As one of the newer sectors of Controlled-Environment Agriculture (CEA), Vertical Farming (VF) is an innovative form of Indoor Farming (IF) in which plants typically are grown either hydroponically or aeroponically in vertical stacks. Light-emitting diodes (LEDs) are considered the best sole source of crop lighting in a warehouse setting.

Vertical farming is proliferating as demand for local, fresh, year-round produce rises, especially in seasonal, urban areas. Based on recent Indoor Farming market analysis, VF has a projected 25.7% compound annual growth rate from 2020 to 2027, which compares favorably with other CEA sectors in the United States (Grand View Research, 2020). Other benefits of IA/VF systems include not contributing to scarcity of arable farm land or freshwater resources (Benke and Tomkins, 2017). The risk of crop loss due to extreme weather conditions is avoided, and fossil-fuel use for transporting produce to urban markets is minimized. On the other hand, there are economic concerns regarding the VF industry. This emerging industry is typically considered expensive entrepreneurship due to high Capital Expenses (CAPEX) and significant Operational Expenses (OPEX). Two of the most important CAPEX components are expensive land in urban areas and the high initial cost of installing LED lighting systems (Qiu et al., 2020).

Electric lighting is one of the most expensive OPEX costs of vertical farming. Even though LEDs are more energy efficient than other electric lighting sources, sole-source electric lighting is still a major input cost, and seeking ways to save energy for lighting is a significant issue for commercial growers. Electricity typically accounts for 25-30% of total OPEX (Kozai, 2013; Kong et al., 2019; Kozai and Niu, 2020). Although the VF industry is expanding worldwide thanks to enthusiastic financial investors, high OPEX is keeping profitability potential of this nascent industry fragile and often elusive. As of 2019, based on a CEA Census report, only 37% of VFs were reported to be profitable (Global CEA Census Report, 2019). Although 58% of VFs reported to experience financial improvement based on the Global CEA Census Report (2021), indoor farms began ceasing operations in 2022 and 2023.

A cost-efficient lighting system is one of the most critical determinants of indoor-farming profitability (Davis & Burns, 2016; Cocetta et al., 2017). An efficient lighting system is a combination of both efficient fixtures as well as effective canopy photon capture (Nelson and Bugbee, 2014).

Over the past decade, notable success has been achieved improving LED electrical efficiency and photon efficacy (Mitchell and Sheibani, 2020). It has been estimated that these parameters are approaching their maximum theoretical values (Kusuma et al., 2020). Electrical efficiency is defined as the ratio of optical power output to electrical power input (WW^-1). As of 2020, based on Kusuma et al. (2020), blue LEDs had a maximum theoretical efficiency of 0.93, followed by red (0.81), far-red (0.77), phosphor-converted cool-white (0.76), phosphor-converted warm-white (0.69), and green (0.42). On the other hand, photon efficacy is a metric to characterize the suitability of LED lighting systems for horticultural purposes. Photosynthetic efficacy is the ratio of micromoles of photosynthetic photons (400-700 nm) generated to the electrical energy applied (µmole J^-1). In 2020, far-red LEDs had the highest photon efficacy (4.7), followed by red (4.5), blue (3.5), cool-white, warm-white, and green with photon efficacies of 2.9, 2.6, and 1.9 (µmole J^-1), respectively (Kusuma et al., 2020).

Other advantages that make LEDs the choice for sole-source lighting include a significantly longer operational lifetime: They do not typically burn out. Instead, performance weakens gradually, and time of decline to 70% of its original intensity typically is 50,000 hours for horticultural LEDs (Kusuma et al., 2020; Paucek et al., 2020), at which time they are candidates for change.

Compared with other electric-lighting sources that have been used for plant growth, LEDs are not bulky, enabling flexibility of fixture design and straightforward installment and replacement. LEDs are solid-state devices that integrate easily with digital-control systems.

Studies of LED indoor illumination patterns have emphasized uniformity of intensity for human perception (Moreno & Tzonchev, 2004; Moreno et al., 2007; Yang et al., 2008; Sun & Lee, 2022). A generalized Lambertian function, estimated by Gaussian distribution, characterizes the typical geometric illumination pattern of LEDs (Yang et al., 2008; Sun and Lee, 2022). Although the illumination pattern of a Lambertian distribution is 180°, 87% of that pattern occurs in a half-width of ± 60°C and a view angle of 120 ° ( Figure 1 ). For indoor crop lighting, an LED lighting source can be a “bulb-type” emitter or a planar/filamentous array positioned horizontally (overhead) or vertically (intra-canopy) (Tsao et al., 2015; Cocetta et al., 2017). As a recently-adopted technology for indoor crop lighting, available LED choices also follow that general Lambertian distribution (Stutte, 2009; Nicole et al., 2016; Lee et al., 2023). The relative intensity of photon emissions taken from a data sheet for the LED arrays used for the present study was plotted by the authors from a data sheet for the Lumi-LED LEDs used by the Orbital Technologies Corporation (ORBITEC, now Sierra Space), to construct the Biomass-Production-System-for-Education (BPSE) adjustable-height LED arrays and was found to be Lambertian in nature ( Figure 1 ). If intensity at 0⁰ angle from normal is 160 μmol/m²/s (as per this study), then 80 μmol/m²/s would be measured by a cosine-corrected light sensor at ± 60⁰ (a cone) with further decreasing relative intensities extending to 90⁰ (a hemisphere) in all directions.

Although “beam spread” or “view angle” of individual LED packages is available in manufacturer catalogs and specification sheets, there is not enough working knowledge about how to best use the irradiation pattern of Lambertian LEDs as the sole-source for plant lighting, and the pattern of irradiation when multiple packages are installed in either bulb or filament type is neither easy to estimate nor is widely reported. Inspection of several online catalogues offering LEDs for horticultural applications advertised view angles of 120⁰ to 130⁰.

In a vertical-farm setting, various lighting-installment plans have been used by growers. Depending on growth-compartment layout, tube-type LED modules may run along or across the growing area. In each case, an adequate number of LEDs must be installed to ensure uniformity of light intensity up to the edges of benches, thereby avoiding low-light-intensity edge effects and crop-growth gradients across benches.

Different factors affect irradiation pattern for crop-growth applications, including density of LED packages/modules on fixture undersurfaces, and separation distance between the LED lighting system and crop surface. In theory, LED packages are basically hemispherical radiators of light energy (radiating at 180°) after mounting on a flat surface for overhead-lighting applications, although most photons radiate within the 120° Lambertian beam spread. In that context, some photons radiate directly downward toward cropping surfaces, but also outward, beyond cropping surfaces at all angles up to maximum beam spread, leading to considerable loss of photon emissions beyond bench dimensions, especially when integrated over entire photoperiods and cropping cycles. Primary and secondary optics installation further shape beam angle, with accompanying energy-absorbance loss for every photon reflected off of a solid surface. A primary optic can be a protective lens made with epoxy or silicone resin to enhance the lifetime of LEDs by protecting them from moisture and dust (Massa et al., 2006; Alim et al., 2021). Moreover, an LED package also can be enclosed and protected by a “pit primary optic”. The pit might be covered with an encapsulation, into which a phosphor material is incorporated to create white radiation from blue LEDs (Personal Communication, Elio Jin-Ha Kim, PhD, Samsung Corporation, 2022). The wall-coating material within the pit typically is titanium dioxide (TiO2) because of its high reflectivity index (Silva et al., 2022).

Beam intensity decreases sharply beyond 120°-130° out to 180° in current LED module design for “bulb-type” emitters after installing a primary optic (OSRAM, 2021; Samsung, 2021). Secondary-optic lenses, such as a cylinder around each LED module, die, or engine, typically lined with reflective materials, sharply attenuate beam spread and are used to enhance spectral uniformity and intensity, as LEDs are brighter in the center of an attenuated beam (Kuo et al., 2011). Depending on secondary-optic lens design, beam spread may be cut to half the original viewing angle (Zhenrong et al., 2009). The energy cost of attenuating beam spread with secondary optics is reduced electrical efficiency and photon efficacy due to absorbance losses whenever photons reflect off of solid surfaces.

Among other LED advantages, low radiant heat production at photon-emitting surfaces allows LED fixtures to be placed close to crop canopies (Bourget, 2008; Massa et al., 2008; Singh et al., 2015; Davis & Burns, 2016). Another advantage is moderate dimmability without significant power loss (Pattison et al., 2016).

Present indoor-crop-lighting practices do not take full advantage of vastly improved LED electrical efficiency and photon efficacy because a significant portion of fixture light beams fall outside cropping areas, especially from LEDs mounted above or beyond bench edges, thereby negating many energy-use improvements that have been made in recent years.

By leveraging the low radiant heat and dimmability properties of LEDs, we postulated that reducing the vertical separation distance between lighting source and crop surface would enhance energy-use efficiency without heat-scorch damage to the crop. The “coolness” factor of LEDs positioned closer to plant tissues allows the electrical power needed to deliver a given light intensity to be tuned down compared to what would be required for a hot light source to achieve the same light intensity at a considerably larger separation distance. At closer separation distance, the chance of obliquely emitted photons escaping the growth compartment is reduced ( Figure 2 ). At the standard separation distances used in vertical farms, photon loss is considerable as a result of wide-angled “beam spread” falling beyond growth-compartment boundaries. The current-driven LED property allows lower energy expenditure by dimmed LEDs still achieving the same photon intensities at closer separation. The minimum separation distance that can be used will be determined by LED spacing and resulting spectral-composition uniformity (Mitchell, 2015).

For the present project, we postulated that by taking advantage of unique physical properties of LEDs, crop-canopy photon-capture efficiency (CCPCE), equivalent to the light-engineering term “utilance”, can be enhanced, thereby improving much-needed energy-use efficiency of the indoor-farming sector. In this study, we characterized two close-canopy-lighting (CCL) scenarios: energy-efficiency and yield-enhancement. The laboratory conducting this CCL proof-of-concept study has a long history of stepwise, related investigations. The productivity/energy-saving advantages of allowing plants to grow up, around lights within a crop foliar canopy was initially demonstrated using low-wattage fluorescent kitchen lamps arrayed 3-dimensionally in growth compartments, providing “intra-canopy lighting” (ICL) to vertically growing plants (Frantz et al., 1998, Frantz et al, 2000; Frantz and Mitchell, 1997). Relative coolness of LEDs at light-emitting surfaces and moderate dimmability without major power loss made LEDs the undeniable choice for sole-source ICL (Massa et al., 2006; Massa et al., 2008), as well as supplemental ICL of high-wire tomatoes growing in greenhouses (Gomez et al., 2013; Gómez & Mitchell, 2016). Previous work from the PI’s lab also demonstrated the substantial energy savings of targeted vs. full-coverage overhead lighting of leaf lettuce (Poulet et al., 2014). The present study is a next-step systematic characterization of the productivity and/or energy-saving effects of close-canopy overhead lighting on densely seeded young lettuce plants with rapid canopy closure.

The standard separation distance between the photon-emitting surface of LEDs and a standing crop in a vertical farm setting is 40-50 cm (Kubota, 2020). Considering 45 cm as a standard control separation distance, three closer separation distances of 35, 25, and 15 cm were tested and compared, using a typical 120° beam-spread angle of bulb-type Lumi-led LED modules.

On day 15, as the crop foliar canopy closed ( Figure 3 ), end-point growth parameters including total shoot fresh and dry biomass were collected on an experimental unit basis of trays. kWh of energy expenditure for each LED lighting system over 15 days were recorded. Energy Utilization Efficiency (EUE) was calculated as grams fresh or dry biomass produced per kWh of electrical energy expended on the LED lighting system for all tested separation distances.

LEDs currently available for indoor farming have much-improved electrical efficiency and photon efficacy compared to other plant-growth-lighting sources. The combination of significantly higher efficacy, extended lifetime, and relative lack of thermal radiation from photon-emitting surfaces make LED lighting systems the only real choice for sole-source indoor lighting.

However, LED physical capabilities can be further leveraged to enhance crop canopy photon capture efficiency (CCPCE) or utilance, efficiency metrics of the fraction of emitted photons incident upon photosynthetic surfaces of plants (Balasus et al., 2021). If CCPCE also is high, overall EUE of an indoor crop-lighting system can be further enhanced. Otherwise, LED efficiency and efficacy improvements are minimized by significant wastage of photons and energy.

In the present small-scale proof-of-concept study, we demonstrate enhanced CCPEC and EUE by implementing CCL strategies. Significant energy is saved as LEDs are dimmed at closer separation distances (energy-efficiency strategy), or significantly higher biomass is produced while the same energy is expended (yield-enhancement strategy). In support of our hypothesis, biomass produced at different separation distances was not significantly different in the energy-efficiency CCL strategy, although the closest separation distance tended to have higher biomass followed by the other three separation distances in order. LED beam-spread distribution was Lambertian with the highest light intensity at the center of a 120° beam. Although beam spread was fixed at various separation distances between emitting and absorbing surfaces, closer separation distances benefitted from the brighter center, which tended to have slightly higher fresh and dry biomass.

The present study highlights the need for deeper knowledge of the beam-spread pattern of LED lighting sources prior to making lighting maps for IF/VF. Better understanding in this particular area would enhance energy-use efficiency by saving significant energy expended on LED lighting systems. Upscaling of this study’s findings would reveal design challenges that need to be addressed.

Since LEDs as presently constructed typically follow Lambertian photon distribution, beam spread is wide, causing measured light intensity to decline with larger incremental distance between lighting source and illuminated surface due to photons that escape sensors (Wang et al., 2019). Thus, separation distance between photon-emitting surface area and photon-absorbing surface area matters.

At a standard vertical-farm separation distance of 40-50 cm between photon-emitting surface and light-absorbing crop surface, chance of escape is higher for photons emitted obliquely outward from LEDs positioned above or near the edges of cropping benches, but some escape is likely, even from LEDs positioned near the center of benches ( Figure 2 ).

“Current droop” is a mechanism by which a significant loss of efficiency occurs from GaN-based LEDs (Piprek, 2010; Oh et al., 2019). As the forward current of LEDs increases, light output increases, but with a significant decrease of output efficiency. The current-droop phenomenon occurs as a result of an internal quantum-well-loss process, regardless of heat generation at the emitting surface (Laubsch et al., 2009). At ambient temperature, external quantum efficiency drops significantly at higher currents as a result of current droop (Meyaard et al., 2012). LED packages must operate at higher current to achieve desired intensity at farther separation distances, which in turn lowers efficiency. Therefore, reducing separation distance will improve photon absorption, as lower forward current can be applied to achieve the same PPFD. LED efficacy at reduced current density also can be improved by increasing the size of an LED chip (Kusuma et al., 2020). However, increasing chip size to reduce current droop specifically works for high-power LED devices (Meyaard et al., 2012), whereas most commercially available LEDs used for IF/VF are mid-power.

LED fixture design will be one important factor determining the success of close-canopy-lighting applications. Number of engines and their mounting patterns on a fixture will affect uniformity of both photon intensity and spectrum. Uniformity of intensity is impacted when cropping areas are affected by shade bands from other electrical components. A shade-band effect might not be pronounced under standard separation distances of 45-55 cm; however, it becomes more problematic at closer separation distances. A crop growing under shade-band areas would react by shade-avoidance syndrome, growing away from shade (Elkins & van Iersel, 2020), and marketability and quality are compromised. A reasonable blend of wavelengths is crucial to avoid specific photomorphogenic effects of individual wavebands on plant morphology.

At the beginning of the current study, aluminum tables with support frames for adjustable-height overhead LED lighting fixtures were constructed within a walk-in controlled-environment chamber. A commercial partner provided multi-waveband, dimmable LED fixtures that we mounted at various heights above the tables containing plant trays. Following preliminary CCL testing, difficulty was noted achieving expected results, including uniform plant growth and leaf pigmentation within trays at closer separation distances. Spectroradiometer readings indicated spacing of individual LEDs to be too wide apart to provide adequate beam overlap among individual LEDs at closer separation distances. However, already present in our laboratory from a previous project were LED arrays constructed to provide ground-based control experiments for the Veggie plant-growth unit being used to grow leafy greens for astronauts on the International Space Station, which already were mounted on height-adjustable scissors jacks. Light engines were arrayed densely and uniformly on the underside of those BPSE fixtures, the LEDs were dimmable by waveband, and spectroradiometer scans indicated uniform spectra at all lamp/crop separation distances between 45 and 15 cm, meaning that light-engine density and LED distribution were sufficient to provide adequate proof-of-concept testing for the scenarios of our CCL study. Not all LED arrays currently in use at standard 40- to 50-cm separation distances between cropping surfaces and LED-lighting fixtures in the indoor-farming industry may work for the energy-saving or yield-enhancing advantages of CCL. Lighting-system designers will have to weigh those advantages of CCL against costs of re-designing lamps, supports, and benches for productivity, profitability, and sustainability.

Secondary-optics installation is typically considered to improve light-intensity uniformity as the light distribution from an LED package without secondary-optic lenses falls under Lambertian type and lacks uniformity of intensity (Jiang et al., 2010; Nian et al., 2019). It also truncates beam spread and photon losses beyond the edges of growth areas would be less (Zhenrong et al., 2009). However, a higher number of engines need to be installed to reach the desired uniformity of intensity, and reflectance losses from secondary optics further add to lower efficiency and efficacy. Therefore, secondary-lens installation is not considered an efficient strategy compared with CCL.

Among different overhead LED fixtures, planar tube types are most commonly used in vertical farm settings (71%), followed by panel LEDs (21%), rounded LEDs (6%), and other types (2%) (Paucek et al., 2020). Regardless of fixture type, the LED luminaire/lighting width is either the same as or beyond bench width ( Figure 2 ). The longer width of LED lighting structures that are common in current VF settings ensures light uniformity to bench edges and avoids gradients of crop growth, which otherwise would be drawbacks for automated harvest and marketability. Photon loss from cropping edges is inevitable as installing fewer tubes across benches is not the solution to the light-uniformity issue. In that setting, if saving energy is desired, CCL is the more effective strategy as the chance of photons escaping from growth compartments can be reduced.

In general, wider benches are not a grower’s desired solution as traditional lateral air flow and CO₂ enrichment occur less efficiently at the center of benches.

As a supplement to CCL, deploying reflective curtains around growth compartments further improved efficiency of light utilization. Even at the closest CCL separation distances, some photons still escape the growth area and are wasted. Our findings suggest that the outcome of this approach is highly dependent upon light intensity, since reflective curtains were more effective for yield-enhancement scenario. Effectiveness of curtains was evaluated at the second-most efficient CCL separation distance of 25 cm to promote adequate air circulation. However, at higher separation distances, the chance of photon incidence on crop surfaces is higher with curtains, which might result in a different outcome.

Although a combination of CCL and reflective curtains should lead to the most robust energy-efficient production system, CCL will cause overhead space limitation for horizontal air movement across benches. Traditional air movement and thermal control, both of which are essential in a VF setting, will be disrupted, especially if curtains also enclose growth compartments. A considerable water-vapor boundary layer can accumulate above crop stands in stagnant air as a result of crop transpiration. Such humidity must be removed to enhance nutrient absorption and allow evaporative cooling. CO₂ refreshment allowing continued photosynthesis is another factor requiring adequate air turnover above crop surfaces.

Although we did not encounter air-movement issues in a small-scale experimental growth-chamber setting, innovative engineering solutions will be needed to negate possible negative effects of combined CCL and reflective curtain installment in future vertical-farm lighting configurations. For this CCL proof-of-concept study, we did not develop a specific LED lighting system, but leveraged the use of height-adjustable LED lighting systems that were available in our laboratory. The height-adjustable property and reasonable uniformity of intensity and spectra enabled us to demonstrate proof of concept for two CCL strategies. We did not investigate the optimal spectral composition for indoor lettuce production; instead, a general lighting recipe mimicking white light was used.

The lawn-of-baby-greens production system developed for this study resulted in higher EUE as the planting distance between small seedlings was minimized and foliar-canopy closure occurred within a 15-day cropping cycle. Tip burn or other physiological disorders that might result from restricted air movement are non-issues for baby-stage lettuce. When CCL is tested for more mature stages of leafy greens, appropriate air movement will be become more essential and the need for engineering solutions necessary.

The EUE and yield-enhancement findings of this CCL proof-of-concept study can be of near-term benefit to indoor production of leafy greens. In addition to the need for re-design of light fixtures for appropriate light-engine mounting patterns and densities on the underside of LED light fixtures, manufacturers of vertical mounting racks need to re-design structural supports to address airflow constraints imposed by CCL. Current standard separation distances between the underside of light fixtures and the top of crops (40 – 50 cm) is adequate to support unconstrained horizontal movement of fresh air across benches needed for thermal and humidity control, as well as CO₂ refreshment needed for rapidly photosynthesizing crop biomass. However, close separation distances (15 – 25 cm) may constrain horizontal air movement across benches, requiring innovative re-design involving controlled-velocity vertical air movement around lamps and through tray-support infrastructure (Zhang and Kacira, 2022). In fact, vertical air movement has been found to raise vapor-pressure deficit (reduce humidity) and prevent tip-burn disorder in lettuce (Ertle, 2023), allowing multiple advantages of implementing CCL. The authors encourage consideration of adjustable-height light supports above stationary crop supports at each tier within vertical cropping areas for flexibility of access and to accommodate different CCL needs of different crops or at different stages of crop production. Such advancements will help improve crop productivity and EUE, reduce OPEX for lighting, and improve profitability potential of the indoor-agriculture vertical-farming industry.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

FS: first author, major contribution. MB and RM: middle authors provided supporting services, LED hardware and interpretation of results. CM: last authorship, lab director. All authors contributed to the article and approved thesubmitted version.