Volume 1 photovoltaic solar energy 1 23 – application of micro and nanotechnology in photovoltaics

Volume 1 photovoltaic solar energy 1 23 – application of micro and nanotechnology in photovoltaics Volume 1 photovoltaic solar energy 1 23 – application of micro and nanotechnology in photovoltaics Volume 1 photovoltaic solar energy 1 23 – application of micro and nanotechnology in photovoltaics Volume 1 photovoltaic solar energy 1 23 – application of micro and nanotechnology in photovoltaics Volume 1 photovoltaic solar energy 1 23 – application of micro and nanotechnology in photovoltaics

L Tsakalakos, General Electric – Global Research Center, New York, NY, USA

© 2012 Elsevier Ltd All rights reserved

1.23.2 Application of Micro and Nanotechnologies to Conventional PV

1.23.2.1.3 Antireflective layers

1.23.2.2.1 Thin films by nanoparticles

1.23.2.2.2 Lift-off of thin films

References

1.23.1 Introduction

Solar photovoltaics (PVs) is widely considered to be one of the leading class of renewable energy technologies as the mix of energy-producing technologies evolves from primarily carbon-based sources to a more balanced distribution [1] As has been noted

in prior chapters, the PV market has grown tremendously in the last decade with steady expansion expected in the coming decade Key terrestrial markets are utility-scale power plants (flat-plate or concentrator configurations), commercial rooftops, and residential rooftops (on-grid or off-grid) PV modules typically comprise a semiconducting PV device that is packaged in a construction typically consisting of one or two glass sheets and/or a polymer backsheet, with encapsulation used to protect the solar cells Electrical cables are used to connect multiple panels together to form large area strings that ultimately connect to inverters and the load of interest The majority of PV modules today (ca 80–85%) are based on silicon technology These have the benefit of a well-established Si technological base, materials abundance, and the ability to produce in large-scale processes The typical power conversion efficiency (PCE or η) of Si PV modules is 14–16%, with some higher-cost technologies achieving 18–19% The costs of Si PV technology are relatively high (ca US$1.10–2 (Wp)−1) compared with other renewable energy technologies as well as other PV technologies, yet have been reduced in the last few years due to economic and supply-side factors The second largest technology segment in the PV industry today is based on thin films [2] These are primarily thin-film (1–5 μm), CdTe-based modules that are monolithically deposited and integrated on low-cost glass substrates This allows for achievement of the lowest-cost PV technology available on the market at under US$0.8 (Wp)−1 Another emerging thin-film PV material system is the Cu(In,Ga)(Se,S)2 (CIGS) class of absorber materials Record CdTe cells have achieved a PCE of 17.1% with production modules at the ∼11–12% level, whereas CIGS record cells have achieved 20.3% with modules at the 12–14% level [3]

There are three primary metrics that are of importance to PVs: (1) cost, (2) PCE, and (3) stability The first and third are somewhat beyond the scope of this chapter, yet very critical to deployment of PVs Within PCE, there are second-level performance parameters that are considered, namely, short-circuit current density (JSC), open-circuit voltage (VOC), fill factor (FF), series resistance (Rs), and shunt resistance (Rsh) The PCE is defined as:

JSC  VOC  FF

in

Sub-band gap loss Thermalization loss

0.00E+00

2.00E-01

4.00E-01

6.00E-01

8.00E-01

1.00E+00

1.20E+00

1.40E+00

1.60E+00

2 nm

1 )

1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05 1.00E+06 1.00E+07 1.00E+08 1.00E+09

1 )

Energy (eV)

C = 1

Nph 〈hν〉

40

EFF = 100

30

20

10

0

Eg (eV)

3

where Pin is the input solar power, which is 1000 W m−2 for the standard terrestrial AM1.5 solar spectrum [4] The key challenge with

PV cells is to simultaneously increase JSC and VOC, while minimizing extrinsic losses such as shunts and energy barriers that reduce FF

In addition to extrinsic loss mechanisms such as grain boundary and contact recombination, and parasitic absorption in front contact/electrode layers, there are two fundamental loss mechanisms that limit the performance of solar cells [5] The first is thermalization of high-energy photons that are excited by photons of energy higher than the band gap (Figure 1) These photons lose energy by emission of photons (heat) and hence lead to a loss of energy The second fundamental loss mechanism is loss of photons below the energy band gap of the semiconductor due to lack of absorption This can account for a significant fraction of energy loss (ca 30–40%) depending on the band gap of the solar absorber materials used These two basic loss mechanisms lead to

a detailed balance limit (assuming unity quantum efficiency (QE)) for a single band gap solar cell of ∼33%, as was first calculated by Shockley and Queisser [6] This leads to an optimum band gap for the AM1.5 terrestrial solar spectrum of ∼1.45 eV (Figure 2)

Figure 1 Fundamental loss mechanisms in silicon solar cells at low- and high-photon energies as related to the AM1.5 solar spectrum

Figure 2 Detailed balance limit efficiency as a function of band gap showing that the optimum for the AM1.5 spectrum is ∼1.45 eV at 1 sun leading to

∼31% PCE, and ∼37% PCE at 1000 suns [7]

US$0.10 W−1 US$0.20 W−1 US$0.50 W−1

100

Thermodynamic limit

80

US$1.00 W Present limit

40

20

‘Gen 2’ – thin films Cost (US$ m−2)

‘Gen 1’ – Crystalline Si

To date, the record for a single junction has been achieved by direct band gap GaAs solar cells at ∼27.6% [8], whereas indirect band gap Si solar cells have achieved a record laboratory cell efficiency of 25% [3]

This begs the question of what PCE is required in a solar PV module technology Before providing a possible answer to this question, note that this specifically refers to PV module efficiency and not solar cell efficiency Note that in most technologies, the module efficiency is 2–3% lower in absolute terms than the solar cells from which the module is constructed, though some module technologies are 5–10% lower in absolute efficiency than the corresponding record laboratory solar cell efficiency reported

in the literature Within this context, the answer to the initial question is dependent on the particular application, but there is a minimum value of ∼10% that has emerged in recent years If the application is space-constrained (e.g., residential rooftops or extraterrestrial applications), then it may be beneficial to pay more for a higher-efficiency technology, such as high-performance silicon in the 16–19% module efficiency range or for space application the use of III–V technologies that are orders of magnitude more expensive For applications with a lesser constraint of space, it is of interest to achieve as low a levelized cost of electricity (LCOE); hence, a combination of efficiency and module cost will be required to achieve the desired goals, which are typically in the range of US$0.15 to US$0.25 kWh−1 depending on location (solar insolation) Of course, lower LCOE is required to compete with conventional power sources (ca US$0.05 kWh−1 –US$0.15 kWh−1) It is in such utility-scale and large-scale commercial rooftop applications that medium efficiency (ca 10–16%), yet, low-cost module technologies such as CdTe (and CIGS) thin-films or lower-cost polycrystalline silicon module technologies are more important

It must be noted that cost is a significant driver in the solar industry Indeed, efficiency is a parameter that impacts costs at the module and system level Costs that must be considered include the capital expenditure (CAPEX), the overall module bill­ of-materials (BOM) cost (US$ m−2), the module manufacturing cost (US$ (Wp)−1), total installed system cost (US$ (Wp)−1), and ultimately the LCOE discussed above Indeed, one cannot consider cost without considering efficiency and vice versa Hence, as has been pointed out by Green [5], there is a strong need to produce module technologies that are both low cost, that is, with cost structure similar to thin films, yet with efficiency that is truly breakthrough (ca 20–40%), the so-called Third Generation or Generation III PV technology (Figure 3) Such PV technologies simply do not exist today and would truly revolutionize the PV industry, leading to wide-scale adoption of PV technologies

While efficiency and cost are widely discussed, another significant parameter for PV module technologies is the stability of the module performance A related parameter is reliability, which is related to complete failure of the module by a number of mechanisms; however, stability is related to the rate of degradation of the module performance with time Stability can in fact be

a critical parameter in determining the success of a PV technology If a high-efficiency module degrades in performance very quickly, then the value of this technology is lesser since it may become difficult to predict the power output (and hence financial payback) of the system as a whole The scientific mechanisms that impact stability are indeed often not well understood for many technologies While this topic will not be discussed in detail, it should be a factor that must be considered in more detail by the research community as new solar cell technologies, such as those based on micro- and nanotechnology, are developed

1.23.2 Application of Micro and Nanotechnologies to Conventional PV

Within the framework of the solar efficiency–cost landscape, as shown in Figure 3, there is a strong need to develop Generation III

PV technologies with breakthrough efficiency and low cost However, there are opportunities to apply new technologies such as

Figure 3 Solar cell landscape as defined by M Green showing the desire to develop Generation III technologies with high efficiency and low costs, as compared to Generation I (silicon) and Generation II (thin films) [5] Reproduced with permission from Green MA (2003) Third Generation Photovoltaics: Advanced Solar Energy Conversion Berlin, Germany: Springer Science+Business Media, Figure 1.2

nanotechnology to conventional solar technologies in the Generation I and Generation II categories Silicon continues to be the dominant solar material in the market (ca 80–85% in 2011) and will not be fully replaced in the near future, whereas CdTe thin films make up approximately 15–20% of the 2011 market with the remainder based on emerging CIGS modules and amorphous silicon (a-Si) thin films for certain applications Micro- and nanotechnologies can be applied for several purposes, the main reasons being to increase the PCE of the technology or to improve the processing of conventional PV modules for reduced costs, yield, and related manufacturing targets Several examples of such approaches will be described below

1.23.2.1 Improved Performance

We first discuss various approaches to improve the performance of conventional solar cells The general strategy here is to improve the performance by overcoming various intrinsic or extrinsic losses in the device In the former case, this may include undesired reflection losses, parasitic absorption in various device layers/films, or recombination losses at grain boundaries, contacts, or in the bulk of the absorber volume In the latter case, this include mechanisms to reduce thermalization loss of carriers excited by high-energy photons, or to allow absorption of portions of the solar spectrum that are below the band gap of the absorbing semiconductor (we implicitly assume conventional conversion mechanisms are operable for Generation I and II devices) Since we are dealing with established solar module technologies, the obvious question is how much of a gain is relevant, followed by what type of gain we can expect In other words, what solar cell parameters can be impacted by various enhancement mechanisms? With regard to efficiency gains, it is noted that absolute efficiency of ∼0.25% or greater can have significant cost and market impact for a particular technology Depending on the starting efficiency of the technology, this could correspond to a gain of approximately 1%, 2%, 3%, or even higher in relative terms Of course, the gain mechanism typically comes with some added cost that should be less than the relative gain Each technology should be evaluated on its own terms to determine whether the gain is worth the cost of introducing the gain mechanism Interestingly, most of the gain mechanisms to be discussed below lead to a gain

in the JSC of the PV cell/module This is important to understand because for some technologies it is well known that the main challenge is in fact not the JSC but rather the VOC (or in some cases FF) This is indeed well known in the case of CdTe thin-film technologies [9] We stress here that the performance enhancement mechanisms we are referring to are based on external, typically optically active, films that by their nature would not affect VOC Parameters that affect VOC are generally directly related to the inherent device structure and, hence, would require detailed device engineering, in addition to efforts at defect passivation We begin the discussion with efforts to improve intrinsic losses by shifting photons either up or down in energy to a region of the solar cell spectral response curve that is maximal

1.23.2.1.1 Upconversion

As the name implies, upconversion (UC) involves the conversion of low-energy photons to high-energy photons Put another way, long-wavelength photons are converted to shorter-wavelength photons For most PV technologies, this implies that infrared (IR) photons are being converted to photons in the near-IR or visible range For example, the band gap of Si is at ∼1100 nm, whereas the peak EQE for Si solar cells typically lies in the range of 550–900 nm Therefore, it would be most useful to convert photons in the range of 1100–1800 nm to the aforementioned regime, preferably in a broadband manner

UC is necessarily a multiphoton process [10, 11] To first order, two photons are utilized in sequential order by initial excitation

of a ground-state electron to an intermediate state followed by subsequent excitation by another low-energy photon to a further energy level at twice or higher the energy of the photons (Figure 4) Relaxation back to the ground state yields a single high-energy photon that can be absorbed by the active layer to produce electron–hole pairs As is evident, this process has an inherent maximum luminescent quantum yield (LQY) of 50% Furthermore, this must be achieved over a broadband utilizing unpolarized, diffused light from the sun The theoretical efficiency of such a process has been calculated by Trupke et al and it was shown to provide an increase in the detailed balance limit efficiency from ∼31% to 47.6% [13]

Two-Photon absorption

APTE, absorption, Cooperative Cooperative or multiphoton ETU GSA+ESA sensitization luminescence SHG excitation

Figure 4 Mechanisms of upconversion [12]

(a) (b)

λexc < λem

There are several other UC mechanisms that have been described in the literature For example, the energy transfer upconversion (ETU) mechanisms involved simultaneous absorption of two photons followed by energy transfer to two separate adjacent states with subsequent relaxation from the higher energy state Alternatively, both carriers can resonantly transfer to the same high-energy state, the so-called cooperative sensitization mechanism Cooperative luminescence occurs via virtual states and hence has a relatively low LQY, whereas second harmonic generation (SHG) also is based on virtual states obtained at extremely high intensities, ∼1013

Wm−2 [12]

Various micro- and nanoparticle types have been reported in the literature with measurable UC Wang et al synthesized upconverting nanoparticles of various compositions and showed that doping of nanoparticles of complex composition such as NaYF4:Yb/Er can control both the phase of the material as well as the size of the particles [14] The nanoparticles were embedded in

a polydimethylsiloxane (PDMS) polymer matrix and UC was observed However, these were not applied to solar cells De Wild et al applied a powder of β-NaYF4:Yb3+ (18%) Er3+ (2%) embedded in poly(methyl methacrylate) (PMMA) to the back of an a-Si device [15] Upon application of a 980 nm laser (10 mW) to the device, a gain in photocurrent of 10 μA cm−2 was measured This shows that it is possible to apply UC to a thin-film device, though more research is needed to improve this gain Similarly, Fischer

et al demonstrated UC in a Si solar cell using trivalent erbium-doped sodium yttrium fluoride [16] An UC gain of 5.1% was measured under monochromatic irradiance of 1880 W m−2 at 1523 nm This led to an EQE gain of 0.34% at an irradiance of

1090 W m−2 at 1522 nm Again, this shows that UC is possible, though more work is required to develop materials with suitable UC efficiency, as well as new methods of device integration

Interestingly, recent reports have focused on the use of molecular-based UC These are typically complex organic molecules in which UC is demonstrated in solution, or in limited examples in a polymer matrix Schmidt and co-workers demonstrated the mechanism of triplet–triple annihilation upconversion (TTA-UC) in a bimolecular system of palladium porphyrin and a polycyclic aromatic hydrocarbon [17] Through a rigorous analysis, it was shown that the efficiency of this UC mechanism is not 11% as has been surmised in prior literature, but as high as 60% theoretically, and they demonstrated a TTA-UC efficiency of up to 40% While this was not applied to a solar cell, it shows the potential to apply such an UC mechanism to solar cells, especially those that show a poor near-IR response such as a-Si and dye-sensitized solar cells A similar mechanism was also demonstrated by Castellano and co-workers [18]

1.23.2.1.2 Downconversion

The above discussion has highlighted the fact that more basic research is required to implement UC in solar cells Perhaps, a mechanism that is slightly more mature from a materials perspective is so-called downconversion (DC) The standard mechanism described in the literature as DC or quantum splitting is the emission of more than one long-wavelength photon per absorbed higher-energy photon A related spectral DC mechanism is based on photoluminescence (PL), that is, a single low-energy photon emitted per absorbed higher-energy photon This is also typically referred to in the literature as downshifting (DS) Examples of spectral DC as applied to solar PVs are described below

1.23.2.1.2(i) Conversion

DC is typically based on mechanisms of quantum cutting This requires a band structure that contains one or more intermediate levels, such that electrons excited into relatively high-energy states relax to one or more of these intermediate states, thus producing more than one lower energy photon A band diagram illustrating this basic mechanism, which is also referred to as quantum cutting,

is shown in Figure 5 This mechanism inherently has an LQY associated with it of greater than 100%, and theoretically up to 200%

DC was first proposed in 1953 by Dexter [20], and experimentally demonstrated in the 1970s by Piper et al [21] Most work has focused on DC from very high-energy photons (< 200 nm) that are useful in fluorescent lighting applications Indeed, LQYs higher than 190% have been shown excitations high in the conduction band of semiconducting materials such as ZnS [22] However, another major approach to DC is to utilize rare-earth based compounds, particularly those in the lanthanide series Due to the electronic structure of trivalent lanthanides, such as Pr, Er, and Yb, that is based on filling of [Xe]4fn5s25p6, there are multiple energy levels associated with these ions that give rise to multiple transitions Transition metal ions can be added to serve as sensitizers, thus both improving photon absorption (oscillator levels in these ions are low and lead to low absorption coefficients) and providing additional energy levels via resonant energy transfer [10]

Photoluminescence Downconversion

Figure 5 Simplified band diagram for (a) photoluminescence/downshifting and (b) for quantum cutting/downconversion [19]

Application of DC to solar cells is quite a challenging problem While up to 40% of the available solar energy is available above the band gap, the photon energy required to achieve high QY as noted above is limited in the solar spectrum Often it is necessary to concentrate The theoretical efficiency of DC was analyzed by Trupke et al [19], who showed that the detailed balance efficiency can increase from ∼31% to ∼39% However, as shown below, the experimental demonstration of downconverting materials that are of relevance to the solar spectrum is quite limited, and experimental LQY values are very low The application of micro- and nanoparticles to DC is also discussed

As noted above, several lanthanide-based phosphor compositions have been shown to yield DC with spectral feature that are relevant to solar PV cells Meijerink and co-workers have shown that YF3:Nd3+, Yb3+ can achieve spectral conversion of photons at

360 nm through a quantum cutting mechanism, which is a very promising development [23] The LQY was found to be ∼140% at low temperature However, the LQY was found to be lower at room temperature Loureiro et al demonstrated DC in nanoparticles based on SrAl12O19:Pr, Mg [24] The impact of surface states was also found to be critical in achieving a high LQY There have been experiments to also demonstrate DC using quantum dot (QD) semiconductors Among the first to explore this were Van Sark and co-workers [25], who showed gains in Si solar cells using a QD layer with spectral conversion

1.23.2.1.2(ii) Shifting

Luminescent DS is a related mechanism that involves the emission of one lower energy photon for each absorbed high-energy photon This involved energy loss subsequent to the initial photon absorption event due to a Stokes shift followed by emission of the lower energy photon Therefore, DC is limited to an LQY of no greater than unity, and it does not contribute to the fundamental limiting efficiency gain for a solar cell Therefore, its practical application is to alleviate parasitic losses associated with layers in a solar cell that lead to a drop in EQE at short wavelengths This can be related to say window layers, as in the case of CdTe, or perhaps highly doped emitter layers in Si

As in the case of DC, lanthanide-doped phosphors have been extensively studied for DC and luminescence in many applications Perhaps the most successful example is the work of Hong and Kawano, who showed that it is possible to achieve a relative gain of up

to 5% on CdTe solar cells based on this material [26] This work also highlights the potential limitation of such materials, namely, that the absorption coefficient for such ions in wide-band gap host lattices is very low (ca 1–100 cm−1), and hence very thick crystals were required to achieve such gains Practical implementation without a thinner layer that can be readily integrated will be quite limited Recently, it was shown by Klampaftis and Richards that it is possible to achieve gains in Si solar cells using DS phosphors, with relative gains of ∼0.2% [27, 28] A cautionary note is provided, in that great care must be taken to demonstrate gains on solar cells, as thermal and optical phenomena that are not related to DS (or DC) can easily manifest in gains that are not truly due to these mechanisms

QDs have also been applied to solar cells as a means of improving efficiency via DS Svrek et al demonstrated gains in IQE with

Si QDs embedded in a dielectric matrix applied to a silicon solar cell [29] The absolute gain in efficiency obtained was ∼0.4%, and it was estimated that an absolute gain of up to 1.4% is possible, assuming a DS LQY of 100%

1.23.2.1.3 Antireflective layers

Another mechanism to improve the performance of solar cells is to impart an antireflection coating (ARC), layer or film Reflection losses related to the refractive index contrast between the source medium, typically air, and the medium into which the photons are entering To first order, this reflection loss is described by the following equation:

�

n1 − n2 �2

1 þ n2

where n1 is the refractive index of the first medium and n2 the refractive index of the medium into which light is entering The ultimate goal of an AR layer is to reduce the refractive index contrast between air and the medium into which light is entering This can be applied as a single layer with an intermediate index or as a layer that provides a graded refractive index from air to the underlying substrate In the case of solar cells, this may include glass, semiconductor materials (such as Si, GaAs, etc.), and/or polymers such as ethyl vinyl acetate (EVA) Nanotechnology has been applied to solar application of AR layers quite extensively Several examples are shown below

One of the best examples of this is the use of nanoporous films to provide a refractive index between that of air and glass This is typically formed in silicon oxide (silica) films fabricated by a sol–gel method [30] These films have been formed on glass substrates and have shown relative efficiency gains as high as 2.65% on silicon solar cells

Nanoporous films represent a route to achieving an AR effect by use of random nanostructures in a solid film Another approach is

to provide a more controlled gradient in effective refractive index This is best exemplified by work on the so-called moth-eye effect, first observed by Clapham and Hutley in the early 1970s [31] It was found that the retina of a moth is black due to a microstructuring

of the retina with protrusions that are graded in their profile, which tend to gradually match the index of the substrate and hence provide very low reflection loss Reflections of 1% of less are possible with such an approach Various top–down methods have been applied for the formation of such structures, including techniques such as interference lithography [32] These microstructured approaches have led to successful demonstrations of an AR effect, though their applicability to PV remains to be seen

A related approach to the moth-eye effect is to use nanostructured protrusions on a surface to provide enhanced AR effect While these approaches can improve efficiency by decreasing reflectance, at the nanoscale it is also possible to

Figure 6 (a) Scanning electron microscopy image of an ODAR layer and (b) modeled reflectance as a function of angle for various geometries showing

an improvement over all angles Adapted from Tsakalakos L, Dalakos G, Brewer J, Zalyubovskiy S GE Global Research, unpublished

improve the angular response of the surface relative to a planar surface Therefore, such a structure can provide both an increase in efficiency as well as an increase in energy yield of the module due to improved capture of light throughout the course of the day We term this type of a nanotextured pillar/rod surface an omnidirectional antireflective (ODAR) layer

Figure 6 shows an example of such an ODAR layer as well as the improvement in reflectance with angle compared with a planar surface This was initially explored by Hadobas et al in dry etched nanostructures formed on Si wafers after interference lithography [33] It was shown that the reflectance can be reduced to levels as low as ∼1% on silicon wafers using such nanomachined Si structures More manufacturable approaches have also been demonstrated using processes such

as electroless metal-assisted etching, chemical vapor deposition (CVD), nanosphere lithography coupled with dry etching, and related methods Figure 7 shows the effective absorption of silicon nanowire (SiNW) films formed by wet etching and CVD processes It is evident that there is a significant gain in light capture by use of such a nanostructured surface, particularly in Si where the absorption depth is long It can also be useful for direct band gap semiconductors that have high refractive indices leading to relatively high inherent reflection losses Indeed, the area of ODARs remains an active area

of research within the applied physics and PV community, offering the potential for rapid efficiency gains

1.23.2.2 Process Improvements

One of the most mature areas for application of nanotechnology to PV is the use of nanostructures to enable new, improved processing methods that are also lower cost There are multiple processes that can potentially be impacted across various PV material systems Some approaches have already begun to find large-scale application at the commercial scale Before providing some examples, a brief description of the major PV processes is provided

(a)

(b)

Low fill factor Medium fill factor High fill factor Small, medium fill factor Flat glass

0 2 4 6 8 10 12

Incident angle (Degree)

0

20

40

60

80

100

Bulk Si film

Etched SiNW film

CVD SiNW film

(d)

Wavelength (nm)

Figure 7 Optical reflectance of SiNW ODAR films compared with planar Si, showing a broadband improvement in performance Adapted from Tsakalakos L, Brewer J, Fronheiser J GE Global Research, unpublished

Silicon technologies are typically based on mining of Si-containing materials (e.g., sand), a refining processes to remove impurities and oxygen, and a crystal pulling process such as the Czochralski process for single crystals or zone melting for polycrystalline boules The boules are then cut, typically with a wire saw that leads to significant materials loss (though kerfless methods are being explored), and the cells are then processed to add the p–n junction (e.g., diffusion from doped silicate glasses), a back contact via screen printing, passivation/AR layers (e.g., SixNy), and front metallization grid contacts by screen printing [34] Thin-film technologies are processed in a very different fashion Here, the goal is to deposit a thin (typically < 5 μm) film on a low-cost substrate The material is usually a direct band gap semiconductor which allows for maximal light absorption at these thicknesses CdTe, which classically is a superstrate configuration (sunny side is through original growth substrate), is processed at temperatures of between 450 °C and 650 °C using a closed-space sublimation process on glass that has been precoated with transparent conducting oxide layers Since these are monolithically integrated modules, the deposition processes are typically interposed by laser scribing processes, and finished with a back contact metallization process [35] Other thin-film materials, such as CIGS, may be deposited by one of many methods, including sputtering (reactive or co-sputter), co-evaporation, ion-assisted deposition, and related vacuum processes Sometimes, these are followed by a selenization (or sulfurization) process [36]

As opposed to CdTe, CIGS is a substrate configuration (sunny side is opposite to growth substrate), so a metal back contact such

as Mo is first applied to the substrate (glass, metal, or polymer) It is also possible to deposit CIGS by electrodeposition of the constituent metals, followed by selenization, for example, as developed by SoloPower, Inc [37] Finally, III–V-based technologies are fabricated almost exclusively by metal–organic chemical vapor deposition (MOCVD) processes on high-cost GaAs or Ge single-crystal wafers Some examples of how micro/nanotechnology is enabling novel processes are now discussed

1.23.2.2.1 Thin films by nanoparticles

Perhaps the best example of how nanotechnology is impacting PV modules is in the processing of thin films In particular, research and development in the processing of CIGS thin films has become quite intensive in recent years Mitzi and co-workers have demonstrated the solution-based processing of CIGS using nanoparticle inks reaching impressive efficiencies, that is, ∼12% [38] This was achieved by processing nanoparticles of the constituent metals into a thin film by spin-casting or slot-die coating, followed

by a selenization process Alternatively, it is possible to directly form nanoparticles of the required composition and simply sinter the film once deposited Nanoparticle sintering has also been used to form films of the emerging earth-abundant material

Cu2ZnSnS4 (CZTS) [39] The best example of commercial success with a nanoparticle approach is by Nanosolar, Inc., which is in production of CIGS modules based on this method [40]

Another example of successful implementation of nanoparticles-based inks is the application of silicon nanoparticles Here, silicon particles may be formed by vapour-phase or laser-based approaches Care must be taken when processing these particles since they typically contain a thin silicon oxide layer While it is possible to process full solar cells based on Si with these particles

[41], to date, the most success has been with the application of selective emitters to Si solar cells This has been pioneered by Innovalight, Inc., who provides a full process to single- and multicrystalline-based cell manufacturers Application of the selective emitter yields absolute efficiency improvements of up to 1% [42]

1.23.2.2.2 Lift-off of thin films

Another approach to forming high-quality thin films is to grow epitaxial absorber layers on a high-quality, typically high-cost substrate, and then transfer them to a lower-cost receptor substrate It is necessary to reuse the substrates for additional film growth

in order to achieve low-cost modules

This lift-off processing route has been most successfully applied to high-efficiency GaAs solar cells In this process, an ultrathin film of AlAs (ca 1–10 nm) is first deposited on the growth substrate, followed by deposition of the p–n diode-based solar cells Owing to the ∼106 contrast in etch rates between AlAs and GaAs in certain acid solutions, one can readily peel off a fully or partially

Substrate

Absorber

Metal

Transparent p-type semiconductor Transparent

n-type semiconductor

CuSCN TiO2

n

i

p

e

p

Absorber

finished solar cell from a substrate This so-called epitaxial lift-off (ELO) approach was proposed in the early 1980s with some success at the level of small solar cells [43] There is a renewed interest in this technology, and recently a world record (28.1%) single-junction, thin-film, and flexible solar cell was demonstrated by Alta Devices, Inc [8]

Lift-off technology was also applied to silicon solar cells Here the Si film must be removed from the growth substrate in a very different fashion The best example of this is the so-called Smart-Cut® process [44] A partially processed Si wafer is bombarded with high-energy hydrogen and/or helium ions that are deposited at a particular depth of interest The ions locally damage the Si crystal, which allows the upper layer to be fractured from the growth substrate by thermal processing or by crack initiation/Si solar cells with efficiency as high as 11.2% have been demonstrated by such an approach [45]

1.23.3 Nanoarchitectures

So far, we have discussed the application of micro/nanostructures and nano-related processes to form convention solar cells There are also significant efforts in the PV community to create novel solar cell structures that are based on micro- and/or nanostructures The ability to refashion materials at the nanoscale creates several opportunities to create distinct advantages for PV [46]

One of these advantages is that one can control light in new ways [47] This was described above in the case of ODAR layers, for example The use of nanoparticle arrays or nanowire films can create strong improvements in light trapping and absorption Most importantly, this can be achieved without the use of an additional ARC, and there is a possibility to create an improvement over the full solar spectrum

Electrically, the use of nanostructures can create not only new mechanisms of charge separation (see below) but also would allow for an improved charge collection via transport of charge carriers to interfaces that are very close to the solid volume in which photoexcited carriers are generated Here we describe two classes of nanostructured solar cells that are under investigation and that have shown promising results

1.23.3.1 Nanocomposites

The fashioning of bulk nanocomposites is one major approach to improving charge collection in solar cells This has been applied

to both inorganic and organic solar cells Among the first class of solar cells to be demonstrated with such a structure was the extremely thin absorber (ETA) solar cell This concept was first developed as early as 1998 by Siebentritt et al [48] The idea is to sandwich a very thin absorber film (i.e., 10–100 nm) such as CuInS2 between a nanostructured template layer that also acts as a charge carrier and a transparent conductor of opposite doping on the top side (Figure 8) For example, a nanostructured n-type TiO2

Figure 8 Schematic of an extremely thin-absorber (ETA) solar cell [49]

(a)

100–1000 nm

n

e

1–50 μm

h

p

(b)

2 μm

conductor can be formed on a substrate by screen printing or spray pyrolysis of nanoparticles [49] CIS is then deposited by an ion layer gas deposition (ILGAR) process, followed by electrodeposition of a transparent p-type CuSCN layer Early work yielded relatively low performance; however, subsequent work showed that it is possible to obtain solar cells with efficiency as high as 5%

by using a modified nanocomposite approach based on a nanostructured stack of TiO2/In2S3/CuInS2 [50]

The concept of a nanocomposite has also been extensively applied to organic photovoltaics (OPVs) Since it is well known that excitons have a very short diffusion length in organic semiconductors, the strategy adopted by many researchers is to create a very high surface area charge separation junction that is randomly interleaved This is typically referred to as a bulk heterojunction (BHJ) device and has shown efficiency that is on the order of 6% and led to significant advances in the OPV field [51] One challenge with such a concept has been that the two semiconductor layers (typically both organic or a hybrid) are interleaved which can create shunts This has necessitated the development of electron blocking layers to minimize shunting Such structures can also benefit from the use of nanocylinders that help to create a thick layer of one subcomponent on top of the BHJ region [52]

1.23.3.2 Nano/Microwires

A class of nanostructured solar cells that has gained significant interest in recent years is based on micro/nanowires or pillars These have been shown in Si [53], CdTe [54], a-Si [55], and III–V materials, such as InP and GaAs [56] Recent work (by this author) has shown that these architectures, when applied to high-efficiency materials such as GaAs, have the potential for >20% PCE, with experimental devices having achieved efficiency of >10% The benefits of such architectures include the separation of light absorption from charge transport by allowing for lateral extraction of charge from the pillar structure [57] (Figure 9), as well as improved optical absorption through extreme light trapping in these structures that is also omnidirectional [58, 59]

Nanowire/pillar approaches also offer potential for improvements in the performance of polymer-based solar cells It is well known that the so-called excitonic PV cells require dimensions less that 10 nm for effective charge separation due to the low transport lengths in polymer semiconductors [60] Active efforts for developing templates for such approaches are underway [61] Silicon has been the most widely studied micro/nanowire material for PV application This is because Si is the second most earth-abundant material and most dominant in the PV industry, yet the cost remains relatively high compared with thin films If one can develop a thin-film Si device that uses less Si yet has equivalent efficiency to a polycrystalline PV cell (14–16%), this would have

a dramatic impact on the market Efforts to reengineer Si solar cells at the micro- or nanoscale have focused exactly on this point There have been two major approaches to forming Si devices, the first being the use of CVD to grow Si wires on low-cost substrates

Figure 9 (a) Schematic of the operation of a nanowire solar cell (not drawn to scale) and (b) cross-sectional scanning electronic micrograph of templated SINWs