. In this way, the crops receive sufficient and evenly distributed irradiation throughout the year and the need for cooling is drastically reduced. With a setpoint of 18 °C, the electrical demand of the greenhouse was found to be reduced by 73% in the experiment conducted in Shiraz, Iran.

Williams et al. [51] modeled and compared the microclimate of a traditional PV park and an AV system. The model was based on computational fluid dynamics and considered various parameters such as panel height, ground albedo, and crop evapotranspiration. The results of the simulations show that in an AV farm with PV panels mounted 4 m above soybeans plants, the temperature reached by the panels can be up to 10 °C cooler than the typical temperature of a PV panel at a 0.5 m elevation in a traditional PV park, resulting in reduced stress and degradation of the modules over time.

Jamil et al. [52] proposed a numerical simulation performed in Ladybug on an arable field with interspersed rows of vertically mounted PV modules at different distances. The goal of the simulation was to calculate the irradiance on the ground and then identify crops whose light needs were compatible with the light intensity reduction caused by the PV modules. The PV modules considered were bifacial and aligned along the N-S axis, meaning they faced E-W. The simulation was conducted with three different spacings between the rows—5 m, 15 m, and 45 m—across three different regions in Canada: Ontario, Manitoba, and Alberta. The analysis covered the April–October period, which is considered growing season in Canada. The number of crops compatible with the AV system was sixteen for Ontario, four for Manitoba, and seven for Alberta when the AV system had a row spacing of 15 m or 45 m. The simulation suggests that large-scale implementation of this kind of AV system in Canada could generate up to 532 TWh/y, which is more than 84% of Canada’s yearly energy consumption of 632 TWh. The inclusion of wheat fields in AV applications could significantly boost this potential. Figure 6 shows the cumulative irradiance levels for a 15 m spacing between the modules.

Riaz et al. [53] performed a numerical analysis considering an AV field with tilted monofacial N-W PV panels and another AV field with vertical bifacial E-W panels. Their goal was to model the effects on the amount of PAR reaching the ground and the panel efficiency losses associated with soiling of the panels’ surface. The monofacial N-W and the bifacial E-W configurations had a similar PAR and energy yield for low densities of panels in the field (<50% of standard PV parks panel density). On the other hand, at higher densities, the monofacial N-W panels generated more electricity at the expense of PAR available to the crops, while the bifacial E-W configuration was more beneficial for food production but less optimal for electricity generation. Bifacial vertical panels, however, suffer from negligible soiling under standard weekly or biweekly cleaning routines, while in tilted panels, soiling can result in a relative PCE loss of up to a 1%/day. Vertical bifacial modules also allow for significantly lower land coverage and rainfall obstruction.

Jones et al. [54] provided a model for the calculation of beam and diffuse fractions of solar radiation with overhead opaque PV modules. The beam and diffuse fractions are calculated using the module row spacing d, module elevation h, panel height a, and inclination α as input parameters. The model provides a simpler alternative to the more complex GZ model and to ray-tracing approaches, making it very useful for first-order estimates when designing AV systems. It also allows for a normalization of the quantities involved and was independently validated by a research study conducted by Dupraz et al. [22].

Exploiting ray-tracing, Katsikogiannis et al. [55] developed a simulation to optimize an AV system with bifacial opaque PV modules in four configurations: overhead stilt-mounted rows (S1), vertical (E1), overhead stilt-mounted in a checkerboard pattern (S2), and vertical checkerboard panels (E2). Higher module elevation, greater row spacing, and higher module transparency coincided with higher irradiance homogeneity. However, increasing the spacing resulted in a lower generated current. The results also show that deviations in the modules’ tilt and azimuth angles from conventional topologies do not cause significant reductions in the potential of the bifacial AV system. N-S configurations are preferable for summer cultivations of shade-tolerant plants, E-W overhead setups yield the best shading schedule and partial microclimate control, and E-W vertical systems offer the best distribution and intensity of light, making them optimal for permanent crops, especially in winter.

Varo-Martinez et al. [56] developed a simulation for an AV system with PV panels interspersed with olive hedgerows. The simulation employs ray-tracing models, and the study provides two straightforward equations for estimating oil and electrical yield for such AV systems with a significant degree of accuracy.

Despite not always being classified as AV, we also report some results obtained from studies on the application of PV systems in mushroom cultivation and livestock raising. El Kolaly et al. [57] tested the performances of m-Si passivated emitter rear contact photovoltaic cells (Mono PERC PV) mounted on the roof of a greenhouse for cultivating Pleurotus mushrooms. The PV panels were installed to evaluate whether they would be able to produce enough energy to power the entire greenhouse for microclimate control. The Mono PERC modules had an 18.1% PCE and covered a total area of 5.8 m². The PV system was able to provide double the amount of energy needed to maintain the greenhouse’s microclimate at the desired conditions. Mushrooms grown in the AV system had a higher biomass yield and higher biological efficiency (defined as the ratio between the weight of the fresh mushroom and the dry weight of the substrate) than those of the control group.

Regarding livestock farming, Lytle et al. [58] studied a 2.8 ha rabbit farm equipped with overhead opaque PV which had the potential to generate 1 GWh/y. Their analysis highlights that using the same land for both PV energy generation and rabbit pasture might significantly reduce its carbon footprint, especially when compared to single-use land for PV parks and traditional livestock pastures, such as those used for cows. Implementation of such systems could increase meat and dairy production without compromising animal welfare, alleviating the need for intensive animal farming to keep up with consumption demand.

Table 2. Summary of research papers on AV with I generation PV devices. “Exp.” denotes experimental research, “Num. An.” denotes numerical analyses.

Study	Study Type	PV Device and Structure	Tilt Angle	Field Type	Coverage Ratio	Crops	Plant Results	Electrical/Optical/Thermal Results
Apriani et al. [23]	Exp.	Opaque and STO mc-Si modules Overhead	0°	Open	100%	Bok choy	+1.4–1.8% mean crop yield under STPV	Higher irradiance uniformity under STO, better PCE for opaque PV
Blando et al. [24]	Exp.	Opaque and STO pc-Si modules Roof-mounted	19°	Greenhouse	0%, 25%, 100%	Red raspberry, wild strawberry, blackberry	Anthocyanin content: highest in 25% CR raspberries, 100% CR strawberries, and 100% CR blackberries 25% CR strawberries had higher glucose and fructose content than other CRs Glucose in 25% CR blackberries was lower than 0% CR and 100% CR	N/A
Tang et al. [25]	Exp.	Opaque pc-Si panels Roof-mounted	30°	Greenhouse	~50%	Strawberry	+17% yield per plant +10% average fruit mass 16.4% solid soluble content per fruit vs. 13.1% of control	Reduced temperature inside the greenhouse Excess light shielded
Buttaro et al. [26]	Exp.	Opaque pc-Si modules Roof-mounted	19°	Greenhouse	0%, 25%, 100%	Wild rocket	Crop yield and dry weight were lowest under 100% CR, and comparable between under 25% CR and the full polycarbonate cover	Both 25% CR and 100% CR were able to supply more energy than needed by the greenhouse appliances
Colantoni et al. [27]	Exp.	STOPV Roof-mounted, some of them being movable for variable CR	-	Greenhouse	33% (on days with low irradiance) and 66% on days with high irradiance	Iberis, petunia, cyclamen	Iberis: no significant difference in plant height, diameter, and number of flowers between both conditions Petunia: slightly higher plant diameter under PV Cyclamen: under PV, there was a small increase in plant height, significant increase in diameter, and small decrease in number of flowers	N/A
Cossu et al. (2014) [29]	Exp.	Opaque pc-Si PV Roof-mounted	30°	Greenhouse	50%	Tomatoes	Negative effect on yield due to uneven lighting, then balanced with supplementary artificial lighting	Strong shading and uneven lighting caused by the roof configuration PV modules unable to provide enough energy for supplementary lighting and heating in winter
Ezzaeri et al. [30,31]	Exp.	Opaque PV Roof-mounted	10°	Greenhouse	10%	Tomatoes	No statistically significant difference between tomatoes of control group	Lower temperature (up to 1.47 °C) and relative humidity inside the AV greenhouse
Hassanien et al. [32,33]	Exp.	Opaque mc-Si panels Roof-mounted	30°	Greenhouse	20–40%	Tomatoes Lettuce	No change in fresh and dry weight for both tomatoes and lettuce. Decrease in the number of leaves per plant for lettuce. Bigger leaves for both crops.	Lower temperature of the AV greenhouse compared to the polyethylene one only on sunny days between noon and 2:00 PM (1–3 °C difference) 30–35% of the outside irradiance inside the AV greenhouse
Tani et al. [34]	Exp.	STO mc-Si panels Roof-mounted with and without diffusing film	-	Greenhouse	50%	Lettuce	Reduced lettuce growth under STPV with direct light due to uneven lighting Better results under diffuse light (still slightly lower than control results) Lowest ascorbic acid content under diffuse light	N/A
Marrou et al. [35]	Exp.	Opaque PV Overhead at	25°	Open field	50%, 30%, 0%	Crisphead lettuce (2 varieties) Cutting lettuce (2 varieties) Cucumber	WUE higher under 50% CR for every crop except “Bassoon” lettuce when compared to 30% CR Relative dry matter accumulation and actual evapotranspiration higher under 50% CR Overall, saved 14–29% of evapotranspire water depending on the CR	N/A
Gonocruz et al. [36]	Exp.	Opaque mc-Si modules Stilt-mounted	-	Open field	14%, 29%, 30%, 34%, 39%	Rice	Up to ~36% CR the crop yield is >80% of the yield of crops grown with no PV	Large-scale implementation with these characteristics could supply up to 29% of electricity demand in Japan
Sekiyama and Nagashima et al. [37]	Exp.	Opaque PV Stilt-mounted overhead	30°	Open field	Low-density, high-density	Corn	Increased average fresh weight, biomass and yield (+5–6% each) under low-density configuration −3% average fresh weight, biomass. and yield under high-density configuration	More energy produced under the high-density configuration at the expense of sunlight available to the crops
Kavga et al. [38]	Exp.	Opaque pc-Si panels Roof-mounted	-	Greenhouse	20%	Pepper	No difference between AV and control pepper in terms of fruit weight, size, thickness, phenolic content, and antioxidant activity Slightly higher yield in AV greenhouse	No temperature difference between AV and control (glass) greenhouse
Gadhiya et al. [39]	Exp.	Opaque pc-Si PV Overhead	19.75°	Insect net semi-open house	50%	Capsicum	Increased yield in the net house (15.60 t/ha vs. 10.20 t/ha of the open field) LER 1.97	Lower temperature than open field (average decrease of 1–2 °C)
Malu et al. [40]	Num. An.	Opaque pc-Si modules Interspersed	21°	Open field	~26%	Grape	N/A	Potential to produce 3 GW with a large-scale implementation in India
Othman et al. [41,42,43]	Exp. Num. An.	Opaque mc-Si Overhead	7.6°	Open field	N/A	Java tea	Lower heat stress on plants, lower attractiveness to pests with proper drainage	N/A
Trommsdorff et al. [44]	Num. An. Exp.	Bifacial opaque PV Overhead	20°	Open field	Configuration-dependent	Clover grass Celeriac Potatoes Winter wheat	2017: yield decrease of 5% for clover grass and 18–19% for celeriac, potatoes, and winter wheat 2018: −8% yield for clover grass but +12%, +11%, and +3% for celeriac, potatoes, and winter wheat, respectively	−17% energy yield with respect to a standard PV park in the same location
Sforza et al. [45]	Exp.	Opaque Si PV Side-mounted	90°	Photobioreactor	33%	Microalgae	Enhanced microalgae growth	Reduced photoinhibition
Nakatani and Osawa [46]	Num. An.	Opaque PV Overhead	0°, 30°	Open field	N/A	Seminatural grassland	Possibility of preservation of local ecosystem	No particular difference from ordinary PV park
Bambara and Athienitis [47]	Num. An.	STO mc-Si panels Roof-mounted	0°	Greenhouse	10–50% (at increments of 10%)	N/A	N/A	Such a system would not be energetically self-sufficient at the time of study but might be in the future
Cossu et al. (2018) [48]	Num. An.	Opaque mc-Si and pc-Si modules Roof-mounted	20°, 22°, 26°	Greenhouse	25%, 50%, 60%, 100%	N/A	N/A	Solar radiation distribution calculated in the greenhouses for various CRs, tilt angles, module elevation, etc. In general, −0.8% total radiation in greenhouses per additional 1% CR. The N-S orientation allows for up to +24% total radiation than E-W.
Gao et al. [49]	Num. An.	Opaque PV Roof-mounted	Configuration-dependent	Greenhouse	Configuration-dependent	N/A	N/A	No-shading sun tracking produced 6.91% more electricity and yielded more uniform irradiance than other sun-tracking methodologies for high PV module density Quasi-perpendicular tracking yielded the best results for low PV module density
Vadiee et al. [50]	Exp.	Opaque PV Roof-mounted, movable	30°	Solar blind greenhouse	Configuration-dependent	N/A	N/A	Panels move to cover the roof once the interior temperature goes over a threshold and stay there until it cools down The inside of the greenhouse receives homogeneous irradiance Electrical demand of the greenhouse reduced by 73% (experiment conducted in Shiraz, Iran)
Williams et al. [51]	Num. An.	Opaque PV Stilt-mounted	25°	Open field	N/A	Soybeans	N/A	PV panels in this system can be up to 10 °C cooler than in traditional PV parks
Jamil et al. [52]	Num. An.	Bifacial opaque pc-Si modules Interspersed, vertically mounted	90°	Open field	N/A	N/A	The analysis identifies several crop types compatible with this kind of system in various regions of Canada (16 for Ontario, 4 for Manitoba, and 7 for Alberta)	Large-scale implementation could yield up to 84% of Canada’s yearly energy consumption
Riaz et al. [53]	Num. An.	Opaque PV Interspersed monofacial N-S, vertical bifacial E-W c-Si panels	30°, 90°	Open field	Configuration-dependent	N/A	N/A	The two panel configurations have similar amounts of PAR on the ground and energy production with low panel density (<50% of standard PV parks) At high density, bifacial E-W gives more PAR and less energy and vice versa for monofacial N-S Soiling in tilted monofacial can yield up to −1% PCE/day Vertical bifacial E-W suffers from negligible soiling
Jones et al. [54]	Num. An.	Opaque PV Overhead	Configuration-dependent	Open field	N/A	N/A	N/A	Model for the calculation of beam and diffuse fraction of solar radiation on this type of field without ray-tracing approaches
Katsikogiannis et al. [55]	Num. An.	Bifacial opaque PV Overhead stilt-mounted rows, vertical, overhead stilt-mounted checkerboard, vertical checkerboard	90°	Open field	N/A	N/A	N/A	Increasing module elevation, row spacing and/or module transparency results in higher irradiance homogeneity. E-W vertical systems give the best shading schedule, microclimate, and irradiance distribution
Varo-Martinez et al. [56]	Num. An.	Opaque PV Interspersed	E-W tracking	Open field	N/A	Olive hedgerows	Simulation based on ray-tracing that gives an equation to predict oil yield	Simulation based on ray-tracing that gives an equation to predict energy yield
El Kolaly et al. [57]	Exp.	Opaque mc-Si Roof-mounted	30°	Greenhouse	19.4%	Pleurotus mushroom	Increased biological efficiency (weight of fresh mushroom/dry mass of substrate)	Produced more than double the energy needed by the microclimate control system
Lytle et al. [58]	Num. An.	Opaque pc-Si modules Overhead	30°	Open field	≤50%	Rabbits Rabbit pasture	This kind of AV system might reduce the carbon footprint of animal farms	Possibility of generating up to 1 GWh/y on a 2.8 ha field

highlights the key information and findings of AV research on I generation PV devices.

Many of the results obtained from studies on I generation PV devices suggest their potential for successful implementation in AV contexts. However, fixed opaque PV panels pose a significant risk of depriving crops of too much radiation, which can lead to decrease in biomass and/or yields, while STO modules strongly mitigate this risk. The best solutions to this problem may be solar blinds and movable modules, as these offer the possibility of dynamically adjusting irradiance on crops based on weather conditions. In almost all cases, the presence of PV devices creates more favorable temperatures and leads to better WUE.

4.2. II and III Generation

The flexibility and semi-transparency of II generation PV devices open up new possibilities compared to I generation devices, such as spectral separation and different coverage types to more carefully manage shading conditions. III generation PV devices offer even more unique possibilities compared to previous generations— for example, achieving a higher degree of transparency with higher uniformity.

Aroca-Delgado et al. [59] studied a 1024 m² greenhouse with overhead flexible PV panels covering 9.8% of its roof area. Despite the PAR reduction, tomato properties—such as total yield, plant pH, number of flowers per branch, and color—were basically the same in the control group and the AV system. Tomato plants in the control group yielded big fruits but in a smaller quantity, while the plants produced smaller fruits in larger quantities under the AV system. As a result, the total yield was the same for both systems. Additionally, tomatoes grown in the AV system had a higher total soluble content per fruit.

Similarly, Osterthun et al. [60,61] presented a semi-transparent spectrally selective PV cell with a transmission spectrum that was viable for microalgae growth and photosynthesis. The cell consists of six layers: glass, an AZO film (Al-doped ZnO), an n-doped electrode, an intrinsic a-Ge, a p-doped electrode, and a selective back reflector. The active layers are all ultra-thin; thus, the cell is highly transparent, absorbing only resonances. Such resonances are achieved thanks to the back reflector, which only reflects green and IR radiation. This results in a PV cell that only uses green and IR wavelengths for electricity generation, and which is highly transparent to blue and red light. Different resonances can be achieved by varying the back reflector thickness, allowing wavelengths absorbed by the cell to be tuned. A test run was performed on A. obliquus grown under white light, selective solar cells, and an Si layer (which transmits red light). The photon flux under the selective cells was 75% lower than under white light, and this resulted in 55% less biomass than under white light after 72 h of cultivation. However, the selective cell did outperform the Si layer; despite the higher photon flux under the Si-layer, the biomass yield was lower due to the lack of blue wavelengths which instead were present under the selective cell. Algae grown under the selective cells showed higher photosynthetic efficiency than those grown under silicon and white light.

On the topic of energy production, Pérez-Alonso et al. [62] monitored the electrical performances of two greenhouses with thin-film PV cells covering 9.79% of the roof area and placed in two different checkerboard configurations. Using the results obtained from the experiment, along with many other results reported in the literature, an artificial neural network was generated, capable of predicting the instantaneous power production of complex AV and BIPV systems by considering the different parameters. This neural network was able to predict the instantaneous power output with an uncertainty of 20 W.

Semi-transparent III generation PV integrates well with microalgae cultivation. In fact, Barbera et al. [63] experimented with a photobioreactor for microalgae cultivation coupled with a dye-sensitized solar cell (DSSC) with an overall transmittance of 48% in the visible range. The DSSC covered the entire irradiated surface of the photobioreactor (see Figure 7) and the algae growth was tested with (control) and without the presence of the DSSC. The tests were conducted at different irradiation levels. With irradiances smaller than the photoinhibition threshold, the algae biomass productivity was lower in presence of the DSSC, while for irradiance close to or higher than the photoinhibition value (which is around 500 μmol m⁻² s⁻¹), the yields were comparable or greater in the DSSC case than in the control group. Considering the day–night cycle, which creates a constant oscillation between low and high irradiances, the results indicate the possibility of having no net negative effects on the growth of microalgae, even with full DSSC coverage. This is due to the negative effects of low irradiance balancing out with the decrease in photoinhibition caused by high light intensities.

Remarkable results were achieved with STPV III gen. devices by Aira et al. [64], who employed amorphous Si (a-Si) PV glass with 30% AVT as a greenhouse covering (100% CR). The control part of the greenhouse was covered with ordinary glass. Lettuce and beans were planted inside following a quincunx layout. The AV greenhouse had lower daily temperature maxima than the control greenhouse, resulting in the temperature rise being slower. CO₂ levels were lower in the AV greenhouse, as plants grown there produced more spontaneous biomass than those grown under control conditions, whose growth was dampened by excessive PAR levels. The AV greenhouse also proved to be self-sufficient in terms of energy, generating enough power to cover all its needs.

PAR reduction induced similarly positive effects in a study by Barron-Gafford et al. [65], in which the authors used an overhead semi-transparent PV panel on a dryland cultivation of chiltepin peppers, jalapenos, and cherry tomatoes, along with a control group in an open-sky configuration. Despite the AV system receiving a considerably lower amount of PAR, the AV chiltepin peppers exhibited a +33% increase in cumulative CO₂ uptake and a threefold increase in fruit production, with no change in WUE. The jalapenos exhibited an 11% decrease in CO₂ uptake but the WUE increased by a +157%, resulting in no difference in total fruit production between the AV and control groups. Cherry tomatoes exhibited a +65% increase in both CO₂ uptake and WUE, leading to double the fruit production with respect to the control group. The microclimate of the system was also closely monitored, with the AV system having reduced air temperatures and higher air and soil moisture.

However, the effects of PAR reduction are highly dependent on the crop, as evidenced by Chavan et al. [66] in two experiments using “Smart Glass” (SG) in an eggplant cultivation greenhouse. The SG blocked 85% of UV, 58% of far-red, and 26% of red light, leading to an overall 19% reduction in PAR. This reduced the heat load by 8% and improved the WUE and nutrient use efficiency. As a result, the plants had unchanged nutritional quality but experienced considerably lower fruit yields (−28% fruit number, −32% fruit weight) with respect to the control. Nutritional quality was evaluated by the authors based on pH, titratable acidity, moisture, mineral content, elemental composition, metabolites, total soluble solids, sugar, fat, and nitrogen content. These characteristics were basically unchanged, aside from some differences in total sugars (+8%), iron (+28%), and mineral content (−9% and −28% for the first and second experiment, respectively). The lower fruit yield is attributed to the lower PAR, which results in a decreased photosynthesis rate and increased abortion rate. Using an SG with higher PAR wavelength transparency could prevent the decrease in fruit yield.

Besides crops, STPV provided also promising results when implemented on photobioreactors for microalgae cultivation. Cho et al. [67] implemented a spectrally selective PV cell on a photobioreactor for simultaneous microalgae cultivation and energy generation. The absorbance of the PV cells was very high for green wavelengths, high in the blue range, and negligible for red light. The presence of PV cells led to a +40% increase in PE at very low irradiance values (~0.05 sun), an enhancement not seen at other irradiance values. For irradiances of 0.2 sun and 0.6 sun, the algae biomass yield was 85% of the reference cultivation, despite the number of received photons being up to 55% less.

Evidence also points to the strong energetic viability of STPV. Li et al. [68] investigated a solar blind system made of semi-transparent bifacial PV modules on a greenhouse. When irradiance was higher than a certain threshold, the modules were placed parallel to the roof, creating a 42% shading on the crops. When the irradiance was lower than the threshold, the modules were rotated to be perpendicular to the roof. The greenhouse was able to generate more energy than what was needed to run all its appliances.

In terms of energy, analogous results were obtained by Yano et al. [14], who prototyped an STPV module made with spherical solar microcells (see Figure 8) with a 1.8 mm diameter. The microcells were embedded into the modules in a checkerboard pattern. The outer layer of the microcells was an n-type semiconductor and the inner layer was a p-type semiconductor, with the electrodes positioned at the poles. The prototypes were of two kinds: low and high cell density (which is three times the density of the low-density prototypes). The shading caused by the high-density modules at 2 m distance was ~40%, while it was <20% for the low-density modules at the same distance. The high-density modules produced three times the energy of the low-density ones, aligning with the density ratio. However, both types would be able to supply enough electricity to ordinary greenhouses in high-irradiance regions.

Organic photovoltaics (OPV) are novel PV technologies based on organic molecules, a characteristic that not only allows for tuning the bandgap and absorption spectra but also facilities the easy recycling and disposal of PV modules at the end of their lifecycle. Regarding absorption spectra, studies like the one by Chang et al. [69] highlight that, unlike traditional semiconductors, organic semiconductors tend to exhibit well-defined energy levels instead of energy bands, opening up the possibility of tuning the absorption spectra with high precision by combining different polymers.

Rapid degradation is one of the major challenges faced by OPV devices, although some polymers exhibit the fascinating property of self-healing when not exposed to the sun. On this matter, Dos Reis Benatto et al. [70] studied the degradation of OPV modules based on carbon (C-OPV-L and C-OPV-N) and AgNW (AgNW-OPV). The C-OPV-N and AgNW-OPV modules were made with a better-quality barrier foil than the C-OPV-L. The modules were tested in two outdoor experiments and three different greenhouses (the modules were tested in one greenhouse before being transferred into the next greenhouse, with each greenhouse exhibiting different humidity values and irradiances). In a total of 300 days of greenhouse tests, the PCE went from an average of ~2.3% to 1.4% for the C-OPV-N modules, ~1.4% to 0.4% for the C-OPV-L modules, and ~2.2% to 1.8%. for the AgNW-OPV modules. The C-OPV-L modules were rendered unusable after 400 days of outdoor testing, while the PCE of the C-OPV-N modules went from 2.3% to 1.2% in roughly 200 days, after which they maintained a stable PCE for the remaining 250 days. Finally, the PCE of the AgNW-OPV samples dropped considerably after 200 days, coinciding with wintertime, but then saw an increase in late spring, with the best sample recovering up to 90% of its initial PCE. The C-OPV (-L and -N) modules did not show any recovery effect for the PCE.

In an investigation of OPV recovery, Magadley et al. [71] used OPV on a greenhouse, covering 26% of the roof area. In general, it was shown that the OPV modules gave the best electrical performance at lower tilt angles and that the PCE and FF were better in the morning than in the afternoon. This variation is attributed to degradation at high irradiances in the late morning and early afternoon followed by an overnight recovery of the modules.

Similarly, Waller et al. [72] tested a curved-roof greenhouse with arrays of semi-transparent organic photovoltaic (OPV) panels to study their PCE and its evolution over time due to degradation of the modules. The average PCE of the OPV arrays was 1.82%. In addition, the study found that the maximum power point and the short circuit current were not only dependent on the total irradiance on the modules but also specifically depended on the direct irradiance. However, the OPV arrays demonstrated a relative decrease in PCE of 38.6% over a 5-month measuring period.

Despite the challenge of degradation, interactions between OPV and crops seem to range from neutral to beneficial. Y. Liu et al. [73] produced flexible transparent OPV modules with an absorbance of >90% at IR wavelengths (which comprise 51% of the total energy of the solar spectrum) and an overall AVT of ~34%. The OPV module consists of layers of the following materials stacked on top of each other: PET/Ag mesh, PH1000, ZnO, active layer, MoO₃, Au, and Ag. The modules were tested for crop growth, and no significant differences were observed between the control plants and the ones grown under the OPV arrays.

Beneficial results for crop growth were found by Friman Peretz et al. [74,75], who evaluated the viability of OPV modules as greenhouse cover. Two experiments were performed on tomatoes: in the first one, cultivation took place in both the AV greenhouse and a standard control one, while in the second experiment, cultivation took place in the AV greenhouse and in another greenhouse that was partially covered with black sheets. The coverage of the total greenhouse roof area with the OPV modules and black sheets was 39% and 25%, respectively, with the OPV modules having a 20% transmittance in the PAR range. Thanks to the reduced air and canopy temperatures, the cumulative number of tomatoes, their mass, and the average single tomato mass were 9%, 36%, and 21% higher, respectively, than the standard control group and roughly the same as the ones grown in the black sheet control greenhouse.

Technological developments like the one presented by Yang et al. [76] have opened up the possibility of higher PAR transmission and better spectral matching. The authors developed a semi-transparent OPV module based on tandem photonic crystals (TPCs) that allows for improving the absorption and transmission of the OPV. The use of TPCs allows the OPV to have an average transmittance of 40.3% between 400 and 700 nm, which is a relative increase of +20.7% compared to semi-transparent OPV without photonic crystals. The absorption rate is 51.5%, slightly higher than the 50.6% of the same OPV without TPCs.

Ravishankar et al. [77] proposed a simulation of the thermal performance of a greenhouse consisting of semi-transparent organic thermal solar cells to see whether it would be able to achieve net zero emissions (NZE) for its own heating needs. The greenhouse is assumed to have an N-S orientation, with the active layer of the organic cells absorbing IR light and being mostly transparent to visible light. The simulation was performed for three climate types—hot-dry, mixed-humid, and cold (as delineated by the U. S. Department of Energy)—considering many different heat sources, airflows, and heat exchanges. The results highlight that in the hot-dry climate, the greenhouse was able to achieve NZE even during winter. In the mixed-humid climate, the greenhouse failed to maintain NZE conditions in winter and required external energy sources, but over the course of the year, it produced a surplus of energy relative to its own consumption. Finally, in the cold climate, the greenhouse failed to achieve NZE unless modifications were introduced, such as cells with better efficiency or periods when the greenhouse remains unused.

Concentrated photovoltaics (CPV) can be used to increase both the total energy production and efficiency of traditional PV modules while also using significantly less PV-active material. However, traditional CPV is held back by the complexity of its mounting structures, which are often expensive and not suitable for automatic assembly. In this context, Kussul et al. [78] proposed a different approach to traditional CPV, designing a structure with flat triangular mirrors that approximate a parabolic shape when mounted. Such a CPV system would be employable in AV applications, possibly reducing the shading on crops without compromising energy generation. The authors also provided simulations for two possible methods of automatic assembly of the modules, something that has been a major problem for CPV thus far. This combination of a simple structure, flat mirrors, and automatic assembly might significantly lower the costs of CPV systems.

However, the implementation of CPV in AV contexts has sparked creative solutions to meet the necessity of partial solar spectrum transmittance and separation. L. Liu et al. [79] and W. Liu et al. [80] developed an innovative wavelength-selective CPV system making use of a dichroic interference film on parabolic glass with a dual tracking system (shown in Figure 9). The dichroic film transmits light in the 400–500 nm, 600–700 nm, and 900–1000 nm intervals (with transmittances of >90%, >90%, and 20% respectively), reflecting everything else. Along the focal axis of the parabolic glass, a PV strip collects the reflected light. In this way, all the photosynthetically useful radiation is transmitted by the system and everything else is reflected onto the PV cells, which—being a narrow strip—provide minimal shading on the crops. Moreover, the different potential wavelength needs of specific crops might be supplied either with a different film that allows for the transmission of those wavelengths or by using supplementary LED lighting. The system was tested in China, where the maximum PCE obtained was around 8.84% and the crops grown under the system (lettuce, cucumber, and water spinach) yielded better results (in terms of crop height, weight, soluble sugar content, and photosynthetic rate) than the control crops. This result could be attributed to the lower temperature of the crops, the 26% reduction in water evaporation, and the prevention of sunburn.

On a similar note, Zhang et al. [81,82] employed multilayer polymer films (MPFs) to act as spectral splitter for AV purposes. The spectral splitting effect was achieved thanks to interference effects caused by the multilayer structure of the films, where each layer has a different refractive index. The MPFs showed very high transmittance (>95%) at selected wavelengths and high reflectance for selected wavelengths intended to be blocked from reaching the crops. The transitions between high and low transmittance wavelength intervals was also very sharp. Reflected light was then concentrated on a narrow PV cell. Lettuce, potatoes, D. officinale, and N. tabacum batches were cultivated under an MPF and under a reference glass. Parameters such as PPFD, microclimate, leaf area, fresh biomass, and dry weight were monitored and compared between the two groups of each crop. Overall, plants under the MPF showed, among other things, an improved biomass yield (being as high as +71% for tobacco) and a better photosynthesis rate. In addition, the microclimate had more favorable characteristics, with the decrease rate of the relative soil moisture being −0.94%/h under MPF compared to −2.57%/h under glass.

A different technology was proposed by Sato and Yamada [83], who studied two types of highly transparent CPV modules designed to concentrate direct sunlight on high-efficiency III-V multijunction PV cells and allow diffuse sunlight reach the ground instead. Diffuse sunlight accounts for 30–60% of total yearly solar irradiation on Earth depending on the geographical location. The type A module (Figure 10A) makes use of 100 mm × 100 mm PMMA Fresnel lenses, which concentrate light on a 10 mm × 10 mm PV cell, while the type B module (Figure 10B)) uses silicone-on-glass (SOG) planoconvex aspheric microlenses (each one being 14.14 mm × 14.14 mm), which concentrate light on 2.5 mm × 1.5 mm PV cells. None of the lenses had any anti-reflection coatings. The efficiency of the modules was calculated as DNI-based, meaning that they were computed as the ratio between the maximum power output of the modules and the direct normal irradiance impinging on them. The type A module had an average DNI-based efficiency of 26.7% and the type B module had an average efficiency of 18.5% (which can be increased up to 28.9% by changing the aperture of the SOG lenses). Depending on the ratio of diffuse to total irradiance, the amount of light transmitted by the modules ranged from 15.3% to 63.7% for type A and from 38.0% to 63.8% for type B. Such values allow for 160 W/m² of irradiance on the ground, even when fully covering a surface by placing the modules side by side with no gaps in between.

Shalom et al. [84] presented a simulation of two possible types of a spectral beam splitter with an N-S orientation and a tracking system. The two beam splitters have a very high transmittance in the 200–600 nm range and very high reflectance, ranging from 600 nm to 2200 nm. The setup is expected to have two beam splitters mounted perpendicular to each other in a “V” shape with a bifacial PV cell vertically mounted in between, creating a 45° angle with the two beam splitters. When radiation arrives on the beam splitters, the selected wavelengths are transmitted to the ground while the rest are reflected on the PV cell. The simulation indicates an overall loss of PAR on the crops limited to 10%, which is lower than the losses of more traditional AV systems.

The results reported for II and III generation PV devices in AV are further summarized in

Table 3. Summary of research papers on AV with II and III generation PV devices. “Exp.” denotes experimental research, “Num. An.” denotes numerical analyses.

Study	Study Type	PV Device and Structure	Tilt Angle	Field Type	Coverage Ratio	Crops	Plant Results	Electrical/Optical/Thermal Results
Aroca-Delgado et al. [59]	Exp.	Flexible PV Overhead roof-mounted a-Si Thin film	-	Greenhouse	9.8%	Tomatoes	No change in total yield, plant pH, number of flowers per branch, and color	Overall PCE of 4.18% during the first experimental season and 3.67% during the second one
Osterthun et al. [60,61]	Exp.	Semi-transparent spectrally selective AZO and a-Si cells	90°	Photobioreactor	100%	A. Obliquus microalgae	−55% biomass under AZO cell More biomass under AZO cell than under opaque Si Highest photosynthetic efficiency under AZO	The cell is highly transparent to blue and red light and absorbs green and IR light −75% photon flux under AZO cell with respect to direct light
Pérez-Alonso et al. [62]	Exp. Num. An.	a-Si Thin-film PV modules Roof-mounted, checkerboard pattern	Variable	Greenhouse	9.79%	N/A	N/A	Developed a neural network capable of predicting the instantaneous power generation of AV and BIPV systems with an uncertainty of 20 W
Barbera et al. [63]	Exp.	DSSC with 48% transmittance in the visible range	90°	Photobioreactor	100%	Microalgae	Lower algae biomass productivity at PPDF < 500 μmol m−2 s−1 and greater or equal above that threshold thanks to shading effect of DSSC preventing photoinhibition	Day–night cycles allow for no net negative effects in real life applications as there is constant alternation of low and high PPFD
Aira et al. [64]	Exp.	a-Si Roof- and side-mounted	35°, 90°	Greenhouse	100%	Lettuce Beans	Lower CO2 levels in AV greenhouse Increased biomass of plants grown in AV greenhouse	Daily maximum temperature lower in AV greenhouse than control AV greenhouse energetically self-sufficient PV glass shielded excessive PAR
Barron-Gafford et al. [65]	Exp.	STPV Overhead	32°	Open field	-	Chiltepin pepper Jalapeno Cherry tomatoes	Chiltepin pepper: +33% cumulative CO2 uptake, 3× fruit production, no change in WUE Jalapeno: −11% CO2 uptake, +157% WUE, no change in fruit production Tomatoes: +65% CO2 uptake and WUE, 2× fruit production	STPV module presence reduced air temperature and allowed for higher air and soil moisture
Chavan et al. [66]	Exp.	Smart Glass (SG)	Variable	Greenhouse 100%	-	Eggplant	Improved WUE and nutrient use efficiency Unchanged nutritional quantity Lower yield (−28% fruit number, −32% fruit weight) Reduced photosynthesis rate and increased abortion rate	SG blocked 85% of UV, 58% of far red and 26% of red for an overall 19% reduction of PAR −8% heat load
Cho et al. [67]	Exp.	Spectrally selective PV cell	60°	Photobioreactor	6%	Microalgae	+40% PE at 0.05 sun irradiance Between 0.02–0.06 sun, the biomass yield was 85% of control despite having up to 55% fewer photons	Very high absorbance of the PV cell at green and blue wavelengths, negligible for red light
Li et al. [68]	Exp.	Bifacial STPV Roof-mounted, movable	Variable	Solar blind greenhouse	Variable	N/A	N/A	At high irradiance, the STPV modules would move to create 42% shading in the greenhouse The system was able to produce more energy than needed by all the appliances
Yano et al. [14]	Exp.	STPV with spherical microcells Roof-mounted	26.5°	Greenhouse	50–100%	N/A	N/A	40% shading at 2 m distance from modules with high microcell density, <20% shading at 2 m for low-density module Both types of modules would produce enough energy to supply an ordinary greenhouse
Chang et al. [69]	Exp.	OPV	N/A	N/A	N/A	N/A	N/A	Organic semiconductors tend to have more defined energy levels instead of energy bands
Dos Reis Benatto et al. [70]	Exp.	Carbon and AgNW OPV Roof-mounted	-	Greenhouse	N/A	N/A	N/A	In 300 days of testing: 39% PCE reduction for C-OPV-N modules, 71% PCE reduction for C-OPV-L modules, 18% PCE reduction for AgNW modules AgNW modules showed a PCE recovery when going from winter to spring
Magadley et al. [71]	Exp.	OPV Roof-mounted	Variable	Greenhouse 26%	26%	N/A	N/A	Best electrical performances at low tilt angles PCE and FF better in the morning than in the afternoon thanks to overnight recovery
Waller et al. [72]	Exp.	ST OPV Roof-mounted	Variable	Curved roof greenhouse	N/A	N/A	N/A	38.6% decrease in PCE over 5 months Maximum power point and short circuit current depend not only on total irradiance but also on direct irradiance specifically
Y. Liu et al. [73]	Exp.	Flexible transparent OPV	N/A	N/A	N/A	-	No significant difference from control crop growth	AVT ~34% and >90% absorbance of IR wavelengths
Friman Peretz et al. [74,75]	Exp.	OPV Roof-mounted	0°, 22°, 41°, 46°	Greenhouse	39%	Tomatoes	+9% cumulative number of tomatoes +36% tomato mass +21% average single tomato mass	20% PAR transmittance of the OPV modules
Yang et al. [76]	Exp.	ST tandem photonic crystal OPV	N/A	N/A	N/A	N/A	N/A	40.3% average transmittance between 400–700 nm
Ravishankar et al. [77]	Num. An.	Thermal organic solar cells Roof-mounted	27°	Greenhouse	100%	N/A	N/A	Hot-dry climate achieves NZE even during winter Mixed-humid climate does not achieve NZE during winter but overall produces surplus energy across the year Cold climate does not achieve NZE
Kussul et al. [78]	Num. An.	Flat triangular mirrors CPV	N/A	N/A	N/A	N/A	N/A	Lower installation costs Lower shading Automatic assembly
L. Liu et al. [79] W. Liu et al. [80]	Exp.	Dichroic film CPV Overhead	N/A	Open field	100%	Lettuce Cucumber Water spinach	Improved crop height, weight, total soluble content, photosynthetic rate −26% water evaporation Prevented sunburn	>90% transmittance between 400–500 nm and 600–700 nm 20% transmittance between 900–1100 nm 8.84% PCE
Zhang et al. [81,82]	Exp.	Multilayer polymer films CPV Overhead	N/A	Open field	100%	Lettuce Potatoes D. officinale Tobacco	Improved biomass yield (up to +71% for tobacco) Improved photosynthesis rate	Very high transmittance of selected wavelengths Sharp transition between low and high transmittance intervals
Sato and Yamada [83]	Exp.	CPV with multijunction III-V cells Overhead		Open field	100%	N/A	N/A	Using Fresnel lenses or planoconvex aspheric microlenses, direct sunlight is concentrated on the PV cells and diffuse light is directed toward crops This setup allows for a uniform 160 W/m2 irradiance on crops even with 100% CR
Shalom et al. [84]	Num. An.	Bifacial PV cell with beam splitters	Variable	N/A	N/A	N/A	N/A	High transmittance between 200–600 nm High reflectance between 600–2200 nm <10% PAR losses

.

The flexible and adjustable nature of the optical properties of II and III generation PV can circumvent the risk of light-starvation for crops. The ability to perform spectral selection, separation, and enrichment allows for experimenting with unconventional geometries and structures in AV systems, facilitating the optimization of light and temperature conditions to a higher degree compared to I generation PV. This is supported by the fact that most of the analyzed studies involving plants reported better or unchanged crop yields and/or nutritional values. The trade-off of creating more favorable conditions for crop growth is the generally lower PCE and energy generation potential of II and III generation PV compared to traditional opaque PV.

LSCs

LSCs combine the possibilities of other semi-transparent PV devices while also allowing for simplified solar spectrum separation with the implementation of fluorophores. Fluorophores can be either embedded in a plastic film (e.g., EVA, PVB) which is then laminated in the LSCs (as is the case for glass LSCs) or embedded directly into the matrix itself (as is the case for PMMA LSCs and other similar polymers). The embedded fluorophores absorb certain portions of the incoming radiation and emit fluorescence, each with a characteristic spectrum. For these reasons, they make optimal light filters and enrichers, and including PV cells at the edges or front-facing makes it possible to carefully adjust shading levels. Organic dyes are popular fluorophores thanks to their high QE and availability. In fact, Detweiler et al. [85] used Lumogen Red F305 doped LSCs in greenhouses for microalgae and cyanobacteria growth (C. reinhardtii, C. vulgaris, D. salina, B. sudeticus, S. platensis). The microalgae and cyanobacteria were each grown in four batches: control, LSC light, LSC med, and LSC dark (with light, med, and dark denoting increasing concentrations of Lumogen Red in the LSCs). The experiment showed that the cultures under LSC panels generally grew as well or better than the control cultures, as evidenced by the ratio of Chla abundance to the number of cells and the abundance ratio of Car to Chla (with Car being carotenoid) being comparable between the different cultures. In addition, the LSCs were also able to power all the electronics needed for the growth and monitoring of the cell cultures.

On the topic of Lumogen Red, Pedron et al. [86] applied LSCs doped with Lumogen Red to cultivars of plants to be used for the phytoremediation of sites contaminated with arsenic and lead. Several 400 mL pots were planted with B. juncea, L. albus, and H. annuus. The tests were carried out both indoors and outdoors. In the first case, there were two batches of plants: one covered with an overhead LSC and one with no cover. For the outdoor tests, one batch of plants was fully enclosed by LSC panels and the other by transparent polycarbonate. The soil of each pot contained 300 g of As and Pb. Despite the 70% reduction in PAR, the plants grown under the LSCs indoors had +25.12%, +27.05%, and +28.17% biomass in their aerial parts with respect to the control ones. The As/Pb uptake, measured as the mass of As/Pb absorbed per unit mass of the plant [mg/kg] was basically the same for plants grown under LSCs and the control group. Despite having the same uptake, the total contaminant accumulation of the LSC plants was higher than that of the control ones due to their overall higher biomass. The outdoor test yielded analogous results to the indoor test.

A novel combination of red-colored LSCs with an innovative geometry was presented by Raeisossadati and Moheimani [87], who employed red PMMA fluorescent LSCs with a bent and comb-like shape in a microalgae culture. Microalgae growth often leads to a turbidity of the water, resulting in less light reaching the depths of the culture. With this peculiar shape, part of the light trapped in the LSC by the exposed flat part was guided to the bent part that was submerged in water. The comb-like shape is useful because it provides a greater edge surface compared to a normal rectangular shape, allowing light to escape more easily and to be more homogeneously distributed in the water. The results indicate an increase in the biomass productivity, nitrogen assimilation, and lipidic content of the algae grown under LSCs, which is attributed to the decrease in photo-limitation at the depth of the cultures.

In one of their works, Goti et al. [88] synthesized new organic quinoxaline-based fluorophores specifically conceived for use in LSCs for agrivoltaic applications. The DQ-Th dyes, tested in toluene solutions and polymethylmethacrylate (PMMA) films, are characterized by a high absorbance rate of between 300–400 nm and 500–600 nm, as well as 40–60% transmittance in the 400–500 nm window and >80% transmittance from 600 nm onward. The emission spectra range from roughly 600 nm onward, with peaks in the 650–750 nm region. Such absorption and emission spectra allow for the collection of green light, which is less relevant for photosynthesis, and its re-emission in the red region, where it can be harvested by PV cells at the edges of the LSCs—without significantly affecting the more useful portions of the PAR. The authors also found that for these dyes, the optical efficiency (OE) of the LSCs increases with increasing dye concentration up to 0.8% wt% (where the optical efficiency is 4.5%), but then starts to decrease at higher concentrations.

Loik et al. [89] used a greenhouse constructed with wavelength-selective PMMA LSCs with PV strips embedded in them. The theoretical PCE of the system was 9.4%. The reddish-colored LSCs have an absorbance greater than 80% between 300–400 nm and 500–600 nm, and of 40–60% between 400–500 nm. Their photoluminescence ranges between 550 and 700 nm, peaking at 630 nm (although this low Stokes shift may introduce non-negligible reabsorption). The growth of various types of tomatoes under this greenhouse suggests no major differences in the number of fruits yielded by each plant with respect to the control, except for one type of tomato, suggesting that the LSCs’ shading effect does not seem to impact the crops.

QDs are an emerging and exciting type of fluorophore which exhibit very high Stokes shifts and tunable absorption and emission spectra. Keil et al. [90] developed a bilayer LSC made of CdSe/CdS and SiQD-PMMA on a glass substrate. This combination allows for a strong absorption up to around 450 nm, with photoluminescence peaks at 600 nm (from CdSe/CdS) and 800 nm (from Si QDs). This kind of device is particularly suitable for agrivoltaic implementation because of its good transmission at the chlorophyll absorption bands and the overlap between said bands and its emission spectra.

Agricultural activities and outdoor exposure intrinsically lead to rainfall and the presence of dust. Siripurapu et al. [91] studied how soiling affects the efficiency of LSCs for BIPV and AV uses. The LSCs in question were made of PMMA, with an overall transmittance of 70%between 400 and 900 nm. The LSCs were covered with increasing amounts of dust on the surface. The presence of dust resulted in lower transmittance and a decreased power output but had no effect on the waveguiding efficiency, as dust particles do not create a strong optical coupling with the surface of the LSCs. Accumulating dust on the back surface of the LSC (i.e., where light would be transmitted to the crops) causes a slight increase in the power output, as dust particles act as backscattering centers, reflecting some radiation back into the LSCs. The LSCs were then sprayed with deionized water, which did not affect the transmittance of the modules but had severe effects on their waveguiding efficiency due to the interfaces created between the water drops and the PMMA. At high water surface density on the LSCs, the effect saturates as water creates a constant layer above the PMMA, resulting in an average effect over the entire module. The last test involved wetting the LSCs with an aqueous solution of NaCl and letting it evaporate. This led to a combination of dust and water effects: the dried residues acted as scattering centers, similar to dust particles, while also exhibiting a strong optical coupling to the LSC surface, which severely impacted waveguiding efficiency.

Regarding the performance and efficiency of LSCs, Vasiliev et al. [92] conducted a field test using a greenhouse constructed with different highly transparent LSCs to compare them with traditional glazing materials and assess their energy production potential and stability. Indeed, the LSCs proved capable of reducing greenhouse running costs and maintained stable daily energy production, with minor long-term reductions in efficiency.

Following an opposite philosophy, Xu et al. [93] proposed a different approach to the incorporation of LSCs into agriculture. Their study is a numerical analysis of the use of an LSC that converts green light into red light but has an innovative structure: while the face exposed to the sun is conventionally planar, the face directed toward the plants has a micro-cone array pattern, which can point either inward or outward. The presence of these micro-cone arrays disrupts total internal reflection on that side, enabling more radiation to be directed toward the crops. In this way, it is possible to have the benefits of spectral selection while mitigating light losses caused by the concentration effect of LSCs. The external quantum efficiency (EQE) of such an LSC—defined as the ratio of photons escaping from the patterned face to the total number of incident photons on the LSC–could be as high as 37.73%, which is a 53.78% relative improvement compared to traditional LSCs. The light transmitted by such a device would also be much more diffuse. The structure of the LSC is shown in Figure 11.

Two other numerical analyses based on innovative and unconventional LSC geometries were performed by Talebzadeh et al. [94,95], who studied the possibility of elliptic and elliptic paraboloid-shaped LSCs to strongly reduce the escape losses and obtain better waveguiding efficiency. The first geometry is the dual elliptical paraboloid solar spectrum splitter (d-EPSSS), visualized in Figure 12a,b, which consists of two opposite-facing solid elliptic paraboloids that are intersected so as to share the same focal point. The upper and lower faces are coupled to a PV cell. The PV cells have a hole in the center, which acts as an entrance and exit port for a collimated beam of light. The collimated beam of sunlight enters the device and reaches the region around the focal point, which is doped with a luminescent dye. The dye absorbs radiation outside the PAR range and its fluorescence is then directed to the PV cells through total internal reflection. Given the geometry of the device, 90% of the emitted radiation reaches the PV cells with a 90° angle, minimizing reflection losses. On the other hand, PAR radiation is transmitted, traveling through the second half of the device and then exiting into a photobioreactor. The numerical analysis shows that the d-EPSSS itself could reach an OE of 73% but, as previously implied, it can only act as a spectral splitter and requires an external collimator for light concentration. Considering a typical collimator with an OE of 71%, the overall efficiency of the setup is around 52%, while a traditional planar LSC would yield an OE of 39% under similar irradiance conditions.

The second geometry understudy is the elliptical array solar spectrum splitter (EASSS), visualized in Figure 12c–e. The EASSS is composed of an array of elliptical cells, with each cell sharing one focal point with the subsequent one. The device works in tandem with a Petzval lens configuration: light goes through the lenses and is directed toward the focal points of the ellipsoids. From there, a luminescent dye absorbs non-PAR while PAR is transmitted. The emitted light is internally reflected across multiple focal points until it reaches the device’s edges, where it is absorbed by PV cells. The numerical analysis shows that the EASSS could achieve total internal reflection over a broader range of emission angles compared to a traditional planar LSC, reducing the average number of scattering and the average optical path per photon as well as decreasing scattering and reabsorption losses. An EASSS of the considered size achieves an OE of 63%, while a planar LSC of the same dimensions would have an OE of around 47.2%.

Interesting results for solar light management with LSCs can be found in studies by Zdražil et al. [96] and Zhao et al. [97], who focused on LSCs functionalized with carbon quantum dots. Zdražil et al. developed blue, green, and red large-area (64 cm²) LSCs with an external OE of 2.3% and an AVT of 83.4% for various applications. Similar results were obtained by [97] with a different approach (for quantum dots synthesis and LSCs fabrication), yielding an OE of 2.2% and a PCE of 1.13% for a 15 × 15 cm² LSC.

Table 4. Summary of research papers on AV based on LSCs. “Exp.” denotes experimental research, “Num. An.” denotes numerical analyses.

Study	Study Type	PV Device and Structure	Tilt Angle	Field Type	Coverage Ratio	Crops	Plant Results	Electrical/Optical/Thermal Results
Detweiler et al. [85]	Exp.	Lumogen Red LSCs (LSC light, LSC med, LSC dark) Roof- and side-mounted	Variable	Photobioreactor	100%	Microalgae Cyanobacteria	Chla/n. of cells, Chla/weight, and Car/Chla ratios comparable or better under LSCs than under control Growth rates similar between all cultures	The LSCs were able to power all the electronics needed for the growth and monitoring of the cultures
Pedron et al. [86]	Exp.	Lumogen Red LSC	-	Indoor and outdoor cultivations	100%	B. juncea L. albus H. annuus	+25.12%, +27.05%, and +28.17% biomass for the three crops, respectively The crops were used for phytoremediation, absorbed As/Pb per plant unit mass were the same, but higher overall biomass meant higher total contaminant accumulation	−70% PAR
Raeisossadati and Moheimani [87]	Exp.	Red PMMA fluorescent LSCs	Vertical	Photobioreactor	N/A	Microalgae	Increased biomass productivity, nitrogen assimilation, and lipidic content	Reduced photo-limitation at the depth of the cultures thanks to comb-like shape of the LSCs which guided light from the surface to the bottom part. Greater edge surface of the LSC means more light escaping and more light homogeneity
Goti et al. [88]	Exp.	DQ-Th LSCs	N/A	N/A	N/A	N/A	N/A	Synthetized DQ-Th dyes with high absorbance between 300–400 nm and 500–600 nm, 40–60% transmittance between 400–500 nm, and >80% transmittance above 600 nm, useful for AV applications OE of the LSCs increases up to 0.8% wt% dye concentration where OE is 4.5%. At higher dye concentration, OE decreases
Loik et al. [89]	Exp.	PMMA LSCs with embedded PV strips roof and vertically mounted	-	Greenhouse	100%	Tomatoes	No significant difference in fruit yield between LSC cultivation and control	>80% absorption between 300–400 nm and 400–600 nm Fluorescence spectrum in the 550–700 nm range, peak at 630 nm
Keil et al. [90]	Exp.	Bilayer CdSe/CdS and SiQD PMMA on glass LSC	N/A	N/A	N/A	N/A	N/A	Strong absorption up to 450 nm Photoluminescence peaks at 600 nm and 800 nm
Siripurapu et al. [91]	Exp.	PMMA LSCs	N/A	N/A	N/A	N/A	N/A	Soiling on the backside of an LSC with dust might slightly increase PCE as dust acts as backscattering center. Water drops on LSC surfaces reduce waveguiding efficiency by creating additional optical interfaces. Dried water residues with dust give intermediate results between dust and water
Vasiliev et al. [92]	Exp.	LSCs Roof-mounted	22.5°, 90°	Greenhouse	100%	N/A	N/A	Reduced greenhouse running costs Stable daily energy production Small long-term PCE reductions
Xu et al. [93]	Num. An.	LSC with micro-cone array	N/A	N/A	N/A	N/A	N/A	Micro-cones frustrated total internal reflection Having micro-cones on the face pointed toward crops increases the amount of light received by them
Talebzadeh et al. [94,95]	Num. An.	Elliptical paraboloid (d-EPSSS) and elliptical array (EASSS) LSCs	N/A	Photobioreactor	N/A	N/A	N/A	The d-EPSSS would allow an OE of 73%. The geometry of the d-EPSSS makes it so most of the radiation incident on the PV cell has an incidence angle of 90°, minimizing reflection losses The EASSS could achieve total internal reflection over a broader range of fluorescence emission angles An EASSS of the considered size could have 63% OE while a planar LSC of the same size would have 47.2% OE
Zdražil et al. [96]	Exp.	C-QDs LSCs	N/A	N/A	N/A	N/A	N/A	Blue, green, and red LSCs with 64 cm2 surface area 2.3% OE 83.4% AVT
Zhao et al. [97]	Exp.	C-QDs LSCs	N/A	N/A	N/A	N/A	N/A	15 × 15 cm2 LSCs 2.2% OE 1.13% PCE

summarizes the results obtained for LSC employment in AV.

4.3. Other

This section overviews studies that do not directly exploit part of the radiation received for energy generation but still provide interesting results on spectral management and separation.

Hemming et al. [98] implemented blue and red fluorescent films in greenhouses and observed their effects on the growth of the “Elsanta” strawberry. The experiments were conducted in two time periods. In the first one, the films were Reference, Blue-a, Blue-b, Red1-a, Red1-b, and Red2-a, and for the second one, they were Reference, Blue-c, Blue-b & Red1-a, Red1-c, Red2-c, and Red3. The blue films had a slightly higher PAR transmission (+0.2, +0.7%, and +1.1% for Blue-a, Blue-b, and Blue-c, respectively) with respect to the reference film. The cumulative total fruit weight was highest under the Blue-c film (+13% higher than under the reference film) and lowest under the Red3 film (−10% compared to the reference).

Enrichment of PAR via red wavelengths was performed by Xia et al. [99], who observed the growth of S. oleracea (spinach) under a Ca₄₀Sr₅₉S:Eu²⁺ (CSSE) phosphor film with controlled illumination. One film (C-foil) converts green photons into red photons, while the other (R-foil, used as the reference) is doped with MgO to reflect green photons. The spectrum under incident light ranges from 475 to 580 nm, with its peak at 550 nm. Under the R-foil, 4.98% of transmitted light falls within the 400–500 nm range, while the remaining 95.02% is in the 500–600 nm range, with no photons in the 600–700 nm interval. The total PPFD is 5.84 μmol m⁻² s⁻¹. For the C-foil, 2.02% of transmitted light is between 400–500 nm, 45.88% is between 500–600 nm, and 52.11% is between 600–700 nm, resulting in a total PPFD of 6.38 μmol m⁻² s⁻¹. Spinach seems to respond well to the green depletion and red enrichment of light operated by the C-foil, as plants grown under it showed a 25% higher CO₂ assimilation rate compared to plants grown under the R-foil.

Stallknecht et al. [100] studied seven different greenhouse glazing materials: ND91, ND58, ND33, CO770, CO700, CO550a, and CO550b. The NDxx are neutral-density acrylic covers, where “xx” indicates the overall percentage of total transmittance over the 400–900 nm range. The COxxx glazings have very high transmittance up to a 100–150 nm wide range, starting at the wavelength indicated by “xxx”. These glazings were tested on basil, petunia, and tomato plants. Basil yield was comparable between ND91, CO770, and CO700, but was lower for other glazings due to lower PAR transmittance. However, despite CO770 and CO700 having the highest transmittance after ND91, basil leaves exhibited a lighter green color, potentially reducing the aesthetic appeal of the final product. Petunia yield was similar for ND91, CO770, CO700, and CO550a. Tomatoes showed considerable decreases in yield, even under CO770 and CO700 (−25% and −37% with respect to ND91). The study highlights that herbs and floriculture such as basil and petunia might be highly compatible with PV implementation thanks to their low photosynthetic saturation point.

Proposals of alternative uses of peculiar glazing materials were put forward by the conclusions of Kittas and Baille [101], who evaluated the following cladding materials for greenhouse covers: glass (used as reference); a low-density polyethylene (LDPE) film; thermal polyethylene (TPE); an ethyl vinyl acetate (EVA) film; a three-layer film made of EVA between two layers of polyethylene (3L EVA); bubbled polyethylene (BPE) film; a violet polyvinyl chloride-based (VPVC) film; and a rose/pink polyvinyl chloride-based fluorescent (FPVC) film. The LDPE, TPE, EVA, and 3L EVA all have a transmittance of >80% from 400 to 1100 nm, while the BPE has a 60–70% transmittance across the same range. The VPVC has a transmittance of <60% from 400 to 700 nm, with a minimum of 10% around the 550 nm mark, while the FPVC has a transmittance of 20–40% between 400 and 600 nm. This translates to an overall PAR transmittance of 83–89% for glass, LDPE, TPE, EVA, and 3L EVA, 63% for BPE, 39% for VPVC, and 59% for FPVC. The authors suggest that with VPVC and FPVC, which are significantly different from traditional cover materials, morphogenetic processes in plants might be considerably altered in unexpected ways. However, this might also present an opportunity for applications in ornamental plant growth or reducing fungal infections in crops (which seem to suffer from low UV radiation).

On a similar note to [93], Shen et al. [102] developed a microphotonic thin film with a peculiar microdome pattern on the front surface and a flat back surface. This asymmetry between the front and back surface allows 89% of the trapped or emitted light to escape from the back surface and be transmitted to the plants. This increases the amount of light received by the crops compared to traditional films, where light has a lower escape probability and two possible escape directions. The film also operates a spectral shift toward red wavelengths. The film was tested on lettuce in an indoor cultivation with electrical lighting and in an outdoor greenhouse cultivation under natural sunlight. In both cases, it yielded a +20% increase in biomass with respect to control cultivations.

Table 5. Summary of research papers on AV that employ spectral separation for agricultural purposes without energy generation. “Exp.” denotes experimental research, “Num. An.” denotes numerical analyses.

Study	Study Type	PV Device and Structure	Tilt Angle	Field Type	Coverage Ratio	Crops	Plant Results	Electrical/Optical/Thermal Results
Hemming et al. [98]	Exp.	Reference, Blue-a, Blue-b, Blue-c, Red1-a, Red1-b, Red2-a, Red2-c, and Red3 fluorescent films Roof-mounted	-	Greenhouse	100%	“Elsanta” strawberry	Cumulative fruit weight: best result under Blue-c (+13%) and worst result under Red3 (−10%) with respect to Reference	Higher PAR transmission than Reference for blue films (+0.2%, +0.7%, and +1.1% for Blue-a, Blue-b, and Blue-c, respectively)
Xia et al. [99]	Exp.	Two types of CSSE phosphor film: C-foil and R-foil	-	Greenhouse	-	Spinach	CO2 assimilation rate 25% higher than R-foil under C-foil	C-foil converts green photons to red, R-foil reflects green photons away PPFD under R-foil: 5.84 μmol m−2 s−1 PPFD under C-foil: 6.38 μmol m−2 s−1 Spectral composition under R-foil: 4.98% of photons between 400–500 nm, 95.02% between 500–600 nm Spectral composition under C-foil: 2.02% between 400–500 nm, 45.88% between 500–600 nm, and 52.11% between 600–700 nm
Stallknecht et al. [100]	Exp.	ND91, ND58, ND33, CO770, CO700, CO550a, and CO550b	-	Greenhouse	-	Basil Petunia Tomatoes	Basil leaves lighter in color than ND91 for every other glazing Comparable yield between ND91, CO770, and CO700 for basil and petunia Tomato yield lower for every glazing compared to yield under ND91	ND91, CO770, and CO700 were the films with the highest PAR transmittance
Kittas and Baille [101]	Exp.	Glass, LDPE, TPE, EVA, 3L EVA, BPE, VPVC, FPVC films	N/A	Greenhouse	N/A	N/A	VPVC and FPVC might considerably alter morphogenetic processes in plants, with possible applications in ornamental crops or reducing fungal infections	Overall PAR transmittances: 83–89% for glass, LDPE, TPE, EVA, 3L EVA 63% for BPE 39% for VPVC 59% for FPVC
Shen et al. [102]	Exp.	Microphotonic thin film	N/A	Indoor cultivation, outdoor greenhouse	100%	Lettuce	+20% biomass in both cultivation with respect to control	The microdome structure of the film allows 89% of the trapped light inside to escape toward the crop. The film also redshifts incoming light

summarizes the results reported in this section.

The studies reported in this section make use of functionalized films, which are also employed in II and III generation PV devices. For this reason, the results are similar between the two sections, as functionalized films used as glazings act as spectral separators and enrichers via fluorescence.

5. Further Discussions

The comparison of the literature performed in the sections above highlights that the integration of PV technologies into agricultural systems depends heavily on the light requirements of different crops and the optical properties of the PV devices used. I generation PV technologies primarily reduce the total amount of radiation reaching the crops beneath them, making them more suitable for shade-tolerant plants that naturally thrive under lower light conditions. These crops can still achieve satisfactory growth while benefiting from the microclimatic improvements offered by PV installations, such as reduced heat stress and lower water evaporation. II and III generation PV technologies, including thin-film solar cells, OPVs, and quantum dot-based or LSC technologies, offer a more refined approach to optimizing light distribution for crop growth. Unlike first-generation PVs, these advanced technologies allow for varying degrees of transparency, enabling a better balance between electricity generation and light availability for plants. Some of these systems even feature fluorescence emission, which can selectively enhance specific wavelengths of light that are most beneficial for photosynthesis, making them particularly well-suited for crops with higher light requirements.

Among these emerging technologies, LSCs stand out as a particularly versatile and cost-effective solution due to their ability to be customized to match the specific needs of different cultivations. LSCs can be manufactured from a wide range of materials, including glass, PMMA (polymethyl methacrylate), and flexible polymers, each offering distinct optical and mechanical properties. The choice of fluorophores embedded within LSCs plays a critical role in tailoring the spectral quality of transmitted light to suit specific crops. These fluorophores absorb and re-emit light at specific wavelengths, enabling targeted spectral enrichment to optimize plant growth. Furthermore, the industrial processes for fluorophore embedment and LSC lamination are already well-developed and scalable, making it possible to produce large-area LSC devices efficiently.

Another key advantage of LSCs is their design flexibility. By adjusting the size of the LSCs, the dimensions of the PV cells, and the placement of these cells—whether at the edges or front-facing—it is possible to optimize both the PCE of the system and the shading effect, ensuring that the light conditions are ideal for a given crop. However, one of the main challenges that LSCs and other fluorophore-based PV devices face is the photostability of the fluorophores. Over time, prolonged exposure to sunlight can degrade these materials, reducing their efficiency and longevity. Addressing this issue requires ongoing research into more stable fluorophores, protective coatings, and improved encapsulation techniques.

For any kind of AV system, recent advancements in the development of PV cells provide promising outlooks in terms of increased PCE and longer lifetimes of PV devices. Any kind of PV cell with a higher PCE than its currently available counterpart implies the possibility of either maintaining current total power generation while reducing the CR or increasing total power generation while keeping the same CR. An increased lifetime of PV devices means reduced resource intensity for PV manufacturing and lower maintenance costs for AV systems. Perovskites and thin-film PV cells seem to be at the center of these advancements, with various recent studies reporting promising results obtained through the deposition or nanostructuring of passivating layers on the surface of these kinds of cells [103,104,105,106].

To ensure the successful implementation of PV-integrated agricultural systems, a multidisciplinary and transdisciplinary research approach is crucial. Collaboration among experts in agronomy, photovoltaics, material science, and environmental science is essential to identify the best crop varieties for specific AV systems. Understanding how different PV technologies interact with plant physiology, growth cycles, and yield potential will be key to optimizing these systems for large-scale adoption. Through such an integrated approach, agrivoltaics can become a sustainable solution for maximizing both food and energy production.

Economic factors can be another source of important constraints on the implementation of AV systems, especially on a large scale. Assessing the costs of AV fields is quite complex given the high number of PV technologies to choose from, different mounting structures, and factors such as elevation, site preparation, and soil protection. In general, AV fields incur higher capital expenditure (CAPEX) than conventional ground-mounted PV parks due to the additional costs associated with mounting structures. Achieving optimal synergy between agricultural activities and PV components will require specialized machinery designed to operate efficiently around and beneath these structures. Additionally, maintenance and cleaning expenses are expected to be higher, as these tasks must be carried out without disrupting crop cultivation. Expenditures can rise even further for fields where crop rotation is applied, as it might be necessary to modify the configuration of the PV modules to accommodate the different needs of the crops. A report from the U.S. Department of Energy of estimates that the CAPEX per rated power of an AV field can be as high as 2.33 US$/W (Figure 13), which is 52% more than the expenditure of 1.53 US$/W required for traditional PV parks [107]. Some operational expenditures (OPEX) may actually decrease, as the agricultural use automatically removes costs of land maintenance of PV parks. OPEX can be further reduced if land is leased at the same rates as agricultural fields [108].

Levelized cost of electricity (LCOE) assessments of AV, performed in the German context with opaque Si PV, estimate an overall cost of ~0.08 €/kWh for interspaced PV on grassland, ~0.12 €/kWh for overhead PV (elevation > 4 m) on arable farmland, and 0.11 €/kWh for PV at a 2.5 m elevation on horticultural land—all of which lower than the LCOE of a traditional domestic roof mounted PV system [108,109]. Different PV technologies can yield substantially different results that need to be evaluated case by case, weighing potential trade-offs in energy production with benefits for crop production. Economic assessments are also highly dependent on the geographic location of choice, as well as the cost of labor, materials, etc.

Additional economic analyses of greenhouse AV are highly necessary to evaluate their long-term and large-scale economic viability. Research on thermal performances of greenhouses has shown that the cost of heating a greenhouse can be significantly lower (up to almost five times lower) when using CPV compared to standard means such as electricity from the grid, diesel, or kerosene [110,111]. Previous studies [14,57,64,68,92] have shown the capability of greenhouses to be energetically self-sufficient, with surplus energy production helping to recover initial costs. So far the inclusion of high capacity energy accumulation systems [112,113] remains mostly uncharted, though it might improve self-sufficiency in regions with lower irradiance or where it has significant periodic oscillations. Storing energy for winter months or simply having the option to power supplementary lighting at nighttime, where necessary, could significantly reduce grid reliance and carbon footprint. Another innovative green solution to energy storage comes from photoelectrochemical water splitting, which is achieved by using PV devices to separate water into hydrogen and oxygen and using hydrogen as a chemical fuel to store surplus solar energy [114]. Such energy storage options increase capital expenditures but could have considerable benefits for operational costs.

It is hard to make further economic comparisons between AV and traditional PV. Standard metrics of economic evaluation of energy generation systems (like the installed cost system and LCOE mentioned in this section) are insufficient for a complete comparison, as AV fields are not simply a sum of agricultural activities and PV installations. A complete economic analysis of AV systems requires solid knowledge of the interactions between PV devices and plant growth, water usage, and other biological parameters. Without knowing how these interactions influence the quality of the specific crops considered, and thus the effects on their market value, it is impossible to make a complete economic evaluation of AV. In addition, there is also the matter of estimating the change in labor and operational expenditures for crop cultivation in AV. This estimation cannot be performed at a laboratory or prototype scale and can only be carried out following a large-scale commercial implementation of AV. Lastly, there is the matter of generalizing such an evaluation to be crop-independent—a task almost impossible given the number of plant species and all their growth need (which are not limited to light exposure and spectral composition).

In fact, greater attention should be paid to the possible positive effects of AV designs on water usage in agriculture, aligning with the WEFE nexus objectives, as results of WUE and evapotranspiration measurements, reported in the reviewed studies, indicate that there is room for improvement in water consumption for agricultural practices.

Public acceptance of AV is another key requirement for its widespread implementation. As mentioned in the introduction, AV can resolve the land use conflicts between PV and agriculture, raising the overall acceptance of PV installations. However, AV might not be able to address concerns regarding the visual disfigurement of natural and rural landscapes, potentially creating tensions among local communities or lowering the touristic appeal of certain areas. In this case, semi-transparent technologies stand out as a good possible solution, as they are less visually intrusive than traditional PV. Nevertheless, all PV implementations require further design advancements to make them blend more seamlessly with the surrounding environment. Beyond technological or design changes, public acceptance can also be increased by raising public awareness and knowledge on the topic, directly involving local communities in the planning of AV systems, and ensuring clear and transparent communication between research organizations, governmental agencies, and local populations. Some countries have already started similar projects, like APV-RESOLA in Germany [108].

Future research on AV calls not only for a common methodological and scientific framework but also for a defined government policy on AV in each country. So far, only three countries have developed policies and regulations on AV systems, namely Germany, Italy, and Japan [115,116,117]. The lack of clear land use policies and regulations is a big obstacle to the widespread commercial adoption of AV systems and could make obtaining funding for research more difficult.

6. Conclusions

This review highlights the necessity of a more unified framework to evaluate AV systems by identifying key biological metrics for crop growth and quality, as well as optical, thermal, and electrical parameters to assess the effectiveness and stability of PV devices. In this process, a major challenge is the establishment of standardized experimental conditions, which is further complicated by the diverse climate conditions across different geographical regions and the specific requirements of plant species. The absence of a common evaluation criteria and the difficulty of ensuring experimental reproducibility significantly hinder progress toward large-scale AV system implementation. A potential standardized framework could be based on PCE and biomass yield—two of the most frequently reported metrics—which serve as straightforward indicators of both energy production potential and food output. However, food production should remain the priority, as clean energy demands can be met through alternative means.

While scientific advancements in AV research have been promising, further in-depth studies are essential to fully understand the advantages and trade-offs of these systems. Future research should follow two parallel paths. First, it should expand experimental AV systems to larger scales (e.g., large-area greenhouses) to account for the numerous interactions involved—such as crop growth and quality, thermal and optical conditions, and performance across different plant species—bringing research closer to real-world commercial applications. Second, it should focus on micro-level aspects to precisely characterize and model specific variables, such as the response of plant varieties to specific light spectra or the optical performance and stability of a given PV technology. By integrating insights from both large-scale and micro-scale studies, researchers can create a feedback loop where detailed findings inform broader experiments, ultimately enabling the prediction of high-resolution data even in large-scale implementations.

Hopefully, this review has contributed to summarizing and assessing the state of the research in AV so far, and can serve as a basis for understanding the next steps toward significant commercial applications of AV. Many of the results presented in this work show that the sunlight exposure needs of crops must be carefully managed. Given these conditions, opaque PV devices, despite being more energetically efficient, lack the necessary flexibility in terms of spectral separation and light management. On the other hand, semi-transparent PV devices, and more specifically LSCs, allow for easy tuning of the spectrum and intensity of transmitted light. The supply chain for the materials that make up LSCs is already strongly consolidated, although there is no proper LSC manufacturing at an industrial level. The results presented in this review show that the advantages of LSCs and semi-transparent devices can justify the development of an industrial infrastructure for their manufacturing, similar to what has been done for traditional PV over the last 20 years.