WO2011042682A2 - Intermediate band semiconductor photovoltaic devices, uses thereof and methods for their manufacture - Google Patents

Abstract

A photovoltaic device such as a solar cell is disclosed having an n-base layer of Inx (AlyGai-y) _1-xP and a p-emitter layer of In_x (Al_yGa_1-y) _1-xP. x is typically 0.48 for the n-base and for the p-emitter. The n-base layer comprises a superlattice structure, or a superlattice structure is provided between the n-base layer and the p-emitter layer. Each repeating unit of the superlattice structure has quantum dots of InAs_ZP_1-Z where 0 ≤ z ≤ 1 with a strain balancing layer of In_x (Al_yGa_1-y) _1-xP. x and y in the strain balancing layer are selected to reduce the accumulated strain due to lattice mismatch between the quantum dots and one or more preceding layers of the device. The device therefore forms an intermediate band solar cell.

Description

INTERMEDIATE BAND SEMICONDUCTOR PHOTOVOLTAIC DEVICES, USES

THEREOF AND METHODS FOR THEIR MANUFACTURE

Field of the invention

The present invention relates to semiconductor photovoltaic devices, methods for making semiconductor photovoltaic devices and uses of semiconductor photovoltaic devices. The invention is of particular, but not exclusive, interest to semiconductor solar cell devices, e.g. devices based on In- Al-Ga-As-P semiconductor materials.

Background to the invention and related art Semiconductor photovoltaic cells are known for use in renewable power generation, both for terrestrial and non- terrestrial applications.

The principles of operation of semiconductor photovoltaic devices are well understood. A typical simple photovoltaic device consists of a p-n (or p-i-n) junction semiconductor diode in which the front surface metallization is

discontinuous in order to let light pass into the active device layers. Electrons and holes that are generated by the absorption of photons in the emitter and base regions, and that are separated by the built-in electric field that exists at the p-n junction, give rise to a potential difference between the output terminals of the diode.

One problem with using a single junction device is that its efficiency tends to be limited. In particular, the device can only make use of photons having an energy above a certain threshold, corresponding to the energy of the band gap at the junction. Photons with energy lower than this are wasted. Furthermore, the best use is not made of the additional energy of photons having energy significantly greater than the threshold.

Accordingly, it is also known to use devices having, for example, two or more junctions providing different band gap energies, in order to make more efficient use of the available photon energies (e.g. in the solar spectrum) .

Such devices are referred to as tandem junction devices. The junctions are connected in series, usually by tunnel diodes. It is known to use lattice matched InGaP on GaAs in tandem junction solar cells, such as double junction and triple junction solar cells, InGaP being used for its relatively high bandgap (E_G = 1.89 eV) to capture high energy photons in the solar spectrum. Where a Ge substrate is used (instead of a GaAs substrate), the Ge may form part of the third, and "lowest" junction. InGaP lattice matched to Ge tends to have a slightly lower bandgap than 1.89 eV. An alternative approach is set out in US-B-6, 44 , 897 (of the Universidad Politecnica de Madrid) in which the concept of the intermediate band solar cell is introduced. Such devices are considered to have the potential to be more efficient than single junction and tandem junction devices.

US-B-6, 444, 897 discloses a solar cell in which an

intermediate band material (referred to as the base) is located between two "ordinary" semiconductors (referred to as n-type and p-type emitters respectively) . This results in the formation of an intermediate band between the valence and conduction bands. This allows electronic transitions not only directly between the valence and conduction bands, but also indirectly, via the intermediate band. An

electronic transition from the valence band to the

intermediate band may be achieved via the absorption of a photon of energy lower than the valence-conduction band gap. A subsequent electronic transition from the intermediate band to the conduction band may be achieved via the

absorption of another photon of energy lower than the valence-conduction band gap. In this way, an electron can be promoted from the valence band to the conduction band via the absorption of relatively low energy photons. The maximum theoretical efficiency for the intermediate band solar cell in US-B-6, 44, 897 is estimated at 63.1%, which compares favourably with the maximum theoretical efficiency estimated for single junction devices (40.7%) and the maximum theoretical efficiency estimated for dual junction devices (55.4%) . The theoretical calculations for this are set out in Luque and Marti (1997) [ . Luque and A. Marti "Increasing the Efficiency of Ideal Solar Cells by Photon Induced Transitions at Intermediate Levels" Physical Review Letters, Vol. 78, pp. 5014-5017, 1997], which contains no details of a proposed materials system for an intermediate band solar cell. For the sake of completeness, the

theoretical values for efficiency for various structures under different concentrations of sunlight are:

One sun illumination: single junction 31%; dual junction 42%; triple junction 49%; infinite number of cells 68%.

Maximum concentration illumination: single junction 40.7%; dual junction 55%; triple junction 63%; infinite number of cells 86%.

This data is taken from A. De Vos, J. Phys D: Applied Phys Vol 13(5) 839 (1980) .

In US-B-6, 444, 897, it is proposed to form the intermediate band semiconductor using quantum dots of one material formed in a matrix of a different material. For example, US-B- 6,444,897 suggests that the quantum dots may be formed of GajjIni-x syPi-y and the matrix material may be Al_xGai-._xASySbi-y . x and y are between 0 and 1 [note that the use of x and y here is different from the use of x and y in relation to the present invention] . US-B-6, 44, 897 does not describe the substrate or other layers that should be used with these quantum dot layers .

US-B-6, 444, 897 also sets out an alternative materials system. In this system, the substrate is Sio.95Geo.05 doped with B with a lattice parameter close to that of Gao. 9 Ino.1P. A p-type lattice parameter adaptation layer of Gao. 9 I no.1P doped with Zn (thickness 3 micrometres) is formed on this substrate.

Next, a Gao. 9 I o.1P layer strongly doped with Zn is formed - this has a thickness of 2 micrometres. Then a Gao.9Ino.1P layer without Zn doping is formed - this has a thickness of 1 micrometer. Next, the intermediate band material is formed. This is Gao.5Geo.5P, of thickness 10 micrometers.

Next a 1 micrometre layer of Gao.5Geo.5P (without doping) is formed, in order to form a space charge zone of the front emitter. On this, a 2 micrometre layer of n-type Gao.9Ino.1P doped with Se is produced. A 0.5 micrometre window layer of AlN is then formed. A metallic grid of Au-Ge is formed through the window layer and an Al-Ag metallic contact is applied to the back of the substrate. Finally, an anti- reflective coating (e.g. T1O2 or ZnS-MgF2) is deposited over the window layer.

Marti et al (2008) [A. Marti, E. Antolin, E. Canovas, N.

Lopez, P.G. Linares, A. Luque, C.R. Stanley, CD. Farmer "Elements of the design and analysis of quantum-dot

intermediate band solar cells" Thin Solid Films, Vol. 516, pp6716-6722, 2008] discuss the design of quantum dot intermediate band solar cells. The devices disclosed are based on the InAs/GaAs system in view of this materials system being known for implementing semiconductor quantum dots. The limit of the number of quantum dot layers is said to be limited due to the build-up of strain in the layers otherwise causing the formation of dislocations. In the specific devices disclosed, this limit of the number of quantum dot layers is 10. In order to maximise the useful absorption of light using the small number of quantum dot layers that can be grown in this system, Marti et al (2008) devised the concept of field damping layers, in which an n- type layer is inserted between the p-type emitter and the quantum dot region, and an undoped layer is inserted between the quantum dot region and the n-type base (note that the term "base" here is used in the sense of the present invention discussed below, whereas Marti et al refer to this n-type layer as an emitter layer and the intermediate band layer as the base) . The purpose of the n-type layer between the p-type emitter and the quantum dot region is to sustain most of the junction build-in potential so that the quantum dots can be driven to a flat-band potential region. The purpose of the undoped layer between the quantum dot region and the n-type base is to prevent tunnelling from the n-type base conduction band to the intermediate band. The paper also notes that the efficiency of quantum dot intermediate band solar cells is likely to increase when illuminated with concentrated solar light.

D. Alonso-Alvarez et al (2008) [D. Alonso-Alvarez, A. G.

Taboada, J. M. Ripalda, B. Alen, Y. Gonzalez, L. Gonzalez, J. M. Garcia, F. Briones, A. Marti, A. Luque, A. M. Sanchez and S. I. Molina "Carrier recombination effects in strain

compensated quantum dot stacks embedded in solar cells"

Applied Physics Letters, Vol. 93, 123114, 2008] disclose the formation of a quantum dot intermediate band solar cell using 50 InAs/GaAs quantum dot layers with a stack period of 18 nm using GaP monolayers for strain compensation. The overall structure of the device was as follows: n-type GaAs substrate; back surface field layer (1 micrometre n-GaAs 2 x 10¹⁸ cm^"3) ; base layer (3 micrometre n-GaAs 5 x 10¹⁷ cm^-3) ; barrier layer (100 nm undoped GaAs); stacked quantum dot layers (explained in more detail below) ; field damping layer (170 nm n-GaAs 10¹⁷ cm^"3); emitter layer (900 nm p-GaAs

2 x 10¹⁸ cm^"3); AlGaAs window layer. Note that this document does not specify that there is a contact layer, but it is assumed that a p-GaAs contact could be provided. Each repeating unit of the stacked quantum dot layers was as follows: GaP monolayer; GaAs 1.53 nm; InAs quantum dots formed by 2 monolayers of InAs; GaAs 12.6 nm; GaP monolayer; GaAs 3.33 nm. It is considered likely that there is

included some Si doping in the GaAs layer formed over the InAs quantum dots. An alternative approach to strain compensation of InAs quantum dots is set out by Oshima et al (2006) [Ryuji Oshima, Takayuki Hashimoto, Hidemi Shigekawa and Yoshitaka Okada "Multiple stacking of self-assembled InAs quantum dots embedded by GaNAs strain compensating layers" Journal of Applied Physics, Vol. 100, 083110, 2006]. In this document, each quantum dot layer is covered by a 40 nm thick

GaNo.005Aso.995 strain compensation layer.

However, basing an intermediate band photovoltaic device on GaAs is not ideal, since the relatively low bandgap

(1.42 eV) of GaAs reduces the potential efficiency of the photovoltaic device.

Summary of the invention

The present inventor has realised that it is possible to use alternative materials systems for a quantum dot intermediate band photovoltaic device. Specifically, the present

inventor has realised that a photovoltaic device can be formed utilising In-As-P quantum dots in a superlattice structure. Furthermore, the inventor has realised that such a photovoltaic device can build on existing knowledge of known devices using a wide band gap p-n junction based on In-Al-Ga-P layers. Wide band gap junctions are known for their use in multi-junction photovoltaic devices, although typically these devices are n-on-p structures rather than p- on-n structures. The inventor has realised that such a wide gap junction can be modified by including In-As-P quantum dots in a superlattice structure. This constitutes a general aspect of the present invention.

According to the knowledge of the inventor at the time of writing, it is considered that the present disclosure represents the first realisation that In-As-P quantum dots can be utilised in a photovoltaic device. However, it is acknowledged that InP quantum dots per se, for example, are known in other types of device. Examples of relevant disclosures in this regard are set out below. Heller et al (2003) [R.D. Heller, G. Walter, N. Holonyak, D.T. Mathes, R. Hull and R.D. Dupuis "Low-threshold room temperature continuous-wave InP quantum dot coupled to InGaP quantum well heterostructure lasers grown by metalorganic chemical vapour deposition" I.E.E.E. International

Conference on Indium Phosphide and Related Materials, 12-16 May 2003 pp89 - 91, ISBN: 0-7803-7704-4.] disclose the formation of a semiconductor laser device incorporating InP self-assembled quantum dots embedded in Ino.49 (Al_xGai-_x) 0.51P · M.E. Pistol (2004) [M.E. Pistol "InP quantum dots in GalnP" Journal of Physics: Condensed Matter, Vol 16, ppS3737-S3748, 2004] sets out a summary of InP quantum dots grown on n-type GalnP by the Stranski-Krastanow technique.

We note here that although the quantum dots form at least intermediate energy levels between the conduction and valence bands of the photovoltaic devices according to preferred embodiments of the present invention, it is not yet clear that the formation of an intermediate band is essential to the operation of such' a device. Accordingly, in what follows, although the term "intermediate band" is used in order to retain the conventional wording, it is to be understood that the present invention may utilise one or more uncoupled intermediate energy levels rather than an intermediate band.

In a first preferred aspect, the present invention provides a photovoltaic device having an n-base layer of

In_x (AlyGai-y) ι-χΡ and a p-emitter layer of In_x (Al_yGai-_y) i-_xP, where x and y are independently selected for the base layer and for the emitter layer to satisfy 0 ≤ x < 1 and 0 ≤ y ≤ 1, wherein the n-base layer comprises a superlattice structure, or a superlattice structure is provided between the n-base layer and the p-emitter layer, optionally with one or more intervening layers, wherein each repeating unit of the superlattice structure has quantum dots of InAs_zPi_₂ where 0 ≤ z ≤ 1 with a strain balancing layer of In_x ( Al_yGai-_y) i-_xP, where x and y in the strain balancing layer are selected to reduce the accumulated strain due to lattice mismatch between the quantum dots and one or more preceding layers of the device .

In a second preferred aspect, there is provided a method for manufacturing a photovoltaic device, the method including the steps:

forming an n-base layer of In_x (Al_yGai__y) i-_xP, where x and y are independently selected for the n-base layer to satisfy 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1, wherein the n-base layer comprises a superlattice structure or a

superlattice structure is formed over the n-base layer; forming a p-emitter layer over the superlattice structure, optionally with one or more intervening layers between the superlattice structure and the p- emitter layer, the p-emitter layer being formed of In_x (Al_yGai-y) i-_xP, where x and y are independently selected for the p-emitter layer to satisfy 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1,

wherein each repeating unit of the superlattice structure is formed by depositing at least one layer of InAs_zPi-_z where 0 ≤ z ≤ 1 for forming quantum dots and by depositing at least one layer of In_x (Al_yGai-_y) i-_xP to form a strain balancing layer, where x and y are selected for the strain balancing layer to reduce the accumulated strain due to lattice mismatch between the quantum dots and one or more preceding layers of the device. In a third preferred aspect, there is provided a use of a photovoltaic device according to the first aspect to

generate electrical power when illuminated by sunlight.

Preferred and/or optional features of the present invention are set out below. These may be applied singly or in any combination with any aspect of the invention. In some preferred embodiments, the superlattice structure is provided between the n-base layer and the p-emitter layer, optionally with one or more intervening layers.

Where the superlattice structure is formed in the n-base layer, it is preferred that there is a part of the n-base layer which does not include any superlattice structure.

Preferably, the photovoltaic device is an intermediate band photovoltaic device, such as an intermediate band solar cell.

Each layer of the device is preferably an epitaxial layer. Accordingly, it is preferred that the device has a substrate which is a single crystal substrate. For example, the substrate may be GaAs (or based on GaAs) . Alternatively, the substrate may. be Ge (or based on Ge) . Still further, the substrate may be a silicon substrate with a sheet-like layer of Ge located over it. The layer of Ge may be separated from the silicon substrate by an insulating

dielectric layer such as silicon oxide or silicon nitride. The layer of Ge is typically not epitaxial with the Si substrate or intervening insulating layers but is

nevertheless monocrystalline to allow epitaxy for the device layers .

The substrate may be a Gei-_qSi_q metamorphic substrate.

Suitable substrates and substrate manufacturing techniques are disclosed in WO 2010/094920, the entire content of which is hereby incorporated by reference. For acceptable lattice matching of Gei-_qSi_q, it is suggested that the value of q should be below 0.06, more preferably below 0.04 and more preferably still in the range 0.01 to 0.03. The process described in WO 2010/094920 is an epitaxial lift-off process in which a thin layer of Gei-_qSi_q material is formed on a GaAs donor substrate and transferred to a receiver substrate. The receiver substrate can be, for example, a metal,

semiconductor, insulator, glass or combination of such materials.

The number of repeating units in the superlattice can be designated n. Preferably, n is at least 5. More preferably, n is at least 10, at least 20, at least 30 or at least 40. n may be at most 1000, but is typically lower than 200. The overall thickness of the superlattice structure may be at least 0.1 micrometres, more preferably at least 0.5 micrometres. The maximum thickness is not particularly limited but may be about 5 micrometres. For example, the maximum thickness may be about 3 micrometres.

The thickness of each repeating unit is typically at least 10 nm. This thickness affects the separation between adjacent quantum dot layers and so can affect the electronic coupling between the quantum dot layers. In turn, this can affect the formation of an intermediate band.

During manufacture of the device, in the deposition of the material (In-As-P) for the quantum dots, there may be deposited at least 1 monolayer of In-As-P. More preferably, there is deposited at least 1.5 monolayers of In-As-P. Up to 3.0 monolayers of In-As-P is considered to be suitable. Preferably, the quantum dots self-assemble via Stranski- Krastanow growth using MBE or MOVPE (MOCVD) .

In the embodiments described herein, the superlattice structure is formed over the n-base layer, with the p- emitter layer formed over the superlattice structure.

However, it will be understood that the superlattice structure could be formed over the p-emitter layer, with the n-base layer formed over the superlattice structure. In the In_x (Al_yGai-_y) i-_xP emitter and base layers, x and y are independently selected. The choice of x (i.e. the relative proportion of indium to (gallium + aluminium) affects the lattice parameter of the resultant material. Thus, the choice of x is an important factor in ensuring that each layer is lattice matched to its underlying layer. The choice of y, on the other hand (i.e. the relative proportion of aluminium to gallium) tends to have only a small effect on the lattice parameter of the resultant material. However, y has an important effect on the direct bandgap energy EG, and so y is chosen in order to provide a suitable bandgap energy EG of at least 1.5 eV, more preferably at least

1.55 eV, at least 1.6 eV, at least 1.65 eV, at least 1.7 eV, at least 1.75 eV, or at least 1.8 eV. For example, the band gap energy E_G of the material of the n-base layer and/or p- emitter layer may be at least 1.5 eV, more preferably in the range 1.7-2.1 eV. For example, a bandgap energy of about 1.85-1.95 eV is suitable. In_xGai-_xP is particularly suitable, having E_G = 1.89 eV (at room temperature) when x has a value (x=0.48) corresponding to In_xGai-_xP lattice matched to GaAs, x here being selected independently of x in the strain balancing layer in the superlattice structure. Note that all band gap values given herein are taken at room

temperature. According to the full theory, the optimum value of E_G for intermediate band solar cells is 1.95 eV.

Accordingly, in some embodiments, it is preferred to

engineer the value of E_G for the host semiconductor (typically this includes p-emitter, and the n-base within which the quantum dot superlattice is incorporated or on which the quantum dot superlattice is formed) . Al can be added (i.e. y may be non-zero, e.g. up to about 0.1) in order to increase EG whilst retaining lattice matching.

For the emitter and base layers, it is possible that x and y are selected independently. However, it is preferred for the emitter and base layers to utilise substantially

identical values for x and y.

In of the emitter or base layer (or in both of the emitter and base layers) , x may be 0, but is preferably at least 0.05, at least 0.1, at least 0.15, at least 0.2, at least 0.25, at least 0.3, at least 0.35, at least 0.4, at least 0.45, or at least 0.5. It is possible for the lower limit for x to be greater than this, for example at least 0.55, at least 0.6, at least 0.65, at least 0.7, at least 0.75, at least 0.8, at least 0.85, at least 0.9, at least 0.95, or x may even be 1 (although this is not preferred) . Furthermore, it is possible for x to be at most 1, more preferably at most 0.95, at most 0.9, at most 0.85, at most 0.8, at most 0.75, at most 0.7, at most 0.65, at most 0.6, at most 0.55, at most 0.5, at most 0.45, at most 0.4, at most 0.35, at most 0.3, at most 0.25, at most 0.2, at most 0.15, at most

0.1, or at most 0.05. In order to provide specific required properties for the emitter layer (e.g. a required lattice parameter) , x may lie in a range selected by combining any one of the above ranges for the lower limit for x with any one of the above ranges for the upper limit for x. For example, x may be in the range 0.45 to 0.55.

In one of the emitter or base layer (or in both of emitter and base layers) , y may be 0, but is preferably at least 0.05, at least 0.1, at least 0.15, at least 0.2, at least 0.25, at least 0.3, at least 0.35, at least 0.4, at least 0.45, or at least 0.5. It is possible for the lower limit for y to be greater than this, for example at least 0.55, at least 0.6, at least 0.65, at least 0.7, at least 0.75, at least 0.8, at least 0.85, at least 0.9, at least 0.95, or y may even be 1 (although this is not preferred) . Furthermore, it is possible for y to be at most 1, more preferably at most 0.95, at most 0.9, at most 0.85, at most 0.8, at most 0.75, at most 0.7, at most 0.65, at most 0.6, at most 0.55, at most 0.5, at most. 0.45, at most 0.4, at most 0.35, at most 0.3, at most 0.25, at most 0.2, at most 0.15, at most 0.1, or at most 0.05. In order to provide specific required properties for the emitter layer (e.g. a required value for E_G) , y may lie in a range selected by combining any one of the above ranges for the lower limit for y with any one of the above ranges for the upper limit for y.

Independently of the values of x and y selected for the n- base and the values of x and y selected for the p-emitter, the values of x and y selected for the In_x (Al_yGai-_y) i-_x? strain balancing layer are typically chosen in order to provide suitable strain balancing and EQ . It is to be noted that the expression In_x (Al_yGai-_y) i-_xP is used here for the sake of simplicity to describe the fractional composition of this quaternary alloy. A more accurate (but less commonly used in the art) description of the

proportions of InAlP and InGaP required to form the InAlGaP quaternary alloy lattice matched to GaAs is given by

(In_X'Ali-_X'P) (In_y'Gai_y'P) 1 where, in the preferred

composition, x'=0.46 and y'=0.48, and 0<z'<l.

In the strain balancing layer, x may be 0, but is preferably at least 0.05, at least 0.1, at least 0.15, at least 0.2, at least 0.25, at least 0.3, at least 0.35, at least 0.4, or at least 0.45. Furthermore, it is possible for x to be at most 0.5, at most 0.45, at most 0.4, at most 0.35, at most 0.3, at most 0.25, at most 0.2, at most 0.15, at most 0.1, or at most 0.05. Thus, in order to provide specific required properties for the strain balancing layer (e.g. a required lattice parameter) , x may lie in a range selected by combining any one of the above ranges for the lower limit for x with any one of the above ranges for the upper limit for x. At x=0.48, it is considered that In_x (Al_yGai-_y) i-_xP is lattice matched to GaAs. At x=0.49, it is considered that In_x (Al_yGai-_y) i-_xP is lattice matched to Ge . Thus, depending on the substrate, it is preferred that x is less than or equal to the lattice matching value (0.48 for GaAs or 0.49 for Ge) in order to provide strain balancing for the quantum dots. Preferably, x is at least 0.3. At x=0.3, the lattice constant of In_x (Al_yGai__y) i-_xP is lower than the GaAs lattice matching value, and so the strain balancing layer is in tension compared with GaAs. More preferably, x may lie in the range 0.4 < x < 0.5. It is considered that the lower the value of x, the thinner the layer needs to be to achieve a strain-balance. As is shown below, values of x in the range 0.4 < x < 0.5 provide a wide range of suitable strain balancing layer thickness. Indeed, for thick strain balancing layers, it is considered that the value of x becomes less important, since the contribution to the total amount of In in the superlattice repeat unit from the quantum dots becomes less significant and so the risk of strain accumulation reduces, provided that the strain balancing layer is close to being lattice matched (and preferably slightly in tension when lattice matched) with the underlying layer.

In the strain balancing layer, y may be 0, but is preferably at least 0.05, at least 0.1, at least 0.15, at least 0.2, at least 0.25, at least 0.3, at least 0.35, at least 0.4, at least 0.45, or at least 0.5. It is possible for the lower limit for y to be greater than this, for example at least 0.55, at least 0.6, at least 0.65, at least 0.7, at least 0.75, at least 0.8, at least 0.85, at least 0.9, at least 0.95, or y may even be 1 (although this is not preferred) . Furthermore, it is possible for y to be at most 1, more preferably at most 0.95, at most 0.9, at most 0.85, at most 0.8, at most 0.75, at most 0.7, at most 0.65, at most 0.6, at most 0.55, at most 0.5, at most 0.45, at most 0.4, at most 0.35, at most 0.3, at most 0.25, at most 0.2, at most 0.15, at most 0.1, or at most 0.05. In order to provide specific required properties for the strain balancing layer (e.g. a required value for EG), y may lie in a range

selected by combining any one of the above ranges for the lower limit for y with any one of the above ranges for the upper limit for y. For example, when the composition for the lattice matched strain balancing layer is determined, this composition may produce a "+ve" conductor band off-set with one or more other layers, and in that case a small concentration of Al could be introduced in compensation, by selecting an appropriate value for y. Another way to change E_G is to reduce the amount of In (i.e. by reducing x) . However, changing x affects the lattice parameter in a significant way. It is considered that the value of y to achieve the optimum band gap of 1.95 eV in In_x (Al_yGai-_y) i-_xP is about 0.1 (see Vurgaftman et al., J. Appl . Phys. Vol 89, pp5815-5875, 2001) . In each superlattice period, there may additionally be formed a capping layer. The capping layer may have a

composition substantially identical to the composition of the strain balancing layer, set out above. However, in some embodiments, the composition of the capping layer may be In_xGai-_xP, where x = 0.48.

In each superlattice period, there may also be included δ- doping, e.g. Si δ-doping. The purpose of δ-doping is to provide one charge carrier (e.g. electron) per quantum dot to half-fill the intermediate band. Preferably, therefore, the δ-doping density is set to equal the quantum dot density. This is typically an area density of 10¹⁰-10ⁿ cm^"2. For each repeating unit of the superlattice structure, the δ-doping layer is typically located 5-10 nm from the quantum dot layer .

Alternatively, the doping of the quantum dots may be

achieved via slab doping, i.e. a proportion of the strain balancing and/or cap layer in each superlattice period may be uniformly doped with an n-type dopant, e.g. Si, so that the area density of the n-type dopant within this thin layer, or slab of semiconductor, is substantially equal to the QD density. The volumetric density may be adjusted to achieve this aim, and the dopant may be supplied concurrent with the host atoms, unlike delta-doping where growth is interrupted and only the dopant atoms are supplied to the surface (plus the group V element (e.g. P) ) .

As stated above, the composition of the quantum dots is InAs_zPi-₂ where 0 ≤ z ≤ 1. z is selected in order to tune the energy gap (E_H) from the valence band to the

intermediate band and the energy gap (E_L) from the

intermediate band to the conduction band. z can be 0 (in some cases this is preferred) or z can be 1 (but this is not preferred) . More generally, in the quantum dots, z may be at least 0.05, at least 0.1, at least 0.15, at least 0.2, at least 0.25, at least 0.3, at least 0.35, at least 0.4, at least 0.45, at least 0.5, at least 0.55, at least 0.6, at least 0.65, at least 0.7, at least 0.75, at least 0.8, at least 0.85, at least 0.9, or at least 0.95. Furthermore, it is possible for z to be at most 1, more preferably at most 0.95, at most 0.9, at most 0.85, at most 0.8, at most 0.75, at most 0.7, at most 0.65, at most 0.6, at most 0.55, at most 0.5, at most 0.45, at most 0.4, at most 0.35, at most 0.3, at most 0.25, at most 0.2, at most 0.15, at most 0.1, or at most 0.05. In order to provide specific required properties for the quantum dots (e.g. a required value for E_H and/or E_L) , z may lie in a range selected by combining any one of the above ranges for the lower limit for z with any one of the above ranges for the upper limit for z. It is preferred that z is selected in order to provide a value for E_L which approaches the theoretical ideal value of 0.71 eV. For example, z may be selected to provide E_L in the range 0.5-0.9 eV, more preferably 0.6-0.9 eV, 0.7-0.9 eV, 0.6- 0.8 eV or 0.7-0.8 eV. In a fourth aspect, the present invention provides a photovoltaic device having an n-emitter layer of

In_x (Al_yGai-y) i-_xP and a p-emitter layer of In_x (Al_yGai-_y) i-_xP, where x and y are independently selected for each layer to satisfy 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1, a superlattice structure being provided between the n-emitter layer and p-emitter layer, optionally with one or more intervening layers, wherein each repeating unit of the superlattice structure has quantum dots of InAs_zPi__z where 0 ≤ z ≤ 1 with a strain balancing layer of In_x (Al_yGai-_y) i-_xP, where x and y in the strain balancing layer are selected to reduce the

accumulated strain due to lattice mismatch between the quantum dots and one or more preceding layers of the device.

The wording of the fourth aspect of the invention refers to an n-emitter layer and a p-emitter layer, in the style of US-B-6, 444, 897.

In a fifth preferred aspect, there is provided a method for manufacturing a photovoltaic device, the method including the steps : forming an n-type or p-type emitter layer of

In_x (Al_yGai__y) i-_xP, where x and y are independently

selected to satisfy 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1;

forming a superlattice structure over the emitter layer, optionally with one or more intervening layers between the superlattice structure and the emitter layer, wherein each repeating unit of the superlattice structure is formed by depositing at least one layer of InAs_zPi-_z where 0 ≤ z ≤ 1 for forming quantum dots and by depositing at least one layer of In_x (Al_yGai-_y) i__xP to form a strain balancing layer, where x and y are selected to reduce the accumulated strain due to lattice mismatch between the quantum dots and one or more preceding layers of the device. The wording of the fifth aspect refers to an n-emitter layer or a p-emitter layer, in the style of US-B-6, 4 , 897.

The fifth aspect allows, in some embodiments, the formation of the superlattice structure over a p-emitter layer. This allows the structure of the device to be grown using an epitaxial lift-off technique. In such a technique, parts of the device (e.g. at least the p-emitter layer and the

superlattice structure) are grown upside-down, these parts subsequently being transferred from a donor substrate to a receiver substrate. Optional features set out above with respect to the first, second and third aspects are applicable singly or in any combination with respect to the fourth and fifth aspects, with the proviso that references to the n-base layer with respect to the first, second and third aspects are to be understood as being applicable to the n-emitter layer of the fourth and fifth aspects and references to the p-emitter layer with respect to the first, second and third aspects are to be understood as being applicable to the p-emitter layer of the fourth and fifth aspects.

Further optional features of the invention are set out below.

Brief description of the drawings

Preferred embodiments of the invention are described below, by way of example and with reference to the drawings, in which :

Fig. 1 shows a schematic illustration of the promotion of charge carriers from the valence band to the conduction band, optionally via an intermediate band.

Fig. 2 shows a schematic illustration of the layers of a single junction, wide gap p-on-n photovoltaic cell based on In_x (Al_yGai-_y) P on a GaAs substrate.

Fig. 3 shows a schematic cross sectional view of the layers of a first embodiment of a device according to the present invention . Fig. 4 shows a schematic cross sectional view of the layers of a second embodiment of a device according to the present invention.

Fig. 5 shows a schematic illustration of the quantum dot layer plus cap layer (QD+cap) building block, used for the purposes of calculating the relationship between composition and thickness shown in Fig. 6.

Fig. 6 shows plots of the estimated In_xGai-_xP cap composition, x, against the thickness, c, of the cap layer (in nm) for various different thickness of InP QD layer (in ML) to maintain a strain balance.

Fig. 7 shows a plot of estimated cell efficiencies against energy of CB-IB transition (E_L) for a lattice matched

In_xGai__xP "host" with E_G = 1.89 eV (x=0.48) . This plot assumes that the intermediate band is ideally formed with the relevant E_L.

Fig. 8 shows, in a similar manner to Fig. 7, shows a plot of estimated cell efficiencies against energy of CB-IB

transition (E_L) for a lattice matched In_x (Al_yGai-_y) i-_xP "host" with E_G = 1.95 eV (x=0.48, y = 0.1) .

Detailed description of the preferred embodiments, further optional features of the invention The intermediate band solar cell (IBSC) was conceived by Luque and Marti of the Universidad Politecnica de Madrid (UPM) , as discussed above. It consists of a semiconductor p-n junction within which modified semiconductor is embedded such that an intermediate band (IB) is formed. This IB section is electrically isolated from the n-base and p- emitter of the solar cell (although Luque and Marti refer to these layers as the n-emitter and p-emitter, in a manner corresponding to the fourth and fifth aspects of the

invention) .

The purpose of the IB is to provide a "route" whereby photons with sub-bandgap energies within the solar spectrum, which would otherwise be wasted, may be absorbed by the semiconductor to create additional photocurrent via a two- photon absorption process . This is illustrated

schematically in Fig. 1. Photon 1 has an energy greater than the bandgap, E_G, and creates an electron-hole pair by the normal means. It requires the absorption of both photon 2 and photon 3 to produce another electron-hole pair. One electron is excited from the VB to the IB by photon 2, and photon 3 then excites another electron from the IB to the CB. The additional photocurrent is achieved in theory without degradation of the open-circuit voltage of the solar cell.

Efficiencies in excess of 60% under maximum concentration of light have been predicted for the device. However, note that the illumination used in such predictions is at

unrealistically high concentrations (e.g. 46,000 times) and using non-AMl.5D light (instead using ideal black body illumination with a temperature of about 5800K) . The modification of the semiconductor may be achieved in a number of different ways. One method which has been pursued for several years is to make QD-IBSCs based on InAs QDs in (Al)GaAs, as discussed above. Here, a number of factors means that the maximum theoretical efficiency is reduced to about 50%. Design constraints on the efficiency of QD-IBSCs have been discussed by A. Marti, L. Cuadra and A. Luque ("Design constraints of quantum-dot intermediate band solar cell", Physica E, 14, ppl50-157, (2002))

From theory, the optimum 300 bandgaps for an IBSC are as follows: E_G=1.95 eV, E_H=1.24 eV and E_L=0.71 eV. This means that the 1.42 eV bandgap of GaAs (the host for InAs-GaAs QD IBSC structures) is lower than the ideal value. It is further reduced by broadening of the QD VB states. Another drawback to using InAs-GaAs QD structures is that a large number of QD layers is needed to increase the absorption of sub-bandgap photons. However, the number of layers that can be grown in a stack is limited by the progressive

accumulation of strain which eventually leads to strain relaxation and a dramatic fall in the overall conversion efficiency of the QD-IBSC as dislocations thread into the p- emitter, forming non-radiative recombination centres. This is explained, for example, in Marti et al 2007 [A. Marti, N. Lopez, E. Antolin, E. Canovas, A. Luque, C. R. Stanley, C. D. Farmer, and P. Diaz "Emitter degradation in quantum dot intermediate band solar cells" Appl . Phys . Lett. 90, 233510 (2007) ]

One solution is to cap the QDs with a strain-balancing layer of, say, GaP or GaAsN as discussed above and then the total number of layers is essentially unlimited. An alternative is to "dilute" the strain by increasing the thickness of the capping layer. However, a consequence of this approach is that adjacent layers of QDs will be electronically uncoupled, and the formation of a band of (intermediate) energies is unlikely. There is still debate as to whether or not an IB is essential for the operation of the IBSC, but as discussed above, the present disclosure adopts the conventional

nomenclature of "intermediate band solar cell" even where intermediate bands are not formed by the arrangement of quantum dots .

In providing a more ideal materials system for QD-IBSCs, two factors are worth noting: (1) the bandgap of In_xGai-_xP

lattice matched to GaAs is about 1.89 eV at 300 K, which is very close to the optimum IBSC value; and (2) In_xGai-_xP is well developed as the wide-gap section of multi-junction tandem solar cells. The growth of InP QDs with In_xGax-xP capping layers on GaAs substrates has been studied for a number of years, as

mentioned above. However, to the inventor's knowledge, such materials have not been used for photovoltaic devices but instead for LEDs and lasers operating in the visible range. M.E. Pistol (2004) [M.E. Pistol "InP quantum dots in GalnP" Journal of Physics: Condensed Matter, Vol 16, ppS3737-S3748 , 2004] provides a useful summary of InP-InGaP QD growth and properties .

The present inventor considers that the value of EG is close to optimum with In_xGai-_xP, and that it is possible to tune EH and E_L towards the optimum values by appropriate selection of the composition of the quantum dots (i.e. appropriate selection of z in InAs_zPi-_z) . This is discussed, in a different context, by other workers. See, for example, Ribiero et al 2002 [E . Ribeiro, R. L. Maltez, . Carvalho, Jr., D. Ugarte, and G. Medeiros-Ribeiro "Optical and structural properties of InAsP ternary self-assembled quantum dots embedded in GaAs" Appl . Phys . Lett. Vol 81 No. 16, 2002, p. 2953] and Miyake et al 2006 [S. Miyake, W. S. Lee, T. Ujihara and Y. Takeda "The wideband light emission around 800 nm from ternary InAsP quantum dots with an intentionally broadened size and composition distribution", 2006 IEEE Int. Conf. on Indium Phosphide and Related

Materials, p. 208] . Miyake et al 2006 fabricated InP and InAsP quantum dots on InGaP lattice matched to GaAs (001) .

The present inventor here sets out two alternative

structures for the preferred photovoltaic devices of the present invention. These are based on the growth of

InAs₂Pi-_z QDs with In_x (Al_yGai__y) i__xP strain-balancing capping layers. The structures are in essence wide-gap solar cell structures but with the normal (p/n) -base region of uniform composition and doping level replaced by a QD superlattice (SL) .

Fig. 2 shows a schematic cross sectional view of a p-on-n adaptation of a wide bandgap solar cell. This device itself is not necessarily part of the prior art. It is presented here not as an embodiment of the invention, but as a basis from which the invention was developed, and in order to assist in understanding the invention. The bandgap EG of the device of Fig. 2 is about 1.89 eV. We note here that on single junction and dual junction cells in the late 1980' s used the p-on-n configuration. However, at the time of writing (and for many years prior to the time of writing) , the n-on-p structure is the widely preferred choice, partly because of the way that the lowest, Ge junction in a triple junction device is usually made.

The device of Fig. 2 has an n⁺ GaAs substrate 10 and, in order from the substrate: an n-GaAs buffer layer 12; an n- In_x (Al , Ga) i-_xP back surface field layer 14 with x=0.48; an n- In_xGai-_xP base layer 16 with x=0.48; a p-In_xGa_!__xP emitter layer 18 with x=0.48; a p-In_x (Al, Ga) i-_xP window layer 20 with x=0.48; and a p⁺-Ga(In)As contact layer 22. Fig. 3 shows a schematic cross sectional view of the layers of a first embodiment of a device according to the present invention, where a proportion or all of the n-base layer of the device of Fig. 2 is replaced by the superlattice structure. E_G in this example is about 1.89 eV.

The embodiment of Fig. 3 has: an n⁺-GaAs substrate 30; n- GaAs buffer layer 32; n-In_x (Al_yGai-_y) i__xP back surface field layer 34 (x=0.48 in this example) which is lattice matched (or nearly lattice matched) to GaAs for (0 < y < 1); n- In_xGai-_xP base layer 36 (x=0.48 in this example); repeating quantum dot layers forming a superlattice (described in more detail below); p-In_xGai-_xP emitter layer 46 (x=0.48 in this example); p-In_x ( l_yGai-_y) i-_xP top window layer 48 (x=0.48 in this example, and y=1.0); p⁺-Ga(In)As contact layer 50. Not shown are the metallic contacts to the substrate and anti- reflective coating layers over the top window layer. The window layer 48 may be modified to be formed with a high-Al content window layer of Al_xGai-_xAs, with optional etch stop layers, as described and defined in WO2009/04 171, the content of which is hereby incorporated by reference in its entirety. This is preferred where the blue/UV

transparency and/or electrical resistivity of the p-Al_xGai- _xAs window layer is superior to that of a p-In_x (Al_yGai-_y) i-_xP window layer. Each repeating unit of the quantum dot superlattice in the embodiment of Fig. 3 is formed as follows: wetting layer and In_zAsi-_zP quantum dots 38; In_x (Al_yGai-_y) i-_xP (x<0.48, y=0) strain balancing layer 40; Si δ-doping layer 42 (or

uniformly n-doped In(Al,Ga)P layer); In_x (Al_yGai-_y) i-_xP (x=0.48, y=0) cap/spacer layer 44. For each repeating unit, the sum of the thickness of the strain balancing layer and the cap layer is typically at least 10 nm and typically at most 100 nm.

The self-assembled In_zAsi__zP QDs are formed by Stranski- Krastanow growth by MBE or MOVPE ( OCVD) . The amount of In_zAsi__zP to be deposited into the QDs is at least 1

monolayer (ML) and preferably 1-10 monolayers (MLs) , typically 1-5 MLs.

The strain associated with the QDs/wetting layer is relieved by the strain-balancing layer 40 of In_x (Al_yGa_!-y) j-_xP (in which y=0 in this embodiment) . The value of x is preferably less than 0.48 (the composition for lattice matching to GaAs) and can be as low as 0.4 so that the lattice constant of the layer is less than the lattice matching value and it is therefore in tension. The lower the value of x, the thinner the layer needs to be to achieve a strain-balance. This layer has a bandgap slightly wider than that of lattice matched In_xGai-_xP. The size of the QDs when capped with In_x (Al_yGai-_y) i__xP (again with y=0 in this embodiment) may differ from uncapped QDs. The area density of the n-type dopant used for δ-doping, for example Si or Te, is set to equal the QD density, i.e. in the range 10¹⁰-10ⁿ cm^-2, and thus provide the one electron per QD to half-fill the IB. The δ-doping should be

separated from the QDs by a thin, undoped layer of

In_x (Al_yGai-y) i-_xP with a thickness which ensures complete transfer of electrons from the n-type dopants into the QDs by modulation doping. A thin In_x (Al_yGai-_y) i-_xP supply layer, uniformly doped n-type can also be used, also separated from the QDs by a thin, undoped layer of In_x (Al_yGai__y) i-_xP .

The capping layer of In_x (Al_yGai-_y) i-_xP (x=0.48, y=0) is lattice matched. Its thickness is adjusted to produce coupling between adjacent QD layers. Alternatively, it can be

increased in thickness to minimise or eliminate the coupling, as required. The total thickness of layers 40 and 44 is typically at least lOnm.

The number of periods n in the superlattice is dependent on the QD absorption coefficients. n is typically at least 10. n may for example be in the range 10-300 or 10-100. Fig. 4 shows a schematic cross sectional view of the layers of a second embodiment of a device according to the present invention. E_G in this example is about 1.95 eV. The embodiment of Fig. 4 has a structure similar to that of the embodiment of Fig. 3. Therefore many of the reference numerals used are the same and these features are not described again here. However, the quantum dot superlattice is modified as follows.

The In_x (Al_yGai-_y) i-_xP (x<0.48, y=0.1) layer 60 is used as a single strain-balancing and cap/spacer layer. The Si δ- doping layer 62 (or uniformly n-doped In(Al,Ga)P layer) is located in this layer, e.g. within 5-10nm of the InP quantum dots 64. The thickness and/or composition of the layer is adjustable to achieve an overall strain-balance either with or without electronic coupling between the QD layers.

Processing of the epitaxial structures into solar cell die is similar to the steps undertaken with known In_x (Al_yGai__y) i-_xP structures .

It is of interest here to provide estimates of the required In_x (Al_yGai-_y) i__xP cap thickness and composition needed to maintain strain balance in a multi-layer In_x ( Al_yGai-_y) i-_xP - InAs_zPi-₂ QD stack, when considering also the thickness of the InAs_zPi-2 QD layer. The inventor has estimated the thickness and composition of In_x (Al_yGai-_y) i-_xP required to cap InP quantum dots (i.e. z=0) whilst maintaining an overall balance in the system strain, as the amount of InP deposited into the QDs is increased from 1.5 ML to 3.0 ML. The thickness of 1 ML is about 0.283 nm (one half the lattice constant at 300K of GaAs) . The proportions of InP:GaP, are in a lattice matched ratio of 0.48:0.52. The assumption is that the structures are deposited on and lattice matched to GaAs substrates. The QD+cap building block, to be repeated as many times as is practical, is shown in Fig. 5. The thickness of InP is n ML and the cap thickness is c ML. For a strain balance, it is estimated that there should be 0.48 x ML (InP) for every 0.52 x ML(GaP) .

In the following discussion, the parameter y is used to denote the amount of In relative to the amount of Ga . Here, y does not relate to an amount of Al . For a In_yGai-_yP cap of thickness c ML and compositional fraction y, the amounts of InP and GaP are yc ML and (l-y)c ML, respectively. The total amount of InP in the structure is n + cy ML. For lattice matched proportions of InP and GaP overall in the QD+cap structure,

. °-⁴⁸

(l - y)c 0.52

This expression can be re-arranged for compositional fraction y and thickness c:

and

0,52 x n

C ~ { AS - y)

The information is most usefully shown in a plot of In_yGai-_yP cap composition, shown in Fig. 6, in which y is plotted against thickness c of the cap (given in units of nm in the graph) . In effect, Fig. 6 shows the composition of the cap layer against cap thickness to maintain a strain balance for different InP guantum dot layer thickness.

The legend in Fig. 6 indicates the relative ordering of the plots in relation to the thickness of the InP QD layers. For example, the highest curve on the graph is for an InP QD layer of thickness 1.5 ML. As the thickness of the QDs increases, at fixed cap layer thickness, so the amount of In that can be incorporated into the cap layer must decrease.

The plots of Fig. 6 show that a wide variation in the cap thickness (and hence coupling between InP QD layers) can be accommodated simply by adjusting the cap composition by relatively small amounts. This is a desirable feature since it means that the bandgap of the cap will be very similar to those of the p-emitter layer and n-base layer in the complete QD-IBSC structure. It is next of interest to consider estimates of the

efficiency of QD-IBSCs according to embodiments of the present invention. The efficiencies of QD-IBSCs based on two band gaps for

In_x (Al_yGai-_y) i-_xP have been analysed:

(i) InP QDs in In_x (Al_yGai__y) i__xP lattice matched to GaAs (x = 0.48, y = 0, host E_G of about 1.89 eV)

(ii) InP QDs in In_x (Al_yGai-_y ) i__xP with a composition adjusted to give the optimum bandgap for an IBSC of 1.95 eV (x = 0.48, y of about 0.1)

The effect of absorption on the solar spectrum by a 30nm thick 90% AlGaAs window layer is included in the present analysis (note that it is not apparent that the photovoltaic window is usually included in such calculations in the art) . Note that the solar irradiance is for one square metre of area, and it is normalised to a total of 1000 W/m². The total current generated by an IBSC comprises two

components: (a) the "normal" current produced by the host semiconductor through the absorption of above bandgap

photons (hf > E_G ) , and (b) an additional current resulting from the creation of an electron-hole pair due to the

absorption of two sub-bandgap photons mediated by the

intermediate band (IB) . The additional current and hence the efficiency of an IBSC is at a maximum when the excitation rates of electrons from the valence band to the intermediate band (VB-IB) and from the intermediate band to the conduction band (IB-CB) are equal. When these two rates are unequal, the additional current produced by sub-bandgap photons is limited by the lower of the two (this is consistent with the circuit model for an IBSC - see Marti et al 2004 [A. Luque and A. Marti, C. Stanley, N. Lopez, L. Cuadra, D. Zhou, J. L. Pearson and A. McKee "General equivalent circuit for intermediate band devices: Potentials, currents and electroluminescence" J.

Appl. Phys. Vol. 96, No. 4, p. 903, 2004]) . Further, the calculations of efficiency assume there is no overlap of the VB-IB and IB-CB absorption coefficients (e.g. a photon which is capable of exciting an electron from the VB to the IB cannot excite an electron from the IB into a state well above the CB minimum) . The absorption processes have unity quantum efficiency. Assumptions are also made about the open circuit voltages under 1 sun and concentrated

illumination, namely the presence of the QDs does not lead to voltage degradation. The values for the open circuit voltage V_oc are those which a cell manufactured from

unmodified host material would be expected to generate (e.g. see the structure of Fig. 2) . V_oc under 1 sun illumination is taken as about 0.7 times E_G, the value increasing to about 0.85 times E_G under high concentration illumination. Voc is fixed throughout the calculations, although this will probably not be the case in practice since the short circuit current varies with the energy separation, E_L, between the CB and IB. The final parameter to be set for the calculation of

efficiency is the fill factor, ff. This has been fixed at 0.85 although it could in practice be marginally higher (e.g. 0.87-0.88) . Notwithstanding the numerous simplifications made, the results are close to those derived from the "full" Luque and Marti theory. There are some interesting and potentially important outcomes. The inventor has found that the overall efficiency shows little variation once the thickness of the host semiconductor exceeds about 1.0 micrometre, although small adjustments to E_L are required to maintain an optimum. Therefore Figs. 7 and 8 show results for a thickness of the superlattice structure plus n-base layer of 1.5 micrometre. This is a desirable outcome since it means that the

thickness of the QD section of the IBSC structure is not restricted. Also, the maximum efficiency is not too

sensitive to the value of E_L.

(i) InP QDs in In_x (Al_vGai-_v) i-_xP lattice matched to GaAs (x = 0.48, y = 0, host E_G of about 1.89 eV) In this estimate, which is an embodiment of the present invention, it is noted here that E_G = 1.89 eV is slightly smaller than optimum for an ideal IBSC. The absorption data for Al₀.36Gao.64As is used here in order to model the In_xGai-_xP used to provide E_G = 1.89 eV, being the nearest equivalent since absorption data for In_xGai-_xP was not available at the time of writing.

Fig. 7 shows a plot of estimated intermediate band solar cell efficiencies (illumination due to unconcentrated sunlight ("1 sun"), and illumination due to concentrated sunlight ("concentration"), against energy of CB-IB

transition (E_L) for a lattice matched In_xGai-_xP "host". We note here that E_L may be varied by varying the value of z in the InAs_zPi-z QDs . A thickness of 1.5 micrometres lattice matched In_xGai-_xP is used. Included in this value is the overall thickness of the superlattice structure and the thickness of the p-emitter (whose thickness is 0.5-1.0 micrometres) . Note that the n-base region is part of the superlattice structure, whose thickness is 0.5-1.5

micrometres .

Voc values are based on data given as follows:

1.3 V for 1 sun [see J. van Deelen et al., J. Crystal Growth Vol 298, pp772-776, 2007, which explores the performance of a single junction, wide gap InGaP cell.] 1.6 V under concentration [see I. Garcia et al., Appl . Phys . Lett, Vol. 94, p053509, 2009. 2.76 V was measured for a dual junction cell at xlOOO suns, and it is reasonable to interpolate a contribution of about 1.6 V from the wide gap In_xGai__xP section.]

The inventor has found that the maximum efficiencies under 1 sun and concentrated sunlight are insensitive to the total thickness of In_xGai-_xP in the photovoltaic structure for the thicknesses considered. A further point to note is that, despite the somewhat simplified nature of the calculations, the optimum value of E_L is about 0.68 eV, close the 0.71 eV derived from the rigorous theory of Luque and Marti for the optimum bandgap of 1.95 eV.

(ii) InP QDs in In_x (Al_yGai__y) i__xP with a composition adjusted to give the optimum bandgap for an IBSC of 1.95 eV (x = 0.48, y of about 0.1) In this estimate, which is also an embodiment of the present invention, it is noted here that the absorption coefficients for 41% AlGaAs are used. This is the nearest equivalent to In_x (AlyGai-_y) i-_xP, since it has a bandgap close to 1.95 eV.

Absorption data for In_x (Al_yGai-_y) i-_xP is not reported in the literature and is not available at the time of writing. Fig. 8 shows a plot of estimated cell efficiencies (1 sun and cone.) against energy of CB-IB transition ( E_L ) for an In_x (Al_yGai-_y) i-xP "host", in a similar manner to Fig. 7. Voc values:

1.37 V for 1 sun

1.6 V under concentration

The maximum efficiencies under 1 sun and concentrated sunlight shown in Fig. 8 are only fractions of a percent different from those estimated for lattice-matched In_xGai-_xP. The optimum value of E_L is about 0.81 eV, fractionally larger than 0.71 eV derived from the rigorous theory of Luque and Marti for the optimum bandgap of 1.95 eV.

The preferred embodiments of the invention have been described by way of example. On reading this disclosure, modifications of these embodiments, further embodiments and modifications thereof will be apparent to the skilled person and as such are within the scope of the present invention.

In particular, the incorporation of suitable fractions of As into the InP quantum dots described above will be apparent · in order to tune EL -

Claims

1. A photovoltaic device having an n-base layer of

In_x (Al_yGai-y) i-_xP and a p-emitter layer of In_x (Al_yGai-_y) i-_xP, where x and y are independently selected for the base layer and for the emitter layer to satisfy 0 ≤ x ≤ 1 and 0 ≤ y < wherein the n-base layer comprises a superlattice structure or a superlattice structure is provided between the n-base layer and the p-emitter layer, optionally with one or more intervening layers, wherein each repeating unit of the superlattice structure has quantum dots of InAs_zPi-_z where

0 ≤ z ≤ 1 with a strain balancing layer of In_x (Al_yGai-_y) i-_xP, where x and y in the strain balancing layer are selected to reduce the accumulated strain due to lattice mismatch between the quantum dots and one or more preceding layers o the device .

2. A device according to claim 1 wherein the device has a single crystal substrate selected from GaAs, Ge or Si or

Gei-_qSi_q.

3. A device according to claim 2 wherein when the substrate is Si, there is located a sheet-like layer of Ge separated from the Si substrate by a layer of insulating dielectric such as silicon oxide or silicon nitride.

4. A device according to any one of claims 1 to 3 wherein the number of repeating units in the superlattice is designated n, and n is at least 5.

5. A device according to any one of claims 1 to 4 wherein the overall thickness of the superlattice structure is at least 0.1 micrometres.

6. A device according to any one of claims 1 to 5 wherein the thickness of each repeating unit is at least 10 nm.

7. A device according to any one of claims 1 to 6 wherein in at least one of the In_x (Al_yGai-_y) j__xP emitter and base layers, x is selected in order to provide substantial lattice matching to the underlying layer.

A device according to any one of claims 1 to 7 wherein selected to be in the range 0.40 to 0.50.

9. A device according to any one of claims 1 to 8 wherein in at least one of the In_x (Al_yGai-_y) i__xP emitter and base layers, y is at least 0.01, more preferably about 0.1.

10. A device according to any one of claims 1 to 9 wherein z is selected in order to tune the energy gap (E_H) from the valence band to the intermediate band and the energy gap (E_L) from the intermediate band to the conduction band.

11. Use of a photovoltaic device according to any one of claims 1 to 10 to generate electrical power when illuminated by sunlight.

12. A method for manufacturing a photovoltaic device, the method including the steps:

forming an n-base layer of In_x (Al_yGai-_y) i-_xP, where x and y are independently selected for the n-base layer to satisfy 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1, wherein the n-base layer comprises a superlattice structure or a

superlattice structure is formed over the n-base layer; forming a p-emitter layer over the superlattice structure, optionally with one or more intervening layers between the superlattice structure and the p- emitter layer, the p-emitter layer being formed of In_x (Al_yGai__y) i-_xP, where x and y are independently selected for the p-emitter layer to satisfy 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1,

wherein each repeating unit of the superlattice structure is formed by depositing at least one layer of InAs_zPi-_z where 0 ≤ z ≤ 1 for forming quantum dots and by depositing at least one layer of In_x (Al_yGai-_y) i-_xP to form a strain balancing layer, where x and y are selected for the strain balancing layer to reduce the accumulated strain due to lattice mismatch between the quantum dots and one or more preceding layers of the device.

13. A method according to claim 12 wherein in the

deposition of the material InAs_zPi-_z for the quantum dots, there is deposited at least 1 monolayer of InAs_zPi-₂.

14. A method according to claim 13 wherein the quantum dots self-assemble via Stranski-Krastanow growth.