Extremely thin absorber solar cells based on nanostructured semiconductors

Abstract

Extremely thin absorber (eta) solar cells aim to combine the advantages of using very thin, easily and cheaply produced absorber layers on nanostructured substrates with the stability of all-solid-state solar cells using inorganic absorber layers. The concept of using nanostructured substrates originated from the dye-sensitised solar cell, where having a very high surface area allows the use of very thin layers of dye while still absorbing sufficient sunlight. However, both the dye and liquid electrolyte used in these devices demonstrated poor stability, and efforts were made to replace them with very thin inorganic absorber layers and solid state hole collectors respectively. The combination of these concepts – a nanostructured substrate coated with a very thin inorganic absorber and completed with a solid state hole collector – is known as an eta solar cell. This review summarises the development of both the inorganic absorbers and solid state hole collectors in porous TiO₂ and ZnO nanorod based cells, focusing on the material properties and growth/deposition methods. Future possibilities for eta solar cells are discussed, including utilisation of a wider range of materials, synthesis methods and novel materials such as quantum dots to produce tuned band gap and multijunction solar cells.

Keywords

Photovoltaic ZnO nanowires 3D solar cell CuSCN Hybrid solar cell

Introduction

Photovoltaic cells use materials that generate photoexcited carriers (electrons and/or holes) upon absorption of incident photons. A built-in asymmetry in the cell leads to a separation of the photoexcited carriers through the material so that a current flows and a voltage is generated. In a traditional semiconductor solar cell, this asymmetry is provided by a p–n or a p–i–n junction, and these layers are built-up on top of one another with one or both layer(s) absorbing the light. Thin film solar cells use direct band gap semiconductors with high absorption coefficients, and so can absorb a large portion of the incident light in layers 2–5 μm thick.1 The photogenerated carriers must then drift or diffuse through these layers to reach the front and back contacts. Thus, the average diffusion length of the carriers must match the thickness of the cell, i.e. a few micrometres, so that they do not recombine before being extracted. Therefore, high quality, expensive materials must be used in these cells, which achieve light-to-electrical energy conversion efficiencies in the range of 10–20.2

The principle of the extremely thin absorber (eta) solar cell design is to completely separate the function of a solid state solar cell into separate materials, with at least one layer being structured to increase the surface area (see Fig. 1). The absorber material is coated onto this structured surface, so that with a local thickness of <100 nm,3 an effective optical thickness of >1 μm is created: sufficient to absorb most of the incident light. The thinness of the absorber layer relaxes the requirements of material quality, as the photogenerated carriers need only to diffuse a few tens of nanometres to reach the junction with either the electron transport (n-type) or hole transport (p-type) material. A theoretical study of eta cells calculated that to achieve efficiencies of 10, an eta cell needed an absorber with a carrier lifetime 14 times lower than a thin film cell.4 This low material quality requirement allows greater flexibility in both the type of material and the deposition method. Low cost chemical synthesis methods can therefore be used as inherent high defect densities are less significant when such low diffusion lengths are required. 5 5,6 In principle, the absorber material therefore needs to only be chosen for properties relating to superior light absorption and ease of deposition, and the transport properties of the n- and p-type materials can be optimised separately. The aim of the eta concept is therefore to be able to produce low cost solar cells using stable materials with efficiencies approaching those of thin film solar cells.

Figure 1

Schematic of an extremely thin absorber solar cell. Nanostructuring of n-type layer by using either porous TiO₂ or ZnO nanorods allows an extremely thin absorber layer to be used. The increased surface area combined with enhanced light scattering from this structure allows a thin layer of absorber material to absorb a large portion of the incident light. The structure is filled with a p-type material to collect photogenerated holes

This review gives details of eta solar cells that have been reported to date, focusing on the materials used in the cells and the deposition methods used to create them. The properties and structure of the materials and the interactions between them are related to the performance of the final solar cells. The review begins by covering the history of eta solar cells from the first use of nanostructured substrates in dye-sensitised solar cells (DSSCs), through the replacement of molecular and liquid state components in these cells with solid state alternatives. Details of particular cells are then given, separated into cells based on porous TiO₂ layers, and ZnO nanorods. The review finishes by comparing the different types of cell, and suggesting potential areas for future work in this technology.

Background: Dye-sensitised solar cell

In 1991, O'Regan and Grätzel published a report of a solar cell using the wide band gap semiconductor titanium dioxide (TiO₂) sensitised with a layer of light absorbing organic dye.7 The operation of this cell can be seen in Fig. 2a. This cell demonstrated light-to-electrical energy conversion efficiencies of 7·12 under simulated sunlight, which was a large improvement on previous DSSCs, which had efficiencies <1.7 The reason for this improvement was the nature of the TiO₂ layer: the cell was based around a highly porous layer of TiO₂, which could adsorb 780 times more dye than a flat electrode due to the high surface area (see Fig. 2b). This greatly increased the effective optical thickness of the dye compared with the monolayer in a planar cell. The optical thickness was further enhanced by scattering, which had been demonstrated previously when texturing amorphous silicon solar cells.8 The optical thickness enhancement allowed a greater portion of the incident light to be absorbed by the cell, leading to the large increase in efficiency. Further details of the DSSC and other photoelectrochemical cells are given in Ref. 9.

Figure 2

Nanostructured DSSC: a operation under illumination; b schematic. In a, arrows show flow of electrons through cell. On absorption of a photon of energy hν, the dye is excited from the ground (D) to excited (D*) state, and transfers an excited electron to TiO₂. The dye is regenerated by an electron from the redox couple, A/A⁻, which receives an electron from the back contact

Despite the improvement in efficiency achieved in the DSSC by using highly porous TiO₂, there were still a number of limitations linked to stability of the iodide/triiodide (I/I₃⁻) redox electrolyte. Problems included containment to avoid evaporation10^–13 and long term stability due to the ionic conductivity, including the possibility of irreversible reactions.12 In order to address some of these limitations, work was undertaken to replace the liquid electrolyte in the DSSCs with solid state alternatives, while still utilising the advantages offered by a highly porous electrode. These alternative materials will be explored in the following section.

Solid state hole collectors

To replace the liquid electrolyte in a DSSC, a possible alternative was to use an appropriate solid state p-type semiconductor. This semiconductor material had to match the valence band level for hole transfer from the dye, but not degrade it. In addition, the material needed high transmission in the visible spectrum to allow incident light to the absorber layer (dye), 11 11,14 which required a wide band gap semiconductor. In the early work on ‘solid state DSSCs’, as they were known, copper iodide (CuI) and copper thiocyanate (CuSCN) were trialled. Their properties and deposition methods are considered below.

CuI

The first complete solid state DSSC was demonstrated in 1995 in which the electrolyte of a DSSC was replaced with the p-type semiconductor CuI.14 CuI has a band gap of 3·1 eV (Ref. 14; transparent to wavelengths above ∼400 nm). It was also expected to have a band alignment to the dye and TiO₂ that was favourable for efficient charge extraction (Fig. 3). CuI was deposited from a solution in acetonitrile, which filled the pores leaving CuI on drying. No high temperature or high energy steps were needed that would damage the dye. The initial TiO₂/dye/CuI cell achieved an energy conversion efficiency η of 0·8, and later η of 1·8 was achieved using a different dye.12 This is significantly lower than for DSSCs using an electrolyte. In addition, the stability sought by using a solid state hole collector was not achieved; the performance of the cell degraded with time under illumination due to degradation of both the CuI film and the dye layer. However, the stability of the dye was improved when UV light was filtered out of the incident radiation.14

Figure 3

Energy band diagram of band alignment of TiO₂/dye/CuI and TiO₂/dye/CuSCN system showing photoexcited electrons (filled) and holes (unfilled) in dye12

After this early work, CuI was used in some further solid state DSSCs. Some improvement was achieved using a buffer layer of MgO between the TiO₂ and the dye, which gave an efficiency of 2·90 compared with 2·13 without the buffer layer.13 This buffer layer also limited the degradation of cell parameters; after 72 h of illumination, they were within 70 of initial values, compared with 2 without the buffer layer. It was suggested that the MgO layer created a physical and potential barrier between the TiO₂ and CuI, suppressing the charge recombination and oxidation at the interface caused by injection of photogenerated holes from TiO₂ to either the dye or CuI.

CuSCN

CuSCN is a p-type semiconductor (due to a stoichiometric excess of SCN⁻ ions16) with a band gap of 3·6 eV (see Fig. 3). 16 16,17 In early work, CuSCN was investigated as a wide band gap material for dye sensitisation where the CuSCN was coated with a dye and the cell completed with a liquid electrolyte.18^–21 Later CuSCN was used with dye-sensitised TiO₂ in place of an electrolyte to form a solid state DSSC. The TiO₂/CuSCN heterojunction has been studied and shown to have good rectification,22 suggesting that it has potential for efficient charge separation in such a device. The first working solar cell to be made using the TiO₂/dye/CuSCN structure was produced in 1998 by O'Regan and Schwartz.23 The dye coated pores of the TiO₂ were filled with CuSCN using electrochemical deposition. It was shown that by post-treatment of the cell with KSCN and UV illumination, the internal efficiency of the cell could be improved. The authors suggest that this increase in efficiency is caused by oxidation of SCN⁻ at the interface of TiO₂/CuSCN to and/or (SCN)_x (a conductive polymer, parathiocyanogen), which are able to reduce the dye more rapidly to its original state than CuSCN.

CuSCN has also been deposited directly from solution. This was performed by dissolving CuSCN in propyl sulphide, successively spreading the solution on the surface and allowing it to dry at 80°C. 24 24,25 The best TiO₂/dye/CuSCN cell produced by this method had η = 2·1 at 100 mW cm⁻² illumination (1 sun).25 Again, this is lower than cells using a liquid electrolyte, but these CuSCN cells show an improvement in performance with storage. This was linked to continued evaporation of the propyl sulphide, which suggests that the initial drying during deposition was insufficient to allow all of the solvent to evaporate.

In 2004, both CuI and CuSCN were produced by successive ionic layer adsorption and reaction (SILAR) methods, in which substrates are dipped in aqueous solutions containing Cu ions, followed by solutions with either iodine or thiocyanate ions.15 These ions adhere to the substrate surface and then react with the counter ions to form the final products. SILAR was developed in 1990 by Nicolau et al. to deposit films of ZnS and CdS.26 Neither CuI nor CuSCN produced by this method has been used for a complete solar cell, therefore their performance relative to different production methods cannot be compared. CuSCN produced in this way is comprised of vertical crystallites with gaps apparent between the grains in scanning electron microscopy (SEM) images (see Fig. 4),15 which could lead to pinholes when top contacts are added. Such pinholes allow short circuit current routes that reduce the overall efficiency of the cell.

Figure 4

Images (SEM) of CuSCN produced by SILAR method15: a on glass; b on fluorine doped tin oxide

Further solid-state DSSCs have been reported using CuSCN and other solid state materials as hole collectors,27^–30 and have been covered in previous reviews. 31 31,32 To progress towards full eta solar cells, it is important to look at the replacement of the light absorbing dyes with solid state semiconductor films, as described below.

Inorganic absorbers

Narrow and medium band gap E_g semiconductors, with E_g in the range 1–2 eV, are able to absorb a large portion of the solar spectrum (Fig. 5). They therefore have the potential to produce solar cells with the highest possible efficiencies (inset, Fig. 5). This maximum efficiency results from a balance between maximising the photocurrent and the photovoltage in the cell to achieve the maximum power point: smaller E_g semiconductors can absorb more of the solar spectrum, and therefore generate higher photocurrent, but photovoltage is always less than E_g/e, where e is the electronic charge.1 Fulfilling the same function as molecular dyes, inorganic semiconductors have been investigated as light-sensitising materials for porous TiO₂ layers. In 1990, it was demonstrated that porous TiO₂ could be sensitised to visible light by coating with CdS.33 CdS was grown in situ by dipping the TiO₂ in a bath of Cd(ClO₄)₂, followed by a Na₂S solution, washing in water after each dip. This process was repeated up to 30 times. This is essentially an SILAR method, as mentioned in the section on CuSCN, although this name was not used in this study. Using this process, small particles of CdS were built up on the surface of TiO₂, which led to increased absorption of light incident on the composite. The absorption onset of the CdS shifted to longer wavelengths with increasing numbers of coating cycles. This shift was ascribed to decreasing quantum confinement effects in the particles as their size increased, calculating an increase from 4 to 7 nm between one and five coating cycles. An electrolytic cell was tested using a TiO₂ electrode coated with five cycles of CdS under monochromated light at 450 nm, giving η = 6. However, this wavelength was near to the peak in the incident photon-to-current conversion efficiency (IPCE) and without characterisation under full spectrum simulated sunlight it is difficult to compare the performance with other devices. As the bulk band gap of CdS is 2·4 eV, 34 34,35 it is transparent to a large amount of the solar spectrum, therefore is not an ideal absorber material (see Fig. 5).

Figure 5

Air mass (AM) 1·5 solar spectrum. Inset shows maximum theoretical efficiency of a single band gap semiconductor solar cell under AM 1·5 illumination as a function of band gap energy. Top scale shows photon energy corresponding to wavelengths on lower scale, and is annotated with band gap energies of semiconductors mentioned in the current review, as well as common solar cell materials. Inset is from Ref. 1

Later, successive dip coating was used to sensitise porous films with other sulphides by coating TiO₂ with nitrates of Cd, Pb, Ag, Sb and Bi, and then forming the sulphide using Na₂S.37 It was found that only PbS and CdS coatings were stable under illumination. Additionally, it was shown that the reduction in electron affinity of PbS caused by quantum confinement was required for injection of photogenerated electrons into the TiO₂ conduction band (CB); after three coatings of PbS, the IPCE began to drop corresponding to the CB level of PbS dropping to lower energy than that of the TiO₂ CB (see Fig. 6). This demonstrated the additional degree of flexibility that is accessible through the use of quantum confined particles: varying the particle size allows tuning of band alignment. CdS cells showed a drop in IPCE to half their original value after weeks or months of illumination, which occurred with PbS after hours. This degradation was linked to reactions with the electrolyte as well as particle growth or detachment.

Figure 6

Energy band diagram showing change in alignment between CBs of TiO₂ and PbS with and without quantum confinement effects36

In 1993, a porous TiO₂ film was coated with CdSe (E_g = 1·7 eV) using electrochemical deposition.38 The thickness of the deposited film was controlled between 3 and 20 nm by varying the number of cycles in the electrochemical deposition. A blue-shift was observed in the absorption onset of films <7 nm. The onset of photocurrent in the cell changed from ∼350 nm in a TiO₂ only cell, to ∼700 nm in a CdSe coated cell; the coating had successfully sensitised the TiO₂ to visible light through transfer of photogenerated electrons from CdSe to TiO₂. However, the photocurrent decayed rapidly in the first 5 s of illumination. This was attributed to a build-up of oxidised electrolyte at the interface with CdSe, limiting hole extraction and increasing recombination in CdSe. Again, the short or long term stability of the cell suffered due to the use of a liquid electrolyte, which demonstrates why more stable, solid state alternatives were sought as discussed in the section on solid-state hole collectors.

Many other cells have been produced using inorganic absorbers coated on porous TiO₂ regenerated using an electrolyte.39^–46 However, after initial investigations into solid state alternatives to the electrolyte or light absorbing layer in the DSSC, solar cells were produced that combined both modifications. These extremely thin absorber (eta) solar cells will be explored in the subsequent sections.

TiO₂ based eta solar cells

As in DSSCs, the TiO₂ layer in eta cells extracts photoexcited electrons generated in the absorber layer. An eta cell can be thought of as a nanostructured p–i–n cell, with the absorber being the intrinsic layer, and the TiO₂ is the n-type layer. TiO₂ is generally assumed to be n-type due to oxygen vacancies arising from non-stoichiometry, and therefore is more accurately written TiO_2−x, where x represents the oxygen deficiency of the material. Therefore, in this review, as in most papers TiO₂ should be considered as shorthand for TiO_2−x. In all papers describing TiO₂ eta cells, this non-stoichiometry is not considered, and the stoichiometry of the TiO₂ used is not measured. Oxygen deficiency generally results from the loss of oxygen in high temperature or low oxygen partial pressure processing. Thus, the exact stoichiometry, and therefore carrier concentration and Fermi level position of the material will vary depending on processing. However, as the stoichiometry is not measured, the impact of this variation cannot be compared between TiO₂ based cells.

Three-dimensional (3D) solar cells

A number of solar cells have been produced using a nanostructured TiO₂ electrode sensitised with semiconductor material, but without a separate hole collecting material. In these cells, the absorber material is deposited thickly enough to protrude from the porous TiO₂ and is contacted directly, thus acting as both the absorber and hole transport material. A cross-section SEM image of such a device is shown in Fig. 7, where CdTe can be seen coating the structured TiO₂ layer and protruding above it. The junction extends through a large portion of the depth, unlike a standard thin film solar cell where the junction only extends across the two-dimensional interface between layers. Although in earlier reports these cells have been referred to as eta solar cells, 47 51 47,51,52 they should not technically be included in this category as the absorber layers are always >100 nm thick so that they protrude from the TiO₂, and can be ∼1 μm thick. Additionally, in all other eta devices, the absorber layer is sandwiched between wide band gap materials, and not contacted directly. In later reports, the cells are referred to as 3D solar cells48^–50 and also as all-solid, inorganic bulk heterojunction solar cells.48 In this review, they are referred to as 3D solar cells, which differentiates them from the three-layer devices (excluding buffer layers) containing an absorber <100 nm thick, which are referred to as eta solar cells.

Figure 7

Cross-section SEM image of a 3D solar cell. Junction formed by electrochemically depositing CdTe into structured TiO₂. CdTe has completely covered the TiO₂ so that it protrudes above the TiO₂ layer and a top contact can be applied directly to it. Interface highlighted with black line51

The first examples of 3D solar cells were produced by electrochemically depositing a 150–250 nm thick layer of either CdTe (E_g = 1·5 eV) (Refs. 47 and 52) or Cd_xHg_(1−x)Te (Refs. 47 and 51) onto a porous TiO₂ film. These cells produced high open circuit voltages V_oc, and short circuit current densities J_sc, 47 47,52 but the low fill factor (FF) limited the energy conversion efficiency (η) of the cells to 1·2 (see Table 1). It is demonstrated that using porous instead of planar TiO₂ in this cell enhances quantum efficiency. This is because a local thickness of CdTe of only 150 nm on porous TiO₂ achieves the same optical thickness as a 2 μm film on a planar electrode. This leads to higher charge extraction as the diffusion length is ∼130 nm in electrochemically deposited CdTe.47 It is calculated that light scattering in the porous TiO₂ film leads to a path length increase of a factor of 5. The use of Cd_xHg_(1−x)Te as an absorber increased J_sc by >50, but also led to a slight reduction in V_oc.

Table 1

Three-dimensional heterojunction solar cells: growth methods and performance*

TiO₂ method and thickness	Absorber material, growth method and thickness	J_sc/mA cm⁻²	V_oc/mV	FF	η/	Year	Reference
SP; 2 μm	CdTe: ECD; 150 nm	8·9	670	0·2	1·2†	2003	47
DB; 2 μm	CuInS₂, and Al₂O₃ and In₂S₃ buffer layers: AL-CVD	18	490	0·44	4	2004	48
DB; 2 μm	CuInS₂ and In₂S₃ buffer layer: SP	17	530	0·55	5	2005	49
DB; 2 μm	CuInS₂ and In₂S₃ buffer layer: SP	23	710	0·43	7	2008	50

*Details of growth methods, film thicknesses where provided and performance data of 3D heterojunction solar cells using porous TiO₂ are given. SP = spray pyrolysis; ECD = electrochemical deposition; DB = doctor blading; AL-CVD = atomic layer chemical vapour deposition.

†Calculated from other parameters given using η = J_scV_ocFF/P_in, where P_in is the illumination intensity.

The 3D solar cells using porous TiO₂ electrodes have also been produced using CuInS₂ as an absorber material, which has a direct band gap of 1·55 eV (see Fig. 5).48 CuInS₂ was deposited into the pores of TiO₂ by atomic layer chemical vapour deposition in an earlier study,48 and later by chemical spray deposition. 49 49,50 Both methods were chosen because they can produce a conformal coating on a textured substrate. With both methods, an In₂S₃ buffer layer was added before CuInS₂ coating. 48 48,50 This material was chosen because its band alignment and band gap of 2·1 eV allowed photogenerated electrons to be transferred to TiO₂, while limiting recombination between CB electrons in TiO₂ and valence band holes in CuInS₂ (see Fig. 8).48 Theoretical consideration of blocking layers in 3D solar cells has shown that such non-insulating interface layers are important for reducing recombination across the interface and therefore increasing V_oc of the cell.53 An additional buffer layer of Al₂O₃ can be seen in Fig. 8, which was added before atomic layer chemical vapour deposition processing to reduce reactions between the TiO₂ and the In₂S₃ precursors during film growth. This was not needed when using the spray process.48 The theoretical study mentioned above suggests that such an insulating buffer layer does not provide the same V_oc enhancement as a non-insulating layer such as In₂S₃, as the photocurrent is reduced by an equal amount to the dark current.53 For spray deposited composites, elemental analysis in a transmission electron microscope (TEM) showed that the TiO₂, In₂S₃ and CuInS₂ layers are in intimate contact, and the pores are almost completely filled (see Fig. 9).49 Under AM 1·5 illumination, the best TiO₂–CuInS₂ solar cell was produced using spray deposition processing, which had a very high efficiency of 7 (see Table 1).50 It was also found that these devices gave the same characteristics after several months of storage, indicating that they were stable.49

Figure 8

Band diagram of TiO₂–CuInS₂ 3D solar cell48

Figure 9

Image (TEM) of porous TiO₂–CuInS₂ 3D solar cell. Contrast can be seen between different layers. a–e are areas where elemental analysis was performed and show a Sn from conductive oxide layer, b Sn, In and S, c Sn, Ti, In and S, d Ti, In and S and e In and S. Cu could not be measured due to a Cu grid being used for measurement. High resolution TEM image of area d shows intimate contact of lattices of different materials49

Thin film absorbers

The first fully solid state eta solar cell was made in 1998, and comprised a 6 μm thick porous TiO₂ film coated with selenium (Se) and filled with CuSCN (see Table 2).17 A 23 nm thick film of the grey, semiconducting form of Se was produced by annealing an electrochemically deposited film of Se and aging for several hours. This form of Se is p-type with a band gap of 1·8 eV. The Se coated porous TiO₂ was filled and covered with CuSCN from acetonitrile solution until it protruded 10 μm above the TiO₂. The band positions of Se and CuSCN were measured by depositing these materials individually onto fluorine doped tin oxide glass and performing Mott–Schottky measurements (see Fig. 10a). The TiO₂/Se/CuSCN solar cell had an efficiency of only 0·13, but did remain stable after prolonged illumination. It was suggested that the high surface area of the Se film probably increased surface recombination, reducing the photocurrent. Also, voids in the TiO₂ film allowed contact between the Se and the fluorine doped tin oxide, which led to short circuiting.

Figure 10

Band alignment for TiO₂/Se/CuSCN solar cell, showing transfer of photoexcited electrons and holes from Se into TiO₂ and CuSCN respectively17

Table 2

TiO₂ based eta cells: growth methods and performance*

TiO₂ method and thickness	Absorber material, growth method and thickness	Hole collector and method	J_sc/mA cm⁻²	V_oc/mV	FF	η/	Year	Reference
MD from colloid; 6 μm	Se: ECD and annealing; 23 nm.	CuSCN: SP from acetonitrile solution.	3†	600		0·13	1998	17
DB; 3 μm	CdS: SILAR; 5 nm.	CuSCN: MD from P-S solution+LiSCN.	2·3	860	0·65	1·3	2006	5
SC; 400 nm	In₂S₃: SILAR.	CuSCN: MD.	8	475	0·58	2·3	2009	6
SC; 2 μm	Sb₂S₃ and In(OH)_xS_y buffer layer: CBD.	CuSCN: MD from P-S solution+KSCN.	14·1	490	0·49	3·37	2009	54
SC; 2 μm	Cu₂S: CuI treatment of CdS layer produced by CBD. In(OH)_xS_y buffer layer used.	CuSCN: MD from P-S solution+LiSCN.	0·4	310	0·48‡	0·06	2009	55
SP	PbS and In(OH)_xS_y buffer layer: CBD.	PEDOT:PSS: SC.	7·4	280	0·4	0·83	2006	56
Dip coating	PbS and In(OH)_xS_y buffer layer: SILAR.	PEDOT:PSS: SC.	10·2	170	0·49	0·85	2006	57
Sol–gel; 2 μm	PbS and In(OH)_xS_y buffer layer: CBD.	PEDOT:PSS: SC.	9·24	249	0·34	0·78	2008	58
Screen print; 2·5 μm	PbS QDs: DC (aqueous).	spiro-OMeTAD: SC.	0·4†	240	0·51‡	0·49	2002	59
DB; 2 μm	PbS QDs: DC (methanol).	spiro-OMeTAD: SC.	4·58	560	0·57	1·46	2009	60

*Details of growth methods, film thicknesses where provided and performance data of eta solar cells using porous TiO₂ substrates are given. MD = manual deposition; ECD = electrochemical deposition; SP = spray pyrolysis; DB = doctor blading; SILAR = successive ionic layer adsorption and reaction; P-S = propyl sulphide; SC = spin coating; CBD = chemical bath deposition; DC = dip-coating.

†These cells were tested at less than 1 sun illumination, therefore parameters, especially J_sc, cannot be directly compared. Efficiency η values for these cells do however reflect the lower illumination.

‡Calculated from other parameters given using η = J_scV_ocFF/P_in, where P_in is the illumination intensity.

Further methods to produce eta solar cells include electrochemical deposition for CdTe, 61 61,62 ZnTe62 and CuSCN, 63 63,64 and SILAR for CdS.61 However, it was noted that using SILAR to coat the inside of the nanometre sized pores of TiO₂ can lead to inhomogeneities and incomplete filling due to poor infiltration of the liquid precursors.61 For this reason, a method was developed to form the semiconductor products by reacting adsorbed precursors containing metal cations (e.g. of Cd, Cu and In) with a gas, called ion layer gas reaction (ILGAR).61 In this process, porous TiO₂ was first dipped into a solution containing the metal chloride. Any residual solvent was then removed in an inert gas stream, and the adsorbed precursor was sulphurised with hydrogen sulphide gas, giving CdS, Cu₂S, In₂S₃ or CuInS₂. This process can be repeated to give controllable coating thickness, and produce a conformal and homogeneous coating of CdS61 and CuInS₂. 61 61,63 ILGAR was used to coat porous TiO₂ with CuInS₂, which was then filled with CuSCN deposited by electrochemical deposition.64 The J–V characteristics of this structure were tested and showed good rectification, consistent with the study discussed in the section on CuSCN.22 This also indicates that an eta solar cell produced using this method does not suffer from the problems with short circuiting that affected the Se sensitised cell.17

Following the early studies above, a number of full eta cells based on the porous TiO₂–CuSCN heterojunction and sensitised with sulphides were produced and tested. These cells were sensitised with CdS,5 In₂S₃,6 Sb₂S₃54 or Cu₂S.55 Details of these cells including deposition methods, layer thicknesses where provided and performance data are given in Table 2. The band gaps of these absorbers span almost the whole range of band gaps used in eta solar cells (see Fig. 5), and are 2·4, 2·1, 1·7–1·8 and 1·2 eV for CdS, In₂S₃, Sb₂S₃ and Cu₂S respectively. Although In₂S₃ has been used as a buffer layer with narrower band gap absorber materials deposited on top, it operates here as the sole absorber material, giving an efficiency of 2·3,6 which is one of the highest efficiencies currently achieved for eta cells. This demonstrates that as eta cells are currently operating well below maximum theoretical efficiency for a single band gap cell (inset, Fig. 5), they do not necessarily require the use of optimal band gap absorbers to achieve higher efficiencies. The construction of this cell also demonstrated the significance of the thickness of the TiO₂ layer for optimising the cell; it was found that the effective optical thickness of the In₂S₃ layer could be increased by increasing the thickness of the porous TiO₂ layer, while using the same number of SILAR coatings.6 This led to an increase in J_sc of the solar cells, but only up to about 200–300 nm of TiO₂, after which no further increase was observed. The processing of materials also has been shown to have an impact on the performance: in the CdS sensitised cell, both chemical bath deposition (CBD) and SILAR were used to deposit CdS, with CBD produced cells showing less light absorption and poorer cell performance.5

In both the Sb₂S₃ and Cu₂S sensitised cells, In₂S₃ was used as a buffer layer [referred to as In(OH)_xS_y in Table 2 due to unknown variation in stoichiometry and composition of actual films produced]. The use of these buffer layers improved both the performance55 and the stability54 of these cells. In the case of Sb₂S₃, it was noted that the In₂S₃ layer prevented oxidation of the Sb₂S₃ occurring when in contact with TiO₂.5 This is similar to buffer layers discussed in the section on CuI, which prevented the transfer of high energy, photogenerated holes from TiO₂ into the absorber material. It was found that under continuous illumination efficiency of the Sb₂S₃ cells dropped by no more than 10, which was regained by standing in the dark for a few hours.54 This indicates that with the right combination of materials and processing such solid state eta cells can be very stable under operation.

An additional processing step that was added to some of these cells, as indicated in Table 2, was to soak the coated TiO₂ and aqueous solution of either LiSCN 5 54 5,54,55 or KSCN,54 before filling with CuSCN from propyl sulphide solution. For the Sb₂S₃ sensitised cell, this treatment led to an improvement in J_sc, V_oc and FF and reduced the resistance of the cell, and was greater for KSCN treatment than LiSCN.54 Post-treatment of CuSCN with KSCN was mentioned in the section on CuSCN, and improvements were attributed to possible doping with SCN⁻ ions, increasing the conductivity of the CuSCN. Pretreatment with LiSCN or KSCN in these cells could be explained by a similar mechanism.

This set of cells with sulphidic absorbers is useful for understanding the impact that the absorber layer, TiO₂ film and processing variations have on the efficiency of the cells. The Cu₂S sensitised cell has the lowest efficiency of 0·06.55 The exact reason for this low efficiency is not identified, but the authors suggest that there are a lot of losses at the TiO₂ interface in this cell, despite attempts to passivate with an In₂S₃ buffer layer. The Sb₂S₃ sensitised cell has the highest efficiency of 3·37,54 which could result from the particular combination of materials in this device, but also could be because a large amount of optimisation of film thicknesses and processing was performed in this study. The large difference in photo harvesting ability between these cells can be seen when comparing the IPCE, or external quantum efficiency (EQE), which is shown in Fig. 11.

Figure 11

a EQE and transmission spectra of TiO₂/In(OH)S/Sb₂S₃/CuSCN solar cell: absorption onset and photocurrent generation onset begin at ∼750 nm, due to photon absorption by Sb₂S₃ layer.54 b EQE spectra of TiO₂/CdS/CuSCN and TiO₂/In(OH)S/Cu₂S/CuSCN solar cells: solid line is for CdS sensitised cell, and dotted line is for Cu₂S sensitised cell using In(OH)_xS_y and dodecylphosphonate buffer layers55

Other examples of porous TiO₂ based eta cells also utilised chemical methods to deposit the absorber layer, but used the conducting polymer polystyrene sulphonic acid doped poly(3,4-ethylenedioxythiophene) (PEDOT:PSS) as the hole collecting layer. All of these cells used a porous TiO₂ layer coated with an In(OH)_xS_y buffer layer and PbS [or Pb(OH)_xS_y] absorber produced by either CBD 56 56,58 or SILAR57 (see Table 2). The PEDOT:PSS was deposited into the coated pores of TiO₂ by spin coating from aqueous solution.56^–58 The band gap of the In(OH)_xS_y buffer layer varied between 2·4 and 3·4 eV depending on the pH of the CBD solution; as the composition becomes more S rich or more O rich, the band gap is closer to that of In₂S₃ (2·1 eV), or In₂O₃ (3·7 eV).56 In(OH)_xS_y in tested cells had a band gap of 2·4 eV. The Pb(OH)_xS_y that was deposited had a band gap of 0·85 eV, larger than that of bulk PbS (0·37 eV), attributed to the same alloying process as for In(OH)_xS_y. It was found that the In(OH)_xS_y, not the Pb(OH)_xS_y, was generating the photocurrent in the cell. A later study found that the low contribution by Pb(OH)_xS_y was due to high charge recombination at the In(OH)_xS_y/Pb(OH)_xS_y interface because of an barrier for electrons (see Fig. 12a).58 However, the solar cell performance was severely degraded if the Pb(OH)_xS_y layer was removed.57 The performance of the cell was also limited because it was not completely opaque, therefore not all of the incident light was absorbed. Also, the absorber layers did not completely fill the TiO₂ pores (see Fig. 12b). The authors identified that these structural issues must be addressed in order to improve the performance of the cells. 56 56,57

Figure 12

a band diagram58 and b SEM image56 of TiO₂/In(OH)_xS_y/Pb(OH)_xS_y/PEDOT:PSS solar cell structure

Nanoparticle absorbers

Following early porous TiO₂ solar cells sensitised with inorganic absorbers, cells were reported using inorganic nanoparticles as absorber materials. Nanoparticles are promising absorbers for eta solar cells, because the absorption coefficient and band gap increase as the particles are made smaller.39 Many cells used electrolytes as hole collectors, and can be thought of as similar to a DSSC, with nanoparticles replacing the dye. However, in some cells the electrolyte was replaced with a solid state material. The first such cell was produced by Plass et al. in 2002, and used the hole conductor 2,2′,7,7′-tetrakis(N, N-di-p-methoxyphenylamine)-9,9′-spirobifluorene (spiro-OMeTAD).59 Porous TiO₂ was sensitised using PbS quantum dots (QDs) grown in situ on the surface by dipping it into lead nitrate solution, followed by Na₂S solution. This process was repeated a number of times, producing different sized QDs (6 nm after seven cycles). PbS absorption was blue shifted due to quantum confinement, which reduced with increasing dipping cycles. Plass et al. note that the change in position of the PbS CB is beneficial to electron injection into TiO₂ (discussed in the section on inorganic absorbers).

In 2009, Lee et al. also produced a PbS QD sensitised porous TiO₂ solar cell using spiro-OMeTAD as a hole collector.60 They also coated the TiO₂ with PbS using lead nitrate and Na₂S solutions, but in methanol rather than water. This produced superior performance compared with aqueous solutions, attributed to better wetting and faster drying. The coverage of TiO₂ by PbS was found to be 30–40. Attempts to improve efficiency by increasing coverage were not successful as subsequent reduction in quantum confinement reduced the efficiency of electron transfer.

ZnO nanorod based eta solar cells

Solar cells considered thus far have been based on variations in the porous TiO₂ structure. However, other nanostructured materials have been used for the same purpose, the most common of which is ZnO nanorods. The motivation behind the replacement of porous TiO₂ with ZnO nanorods was to enhance the efficiency of charge transport: ZnO nanorods offer a direct conduction path to the contact65 and have high electron mobility (200 cm² V⁻¹ s⁻¹ for ZnO compared with 10 cm² V⁻¹ s⁻¹ for TiO₂).27 The open structure of the ZnO nanorods allows flexibility for coating procedures as the precursors can readily reach the entire surface.66 ZnO nanorods also produce an increase in optical path through light scattering similarly to porous TiO₂.67

As with TiO₂, ZnO is assumed to be n-type due to oxygen vacancies, and is more accurately considered ZnO_1−x. Again, the actual stoichiometry of ZnO used in eta cells is not measured, therefore cannot be compared. However, it is likely that due to different processing conditions, the carrier concentration of ZnO will vary between cells, which will have an impact on the solar cell performance.

Thin film absorbers

The first demonstration of a semiconductor coating on ZnO nanorods was made in 2000 by depositing amorphous silicon (a-Si) onto the rods using chemical vapour deposition.65 ZnO nanorods in this study were deposited by electrochemical deposition ∼2 μm long and 100–200 nm across (see Fig. 13a). The a-Si coating was uniform and conformal (Fig. 13b). It was noted that ZnO nanorods are more easily coated than porous TiO₂ due to their open morphology. No performance results from ZnO–a-Si solar cells were given.

Figure 13

ZnO nanorods grown by electrochemical deposition: a uncoated; b coated with Si by chemical vapour deposition. arrow indicates joining of coating between two rods65

Solar cells have been produced based on both CdTe and CdSe coated ZnO nanorods (see Fig. 14).68^–70 The growth methods and performance of these cells can be seen in Table 3. CdTe and CdSe were chosen as their narrower band gaps enhanced absorption in the visible region, and band alignment was predicted to favour charge extraction (see Fig. 15). The best cells used a CdSe absorber layer, which was annealed after deposition to improve crystallinity and reduce trap states.70 This cell had η = 2·3 at ∼1/3 sun after 1 week of storage. However, the efficiency dropped back to 1·5 after further months of storage. It was found that vacuum storage enhanced the performance of the cell, attributed to continued evaporation of the propyl sulphide, as found previously with CuSCN deposited by this method (section on ‘CuSCN’). The performance of CdTe based cells was poorer than CdSe based cells, as can be seen in Table 3, but an efficiency value was not given.69 The authors suggested that low J_sc and FF resulted from the small difference between the predicted CB levels of ZnO and CdTe (Fig. 15).69

Figure 14

a uncoated and b CdSe-coated ZnO nanorods grown by electrochemical deposition70

Figure 15

Predicted band alignment for a ZnO/CdTe/CuSCN and b ZnO/CdSe/CuSCN (Refs. 69 and 70)

Table 3

ZnO based eta cells: growth methods and performance*

ZnO method, average height and width	Absorber material, growth method and thickness	Hole collector and method	J_sc/mA cm⁻²	V_oc/mV	FF	η/	Year	Reference
ECD; 1·5 μm, 125 nm	CdSe: ECD; 35 nm.	CuSCN: MD from P-S solution.	4†	500		2·3	2005	68
ECD; 2 μm, 150 nm	CdTe: MOCVD; 65 nm.	CuSCN: MD from P-S solution.	0·03†	200	0·28		2005	69
ECD; 2 μm, 150 nm	CdSe: ECD; 35 nm.	CuSCN: MD from P-S solution.	3·9†	490	0·32	2·3	2005	70
CBD; 1·5 μm, 100 nm	In₂S₃: ILGAR; 25 nm.	CuSCN: MD from P-S solution.	10·5	570	0·56	3·4	2008	3
CBD; 800 nm, 60 nm	In₂S₃: ILGAR; 20 nm.	CuSCN: MD from P-S solution.	7·8	540	0·54	2·5	2008	66
CBD; 1·5 μm, 100 nm	In₂S₃: ILGAR; 35 nm.	CuSCN: MD from P-S solution. Annealed.	9	640	0·45	2·8	2009	71
ECD; 1·5 μm, 150 nm	CdSe: MD of QDs, then annealed.	MEH-PPV	5†	500	0·31^f	0·9	2008	72
ECD; 1·5 μm, 150 nm	CdSe: MD of QDs, then annealed.	P3HT				1·5	2008	72

*Details of growth methods, film thicknesses where provided and performance data of eta solar cells using ZnO nanorod substrates are given. ECD = electrochemical deposition; MD = manual deposition; P-S = propyl sulphide; MOCVD = metal organic chemical vapour deposition; CBD = chemical bath deposition; ILGAR = ion layer gas reaction; P3HT = poly(3-hexylthiophene).

Eta cells have also been produced by coating ZnO nanorods with an In₂S₃ absorber layer using ILGAR (see section on ‘Thin film absorbers’ under ‘TiO₂ based eta solar cells’), and filling with CuSCN. 3 66 71 73 3,66,71,73,74 Growth methods and photovoltaic parameters for these cells are given in Table 3. A cross-section of one of these cells can be seen in Fig. 16a. ILGAR produced a conformal coating of In₂S₃ on ZnO nanorods with controllable thickness (see Fig. 16b–d).3 When the local thickness of the In₂S₃ layer was varied from 10 to 75 nm, J_sc decreased and V_oc increased (see Fig. 17a).3 It was concluded that the diffusion length of carriers in the In₂S₃ layer was ∼10 nm. Therefore, increasing the local thickness led to increased recombination and the observed reduction in J_sc. Conversely, the increase in V_oc with In₂S₃ thickness was due to a reduction in tunnelling recombination through the absorber layer. The optimum solar cell performance was achieved with an In₂S₃ thickness of 25 nm. The influence of ZnO nanorod length on the performance of this cell structure was also studied.66 The rod length was varied between 0 (planar cell) and 3 μm by varying the deposition time between 0 and 150 min, while the local In₂S₃ thickness was maintained at 20 nm. As the rod length increased, it was found that J_sc increased while V_oc and FF decreased. The decrease in V_oc and FF resulted from a drop in the shunt resistance, attributed to an increase in surface recombination due to increased surface area of the longer rods. The optimum rod length was found to be 800 nm. Further studies of ZnO nanorod/In₂S₃/CuSCN solar cells focused on annealing the complete structure, and the effect on the In₂S₃/CuSCN interface. It was found that annealing the cell at 200°C for 2 min produced optimal improvement in J_sc, V_oc and FF.71 Annealing the In₂S₃/CuSCN layers caused Cu to diffuse towards the interface where it reacted with In₂S₃ giving additional sub-band gap states and a shift in the In₂S₃ band gap (see Fig. 17b and c). 71 71,74 This reduced recombination, increasing V_oc,74 and shifted absorption to longer wavelengths, increasing J_sc.71 These studies provide a detailed set of results showing the impact of varying the parameters of such a ZnO nanorod eta cell, and produced a cell with efficiency of 3·4,3 which is the highest efficiency to date for an eta cell.

Figure 16

a cross-section SEM image of ZnO nanorod/In₂S₃/CuSCN solar cell73 and b–d top–down SEM images of ZnO nanorods with 0, 15 and 45 nm of In₂S₃ coating respectively3

Figure 17

a change in J_sc, V_oc and FF in ZnO nanorod/In₂S₃/CuSCN solar cell as local thickness of In₂S₃ layer is increased3 and proposed band structure of In₂S₃–CuSCN interface b before and c after annealing74

ZnO nanorod/semiconductor/polymer solar cells were made using a novel coating method to deposit CdSe films onto ZnO nanorods.72 The films were produced by drop casting solutions of octadecyl amine capped CdSe QDs in toluene onto electrochemically deposited ZnO nanorods. Once deposited, the QD coated nanorods were annealed in an air–CdCl₂ mix at 380°C to form a continuous film of CdSe on the ZnO. Annealing led to a large increase in the quantum efficiency of the cell and a red-shift in the absorption onset due to loss of quantum confinement (see Fig. 18). The solar cell was completed by filling the rods with the conducfotive polymer poly[2-methoxy-5-(2′-ethyl-hexyloxy)-1,4-phenylene vinylene] (MEH-PPV). These cells gave an efficiency of 0·9, which was improved to 1·5 by replacing the MEH-PPV with poly(3-hexylthiophene) (P3HT).

Figure 18

Increase in quantum efficiency of ZnO nanorod/CdSe/MEH-PPV solar cell with different annealing times in air/CdCl₂ at 380°C (Ref. 72)

Nanoparticle absorbers

Solar cells have been produced by coating ZnO nanorods with QDs to sensitise them to visible light. The QDs used include CdSe, 75 75,76 CdTe,77 CdS 35 35,78 and CuInS₂.79 However, these cells are completed with a liquid electrolyte or tested in an electrochemical apparatus, therefore are not true eta solar cells as they are not fully solid state (see ‘Introduction’). They are therefore more similar to a DSSC, where the dye has been replaced with a nanoparticle absorber layer but the liquid electrolyte is retained. The methods for depositing the nanoparticles and performance of the devices is briefly considered below, but the parameters are not included in Table 3 as they are not eta solar cells.

Some nanoparticle coatings have been made by immersing nanorod coated substrates into a chemical bath of nanoparticle precursors and heating to induce nucleation of the nanoparticles directly on the surface. 76 76,78 These particles are too large to display quantum confinement. In the case of CdSe coated nanorods, the coatings are annealed after deposition at 400°C to produce a polycrystalline thin film layer.76 The best performing cell was obtained with a complete covering of the nanorods that appears ∼100 nm thick, giving η = 0·34 at 1 sun.76

Other studies using semiconductor nanoparticles to sensitise ZnO nanorods have retained the quantum confinement of the nanoparticles, thus can be referred to as QDs. These QDs have been grown before deposition on the nanorods, using capping molecules or shells to improve stability and prevent agglomeration in solution. The QDs have then been attached to the nanorods by soaking nanorod coated substrates in the QD solution for up to several days. 75 77 75,77,79 Using this method led to a limited coverage of QDs in the case of CdSe75 and CuInS₂,79 forming less than a monolayer on the surface. These cells thus suffer similarly to ZnO DSSCs80 as the lower surface area of ZnO nanorods compared with porous TiO₂ means that a monolayer of absorber cannot absorb sufficient incident light, and the efficiencies of these cells remain below 1. In the case of CdTe nanoparticle sensitised nanorods, soaking in the nanoparticle solution produced a much thicker coating, estimated to be 10 nm thick.77 The particles appeared to retain quantum confinement as the photocurrent onset occurred ∼550 nm: well above the bulk band gap of CdTe, which is 1·5 eV (≈830 nm).81 The reason for the increased coverage in this case was not investigated, and efficiency was not given so the impact of this increased coverage cannot be compared with other cells. However, it was shown that by increasing the nanoparticle coating thickness, the incident light absorption was increased, which should lead to increased photocurrent in the cell compared with those only using a monolayer of nanoparticles.

Comparison of designs and materials

The various designs of solar cell devices leading up to and including eta solar cells have been explored, and the materials and parameters are summarised in Table 4. This table only includes the designs where a full cell has been completed and tested. Where the same group has published multiple reports on the same design of solar cell, only the results from the best performing cell are included.

Table 4

All cells: materials and performance*

Basis	Buffer layer(s)	Absorber material	Hole collector	J_sc/mA cm⁻²	V_oc/mV	FF	η/	Year	Reference
TiO₂		Dye	Electrolyte	11·5†	685	0·68	7·12	1991	7
TiO₂		Dye	CuI	6†	450	0·53‡	1·8	1998	12
TiO₂	MgO	Dye	CuI	8·74	510	0·54	2·9	2003	13
TiO₂		Dye	CuSCN	7·8	600	0·44	2·1	2002	25
TiO₂		CdS QDs	Electrolyte	0·0175	395			1990	33
TiO₂		CdTe	CdTe	8·9	670	0·2	1·2‡	2003	47
TiO₂	In₂S₃	CuInS₂	CuInS₂	23	710	0·43	7	2008	50
TiO₂		Se	CuSCN	3†	600		0·13	1998	17
TiO₂		CdS	CuSCN	2·3	860	0·65	1·3	2006	5
TiO₂		In₂S₃	CuSCN	8	475	0·58	2·3	2009	6
TiO₂	In(OH)_xS_y	Sb₂S₃	CuSCN	14·1	490	0·49	3·37	2009	54
TiO₂	In(OH)_xS_y	Cu₂S	CuSCN	0·4	310	0·48‡	0·06	2009	55
TiO₂	In(OH)_xS_y	PbS	PEDOT:PSS	7·4	280	0·4	0·83	2006	56
TiO₂		PbS QDs	spiro-OMeTAD	4·58	560	0·57	1·46	2009	60
ZnO		CdTe	CuSCN	0·03†	200	0·28		2005	69
ZnO		CdSe	CuSCN	3·9†	490	0·32	2·3	2006	70
ZnO		In₂S₃	CuSCN	10·5	570	0·56	3·4	2008	3
ZnO		CdSe	MEH-PPV	5†	500	0·31‡	0·9	2008	72

*Summary of performance of cells discussed in papers where sufficient data are given.

‡Calculated from other parameters given using η = J_scV_ocFF/P_in, where P_in is the illumination intensity.

It can be seen from Table 4 that the efficiency values for the cells vary from <1 up to ∼7. None of the solar cells exceed the 7·12 efficiency of the original nanostructured DSSC. The highest efficiency of the other cells is achieved by the CuInS₂ based 3D solar cell. As discussed in the section on ‘3D solar cells’, this should not be considered a true eta solar cell, as the absorber layer is deposited to protrude from the porous TiO₂ layer so that it also functions as the hole collector. The most notable figure for this cell is the short circuit photocurrent J_sc – the highest for all of the cells. The much lower J_sc values for the full eta cells imply that there are losses in the cells that reduce the output current. The simplest explanation for this is insufficient absorption of the incident light, as noted for some In₂S₃–PbS coated TiO₂ cells. 56 56,57 The use of a nanostructured basis for an eta cell allows less absorber material to be used to absorb incident light, but since 3D solar cells (with thicker absorbers) produce a higher photocurrent, a thicker absorber may be needed in eta cells to collect all of the light. However, it was shown for the ZnO–In₂S₃ solar cell3 that increasing the absorber thickness does not necessarily improve J_sc: high recombination losses in the absorber materials can lead to a decrease in current with thicker absorbers. This demonstrates that the optimisation of absorber thickness in eta cells is complex. More studies to establish the optimum balance between light harvesting and charge extraction in such devices would be beneficial. These include further studies on light scattering such as those performed on ZnO nanorods,67 and studies on variation in light absorption with varying absorber layer thickness.

Calculations of the optimum absorber layer thickness as well as the level of nanostructuring in the cell have been made in a theoretical study of eta cells.4 This study modelled an eta cell as a p–i–n junction, with multiple junctions in parallel representing the nanostructuring of the substrate. Empirical parameters from CdTe and CuInS₂ were used in the model to find the optimum absorber thickness and structuring level with each material as the absorber layer. It was confirmed that the use of nanostructured substrates did allow higher efficiencies to be achieved with lower quality materials as the absorber thickness could be reduced to match the low carrier diffusion length. Because of the low carrier lifetime in the materials, the surface recombination had a negligible effect on the performance of the cell. The tunnelling recombination was much more important as a limiting factor on the cell performance, and meant that absorber thickness should not be reduced below 15–20 nm, as this led to very high losses due to tunnelling. This was found experimentally for ZnO/In₂S₃/CuSCN cells3 (see section on ‘Thin film absorbers’ under ‘ZnO nanorod based eta solar cells’). This puts a lower limit on the absorber quality of having a carrier diffusion length of at least ∼10 nm, as it was shown that the average carrier diffusion length of the absorber must be at least half the absorber thickness.4 For CdTe and CuInS₂, an optimum absorber thickness of ∼50 nm was calculated, with a surface enhancement of around tenfold for CdTe, and fivefold for CuInS₂. With these parameters, it was calculated that efficiencies of ∼15 could be achieved, although the omission of series resistance in the cell implies that true values will be limited to below this value. The calculation of optimum parameters for eta solar cells is extremely helpful to guide the design of future cells, although the predictions are limited by the assumptions made about the structure, and the availability of data on the absorber materials used in actual cells.

Other factors not present in 3D solar cells, which could contribute to lower J_sc in eta cells, are the hole collector and hole collector/absorber interface. The In₂S₃/CuSCN interface was identified as a source of loss in cells using ZnO nanorods (section on ‘Thin film absorbers’ under ‘ZnO nanorod based eta solar cells’), and annealing was used to reduce these losses.74 With such optimisation of the interface, in combination with optimisation of the absorber thickness, cells with η = 3·4 were produced,3 the highest for eta solar cells (Table 4). This demonstrates that such systematic optimisation of the parameters in an eta solar cell is necessary to produce higher efficiency values. The hole collector itself also presents opportunities for optimisation. In the section on ‘Thin film absorbers’ under ‘TiO₂ based eta solar cells’, the use of LiSCN or KSCN as a pretreatment before CuSCN deposition was shown to improve solar cell performance through doping with SCN⁻ ions. However, when using CuSCN as a hole collector for ZnO nanorod based eta cells, LiSCN or KSCN was not used. Utilisation of such optimised procedures is important to ensure continued progress in eta cell performance.

There is also potential for optimisation of the TiO₂ or ZnO layer for eta cells. A limited number of groups studied the impact of varying the TiO₂ thickness6 or ZnO rod length,66 but often no indication is given that these layers have been optimised, although this may have been performed before publication and not included in the report. With both materials, there are also other structural factors that could be varied to optimise the design, such as TiO₂ grain and pore size and ZnO nanorod spacing and diameter. Optimising these properties is a complicated process, as they are often difficult to fully control during synthesis, and the best parameters will vary depending on the nature of the absorber and hole collecting layer. Additionally, as mentioned at the beginning of sections on ‘TiO₂ based eta solar cells’ and ‘ZnO nanorod based eta solar cells’, the n-type conductivity of TiO₂ and ZnO results from oxygen deficiency, which is not normally measured. The stoichiometry, and therefore conductivity and Fermi level position will vary depending on the processing conditions, which will have an impact on the cell performance. Greater analysis and control of the composition of the materials use will help to find the optimum material parameters.

When comparing the efficiency of eta cells to DSSCs, it should be remembered that a strong motivation for the replacement of molecular dye and liquid electrolyte was an increase in stability of the solar cell. This increase in stability has been achieved by replacement of both the electrolyte with solid hole conductors and the dye with semiconductor absorber materials. Not all solid state materials used showed an increase in stability, such as CuI, hence the dominance of CuSCN in later cells. Other than CuSCN, the materials most commonly used as hole collectors are conductive polymers (PEDOT:PSS, spiro-OMeTAD and MEH-PPV). These materials are attractive as they can be easily deposited from solution. However, the stability of eta cells using these materials has not been reported. This should be tested in future studies to ensure that the performance of cells using polymers does not degrade with storage or illumination. The improvement in stability of nanostructure based solar cells achieved with many solid state materials in the eta design is a positive step in the search for low cost, stable and efficient solar cells. However, for these cells to become a viable technology, more work is needed to increase their efficiency while still using low cost materials and production techniques.

Future possibilities

In general, the few potential avenues for improvement discussed above, as well as many further ones, will come from more systematic investigations of the optimal parameters such as those undertaken for the ZnO/In₂S₃/CuSCN solar cells. With the complexity of eta cells there are many variable parameters. By varying these properties researchers will not only find the optimal combinations for increased efficiency, but also gain an understanding of the mechanisms behind performance changes. With the input of predicted optimal parameters from models of eta cells, such as those discussed above, the range of parameters trialled may be guided within certain bounds. Expansion of such models to include factors such as series resistance, and data from a wider range of materials would be beneficial to enhance their applicability to real systems. The use of a broader range of materials in eta cells would also be beneficial to find systems that can produce higher efficiencies. Although the selection of suitable wide band gap p-type semiconductors for hole collectors is quite small, trials with materials other than CuSCN and conducting polymers could demonstrate the extent to which these materials limit the performance of the cells. Conversely, there is a wide choice of absorber material, especially since choice is not limited to high mobility materials. Theoretically, any semiconductor with an appropriate band gap for solar energy absorption could be used, and this choice is widened to narrower band gap semiconductors if quantum confined nanoparticles are used. Even if quantum confinement is lost through annealing, nanoparticles may also be useful as precursors to thin absorber layers, as demonstrated for coating ZnO nanorods with CdSe (section on ‘Thin film absorbers’ under ‘TiO₂ based eta solar cells’).72 Such novel techniques will be useful for depositing a wider range of semiconductor absorbers for eta solar cells. Also, multiple materials and/or nanoparticles may be used as absorbers in a single cell. This could allow a greater utilisation of the incident light by tuning the absorption of the cell. This could also be achieved using layer based deposition techniques such as SILAR and ILGAR, as they can produce composition or material variation through a thin absorber. However, although the SILAR and ILGAR techniques have been effective in producing conformal coating of absorbers for eta solar cells, they have almost exclusively been used for the deposition of sulphides. Although this still allows a range of useful materials such as In₂S₃, CuInS₂ and Sb₂S₃, it does restrict the choice of materials. Thus for future work it would be beneficial to either adapt the technique for deposition of materials based on other anions, or to use alternative techniques to deposit a wider range of semiconductors.

Summary

The effectiveness of high surface area nanostructured substrates in increasing solar cell efficiency was first demonstrated in 1991 by O'Regan and Grätzel who produced cells with 7·12 efficiency with porous TiO₂, compared with <1 without. Problems with stability led to replacement of both the dye and the electrolyte. The wide band gap p-type semiconductors CuI and CuSCN were used to replace the electrolyte. CuSCN, produced by electrochemical deposition or from propyl sulphide solution, became dominant due to its superior stability. Later, the hole conducting polymers PEDOT:PSS, spiro-OMeTAD and MEH-PPV were also used for this function. The dyes were replaced with inorganic narrow band gap semiconductors, some as nanoparticles (or QDs). Initially, these innovations were used separately, but later solid hole conductors and inorganic absorbers were combined to produce extremely thin absorber (eta) solar cells. Porous TiO₂ cells were produced using absorbers such as Se, CdS, CuInS₂, In₂S₃, Sb₂S₃, Cu₂S and PbS thin films and QDs. Later, cells were also made using ZnO nanorods as an alternative to porous TiO₂ with CdTe, CdSe or In₂S₃ absorbers. Absorber layers were produced using MOCVD, electrochemical deposition, SILAR, ILGAR and CBD. For eta cells, porous TiO₂ and ZnO nanorods offer different advantages. Porous TiO₂ has a higher surface area, which is beneficial for very thin absorber layers such as monolayers of dyes or QDs. ZnO nanorods provide a more direct conduction path to the back contact, and are more easily penetrated by absorber layer precursors. Thus either material may be more appropriate for each absorber layer and deposition method.

Conclusions

The stability sought in replacing dyes and electrolytes in nanostructured DSSCs has been achieved in many eta cells. Now, further efficiency improvements must be achieved: the highest reported efficiencies for true eta cells are ∼3·4 (TiO₂/In₂S₃/Sb₂S₃/CuSCN and ZnO/In₂S₃/CuSCN): still below the 7–12 levels of the DSSC, or 10–20 levels of thin film cells. An efficiency of 7 has been achieved for a 3D solar cell, which uses porous TiO₂ with CuInS₂ as both absorber and hole conductor. The reasons for this improved performance needs to be understood in order to further improve eta cells. Implementation of wider range of materials for both absorbers and hole conductors would be useful to identify and understand optimal performance. Greater use of nanoparticles and innovative designs such as multijunction cells could also make progress in performance. Hopefully, with new materials combined with incremental improvements, efficiency levels approaching 10 and beyond could be reached in an eta solar cell using low cost materials and deposition techniques. Such devices will achieve the goal of stable, low cost, high efficiency solar cells.

Footnotes

This review was the commended review of the 2010 Materials Literature Review Prize of the Institute of Materials, Minerals and Mining, which is administered by the Editorial Board of Materials Science and Technology.

References

Nelson

: ‘The physics of solar cells’; 2003, London, Imperial College Press.

Green

, Emery

, Hishikawa

, Warta

: Prog. Photovolt.: Res. Appl. , 2009, 17, (5), 320–326.

Belaidi

, Dittrich

, Kieven

, Tornow

, Schwarzburg

, Lux-Steiner

: Phys. Status Solidi Rapid Res. Lett. , 2008, 2, (4), 172–174.

Taretto

, Rau

: Prog. Photovolt.: Res. Appl. , 2004, 12, (8), 573–591.

Larramona

, Chone

, Jacob

, Sakakura

, Delatouche

, Pere

, Cieren

, Nagino

, Bayon

: Chem. Mater. , 2006, 18, (6), 1688–1696.

Herzog

, Belaidi

, Ogacho

, Dittrich

: Energy Environ. Sci. , 2009, 2, (9), 962–964.

O'Regan

, Grätzel

: Nature , 1991, 353, (6346), 737–740.

Deckman

, Wronski

, Witzke

, Yablonovitch

: Appl. Phys. Lett. , 1983, 42, (11), 968–970.

Grätzel

: Nature , 2001, 414, (6861), 338–344.

10.

Hagfeldt

, Didriksson

, Palmqvist

, Lindström

, Södergren

, Rensmo

, Lindquist

S.-E

: Sol. Energy Mater. Sol. Cells , 1994, 31, (4), 481–488.

11.

O'Regan

, Schwartz

: Chem. Mater. , 1995, 7, (7), 1349–1354.

12.

Tennakone

, Kumara

GRRA

, Kottegoda

IRM

, Perera

VPS

, Weerasundara

PSRS

: J. Photochem. Photobiol. A: Chem. , 1998, 117, (2), 137–142.

13.

Taguchi

, Zhang

, Sutanto

, Tokuhiro

, Rao

, Watanabe

, Nakamori

, Uragami

, Fujishima

: Chem. Commun. , 2003, (19), 2480–2481.

14.

Tennakone

, Kumara

, Kumarasinghe

, Wijayantha

, Sirimanne

: Semicond. Sci. Technol. , 1995, 10, 1689–1693.

15.

Sankapal

, Goncalves

, Ennaoui

, Lux-Steiner

: Thin Solid Films , 2004, 451–452, 128–132.

16.

Tennakone

, Jayatissa

, Fernando

CAN

, Wickramanayake

, Punchihewa

, Weerasena

, Premasiri

WDR

: Phys. Status Solidi A , 1987, 103A, (2), 491–497.

17.

Tennakone

, Kumara

GRRA

, Kottegoda

IRM

, Perera

VPS

, Aponsu

GMLP

: J. Phys. D: Appl. Phys. , 1998, 31, (18), 2326–2330.

18.

Tennakone

, Fernando

CAN

, Dewasurendra

, Kariapper

: Sol. Energy Mater. , 1986, 14, (6), 499–506.

19.

Fernando

: Sol. Energy Mater. Sol. Cells , 1993, 28, (4), 375–380.

20.

Fernando

, Kitagawa

, Suzuki

, Takahashi

, Komura

: Sol. Energy Mater. Sol. Cells , 1994, 33, (3), 301–315.

21.

Fernando

CAN

, Kumarawadu

, Takahashi

, Kitagawa

, Suzuki

: Sol. Energy Mater. Sol. Cells , 1999, 58, (4), 337–347.

22.

Rost

, Sieber

, Siebentritt

, Lux-Steiner

, Konenkamp

: Appl. Phys. Lett. , 1999, 75, (5), 692–694.

23.

O'Regan

, Schwartz

: Chem. Mater. , 1998, 10, (6), 1501–1509.

24.

Kumara

GRRA

, Konno

, Senadeera

GKR

, Jayaweera

PVV

, Silva

DBRAD

, Tennakone

: Sol. Energy Mater. Sol. Cells , 2001, 69, (2), 195–199.

25.

O'Regan

, Lenzmann

, Muis

, Wienke

: Chem. Mater. , 2002, 14, (12), 5023–5029.

26.

Nicolau

, Dupuy

, Brunel

: J. Electrochem. Soc. , 1990, 137, (9), 2915–2924.

27.

O'Regan

, Schwartz

, Zakeeruddin

, Grätzel

: Adv. Mater. , 2000, 12, (17), 1263–1267.

28.

O'Regan

, Lenzmann

: J. Phys. Chem. B , 2004, 108B, (14), 4342–4350.

29.

Zhang

X.-T

, Liu

H.-W

, Taguchi

, Meng

Q.-B

, Sato

, Fujishima

: Sol. Energy Mater. Sol. Cells , 2004, 81, (2), 197–203.

30.

O'Regan

, Scully

, Mayer

, Palomares

, Durrant

: J. Phys. Chem. B , 2005, 109B, (10), 4616–4623.

31.

, Wang

, Kang

, Wang

, Qiu

: Sol. Energy Mater. Sol. Cells , 2006, 90, (5), 549–573.

32.

Snaith

, Schmidt-Mende

: Adv. Mater. , 2007, 19, (20), 3187–3200.

33.

Vogel

, Pohl

, Weller

: Chem. Phys. Lett. , 1990, 174, (3–4), 241–246.

34.

Kohtani

, Kudo

, Sakata

: Chem. Phys. Lett. , 1993, 206, (1–4), 166–170.

35.

Zhang

, Xie

, Jiang

, Wei

, Pang

, Wang

: Nanotechnology , 2009, 20, (15), 155707.

36.

Hoyer

, Könenkamp

: Appl. Phys. Lett. , 1995, 66, (3), 349–351.

37.

Vogel

, Hoyer

, Weller

: J. Phys. Chem. , 1994, 98, (12), 3183–3188.

38.

Liu

, Kamat

: J. Phys. Chem. , 1993, 97, (41), 10769–10773.

39.

Zaban

, Micic

, Gregg

, Nozik

: Langmuir , 1998, 14, (12), 3153–3156.

40.

Niitsoo

, Sarkar

, Pejoux

, Rühle

, Cahen

, Hodes

: J. Photochem. Photobiol. A: Chem. , 2006, 181, (2–3), 306–313.

41.

, Zhu

, Norman

, Ferrere

, Frank

, Nozik

: J. Phys. Chem. B , 2006, 110B, (50), 25451–25454.

42.

Chang

C.-H

, Lee

Y.-L

: Appl. Phys. Lett. , 2007, 91, (5), 053503–3.

43.

Lee

, Yum

J.-H

, Leventis

, Zakeeruddin

, Haque

, Chen

, Seok

, Grätzel

, Nazeeruddin

: J. Phys. Chem. C , 2008, 112C, (30), 11600–11608.

44.

Sun

W.-T

, Yu

, Pan

H.-Y

, Gao

X.-F

, Chen

, Peng

L.-M

: J. Am. Chem. Soc. , 2008, 130, (4), 1124–1125.

45.

Lan

G.-Y

, Yang

, Lin

Y.-W

, Lin

Z.-H

, Liao

H.-Y

, Chang

H.-T

: J. Mater. Chem. , 2009, 19, 2349–2355.

46.

Lee

, Chen

, Moon

, Sauvage

, Sivula

, Bessho

, Gamelin

, Comte

, Zakeeruddin

, Seok

, Grätzel

, Nazeeruddin

: Langmuir , 2009, 25, (13), 7602–7608.

47.

Ernst

, Belaidi

, Konenkamp

: Semicond. Sci. Technol. , 2003, 18, (6), 475–479.

48.

Nanu

, Schoonman

, Goossens

: Adv. Mater. , 2004, 16, (5), 453–456.

49.

Nanu

, Schoonman

, Goossens

: NanoLetters , 2005, 5, (9), 1716–1719.

50.

Goossens

, Hofhuis

: Nanotechnology , 2008, 19, (42), 424018.

51.

Belaidi

, Bayón

, Dloczik

, Ernst

, Lux-Steiner

, Könenkamp

: Thin Solid Films , 2003, 431–432, 488–491.

52.

Ernst

, Engelhardt

, Ellmer

, Kelch

, Muffler

, Lux-Steiner

, Könenkamp

: Thin Solid Films , 2001, 387, (1–2), 26–28.

53.

Grasso

, Burgelman

: Thin Solid Films , 2004, 451–452, 156–159.

54.

Itzhaik

, Niitsoo

, Page

, Hodes

: J. Phys. Chem. C , 2009, 113C, (11), 4254–4256.

55.

Page

, Niitsoo

, Itzhaik

, Cahen

, Hodes

: Energy Environ. Sci. , 2009, 2, (2), 220–223.

56.

Bayón

, Musembi

, Belaidi

, Bär

, Guminskaya

, Fischer

, Lux-Steiner

, Dittrich

: Comptes Rendus Chimie , 2006, 9, (5–6), 730–734.

57.

Oja

, Belaidi

, Dloczik

, Lux-Steiner

M.-C

, Dittrich

: Semicond. Sci. Technol. , 2006, 21, (4), 520–526.

58.

Musembi

, Rusu

, Mwabora

, Aduda

, Fostiropoulos

, Lux-Steiner

M.-C

: Phys. Status Solidi A , 2008, 205A, (7), 1713–1718.

59.

Plass

, Pelet

, Krueger

, Grätzel

, Bach

: J. Phys. Chem. B , 2002, 106B, (31), 7578–7580.

60.

Lee

, Leventis

, Moon

S.-J

, Chen

, Ito

, Haque

, Torres

, Nüesch

, Geiger

, Zakeeruddin

, Grätzel

, Nazeeruddin

: Adv. Funct. Mater. , 2009, 19, (17), 2735–2742.

61.

Möller

, Fischer

, Muffler

, Könenkamp

, Kaiser

, Kelch

, Lux-Steiner

: Thin Solid Films , 2000, 361–362, 113–117.

62.

Ernst

, Sieber

, Neumann-Spallart

, Lux-Steiner

, Konenkamp

: Thin Solid Films , 2000, 361, 213–217.

63.

Könenkamp

, Ernst

, Fischer

, Lux-Steiner

, Rost

: Phys. Status Solidi A , 2000, 182A, (1), 151–155.

64.

Kaiser

, Ernst

, Fischer

, Könenkamp

, Rost

, Sieber

, Lux-Steiner

: Sol. Energy Mater. Sol. Cells , 2001, 67, (1–4), 89–96.

65.

Könenkamp

, Boedecker

, Lux-Steiner

, Poschenrieder

, Zenia

, Levy-Clement

, Wagner

: Appl. Phys. Lett. , 2000, 77, (16), 2575–2577.

66.

Kieven

, Dittrich

, Belaidi

, Tornow

, Schwarzburg

, Allsop

, Lux-Steiner

: Appl. Phys. Lett. , 2008, 92, (15), 153107.

67.

Tena-Zaera

, Elias

, Levy-Clement

: Appl. Phys. Lett. , 2008, 93, (23), 233119.

68.

Lévy-Clément

, Tena-Zaera

, Ryan

, Katty

, Hodes

: Adv. Mater. , 2005, 17, (12), 1512–1515.

69.

Tena-Zaera

, Katty

, Bastide

, Lévy-Clément

, O'Regan

, Muñoz-Sanjosé

: Thin Solid Films , 2005, 483, (1–2), 372–377.

70.

Tena-Zaera

, Ryan

, Katty

, Hodes

, Bastide

, Lévy-Clément

: Comptes Rendus Chimie , 2006, 9, (5–6), 717–729.

71.

Belaidi

, Dittrich

, Kieven

, Tornow

, Schwarzburg

, Kunst

, Allsop

, Lux-Steiner

M.-C

, Gavrilov

: Sol. Energy Mater. Sol. Cells , 2009, 93, (6–7), 1033–1036.

72.

Nadarajah

, Word

, VanSant

, Könenkamp

: Phys. Status Solidi B , 2008, 245B, (9), 1834–1837.

73.

Dittrich

, Kieven

, Rusu

, Belaidi

, Tornow

, Schwarzburg

, Lux-Steiner

: Appl. Phys. Lett. , 2008, 93, (5), 053113.

74.

Dittrich

, Kieven

, Belaidi

, Rusu

, Tornow

, Schwarzburg

, Lux-Steiner

: J. Appl. Phys. , 2009, 105, (3), 034509.

75.

Leschkies

, Divakar

, Basu

, Enache-Pommer

, Boercker

, Carter

, Kortshagen

, Norris

, Aydil

: NanoLetters , 2007, 7, (6), 1793–1798.

76.

Tang

, Hu

, Chen

, Luo

, Li

, Zhang

: Electrochim. Acta , 2009, 54, (10), 2742–2747.

77.

Cao

, Chen

, Guo

: J. Phys. Chem. C , 2008, 112C, (51), 20560–20566.

78.

Tak

, Kim

, Lee

, Yong

: Chem. Commun. , 2008, 2008, 4585–4587.

79.

Kuo

, Liu

, Chen

, Lin

: J. Mater. Chem. , 2009, 19, (37), 6780–6788.

80.

Law

, Greene

, Johnson

, Saykally

, Yang

: Nat. Mater. , 2005, 4, (6), 455–459.

81.

Birkmire

, Eser

: Ann. Rev. Mater. Sci. , 1997, 27, (1), 625–653.

Extremely thin absorber solar cells based on nanostructured semiconductors

Abstract

Keywords

Introduction

Background: Dye-sensitised solar cell

Solid state hole collectors

CuI

CuSCN

Inorganic absorbers

TiO2 based eta solar cells

Three-dimensional (3D) solar cells

Thin film absorbers

Nanoparticle absorbers

ZnO nanorod based eta solar cells

Thin film absorbers

Nanoparticle absorbers

Comparison of designs and materials

Future possibilities

Summary

Conclusions

Footnotes

References

TiO₂ based eta solar cells