FIELD AND BACKGROUND OF THE INVENTION

The present invention relates to a photovoltaic or photochemical cell

and, more particularly, to a photovoltaic or photochemical cell made of up

of nanocrystals.

All photovoltaic (PV) cells depend on two principles for their

operation. First, the creation of charges (e-/h+ formation) by photon

absorption. Next, the separation in space by a built-in electric field (the

space charge layer). The first step requires only a semiconductor with a

suitable bandgap. The second step requires a built-in electric field in the

semiconductor. This is formed by generation of a junction which may be

produced in various ways as follows.

One way applies for a p-n junction cell which is used in most types

of photovoltaic cells. Here, either the surface is doped to make it the

opposite conductivity type of the bulk semiconductor (e.g., p-n Si cells) or

a layer of opposite conductivity type, and usually a different

semiconductor, is deposited onto the first semiconductor forming a

heterojunction cell, such as CdS/CdTe.

Another way involves the deposition of a metal onto a

semiconductor of one conductivity type. A careful choice of both metal

and deposition method is needed in these Schottky cells.

A third way calls for the immersion of the semiconductor in an

electrolyte, which is usually a liquid but may be a solid, forming a

photoelectrochemical cell (PEC). A liquid electrolyte has the advantages

of ease of formation and good contact between electrolyte and

semiconductor, but has potential problems in terms of stability and sealing.

Electron/hole recombination in the space charge region is the cause

for loss of quantum efficiency, and can occur by bulk and surface

recombination. The latter is particularly important for polycrystalline

semiconductors where the number of grain boundaries in the direction of

charge transport is inversely proportional to crystal size. For this reason,

polycrystalline films are sought with crystallite size as large as possible,

usually greater than 1 micron.

Semiconductor films having small crystal size are much easier and

cheaper to produce since, for example, they eliminate the high costs

involved with using energy-consuming high temperatures. However, they

normally exhibit prohibitively large recombination losses in photovoltaic

cells. Small crystal size and low recombination losses are normally

considered to be mutually exclusive properties.

Photochemical reactions on semiconductor colloids do show high

quantum efficiencies in many cases, and here the small size is an

advantage, since bulk recombination is negligible. See J. Moser and M.

Gratzel, Helv. Chim. Acta, 65, 1436 (1982), which is incorporated by

reference as if fully set forth herein. However, each colloidal crystal

behaves as a complete 'cell' with both oxidizing and reducing reactions

occurring on the same crystal. They cannot be used to produce electricity

as in a regular PEC, except in the case of slurry electrodes for

semiconductor suspensions which exhibit low quantum efficiencies since,

according to the mechanism postulated for these systems, the charged

crystallites must make electrical contact with a collection grid before

recombination occurs. See W. W. Dunn, Y. Kikawa and A. J. Bard, J.

Am. Chem. Soc, 103, 3456 (1981), which is incorporated by reference as

if fully set forth herein. Also, semiconductor colloids used for

photochemical reactions are normally dispersed as a dilute phase in a

liquid.

SUMMARY OF THE INVENTION

According to the present invention there is provided a cell,

comprising: a substrate and a nanocrystalline semiconductor layer located

on one or both sides of said substrate, said layer having thickness

calculated to minimize recombination losses.

According to further features in preferred embodiments of the

invention described below, especially where the device is used for the

conversion of light into electricity, the conductive substrate is conductive.

The substrate may be made of many suitable materials, such as conductive

glass, and the like.

According to still further features in the described preferred

embodiments, the semiconductor layer is porous and has a thickness of

from about 100 nm to about 1000 nm and, preferably, from about 200 nm

to about 500 nm. The nanocrystalline semiconductor layer is made of

crystals of average size of from about 3 to about 100 nm, preferably from

about 4 to about 30 nm, most preferably from about 5 to about 20 nm.

According to another embodiment the cell also includes a second

nanocrystalline semiconductor layer located on the other side of the

substrate.

One embodiment of a device according to the present invention

successfully addresses the shortcomings of the presently known

configurations by providing a nanocrystalline photovoltaic cell which

features significantly lower recombination losses than has been heretofore

possible.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention is herein described, by way of example only, with

reference to the accompanying drawings, wherein:

FIG. 1 is a photoresponse curve (normalized so that each spectrum

gives approximately the same peak response) as a function of wavelength

for a PEC using a front-wall illuminated, chemically deposited CdSe film

grown at 30°C on conducting glass under illumination from a quartz-iodine

lamp (approximately equivalent to AM2 illumination) on conducting glass

for difference thicknesses of CdSe;

FIG. 2 is as described with regards to Figure 1 but for back-wall

illumination;

FIG. 3 shows the I-N characteristics under solar illumination (920

W/m ² ) of a 0.9 cm ² CdSe film (deposited on both sides of double-sided

conducting glass at 30°C under illumination during deposition; the total

thickness of CdSe is about 120 nm) in a liquid junction photovoltaic cell

configuration using an electrolyte of 1 M Νa ₂ S and 0.1 M S in water. The

solid line indicates the illuminated I-V plot; the broken line indicates the

dark I-V; the inset shows the transmission spectrum of the film;

FIG. 4 is a schematic model of a porous nanocrystalline film

showing electrolyte contact with individual crystallites; illumination is

shown to produce an electron/hole pair in one crystallite; the hole transfers

to the electrolyte and the electron is shown traversing several crystallites

before reaching the substrate;

FIG. 5 is a plot showing the absorption of blue and red light by a

thin film in both front-wall (FW) and back-wall (BW) modes; the use of

blue and red light were selected to illustrate long and short wavelength

light, respectively;

FIG. 6 is as in Figure 5 but using a relatively thick film.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

, The present invention is of a photovoltaic or photochemical cell

which uses a semiconductor film made up of nanocrystals and yet exhibits

low recombination losses. Throughout, the term 'cell' is used to indicate

a device for converting light energy into either electrical or chemical

energy, or which uses light energy to catalyze a chemical reaction.

Porous nanocrystalline semiconductor films of CdS and CdSe can

be used as photoelectrodes in photovoltaic cells with relatively low

recombination losses. Spectral response measurements have been used to

show how the recombination losses depend on film thickness. These

nanocrystalline photovoltaic cells are shown to operate due to charge

separation at the semiconductor-electrolyte interface rather than by a

built-in space charge layer as normally occurs in photovoltaic or

photoelectrochemical cells. The rapid removal of one charge by the

electrolyte explains the low recombination loss.

The electrolyte used may be either liquid or solid. When the

electrolyte is solid, the electrolyte is preferably applied to the

nanocrystalline layer as a liquid, which subsequently solidifies. An

example of this would be the use of polymer electrolytes which are formed

from liquid precursors and subsequently polymerized. See M. Morita, T.

Fukumasa, M. Motoda, W. Tsutsumi, Y. Matsuda, T. Takahashi and H.

Ashitaka, J. Electrochem. Soc, 137, 3401 (1990), and references cited

therein, which is incorporated by reference as if fully set forth herein.

The principles underlying a cell according to the present invention

can be better understood with reference to the accompanying Figures.

Without in any way limiting the scope of the present invention, the

mechanism which gives the photovoltaic cells according to the present

invention their unique and desirable characteristics, and the manifestation

of the mechanism in measurable phenomena, are discussed below, after an

example of the preparation of illustrative photovoltaic cells according to the

present invention.

It is to be noted that in all photovoltaic cells light is absorbed in the

semiconductor. This is technologically distinct from photogalvanic cells

wherein light is absorbed by a sensitizer which then transfers charge to a

semiconductor. Recent advances in the latter technology have been made

whereby the underlying semiconductor layer is highly porous. The porous

semiconductor layers described in N. Vlachopoulous, P. Liska, J.

Augustynski and M. Gratzel, J.A.C.S., JJO, 1216 (1988), which is

incorporated by reference as if fully set forth herein, may be used as

substrate in the devices according to the present invention.

The semiconductor films described here may be used also for

photochemical cells, where a chemical reaction is driven or catalyzed by

light. In such cells, the substrate need not be electrically conducting and

may even advantageously be porous, allowing flow of solution through the

substrate. As for the photovoltaic cells described herein, the

nanocrystallinity of the semiconductor deposited on, or distributed through,

the substrate is of critical importance to the present invention.

According to the present invention, porous, nanocrystalline

semiconductor films can be used as photoelectrodes with relatively low

recombination losses, in spite of the nanocrystalline structure, i.e., crystal

size on the order of 6 nm. Spectral response measurements show that

above a certain layer thickness, which depends on crystallite size,

recombination increases, and this increase depends strongly on the direction

of illumination, whether it is front-wall or back-wall.

Charge separation in these photoelectrodes takes place at the

semiconductor-electrolyte interface which are found throughout the

thickness of the porous film, rather than by a built-in space charge field.

The relatively low recombination losses are due to the rapid removal of

one of the charges or holes allowing electrons to pass through a number of

grain boundaries with a small chance of meeting a hole on its journey to

the substrate. In some cases, the films show 'p-type' photoresponse

behavior as would be expected if the direction of current flow were

determined by the semiconductor surface/electrolyte boundary rather than

bv a field in the bulk of the semiconductor.

EXAMPLES

A film according to the present invention was prepared by

electrodepositing CdS according to the method of Baranski et al. (A.S.

Baranski, W.R. Fawcett, A.C. McDonald, R.M. de Nobriga and J.R.

McDonald, J. Electochem. Soc. 128, 963 (1981)), which is incorporated by

reference as if fully set forth herein, from a DMSO solution of CdCl ₂ (0.05

M) and S (0.1 M) under a constant current density of 1 mA cm ^"2 onto a Ti

substrate at 120°C. The CdSe film electrode was immersed in an aqueous

solution of 1 M each KOH, Na ₂ S and S. A second electrode of Pt was

immersed in the same solution. When illuminated with tungsten-halogen

light of 100 mW/cm ² intensity, the photovoltage between the two

electrodes was measured to be 0.26 V and the short circuit photocurrent

was 0.19 mA/cm ² .

In another example, CdSe was chemically deposited onto SnO ₂

coated glass (<10Ω) from an aqueous solution containing 80 mM CdS0 ₄

nitriloacetate (KNTA) and 80 mM sodium selenosulphate. The Cd NTA

complex solution was adjusted to a pH of 9.5 (±1.5) with KOH solution.

This solution and the Na ₂ S0 ₃ solution (made by dissolving 0.2 M black Se

in 0.4 M Na ₂ SO ₃ for several hours at about 60°C) were brought to the

desired temperature and then mixed.

The substrates were put into the solution, which was then placed in

a thermostatic bath under illumination from an ELH lamp roughly

equivalent to sunlight (AM2). Deposition occurred over several hours and

the film was removed from the bath when a total thickness of 120 nm (60

nm per side) was reached. The film was made into a liquid junction

photovoltaic cell as in the first example, but using an electrolyte 1 M in

Na ₂ S and 0.1 M in S. When illuminated in sunlight (920 W/m ² ), an open

circuit voltage of 0.47 V and a short circuit current of 1.45 mA/cm ² was

generated.

To understand the results described above, it is convenient to start

with the model of the junction between the nanocrystalline semiconductor

layer and the electrolyte junction described previously, as shown in Figure

4. A photon is absorbed by one nanocrystal, forming an

electron hole pair. Since the direction of photocurrent flow is, in most

cases, that of an n-type semiconductor, this means that the hole goes into

solution where it oxidizes an electron donor, while the electron reaches the

back contact, typically a conducting glass. Clearly, the farther the

electron/hole pair is generated from the back contact, the greater the chance

of the pair being lost by recombination.

This visualization of the phenomenon lends itself to the creating of

a model which can be made for photocurrent generation as a function of

average distance of charge generation from the back contact. An

illustration of the model is shown schematically in Figures 5 and 6.

Figure 5 illustrates a thin film of CdSe, where a thin film is taken

to mean a film of a thickness approximately equal to the absorption

coefficient (α) for blue light.

For a thin film of CdSe (Figure 5), FW illumination with blue light

leads to absorption of most of the light relatively close to the substrate

because the film is thin. For red light, the situation is similar except that

the absorption of the light is not as strong. No appreciable difference is

expected for BW illumination, as is evident from Figure 5.

For a thick film, defined as one which is considerably larger than α

for blue light, FW illumination with blue light results in most of the charge

carriers being generated relatively far from the substrate, and the electrons

being as a result more susceptible to recombination loss. With FW red

light illumination, more carriers are generated closer to the substrate. This

should lead to a poor photocurrent response for short wavelength light,

which improves with increase in wavelength until the band edge is reached,

at which point the response drops again. Also, the long wavelength

response should improve with increasing thickness (up to a certain point)

since more of the long wavelength light is absorbed.

Considering now BW illumination of the thick film, the situation is

reversed compared with FW illumination. In the case of BW illumination,

blue light is absorbed close to the substrate while red light is absorbed

further away. This should result in a good response for blue light, with the

close to bandgap response poorer than for FW illumination, since some of

this light is absorbed relatively far from the substrate.

CdSe films deposited on conducting transparent glass were used,

thus allowing illumination from either side of the film, i.e., either front-

wall [FW] or back-wall [BW] illumination.

Crystallite sizes can be measured by electron microscopy or

estimated from the transmission spectra. Using the latter method, it is

estimated that the crystal sizes for the samples used in Figures 1 and 2 are

about 6.5 nm.

The results described above demonstrate that in order to obtain

optimum photocurrents from photovoltaic cells based on these

nanocrystalline films, the optimum configuration will often be a thin film

deposited on both sides of double sided conducting glass. Figure 3 shows

I-V characteristics, both dark and illuminated, of such a CdSe bi-film with

the transmission spectrum of the film shown in the inset.

It is to be noted that attempts to make a solid state Schottky junction

with these films resulted in a high resistance I-V characteristic, which was

virtually collinear with the voltage axis, and little, if any, photoeffect, as

described in Hodes et al.

Two methods were employed to control crystal size in these films.

One way was by controlling the deposition temperature. Another involved

the use of illumination during deposition. An increase in either the

deposition temperature or the illumination during deposition led to an

effective increase in crystal size.

When illumination was used during deposition, the spectral

responses shown in Figures 1 and 2 were obtained. Quantum efficiency

measurements for these films in the region of strong absoφtion and flat

photocurrent response gave values typically between 0.6 and 0.7.

The behavior to be expected based on the above model corresponds

closely to the results in Figures 1 and 2 in all aspects, which attests to the

accuracy of the model. In addition, the thickness at which the peak in the

spectra first begins to appear gives an indication of the thickness of the

active layer of CdSe. For thicknesses greater than this thickness, light

absorbed in the region farthest from the substrate (blue for FW and red for

BW illumination) will be subject to increasing recombination and resistance

losses. For the films with crystal size of about 5.0 nm, this thickness is

about 150 nm, while for the film in Figures 1 and 2 (crystal size of about

6.5 nm), it is about 300 nm.

While these thicknesses may seem thin if close to total absoφtion

of light is desired, as would be the case for solar cell use, it should be

noted that the oscillator strength, and therefore the absoφtion coefficient

of these size-quantized films may be enhanced relative to bulk material.

The α in the region of strong absoφtion is about 2xl0 ⁵ cm ^"1 for the

approximately 5.0 nm crystallites and about lxlO ⁵ cm ^"1 for the 6.5 nm

crystallites. This is to be compared with approximately lxl 0 ⁵ cm ^'1 for

'normal' CdSe. Hence, preferable film thicknesses for maximum

absoφtion would be range from approximately two or approximately three

times those thicknesses at which peak behavior for FW illumination start

to be exhibited. Glass with a conducting layer on both sides, and with the

semiconductor deposited on both sides, would come close to fulfilling this

requirement.

There are two loss mechanisms in films such as those described

herein as the thickness increases. First, there is a resistance loss, leading

to loss in photovoltage and fill factor. Second, there is a recombination

loss, presumably occurring at crystal boundaries, which leads to loss in

photocurrent. These losses are in addition to the possibility of indirect

recombination by electron injection into the electrolyte, which is probably

the predominant loss mechanism for thin films. These recombination

losses amount to 30-40% reduction in efficiency, as described earlier.

Direct recombination is relatively low for LTs in contrast to a solid

junction because the holes are removed rapidly by the electrolyte, which

lowers the recombination probability.

The mechanism for charge separation described herein is quite

different from that in a conventional PEC or PV cell. In effect, these films

behave as colloidal semiconductors. It is generally accepted that the

existence of an appreciable space charge layer in small lightly doped

semiconductor particles is unlikely, since the particle is too small to

support a field in its bulk. See A. J. Bard, J. Phys. Chem., 86, 172 (1982),

which is incoφorated by reference as if fully set forth herein.

In such a case charge separation occurs at the particle surface by

transfer into the electrolyte. For a colloidal system, such a charge transfer

is measured by the degree of chemical reaction occurring due to either or

both hole and electron transfer. The two reactions are different, resulting

in a non-regenerative system. If the reduction and oxidation reactions were

the same, forming a regenerative system, no net change, other than heat

production, would occur.

For the films described here, however, both regenerative and

non-regenerative reactions can be used, and in the former case, the system

acts as a PV cell with electricity being produced. However, in contrast to

other PV systems, and in common with dispersed colloidal photochemical

systems, charge separation occurs not by a space charge layer, but rather

by differing rates of electron and hole transfer into solution.

It will be clear that the methods used to prepare films according to

the present invention may include most of the methods commonly used to

deposit semiconductor films, and, in addition, some less commonly used

methods. Because it is desired to produce crystals of small size, in contrast

with the large crystal size normally required for photovoltaic cells, the

preparation techniques can be based on low temperature techniques, using

temperatures on the order of 0 to 200°C, rather than on the more common

but more demanding higher temperature techniques taking place at

temperatures of 400 to 600°C. Use of low temperature techniques

dramatically the energy consumption and thus the cost of manufacture.

Besides the methods of electrochemical deposition and chemical

solution deposition described in the examples above, vacuum evaporation

and sputtering (vacuum deposition) onto substrates which are at

approximately room temperature or even cooler will give small crystal size.

The presence of an inert gas under low pressure during vacuum

evaporation will help to form the small crystal sizes required. However,

a large variety of low temperature methods, including, but not limited to,

electrophoresis, anodization, and those mentioned above, can effectively be

used. Another fabrication technique involves the successive immersion of

the substrate in solution of the ions making up the semiconductor.

The porosity of the film may be controlled by depositing a

composite of the desired semiconductor together with an easily dissolvable

material. This may be desirable to increase the rate of diffusion of solution

species in the porous film. Thus, as example, CdSe could be evaporated

or sputtered together with Al (from separate sources) and the Al

subsequently dissolved by an alkali metal hydroxide solution (which will

not dissolved the CdSe). As long as the concentration of Al is not too high

(in which case the film would disintegrate) this will trace a film of porosity

depending on the ratio of CdSe to Al.

For porous substrates (which, for photochemical reactions, may be

either conductive or non-conductive and preferably transparent to the

illumination), simple successive immersions of the substrate in solutions of

the ions making up the semiconductor may be sufficient. For example, to

prepare CdS on a porous substrate, the substrate may be dipped first in a

solution of a cadmium salt, then in a sulfide solution (or vice versa).

While the focus throughout the above description has been on use

of devices according to the present invention as photovoltaic cells for the

production of electrical power, it will be clear that nanocrystalline

semiconductor layers according to the present invention may also be used

for both photocatalytic and photoconversion reactions, when the system is

non-regenerative.

Thus, instead of the production of electricity being the primary

product, chemical energy, in the form of desired chemical products,

becomes the primary product. While ostensibly a radical departure from

use of devices according to the present invention as solar cells, it should

be appreciated that the only difference is that, in the case of electricity

production, the electrolyte is chosen so that whatever is oxidized (reduced)

at the semiconductor electrode is reduced (oxidized) back to the original

reactants at the second electrode, while in the case of chemical production,

the electrolyte is so chosen that the reactions at the two electrodes are

different and there is a net production of new chemical products.

An example of this is the photoelectrolysis of water. If dissolved

oxygen is present at the non-semiconductor electrode, the oxygen is

reduced to water, exactly balancing out the evolution of oxygen at the n-

type semiconductor electrode. In this case the only net production is of

electrons. However, if oxygen is absent, hydrogen may be evolved instead,

making the device a chemical cell.

An example of a photocatalytic reaction where such films may be

of used is the photo-oxidation of small amounts of organochlorine or

cyanide contaminants in water. See C. Korman, D. W. Bahnemann and M.

R. Hoffman, Environmental Sci. & Tech., 25, 494 (1991), which is

incoφorated by reference as if fully set forth herein.

The advantage of the porous layers over more usually proposed

suspensions or colloids of the semiconductor is that the semiconductor is

present in a convenient, immobilized form on a substrate which can be

easily immersed and removed from the water, unlike the case with the

removal of a dispersed suspension.

For such puφoses, the substrate itself may also be porous and of

high surface area. A preferred embodiment in this case might be a flow-

through 'membrane' whereby the water to be purified flows through the

illuminated porous semiconductor and porous substrate.

While the invention has been described with respect to a small

number of preferred embodiments, it will be appreciated that many

variations, modifications and other applications of the invention may be

made.