Em Mott 2015
Em Mott 2015
Em Mott 2015
Energy &
View Journal
Environmental
Science
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: C. J. M. Emmott,
J. A. Röhr, M. Campoy-Quiles, T. Kirchartz, A. Urbina, N. Ekins-Daukes and J. Nelson, Energy Environ.
www.rsc.org/ees
Page 1 of 15 Energy & Environmental Science
View Article Online
DOI: 10.1039/C4EE03132F
Organic photovoltaics is an emerging solar power technology which embodies properties such as transparency, flexibility, and
rapid, roll to roll manufacture, opening the potential for unique niche applications. We report a detailed techno-economic analysis
of one such application, namely the photovoltaic greenhouse, and discuss whether the unique properties of the technology can
provide advantages over conventional photovoltaics. The potential for spectral selectivity through the choice of OPV materials
is evaluated for the case of a photovoltaic greenhouse. The action spectrum of typical greenhouse crops is used to determine the
impact on crop growth of blocking different spectral ranges from the crops. Transfer matrix optical modelling is used to assess
the efficiency and spectrally resolved transparency of a variety of commercially available semi-conducting polymer materials, in
addition to a non-commercial low-band-gap material with absorption outside that required for crop growth. Economic analysis
suggests there could be a huge potential for OPV greenhouses if aggressive cost targets can be met. Technical analysis shows
that semi-transparent OPV devices may struggle to perform better than opaque crystalline silicon with partial coverage, however,
OPV devices using the low-band-gap material PMDPP3T, as well as a high efficiency mid-band-gap polymer PCDTBT, can
demonstrate improved performance in comparison to opaque, flexible thin-film modules such as CIGS. These results stress the
importance of developing new, highly transparent electrode and interlayer materials, along with high efficiency active layers, if
the full potential of this application is going to be realised.
2| 1–14
Energy & Environmental Science Page 2 of 15
View Article Online
DOI: 10.1039/C4EE03132F
1.1 Scope
2 Methods
2.1 Modelling crop growth
1–14 | 3
Page 3 of 15 Energy & Environmental Science
View Article Online
DOI: 10.1039/C4EE03132F
ing the plant) will result in a 1% fall in crop production for 5x1018 1.0
3x1018 0.6
2x1018 a 0.4
bs
1x1018 0.2
bsa
Published on 04 February 2015. Downloaded by University of Connecticut on 04/02/2015 12:13:29.
0 0.0
300 400 500 600 700 800 900 1000
Wavelength [nm]
0.030
b)
0.000
300 400 500 600 700 800 900 1000
In order to assess the impact of transparent solar cells on Wavelength [nm]
plant growth, the averaged action spectrum of 27 herbaceous 0.040
c) P3HT
Absorption coeff. [nm-1]
plants (including common greenhouse crops; tomatoes, let- 0.035
PCDTBT
tuce and cucumbers) was used. 25 Figure 3a shows the photon 0.030 PTB7
flux density, bs (λ ), and the photon flux density weighted by 0.025 Si-PCPDTBT
0.020 PMDPP3T
the plant action spectrum, bs (λ )a(λ ). This modified spec-
trum can be understood as the photon flux spectrum which is 0.015
4| 1–14
Energy & Environmental Science Page 4 of 15
View Article Online
DOI: 10.1039/C4EE03132F
Fig. 4 Left: schematic of the solar cell device architecture used in this study: ITO/MoO3 /polymer:fullerene/TiO2 /ITO. For the optical
modeling, a thick layer of glass (1.1 mm) was placed on either side of the solar cell in order to account for encapsulation. Right: energy level
diagram for the modelled device. TiO2 and MoO3 are acting as hole- and electron blocking layers respectively.
900-hepta-decanyl-2,7-carbazole-alt-5,5-(40,70-di-2-thienyl- tion from interfaces between each layer. The refractive in-
20,10,30-benzothiadiazole)] (PCDTBT), poly (4,8-bis[(2- dex and complex refractive index data for the commonly
ethylhexyl)oxy]-benzo[1,2-b:4,5-b]dithiophene-2,6-diyl3- studied blends (P3HT:PCBM (1:1), PCDTBT:PC70 BM (1:4),
fluoro-2-[(2-ethylhexyl)carbonyl]thieno[3,4-b]thiophenediyl) PTB7:PC70 BM (1:1.5), Si-PCPDTBT:PC70 BM (1:1.5) and
(PTB7) and poly [(4,40-bis(2-ethylhexyl)dithieno[3,2- PMDPP3T:PCBM (1:3)) have previously been reported in the
b:20,30-d]silole)-2,6-diyl-alt-(2,1,3-benzothiadiazole)- literature 40,43–48 and the indices for the lesser studied blends
4,7-diyl] (Si-PCPDTBT). Along with the commercially (P3HT:PC70 BM (1:1), PCDTBT:PCBM (1:4), PTB7:PCBM
available materials, the low band gap polymer poly[[2,5- (1:1.5), Si-PCPDTBT:PCBM (1:1.5) and PMDPP3T:PC70 BM
bis(2-hexyldecyl- 2,3,5,6-tetrahydro-3,6-dioxopyrrolo[3,4- (1:3)) were calculated using Bruggemans model. 49,50 All
c]pyrrole-1,4-diyl]-alt- [3,3-dimethyl-2,2:5,2- terthiophene]- complex refractive index data are presented in the supplemen-
5,5-diyl] (PMDPP3T) was also studied due to its high tary information.
transparency at visible wavelengths and good device per- An internal quantum efficiency (IQE) of 100% has been
formance. These polymers are studied in blends with shown to be possible in thin (active layer of 80 nm) devices
either 6,6-phenyl C61- butyric acid methyl ester (PCBM) made from PCDTBT:PC70 BM. 33 The model therefore as-
or 6,6-phenyl C71-butyric acid methyl ester (PC70 BM). sumes that all absorbed photons will contribute a single elec-
PC70 BM is usually employed to enhance the absorption in the tron and a single hole for electrical conduction, i.e., the IQE is
spectral region not covered by the polymer, resulting in higher 100% across the spectrum. At zero bias, all absorbed photons
efficiencies but reducing transparency in the visible region will then contribute to the Jsc . It is further assumed that the
of the spectrum. Polymer:fullerene blend ratios were chosen fill factor, FF, for all the calculated solar cells is 70%. The
from ratios shown in the literature to give high efficiencies for open circuit voltage, Voc , for each device architecture is ob-
both regular and inverted device structures. 31–40 tained from J-V curves, measured under simulated solar irra-
In order to model the short circuit current density, Jsc , diance, of the studied device blends from the literature (table
we employ a transfer matrix model according to Petters- 1). 31–33,35–40
son et al. 41 and implemented using genpro1 in ASA. 42
The model calculates the electric field propagating through 2.2.1 Efficiency limits for available materials The open
a multilayer structure and hence the absorption in the fol- circuit voltage of any solar cell is determined by the energetic
lowing active layers and interlayers, along with the reflec- separation between the quasi Fermi levels for electrons and
1–14 | 5
Page 5 of 15 Energy & Environmental Science
View Article Online
DOI: 10.1039/C4EE03132F
Table 1 Polymer:fullerene blends and experimentally determined values for the Voc of the respective blends, along with energy levels of
materials presented in this study and band-gaps (determined from the absorption onset). Note that blends of a polymer with PCBM or
Material Empirical Voc [V] Blend ratio HOMO [eV] LUMO [eV] Bandgap [eV]
P3HT 51 -5.1 -3.0 1.96
PCDTBT 33,35 -5.5 -3.6 1.94
PTB7 36,39 -5.2 -3.3 1.72
Published on 04 February 2015. Downloaded by University of Connecticut on 04/02/2015 12:13:29.
holes, and this separation is ultimately limited by the band main similar to an existing benchmark. In the present context,
gap in conventional solar cells or, in the case of organic het- this approach could be used to design OPV-greenhouse sys-
erojunction solar cells, by the energetic difference between the tems with the maximum net performance for any given spec-
HOMO of the donor and the LUMO of the acceptor, some- trum required by the crop.
times called the electrical gap. For unconcentrated sunlight,
the Voc is always substantially lower than the electrical gap or 2.2.2 Choice of materials The transparency of a solar
band gap both in the case of organic and inorganic solar cells. cell device is greatly impacted by the interlayer and electrode
The exact difference between band gap and Voc varies between materials used. In our simulations, the device architecture
different materials and depends on the amount of non-radiative and contact materials were chosen for their high transparen-
recombination and on the shape of the absorption edge. 54 For cies and potential for large scale production. In this study,
the case of organic solar cells, qVoc typically lies some 300- a front contact of molybdenum trioxide (MoO3 ) coated onto
400 meV below the electrical gap as probed by electrolu- tin-doped indium oxide (ITO) of 30 nm and 70 nm respec-
minescence or photocurrent spectroscopy. 55–57 qVoc has also tively was chosen. MoO3 is a commonly used material for
been correlated to the value of HOMO(A)-LUMO(D) where inverted solar cells and has been shown to offer good hole col-
the energy levels are probed by cyclic voltammetry and pho- lection and electron blocking properties. Even though ITO
toelectron spectroscopy, 51 but this method is subject to large is a very expensive material (which requires vacuum deposi-
uncertainties in the HOMO and LUMO energies. tion, greatly increasing costs 58 ) and much effort is being put
The experimentally reported energy levels are shown in ta- into removing the need for ITO as a transparent material in
ble 1 alongside empirical Voc values. The data show the ex- organic photovoltaics and organic light emitting devices, 59 al-
pected correlation between HOMO(A)-LUMO(D) and qVoc . ternative materials which offer similar transparency and con-
Table 1 also shows that as the donor materials increase their ductivity have not been demonstrated on large area substrates.
absorption in the infrared, the resulting raise of the LUMO Ag-NWs has been proposed as a potential candidate to replace
level results in a lower Voc . This supports the assumption that ITO, as this material has shown high transparencies along with
a certain threshold between the LUMO of the donor and the low sheet resistances and costs. 58,60,61 However, relevant op-
LUMO of the acceptor is needed for efficient charge separa- tical data for sufficiently transparent electrode materials could
tion. Therefore, only adjustment of the HOMO of the donor not be found for use in this study.
polymer is possible to lower the band gap and thus limiting the For the back contact we use a hole blocking interlayer of
obtainable Voc and hence efficiency of these low band-gap ma- 30 nm TiO2 with a 70 nm electrode layer based on the opti-
terials. However, the development of novel acceptor materials, cal and electrical properties of ITO. This results in a device
with higher HOMO levels, could overcome this problem. using ITO for both electrodes which does not correspond to
The correlation between HOMO(A)-LUMO(D) and qVoc any real cell architecture. However, since only the optical ab-
can also be used to estimate the Voc that might be available sorption and transmission, as well as electrical conductivity
from a new material, assuming that non-radiative losses re- of these layers are relevant for our modelling, ITO provides
6| 1–14
Energy & Environmental Science Page 6 of 15
View Article Online
DOI: 10.1039/C4EE03132F
appropriate proxy for a generalised assumption of electrode two primary absorption peaks at 400 nm and 650 nm do, how-
gion (300 400 nm), and since plants utilise little of the light in
this region (see section 2.1), the primary implication with the 3 Results
MoO3 absorption will be the decrease in the Jsc due to less ab-
sorption in the active layer in that spectral region, rather than Figure 5a shows the crop growth factor (equation 1) as a func-
any major impact on crop growth. TiO2 is however the contact tion of the active layer thickness. The maximum achievable
material with the highest absorption in the spectrum, but since crop growth factor was found to be around 88% for an ex-
TiO2 is used as the back contact interlayer and most light be- tremely thin active layer (∼ 5 nm), given the choice of contact
tween 300 - 400 nm is already being absorbed by the front con- materials (ITO/MoO3 front contact and TiO2 /ITO back con-
tact and the fullerenes, the primary decrease in performance tact). It is seen that films made with PCBM fullerene acceptors
due to TiO2 will be due to a minor reduction in device trans- will in all cases be more transparent than those made with the
parency, although this has little impact on crop growth due to same weight ratio of PC70 BM (at a specific thickness) due to
the low wavelength of TiO2 absorption. Commonly used alter- the greater absorption of the PC70 BM in the 400 nm to 700 nm
native interlayer materials, such as PEDOT:PSS would result region.
in greater impacts on both the device transparency and effi- Figure 5 shows the greatest transparency is obtained using
ciency due to less light reaching the active layer. the low band gap PMDPP3T polymer due to the window of
From figure 3b it is seen that the choice of fullerene greatly transparency between 300 nm to 600 nm. Figure 5b shows
impacts the transparency of the device. The larger PC70 BM the short circuit current density, Jsc , as a function of the ac-
fullerene absorbs much more within the PAR region than tive layer thickness. P3HT and PCDTBT blends are seen to
PCBM. However, the larger absorption in the higher energy absorb the least photons owing to their relatively high band
region of the spectrum observed for PC70 BM could mean gap, allowing only for absorption below around 650 nm, and
that a thinner active layer is needed in order to obtain suffi- correspondingly yield low Jsc ’s. PTB7 is seen to generate
cient efficiency. The choice of fullerene expected to play a an intermediate magnitude of current, and Si-PCPDTBT and
key role particularly in the case of PCDTBT: fullerene (1:4) PMDPP3T are seen to generate the most current due to their
and PMDPP3T:fullerene (1:3) blend since the fullerenes acts low band gaps. Furthermore, it is observed that blends with
as major absorbers. For this reason, the fullerene accep- the larger PC70 BM fullerenes (dashed lines) will in all cases
tor molecules, in conjunction with carefully selected polymer generate more current for a given device thickness, due to the
donors (figure 3c), can be combined to provide an optimal ab- increased absorption of the fullerene in the visible region.
sorption profile for device efficiency along with levels of trans- Figure 6 shows the growth factor, G, as a function of the
parency required for plant growth. estimated power conversion efficiency (PCE). Despite PTB7
The absorption spectra of the five polymers shown in fig- and Si-PCPDTBT devices showing a higher Jsc due to higher
ure 3c show that P3HT absorbs primarily within the PAR re- absorption than the PCDTBT devices, the smaller Voc of PTB7
gion, suggesting devices using this material will likely greatly (0.75 V) and the very small Voc of Si-PCPDTBT (0.55 V)
impact crop growth, despite the peak in P3HT absorption co- means that the overall G as a function of PCE is quite low.
inciding with a slight dip in the action spectra (in the green, In the case of P3HT, a low current and a low Voc results in
where plants absorb less). Contrary to P3HT, PCDTBT shows low performance. The large Voc of PCDTBT:fullerene devices
two profound absorption peaks namely around 400 nm and (0.9 V) resulted in a very large G as a function of PCE even
580 nm, with a reasonable absorption between the two char- though the absorption of these blends is inferior than all other
acteristic absorption peaks of plants. PTB7 shows absorption blends. The largest overall G as a function of PCE is however
peaks around the largest peak in the action spectrum and in observed for PMDPP3T (except within a small region where
the UV, but allows for transparency in the rest of the spec- PCDTBT performs similarly). For all blends studied it is ob-
trum (350 nm - 500 nm). Si-PCPDTBT shows three profound served that the improved absorption gained when using the
absorption peaks, namely below 300 nm, at 400 nm and at more expensive PC70 BM compared to PCBM is not beneficial
650 nm. It shows a large absorption in the infra-red, allowing due to the reduction in transparency of the device. The oppor-
for high absorption without affecting the plant growth. The tunity to utilize PCBM in the most optimum devices, instead
1–14 | 7
Page 7 of 15 Energy & Environmental Science
View Article Online
DOI: 10.1039/C4EE03132F
100
6 0
[% ]
40 5 0
Published on 04 February 2015. Downloaded by University of Connecticut on 04/02/2015 12:13:29.
c o s t O P V s c e n a r io
e c o n o m ic s c e n a r io
4 0
G
20
3 0
B e s t c a s e
V e r y lo w
2 0
0
0 50 100 150 200 1 0
25
b) 0
0 2 4 6 8 1 0
20
P C E [% ]
Jsc [mA/cm2]
15
Fig. 6 Crop growth factor, G, as a function of the power conversion
efficiency, PCE, for active layers ranging between 5 nm and 200 nm.
10
Solid lines are blends with PCBM and dashed lines are blends with
PC70 BM in the ratios of P3HT:fullerene (1:1), PCDTBT:fullerene
5
(1:4), PTB7:fullerene (1:1.5), Si-PCPDTBT:fullerene (1:1.5) and
PMDPP3T:fullerene (1:3). The fill factor was set to 70% in all
0 cases, and the Voc was set according to empirical values found in the
0 50 100 150 200
literature, see table 1. The modelled devices were illuminated from
Active layer thickness [nm]
the MoO3 contact. Straight vertical lines indicate the efficiencies
required for an OPV greenhouse to be economically justified, as
Fig. 5 a) crop growth factor, G, and b) short circuit current density, discussed in the following section.
Jsc , as a function of active layer thickness calculated from the
optical modelling. The solid lines represents the polymer:PCBM
blends and the dashed lines the polymer:PC70 BM blends. Blend
ratios of P3HT:fullerene (1:1), PCDTBT:fullerene (1:4), 4 Economic Analysis
PTB7:fullerene (1:1.5), Si-PCPDTBT:fullerene (1:1.5) and
PMDPP3T (1:3) were studied. The modelled devices were
In order to assess the potential of an OPV greenhouse, an eco-
illuminated from the MoO3 contact.
nomic analysis of such a system was conducted. The eco-
nomic analysis of the photovoltaic greenhouse is based on
a multi-span structure with a planar roof angled at 23 ◦ (as
shown in figure 2), located in southern Spain. The south fac-
ing roof is covered in photovoltaic modules, with the remain-
der of the structure consisting of plastic sheeting over a steel
frame. Costs for the installation include all balance of system
of PC70 BM, could greatly reduce the cost of modules.
(inverter, labour etc. See supplementary information for de-
tails) and replacement costs over a 10 year time-frame. 62 A
Counter-intuitively, figure 6 shows that low band-gap poly- 10 year time-frame is chosen, since OPV has a relatively short
mers do not always outperform polymers which absorb lifetime, and thus the system time-frame is set at the expected
strongly in the PAR. PCDTBT:fullerene blends can demon- lifetime of major electrical components such as the inverter.
strate similar performance in a greenhouse application com- This analysis is based on the added costs of including PV in a
pared with lower band-gap polymers, due to the high effi- greenhouse and thus does not include the greenhouse structure
ciency (largely resulting from the high Voc ) of PCDTBT based itself or the plastic sheeting which would be required regard-
blends being able to compensate for low PAR transparency. In less of whether PV was present on the greenhouse.
the case of the comparison between PCDTBT and PMDPP3T Such analysis has a strong dependency on the cost of the
shown here, this is principally due to the reduced Voc which OPV module. The cost of modules is heavily dependent on the
low-band-gap materials necessitate when used in blends with cost per square metre rather than the level of efficiency, due to
PCBM or PC70BM. costs being dominated by non-active layers. 63 This analysis
8| 1–14
Energy & Environmental Science Page 8 of 15
View Article Online
DOI: 10.1039/C4EE03132F
used baseline and optimistic scenarios for the cost per square plastic sheeting covering a steel frame. Such structures require
in close agreement with reference 63). capacity of the worldwide greenhouse area. This assumes
A model was developed to compute a Net Present Value durable greenhouses become the norm worldwide and is based
(see SI 64 ) for the photovoltaic greenhouse, based on the as- on assumptions of typical insolation and greenhouse design
sumptions described above combined with an assumed eco- (i.e. roof inclination) for each region. This shows that if
nomic environment. The results of this analysis are shown all polytunnels in China could be transformed into PV green-
in table 2. For each scenario, the model calculated the mini- houses, the 415 GWp of PV capacity this could provide, would
mum efficiency of the photovoltaic module which gave a Net supply almost 15% of the national electricity demand. In
Present Value, over a ten year time-frame, of zero. This was Spain and Turkey, this figure would be around 6% of the na-
then translated to a cell efficiency by considering that 85% of tional electricity demand, contributing significantly to green-
the module constitutes functioning cells, as has been seen in house gas emission reductions.
large scale OPV modules using laser patterning techniques. 65
The results of the economic analysis (table 2) show that in
the baseline scenario, where OPV modules remain reasonably 5 Discussion
expensive, a cell efficiency of 10.2% (equating to a module
5.1 Materials for OPV Greenhouses
cost of 0.46 e/Wp ) would be required for a PV greenhouse to
be economically justified. This is far in excess of the efficien- OPV technology may represent a unique solution for PV
cies achievable with the materials studied in this work, whilst greenhouses by providing easily installable PV systems whilst
insuring limited impact to crop growth, as seen in figure 6. maintaining the productivity of agricultural land. OPV mate-
The Best Case and Very Low Cost OPV scenarios both result rials show the potential for limited impact on crop growth,
in greatly reduced device efficiencies, of 2.1% and 1.31%, for particularly in the case of polymers which show minimal ab-
an OPV greenhouse to make financial sense. Reference to fig- sorption in the PAR, such as PMDPP3T. However, we have
ure 6 shows that such efficiencies could be achieved with all also shown that optical absorption within the PAR region by
the polymers studied, including the comparatively poorly per- organic semi-conducting polymers is not a barrier to their ap-
forming P3HT, whilst ensuring crop production is not reduced plication in a PV greenhouse, as very thin devices can allow
by more than 30%. This also suggests that the active layer sufficient transparency, provided such materials yield high ef-
thickness of such devices would be less than 40 nm, in order to ficiency devices (as in the case of PCDTBT). However, signif-
reach the levels of transparency and efficiency required. The icant absorption of PAR in ancillary materials means that all
results of the Best Case and Very Low Cost OPV scenarios device structures will have some impact on crop growth, irre-
also show that OPV cost targets of 1.71 e/Wp and 0.53 e/Wp spective of the active layer absorption. The situation could be
respectively would be required, values which are both either significantly improved with the development of contacts with
already achievable or projected, at least for P3HT based mod- increased transparency, both to increase efficiency and reduce
ules. 18,62,63 the impact on crops.
Polymers with low extinction coefficients often result in
4.1 Global Scalability of PV Greenhouses poor OPV devices due to the need for very thick devices
which subsequently suffer high recombination losses due to
Plastic and glass represent the principle materials for green- poor charge transport. In contrast, the requirements of an OPV
house structures worldwide. Both of these materials can be greenhouse mean that such materials could be well suited to
used as a substrate for OPV technology, making it a potentially this application. Many materials studied in this work would
simple matter to replace such materials with photovoltaic sur- require extremely low thicknesses in order to maintain the
faces. In order for PV greenhouses to make economic sense, required level of transparency, and this may lead to difficul-
the greenhouse structure should outlast the module, i.e. more ties in manufacture. However, thicker devices which could
than 5 years for the case of OPV, or longer for inorganic tech- be more easily manufactured (i.e. with an active layer of
nologies. Currently, much of the global greenhouse capacity more than 50 nm) could be made using material combinations
is based on reasonably primitive structures such as low quality with broad absorption spectra but low extinction coefficients,
1–14 | 9
Page 9 of 15 Energy & Environmental Science
View Article Online
DOI: 10.1039/C4EE03132F
Table 2 Inputs and results of an economic analysis of an OPV greenhouse showing cell efficiencies required to achieve a Net Present Value of
zero over a ten year period
Insolation
Value of Electricity 0.159 e/kWh 1 0.095 e/kWh 2 0.095 e/kWh
Annual Increase in Electricity 0% 3% 3%
Value (above inflation)
Minimum Cell (Module) Effi- 2.1% (1.75%) 10.2% (8.63%) 1.31% (1.11%)
ciency
1 Spanish feed-in-tariff for building integrated projects; 2 Price of electricity for a commercial consumer in Spain
such as PCDTBT:PCBM, whilst maintaining the high levels PMDPP3T : PC60 BM based device ranging in thickness from
of transparency required. 1-200 nm, and covering from 0-100% of the greenhouse roof.
The more expensive PC70 BM is often used in OPV blends A typical greenhouse in Northern Europe consumes be-
to give higher efficiencies, due to the higher absorption of this tween 20 MWh/ha/year and 70 MWh/ha/year of electricity. 69
larger fullerene, compared with the lower cost PCBM. This Figure 8 shows that all PV technologies can provide sufficient
work demonstrates that the use of PC70 BM provides no ad- electricity, on an annual average basis, to meet all the require-
vantage in a PV greenhouse application as the increased effi- ments of greenhouse agriculture with minimal impact to crop
ciency is outweighed by the lower transparency in the PAR. growth. Although electricity is only 10-30% of the energy re-
In order to justify higher costs per square metre for OPV quirement of a greenhouse, 69 electricity generated on a green-
modules, higher efficiencies are required. Using lower trans- house structure also holds substantial value through exporting
parency but higher efficiency devices could be well suited to clean energy to the grid.
use on greenhouses growing speciality crops which require Due to the low efficiencies seen for semi-transparent mod-
lower light levels, such a tropical flowers, which can withstand ules, figure 8 shows that in terms of energy density, OPV pro-
around a 50% reduction in light in certain climates. 30 Alterna- vides no advantage over mature, opaque crystalline silicon PV
tively, lower band-gap materials or tandem cells with polymer technology with partial roof coverage. This demonstrates the
materials absorbing either side of the PAR could increase effi- need for high efficiency polymers and ultra-transparent ancil-
ciencies without impacting crop growth. However, low band- lary OPV materials which can provide increased efficiencies
gap materials are limited by the increasingly poor Voc which without impacting crop growth. Such high transparency OPV
results from the LUMO level of the fullerene acceptor, unless contact materials would result in the shaded green area in fig-
alternative acceptor materials with lower LUMO levels can be ure 8 moving further up the y-axis. If contact materials with
found. 100% transparency in the PAR region could be developed, the
left most point of solid green line at the bottom of the shaded
area of figure 8 would reach close to the top of the y-axis, plac-
5.2 OPV vs. Mature PV Greenhouses
ing the shaded area largely above the crystalline silicon line,
As discussed in section 1, mature PV technologies have been and showing the advantage of OPV of all opaque technolo-
used with some success within PV greenhouses. Through the gies. These results shows that some already available semi-
use of large gaps between cells, opaque crystalline or thin- conducting polymers used in OPV, such as PMPDPP3T and
film cells can create a see-through module, with the degree of PCDTBT, have the potential to perform better than some thin-
transparency being determined by the level of coverage of the film technologies, such as flexible CIGS, in this application.
cells. Figure 8 shows a comparison of such an approach with Other properties of OPV, such as low weight and flexibility
semi-transparent OPV technology, as modelled in this work, could, however, give the technology an advantage over rigid
as well as semi-transparent amorphous silicon technology. crystalline silicon modules in the ease of which it could be
This assumes opaque crystalline silicon cells with a cell ef- implemented within a plastic (rather than glass) greenhouse,
ficiency of 20% and opaque CIGS modules of 12% efficiency, which comprise the vast majority of greenhouses worldwide.
under an insolation of 2200 kWh/m2 and a performance ra- OPV technology could make retro-fitting of existing green-
tio of 80%. The green shaded area shows a semi-transparent houses (which may not be capable of supporting the weight
10 | 1–14
Energy & Environmental Science Page 10 of 15
View Article Online
DOI: 10.1039/C4EE03132F
Fig. 7 The global potential OPV greenhouses assuming 5% efficient OPV modules. Figures represent the countries with the highest reported
levels of greenhouse area in the region, namely: Mexico and the US in America 66 ; Spain, Turkey, Italy, Netherlands and France in
Europe 66,67 ; Japan and North Korea in Northern Asia 66 ; and China. 68
of crystalline silicon modules) possible, potentially using fast low efficiencies could still prove to be economically attrac-
deployment techniques as demonstrated in reference 17. Ad- tive, as seen in the Very Low Cost OPV scenario in table 2. A
ditionally OPV could be implemented through a roll-able comparison of figure 6 with table 2 shows that achieving the
screen, which would allow for OPV modules to be removed required efficiencies to make this application economically vi-
when light levels are lower, such as in the winter, thus re- able could already be achievable even with the worst perform-
ducing impacts on crop growth. This could allow the OPV ing material modelled in this work, P3HT, whilst incurring
modules to act as a NIR screen, which have been shown to re- less than a 30% reduction in crop growth .
duce energy demands of greenhouses without reducing yields,
if used for certain periods of the year 27 . If the modules were By placing an economic value on the crops grown in an
removed for 5 months over the winter months, the electricity OPV greenhouse, the optimum level of OPV transparency
yield would be reduced by around 35% (for a system located could be determined. However, since most crops grown in
in Southern Spain) as well as potentially accelerating degrada- polytunnels are fruit or vegetable crops, the nature of the pro-
tion due to mechanical stress from moving the module. Thus duce, such as the size or shape, greatly impacts its value. The
such a situation would significantly impact the economics of formation of fruits within a crop depends on the nature of the
the installation. However, thin-film technologies, such a CIGS light within the greenhouse, and the effect differs between dif-
also have similar flexible properties and perform almost as ferent crops and between seasons and climates. Therefore, any
well as the best OPV device in a greenhouse application, as such trade-off would need to take such factors into account,
seen in figure 8. which is beyond the scope of this work.
In addition, electricity provided by an OPV greenhouse may
5.3 OPV Greenhouse Economics have an increased value if it enabled climate controlled agri-
culture, such as providing water pumping (or even desalina-
Table 2 shows that when considering the most optimistic as- tion) and other automated processes in a location far from grid
sumptions, module efficiencies would need to be in the re- lines. Table 2 assumes grid connected installations, however,
gion of 2%, a value already achievable in large area mod- off-grid electricity is often far more expensive as alternatives
ules. 70 However, considering a more pessimistic economic are limited to renewable power such as wind or solar, or diesel
and technological environment, efficiencies close to the hero generators. In such a case, the economics of a PV green-
cells currently seen in the lab would be required, a challeng- house would be likely to be greatly improved despite added re-
ing prospect for the technology, particularly if there is a need quirements for battery storage systems, resulting in even lower
to maintain transparency of the module. However, if the huge OPV efficiencies being required for the application to be eco-
cost reductions which are hoped for OPV are realised, very nomically justified.
1–14 | 11
Page 11 of 15 Energy & Environmental Science
View Article Online
DOI: 10.1039/C4EE03132F
P M D P P 3 T :P C B M
8 0
0 % S e m i- tr a n s p a r e n t a - S i
O p a q u e c -S i
Published on 04 February 2015. Downloaded by University of Connecticut on 04/02/2015 12:13:29.
C ro p g ro w th fa c to r (% )
O p a q u e F le x ib le C IG S
6 0 R o
o f
C o
v e
ra g
e
4 0
1 0
0 %
2 0
0 N o rth e rn E u ro p e
0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0 In s o la tio n
S o u th e rn E u ro p e
0 1 0 0 0 2 0 0 0 3 0 0 0 In s o la tio n
E le c tr ic ity Y ie ld ( M W h /h a /y e a r )
Fig. 8 PV electricity generation against impact on crop growth for differing levels of roof coverage and for both OPV and mature PV
technologies. X-axes show electricity output assuming insolation of 2200 kWh/m2 (representative of Southern Europe) and 1300 kWh/m2
(Northern Europe), including a performance ratio of 80% for all cases and a module active area of 85% for OPV technologies. Data on
transparency and PV efficiency for OPV was based on the model described in this work and from amorphous silicon from 13 . Opaque
crystalline silicon cells are assumed to be 20% efficient and flexible CIGS modules 12% efficient. The green shaded area shows modules
using a blend of PMDPP3T:PCBM with a range of device thicknesses
12 | 1–14
Energy & Environmental Science Page 12 of 15
View Article Online
DOI: 10.1039/C4EE03132F
1–14 | 13
Page 13 of 15 Energy & Environmental Science
View Article Online
DOI: 10.1039/C4EE03132F
Physics, 1999, 86, 487–496. 68 J. Chang, X. Wu, A. Liu, Y. Wang, B. Xu, W. Yang, L. A. Meyerson,
14 | 1–14
Published on 04 February 2015. Downloaded by University of Connecticut on 04/02/2015 12:13:29.
Broader Context
for protected agriculture. There is therefore a potential match between areas already covered in plastic and
plastic photovoltaic technology, which could have a large impact in allowing further evolution in the energy
system while avoiding competition for land use in agricultural areas. Although organic photovoltaics (OPV)
have received huge interest in the academic literature, there has been limited analysis of the application of the
technology. This study looks at an innovative use for the technology, namely the photovoltaic greenhouse and
demonstrates the huge potential of this application, whilst also assessing the performance characteristics
required of OPV, and how the technology compares to more mature PV technologies.
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: