For information about Web accessibility, please contact the Webmaster at

Haverford College

Marian E. Koshland Integrated Natural Sciences Center

Biography of an Experiment: The Invention of the Dye-sensitized Solar Cell

As part of the Howard Hughes Medical Institute’s Interdisciplinary Scholars program, we have put together a biography of an experiment related to Michael Graetzel and Brian O'Regan's 1991 Nature article entitled, “A low cost, high efficiency solar cell based on dye-sensitized colloidal TiO2 films.” Included is an annotated version of the original Nature article. The scientific language used in the article may appear intimidating, so we have included helpful tips to facilitate understanding. Additionally, we provide some scientific background on semiconductors and conventional solar cells.


As part of the Howard Hughes Medical Institute’s Interdisciplinary Scholars program, we have put together a biography of an experiment related to Michael Graetzel and Brian O'Regan's 1991 Nature article entitled, “A low cost, high efficiency solar cell based on dye-sensitized colloidal TiO2 films.” Included is an annotated version of the original Nature article. The scientific language used in the article may appear intimidating, so we have included helpful tips to facilitate understanding. Additionally, we provide some scientific background on semiconductors and conventional solar cells.
To better understand the experimental and conceptual breakthroughs involved in the invention of DSSCs, we had a conversation with the inventor himself, Michael Graetzel. Conducted via video conference, segments of the interview can be viewed under the “Interview” tab. As a final project, we demonstrate the ease with which a DSSC can be constructed. In just one afternoon, two fairly rudimentary cells were constructed, a process that can be viewed pictorially in the “How to Make a DSSC” tab.

Reading the Nature Article

Please click on the pictures for their descriptions

A low-cost, high-efficiency solar cell based on dye-sensitized colloidal TiO2 films

Brian O'Regan* & Michael Grätzel

Institute of Physical Chemistry, Swiss Federal Institute of Technology, CH-1015 Lausanne, Switzerland

THE large-scale use of photovoltaic devices for electricity generation is prohibitively expensive at present: generation from existing commercial devices costs about ten times more than conventional methods1. Here we describe a photovoltaic cell, created from low- to medium-purity materials through low-cost processes, which exhibits a commercially realistic energy-conversion efficiency. The device is based on a 10-µm-thick, optically transparent film of titanium dioxide particles a few nanometres in size, coated with a monolayer of a charge-transfer dye to sensitize the film for light harvesting. Because of the high surface area of the semiconductor film and the ideal spectral characteristics of the dye, the device harvests a high proportion of the incident solar energy flux (46%) and shows exceptionally high efficiencies for the conversion of incident photons to electrical current (more than 80%). The overall light-to-electric energy conversion yield is 7.1-7.9% in simulated solar light and 12% in diffuse daylight. The large current densities (greater than 12 mA cm-2) and exceptional stability (sustaining at least five million turnovers without decomposition), as well as the low cost, make practical applications feasible.

Solar energy conversion by photo electrochemical cells has been intensively investigated2-11. Dye-sensitized cells differ from the conventional semiconductor devices in that they separate the function of light absorption from charge carrier transport. In the case of n-type materials, such as TiO2 current is generated when a photon absorbed by a dye molecule gives rise to electron injection into the conduction band of the semiconductor, Fig. 1. To complete the circuit, the dye must be regenerated by electron transfer from a redox species in solution which is then reduced at the counter electrode. The monochromatic current yield

ηi = LHE(λ) x φinj x ηe

where LHE (light harvesting efficiency) is the fraction of the incident photons that are absorbed by the dye, φinj is the quantum yield for charge injection and ηe is the efficiency of collecting the injected charge at the back contact, expresses the ratio of measured electric current to the incident photon flux at a given wavelength. The photovoltage ΔV in Fig. 1, generated by the cell, corresponds to the difference between the Fermi level in the semiconductor under illumination and the Nernst potential of the redox couple in the electrolyte.


Although attempts to use dye-sensitized photoelectrochemical cells in energy conversion have been made before, the efficiency of such devices has been extremely low and practical applications have seemed remote. One problem is that of poor light harvesting. On a smooth surface, a monomolecular layer of sensitizer absorbs less than 1% of incident monochromatic light. Attempts to harvest more light by using multilayers of dye have in general been unsuccessful. The remaining option is to increase the roughness of the semiconductor surface so that a larger number of dye molecules can be adsorbed directly to the surface and can simultaneously be in direct contact with the redox electrolyte. Matsumura et al.12 and Alonso et al.13 have used sintered ZnO electrodes to increase the efficiency of sensitization by rose bengal and related dyes. Willig, Parkinson and colleagues14 have reported high quantum yields for the dye sensitization of SnS2. But the conversion yields from solar light to electricity remained well below 1% for these systems. In addition, the instability of the dyes employed presented a severe practical drawback. By using semiconductor films consisting of nanometer-sized TiO2 particles, together with newly developed charge-transfer dyes, we have improved the efficiency and stability of the solar cell.

High-surface-area TiO2 films were deposited on a conducting glass sheet from colloidal solutions. A transmission electron micrograph of the colloid is shown in Fig. 2. Electronic contact between the particles was produced by brief sintering at 450 °C. The size of the particles and pores making up the film is controlled by the size of the particles in the colloidal solution. The internal surface area of the film is determined by the size of the particles and the thickness of the film. These parameters were optimized to obtain efficient light harvesting while maintaining a pore size large enough to allow the redox electrolyte to diffuse easily. Films of 10 µm thickness consisting of particles with an average size of 15 nm gave linear photocurrent response up to full sunlight and were used throughout. A cubic close packing of 15-nm-sized spheres to a 10-µm-thick layer is expected to produce a 2,000-fold increase in surface area.


Absorption spectra obtained for such nanostructured TiO2 films are shown in Fig. 3. Bare films are transparent and colourless, displaying the fundamental absorption edge of anatase (band gap 3.2 eV) in the ultraviolet region. Deposition of a monolayer of the trimeric ruthenium complex15,16, RuL2(µ-(CN)Ru(CN)L′2)2, 1, where L is 2,2′bipyridine-4,4′-dicarboxylic acid and L′ is 2,2′-bipyridine, results in deep brownish-red coloration of the film. The absorption onset is shifted to 750 nm, the light harvesting efficiency reading almost 100% in the whole visible region below 550 nm. Integration of the spectral overlap between a solar emission of AM1.5 and this absorption band shows that 46% of the incident solar energy flux is harvested by the dye coated film (AM = 1/sin α where α is the angle of incidence, of the solar rays at the Earth's surface). [Explanation]

The optical density of the film at 478 nm corrected for the absorption by the conducting glass support was 2.45. Dividing by the extinction coefficient16 of 1 (e478 = 1.88 x 107 cm2 mol-1) yields the dye surface concentration, Γ = 1.3 x 10-7 mol cm-2, As each dye molecule occupies an area16 of 1 nm2, the inner surface of the film is 780 cm2 for each 1 cm2 of geometric surface, Thus, the roughness factor is 780, which is smaller than the predicted value of 2,000. The difference is attributed to necking between TiO2 particles. In addition, the large size of 1 prevents its access to very small pores, reducing the apparent surface area.


The photocurrent action spectrum obtained with the dye-coated TiO2 film is also shown in Fig. 3. It closely matches the absorption spectrum, indicating that the current is due to electron injection from 1 into the conduction band of TiO2. The photocurrent yield, measured at 520 nm was found to depend on the counter ion of the iodide/triiodide redox electrolyte, increasing from 68% for tetrapropylammonium to 84% for Li+. After correction for the light absorption by the conducting glass, the yields are 80% and 97%, respectively. This shows that the nanostructured TiO2 films used in conjunction with suitable charge transfer dyes can achieve quantitative conversion of visible light photons into electric current.

Figure 4 shows the current-voltage characteristics obtained with the thin layer cell under illumination by simulated AM1.5 solar light. The conversion efficiencies for one tenth and full sunlight are 7.9% and 7.12%, and the fill factors (maximum output power of cell ÷ [short circuit current x open circuit voltage]) are 0.76 and 0.685, respectively. Similar yields were obtained under direct sunlight (measurements performed in June early in the afternoon on the roof of the institute). Under diffuse daylight the efficiency increased to 12%, indicating that under such conditions the cell performance is better than that of conventional silicon devices. This is because the spectral distribution of diffuse daylight overlaps more favourably with the absorption spectrum of the dye-coated TiO2 film than direct sunlight. The fill factor of the cell remains above 0.7 even at very low light intensity (< 5 W m-2). Conventional photovoltaic cells have a much smaller fill factor (<0.5) under these conditions. This indicates that loss mechanisms such as recombination, normally encountered in semiconductor photoconversion, have been minimized. This result might. appear surprising in view of the disordered structure of our film giving rise to defects. But the role of the semiconductor in a dye-sensitizted device is merely to conduct the injected majority charge carriers. There are no minority carriers involved in the photoconversion process. Therefore, surface and bulk recombination losses due to lattice defects, encountered in conventional photovoltaic cells, are not observed in such a device.


The long-term stability of cell performance was tested by illuminating the thin TiO2 film loaded with 1 with visible (λ > 400 nm) light for 2 months. The change in the photocurrent was less than 10% over this period, during which a charge of 62,000 C cm-2 was passed through the device, corresponding to a turnover number of 5 x 106 for the sensitizer. This implies that if any dye degradation had occurred its quantum yield (φdec) is less than 2 x 10-8. As φdec = kdec/Σk, the rate constant, kdec s-1, for excited-state decomposition (due to processes such as ligand loss) must be at least 10-8 times smaller than Σk, the sum of rate constants for all channels of dye deactivation. Because charge injection is the predominant channel, this sum is practically equal to the rate constant for charge injection, which exceeds 1012 s-1 in the case of 1. Therefore, the upper limit for kdec is 2 X 104 s-1, which agrees with the known photophysics of this class of transition metal complexes17. The very fast electron injection observed with dyes such as 1, combined with high chemical stability, renders these compounds very attractive for practical development. [Explanation]

ACKNOWLEDGEMENTS. We are grateful to I. Rodicio for experimental help. to M. Nazzeruddin for a sample of sensitizer and to A. Kay for advice on the preparation of colloidal membranes. This work was supported by the Swiss National Energy Office.

* Present address: Department of Chemistry, University of Washington, Seattle, Washington 98195, USA.

 To whom corresponoence should be addressed.

Received 19 July: accepted 20 August 1991.

1. Bucher. K. & Fricke. J. Phys. Zeit 21, 237-244 (1980).
2. Honda, K. & Fujishima, A. Nature 238, 37-39 (1972).
3. Tufts. B. J. et al. Nature 326, 681-683 (1987).
4. Gerischer, H. Electrochim. Acta 36, 1677 (1990).
5. Licht, S., Hodnaes, G., Tenne, R. & Massen. J. Nature 326, 863-864 (1987).
6. Heller. A. Acc. chem. Res. 14. 154-162 (1981).
7. Nozik, A. J. Phil. Trans. R. Soc. Lond. A295, 453-470 (1980).
8. Tributsch, H. & Bennet, J. C. J. electroanal. Chem. 81, 97 (1977).
9. Wrighton, M. S. Acc. chem. Res. 12, 303-310 (1979).
10. Bard. A. J. Science 207, 139 (1980).
11. Memming, R. Phil. Tech. Rev. 38, 160 (1979).
12. Matsumura, M., Nomura, Y. & Tsubomura, H. Bull. chem. Soc. Japan 50, 2533 (1977).
13. Alonso. N., Beley, V. M., Chartier, P. & Ern, Y. Rev. Phys. Appl. 16, 5 (1981).
14. Willig, F., Eichberger. R., Sundaresan, N. S. & Parkinson, B. A. J. Am. chem. Soc. 112, 2702-2707 (1990).
15. Amadelli, R., ArgazzI, R., Bignozzi. C. A. & Scandola, F. J. Am. chem. Soc. 112, 7099-7103 (1990).
16. Nazeeruddin, M. K., Liska, P., Moser, J., Vlachopoulos, N. & Grätzel, M. Helv. chim. Acta 73, 1788-1803 (1990).
17. Juris, A., Balzani, V., Barigletti, F., Campagna, S., Belzer, B. Coord. Chem. Rev. 84, 85 (l988).
18. Anderson, M. A., Gieselmann, M. J. & Xu. Q. J. Membrane Sci. 392, 43 (1988).
19. O'Regan, B., Moser, J., Anderson. M. & Gratzel, M. J. Phys. Chem. 94, 8720-8726 (1990).


The following is some background information that should help formulate a greater understanding of the actual article.

Introduction to Semiconductors

Due to the Pauli exclusion principle, two electrons cannot exist in identical energy states. Therefore, at absolute zero, the electrons of a substance will pack down into the lowest energy states available in that substance; the highest energy level occupied by an electron at absolute zero is known as the Fermi level of a substance.

Figure 1

A substance will conduct if, directly above the Fermi level of that substance, there exists a high density of energy states that the electrons of that substance can occupy. That is, electrons can jump up to higher energy levels with little to no energy cost; essentially, the electrons can move freely between energy levels. When a potential difference is applied to the substance, the energy levels of the atoms of the substance will be changed such that a gradient will be established in the energy levels of the substance - the unoccupied states on the side of the substance with the more positive elecric potential will be at a lower energy than the occupied states with the more negative electric potential, see Fig 1. For this reason, the electrons will move from the occupied states in the end of the substance with the more negative electric potential to the unoccupied states in the side of the substance with the more positive electric potential; this is allowed because the electrons can move between electric states easily. That is, the electrons are able to flow through the substance; therefore, these substances are called conductors. See Figure 1.

In an insulator, there is a large difference in energy between the highest-energy electron-occupied energy level and the lowest-energy unoccupied energy level; this difference is called a band gap. For this reason, it takes a significant amount of energy for electrons to move between energy levels and for this reason, when a potential difference is applied to an insulator, electrons cannot flow. See Fig 2.

A semiconductor is a material with a band gap that is small enough such that electrons can be excited into higher energy states with small amounts of energy. Semiconductors can be used in solar cells when the size of the band gap is equivalent to the energy of the photons emitted by the sun. When solar photons strike these materials, their electrons can be excited into higher energy levels. As the energy levels above the band gap are not filled by the excited electrons, these excited electrons are allowed to flow freely and can be used to create a potential difference. See Figure 2.

Figure 2

Introduction to Conventional Solar Cells

First, an introduction to charge carriers is necessary. Moving electric charge constitutes a current. This charge must be carried by some particle. There are two main ways of thinking about current flow - in terms of electrons (negative charge carriers) or in terms of holes, which are conceptual particles that indicate the absence of electrons and are considered to be positive charge carriers. To illustrate how this could work: suppose you have a lattice of covalently bonded atoms, and suppose that one electron is removed from one of the atoms. The atom that loses an electron is positvely charged and is left with a hole (the absence of an electron). An electron from a nearby atom can come and fill that hole; however, the electron must be removed from the nearby atom for that to occur. The atom that originally lost an electron becomes neutral again and the nearby atom has lost an electron (gained a hole) and becomes positively charged. Conceptually, we can think that the hole was transferred from the original atom to a nearby atom, and that the hole carried its positive charge with it when it moved. That is, the hole is a mobile positive charge carrier.

Now, conventional solar cells are made of silicon, a semiconductor material (see Introduction to Semiconductors for more information). Silicon, like carbon, has four electrons in its valence shell - for this reason, it will form bonds to four other silicon atoms, forming a lattice. When silicon is bonded to four other atoms, none of its electrons are free to move about the lattice. Due to this lack of free electrons, pure silicon is a poor conductor.

Because of this, pure silicon is doped for use in solar cells; that is, other atoms are purposefully introduced into the lattice structure in order to make the solar cell work. When tiny amouts of atoms like phospohorus, which have five valence electrons, are introduced into the lattice (for example one atom of phosphorus per million atoms of carbon), extra electrons are added to the lattice structure, creating what we call n-type silicon. On the other hand, when atoms such as boron (with only three valence electrons) are introduced into the structure, the structure has room for more electrons (it has an excess of holes, which can be filled with electrons), creating p-type silicon.

When a p-type semiconductor is placed in contact with an n-type semiconductor, some of the excess electrons in the n-type material will flow over into the p-type material in order to recombine with some of the extra holes in the p-type material. This does not occur indefinitely, however; when electrons move from the (normally neutral) n-type semiconductor into the p-type semiconductor, the atoms of the n-type semiconductor near the p-n junction become positively charged (they have lost electrons) and the atoms of the p-type semiconductor near the p-n junction become negatively charged (they have lost holes). This creates an electric field which tends to push electrons (negative) into the n-type material (away from the negative and into the positive); that is, an excess electron in the p-type material would be swept into the n-type material but could not move back across the electric field. The opposite is true for holes (they are swept into the p-type material).

So, when a photon from the sun with enough energy to excite one of the electrons in the semiconductor strikes the solar cell, one of the valence electrons in the semiconductor will be exicted to a higher energy level, leaving a hole in the old energy level. This hole will tend to be swept towards the p-type material while the electron, which has been "knocked loose" and can now travel easily between atoms due to the relative lack of other electrons near its energy level (see Introduction to Semiconductors for more information) will be swept towards the n-type material. The electron will be driven to recombine with its hole, however, and therefore, if the n-type material is connected to the p-type material by an external wire, the electron will travel through the wire from the n-type material to the p-type material, creating moving charge or a current.

The important point to take from the above disscussion is that, in a conventional solar cell, the semiconductor has two main roles - it acts as a source of electrons and provides an electric field in order to separate the charge. Because of this, some of the charge separation is lost when the excited electron recombines with its hole before the two charge carriers have a chance to be separated (that is, sometimes the electron is able to drop down from its excited state, filling a hole in the valence shell, before the hole and the electron can be separated by the electric field). This causes a significant decrease in the efficiency of the conventional solar cell.

Now, note that the article mentions that "dye sensitized cells differ from the conventional semiconductor devices in that they separate the function of light absorption from charge carrier transport" (and it stresses the same point again near the end). Essentially this says that, In a dye-sensitized solar cell, unlike in a conventional solar cell, none of the semiconductor's electrons are actually excited; the semiconductor acts only as a medium for electron transport. For this reason, the valence shell of the semiconductor is always full in a dye-sensitized solar cell (there are no holes) and we therefore don't see the loss of efficiency from electron-hole recombination that we see in a conventional solar cell. This is one of the reasons that dye-sensitized solar cells are able to obtain such high effciencies.

How a dye-sensitized solar cell works

First of all, note that the dye is a "deep brownish-red" colour. This means that the dye is absorbing much of the visible light incident on it.


First, a photon hits a molecule of the dye, exciting an electron in the dye. This excited electron is then injected into one of the energy states of the titianium dioxide above TiO2's band gap (see Introduction to Semiconductors for more information); this energy state is still at a lower energy than the excited state of the electron in the dye. This is steps 1 and 2 of the diagramme in the article (FIG. 1).

However, the dye (which has lost an electron) will quickly decompose if another electron is not provided. For this reason, the dye is given an electron from the "redox species in solution" - in this case, iodide (I3-) - oxidizing it into triiodide (3I-). This is denoted by step 5 in FIG. 1 (in the article).

Now: one very important fact must be noted here. The rate at which the dye injects electrons into the semiconductor is much larger than the rate at which the electron recombines with the hole that it left in the valence shell of the dye (see Introduction to Conventional Solar Cells for more information). For this reason, once an electron has been excited, it will be injected into the energy levels of the semiconductor and the dye will pick up another electron from the redox species in solution before the electron has a chance to recombine with the hole that it left behind. If this were not the case, the solar cell would not work - the electron would simply recombine with the hole in the dye, no electron injection would occur, we would have no separation of charge, and current wouldn't flow.

So, electrons are being injected into the semiconductor and the dye molecules are getting their electrons back from the redox species in solution. In order to complete the circuit, the triiodide must be reduced back into iodide. This requires an amount of energy equal to the Nernst potential of the iodide/triiodide redox couple. This potential is lower than the difference in energy between the lowest energy level of TiO2 above the band gap and the highest energy level of TiO2 below the band gap (that is, the Nernst potential of iodide/triiodide is lower than the band gap energy of TiO2). Therefore, the injected electrons have more energy than they need to reduce the triiodide (note that this energy did not come from nowhere - it was given to the electrons by a photon, which came from the sun). Again, see Introduction to Semiconductors for more information.

Therefore, if a load is hooked up between the semiconductor and the counterelectrode, the electrons will use their extra energy to provide power to the load. Once the electrons reach the counterelectrode, they will reduce the triiodide, forming iodide and completing the circuit.

A Conversation With Michael Graetzel

We had the pleasure of seeing Michael Graetzel at the University of Pittsburgh’s 2009 Science Unplugged conference. He gave a talk entitled “Light and Energy, Mesoscopic Systems for Solar Power Conversion and Storage” about the origins and current scientific progress of dye-sensitized solar cells. Though we had little chance to speak to Professor Graetzel in person at Pittsburgh, we arranged a video-interview presented below. We were particularly interested in the critical breakthroughs that led to the discovery of dye-sensitized solar cells. Furthermore, we had the opportunity to hear the most recent news about current DSSC innovation.

Michael Graetzel is a professor at the Ecole Polytechnique Federale de Lausanne in Switzerland. He is the director of the Laboratory of Photonics and Interfaces at this prestigious institution. He has authored over 800 papers and has received numerous awards, including the Balzan Prize, the Galvani Medal, the Faraday Medal, and the Harvey Prize (Information and photo courtesy of Professor Graetzel,

What attracted you to the field of photochemistry?

Well to be honest, I studied chemistry. And why did I study chemistry? One very important reason is that I thought with chemistry we could learn to understand what was going on around us. Biological processes, medical, life, it’s all chemistry. And one very important part of our environment is the photosynthetic process that happens in nature because that’s where we get all our energy from.

Graetzel describes photosynthesis as a major inspiration

I took an interest in the photosynthetic process of energy conversion which dates back to when I was a post-doc. I always wanted to work on photo-reactions, I don’t know why, that was just a passion I had, and then the natural photosynthetic processes. At the time it wasn’t even known how it functions. Today we have a much better understanding how photosynthesis works. I am talking about the act of charge generation in the leaf. How is light converted to electric charge? Some people don’t know that electric charges are produced first, and then they are used to oxidize water and reduce carbon dioxide. So that was fascinating to me to see how charge separation in the plant happens, how nature does it. It was the starting of our photocell that mimics that reaction. It is the only cell that uses a molecular absorber. In semi-conductor photovoltaic cells you have the semi-conductor that generates charges, and conducts the charges. In our solar cell…the molecule absorbs light, generating charges. These charges are transported by some other material to the collector. So it is a separation of light absorption and charge generation, from carrier transport. That happens in the green leaf also, we are very close to the green leaf.

Were environmental concerns initially motivating you?

Yes, the driving force for all this work came with the first oil crisis in the 70s when we ran out of fuel. You couldn’t get gasoline anymore. Because there was some kind of war going on in the Middle East and the oil producing countries cut their production. Then I started thinking about the oil supply, and it occurred to me that there wasn’t all that much oil left. Just looking at the statistics, it didn’t look like we had a thousand year supply left. It was just like a few decades, and certainly less than a hundred years. So then I became really motivated to do research to help with the energy problem. Very early on, I felt this was a mission which I should take on.

Was this as a post-doc?

It was when I was a post-doc. As a PhD student I had not thought about the energy problem at all. It cropped up in I think 1973. I was a post-doc in the United States (Notre Dame). When I got back to Europe, there was no gasoline, at least not enough. And I thought well how much gasoline do we have actually? What happens if one day there is just no supply anymore? If we ever get into that situation, it’s going to be a disaster. Science has to come up with first analysis and then we have to think about how to remedy the problems. So I felt it was a good idea to take photosynthesis as an example because after all, it worked well….3 billion years!

In Pittsburgh, you mentioned that the fact that electron injection was significantly faster than electron recombination in dye-sensitized materials sparked your interest in pursuing what eventually became DSSCs. Please comment on this breakthrough.

It really took a long time. We were playing around with these colloidals, now their called quantum dots. We had selected titanium dioxide because it is a material that is widely available, non-toxic, has all these desirable properties, except it doesn’t absorb light! How do you make it absorb light? Well you can sensitize it. By chance we found out that if you use those carboxylated dyes, they really would go on the surface and anchor very strongly. One of my students had used an ester, the ester hydrolyzed and then the carboxylates were formed by hydrolysis. These dyes would coat on the surface under mild heating and we could smell the alcohol from the hydrolysis. I told him, ‘you hydrolyzed the acid, that’s the carboxylate that makes it draft to the surface’, and that was correct.

We then did laser experiments. That was actually my vocation. I started out doing laser photolysis, in other words interested in photoreactions on short time scales which you can explore by laser photolysis. Today we have systems, in my own lab I have a system that goes down to the femtosecond (10-12 seconds) range. In those early days we were able to get to the nanosecond (10-9 seconds), that was already a big achievement. When I was appointed as a young professor here, I got myself a laser because I was coming from that end. I like to study the charge separation, recombination, to see how fast they are. But I did have an effort, right from the beginning in photo-electrochemical cells, because I was impressed with those cells.

So I put one of my post-docs on it. First he got familiar with the literature. He did nothing original, just repeated the work in the literature so we would get going and have a system that would be operative. But our main task was in these laser experiments and there we found out that indeed the forward reaction, we couldn’t measure it, it was too fast. The laser flash lasts for a couple nanoseconds, but by then the reaction was already over, we could only say it was instantaneous. Now we know that is about 20 femtoseconds, it’s a very very fast reaction. And so we saw the signal going up for injection, and then we could see the recombination in milliseconds. We thought gosh, there’s a big difference between the charge generation and recombination. That was an important finding, because at the end, when we see that as a scientist you say , ‘well maybe we can collect those charges’. Had it been the opposite with a slow injection, you wouldn’t see any charges, we would not collect anything. The fact that we could see that reaction, that we could see those charges by optical spectroscopy, the electrons in the titanium dioxide have a (characteristic) color. We knew how to interpret those signals. As a laser spectroscopist you know that those transient intermediates have a spectral signature. So we know what we were measuring.

With this curious fact noted, Graetzel and colleagues moved forward to try to collect those charges on a film. In other words, make a cell.

At that time we thought maybe that experiment should be done on a film of particles. Because doing it on a film, we might be able to collect the charges. Because if you have the particles dispersed in solution you cannot collect—the charges go in and out. You can only say they last for a while before they recombine. The question is ‘Do they last long enough for you to collect them?’ We felt there was a chance. So we did that experiment. The student was Jean Desilvestro. He picked up on the photo electrochemical reaction. He had a system going, he had learned from the earlier work of the post doc. I told him to do that experiment. He comes back and I said, ‘well what did you find,’ and he said, ‘well disappointing,’ I said ‘what’s the current you measured?’ and he said ‘microamps.’ But I knew that this was a lot, I had studied the literature. It is very important to study the literature and know what is state of the art. So I said, ‘that’s not bad at all, because everybody else measures nanoamps not microamps.’ Nanoamps and microamps are big difference.

1985 JACs paper, highly efficient sensitization of titanium dioxide and finally the Nature paper

We have a JACS paper in 1985 which mentions those experiments. It’s called highly efficient sensitization of titanium dioxide. People always quote the Nature paper. But in reality, the Nature paper is not the first paper on dye-sensitized cells. The paper already mentioned… it shows the case of a rough electrode where we can get high incident photon to current efficiencies. We were able to show 40% of the incoming (photon to current efficiency). That was done with a three electrode system; it wasn’t really a regenerative photovoltaic cell. If I had published in 84, the full cell, I would have lost millions because later I had a patent application, and it would have been held against the patent application. We didn’t do it deliberately, we just published our observations; we hadn’t made a cell. We published that you could do a three electrode system where one was a working electrode, and a reference electrode. We just looked at the working electrode and how it behaved, how the response was. In a photovoltaic cell, you have no reference electrode; you just put two electrodes smack on top of each other…That paper is published in JACS in the year 1985. The experiment with the laser was mentioned. Also that carboxylated dyes work better than other dyes. That was done with laser photolysis.

Three years past, in ’88 I gave a lecture at the International Conference of Solar Energy Conversion which was held in Chicago. There we had built the first cell. And that was published in a JACS publication that came out in ’88. That was the next paper. Took three years between those two papers. But there I had applied for a patent before, before that publication came out I had to put a patent application out. Because, here in Europe if you publish something, you can’t patent it anymore. In the states its different, you can go back one year. I had been careful and I reported in Chicago. That’s when Brian O’Regan came in, because he was at the Chicago conference. He listened to my lecture, and he got very excited. He came to me after and said, ‘you know I am in the Chemistry department at Madison, but I’m very interested in your work and I make those membranes for ultra filtration, they are porous and have high surface area, maybe they can work too. Please come to see us and give us a talk.’ And so I did. A few months later I was visiting the United States and I went to Madison. Again I had my own electrode on me, and so we went in his lab. I said look, ‘Here’s my electrode, I will measure it with you together.’ He turns the light on, and at this time he used this mechanical recorder, and the mechanical recorder went out of range. It jumped, almost killed the recorder because the current was so big. He didn’t expect such a big current. Then I said to him, ‘well let’s try your thing,’ I put my dye on his membrane and it worked also, I thought it worked very well. I hadn’t expected that. So I said, ‘Brian, you’ve got something interesting.’ We decided to work together and that finally resulted in those nanoparticle membranes that are still being employed today… Brian came over to Switzerland, and I was convinced that my student could show that with these nanoparticle pastes you could just make cells very reproducibly and get the high performance. If it’s so easy, you can do it to reproduce it. Our old method which is kind of a sol-gel method was too complicated, didn’t give the same reproducible effects. It was OK, I mean, it wasn’t bad. But it couldn’t match these nanoparticle networks. So that’s how the Nature paper finally came out…It came out six years after the first laser publication.

I mean we used nanoparticles in the laser work. But when we went to the films we hadn’t used the same nanoparticles, we had used the sol-gel method. That film looked like a swiss cheesecake, it was rough. It certainly had a high surface area…but it wasn’t the nanoparticle film we use today, that came in when Brian listened in Chicago. And he is the first author on the Nature paper. He is still doing that work on DSSCs.

He is now in London at Imperial College, a scientist. I saw him just recently, he is an extraordinary scientist. His career was also built on this cell, now he is perhaps fifty years old, but he certainly made his career on this cell.

The story of Graetzel as a traveling salesman on the skeptical scientific world stage

When we started reporting like in the nature paper, then our colleagues didn’t want to believe it. They said, ‘no that’s crap, cannot be right.’ Because it was against the pervading opinion, everybody in the field thought this was a lost case, to try to sensitize oxides. The reason was that a molecule doesn’t absorb enough light in a monolayer, and if you go to corrugated systems then you don’t capture the charge; they recombine. That was the pervading accepted opinion. We were saying we had 1000 times more output than anyone else with this highly corrugated nanoparticle system. So they thought maybe we are crazy, or we have had a mistake in our experiments.

For a while I was traveling with those cells, always in my briefcase. One time I was in Berkeley as an invited guest of a professor. I had visited Texas to give a talk, and the department chairman said, ‘Mike, we can’t reproduce your experiment.’ So I said, ‘Well, I have my electrode with me,’ I had this electrode but I didn’t have the full setup. So who is the guy who would conduct the measurements? There was this Indian chap post doc. So, I said, ‘Look, I am giving this talk at 11, and we have a plane at 5, so we have enough time in the afternoon to do the experiment together. I said, ‘Fine, all I need is an iodide solution. I have my own dye and my own electrode, just need to put that together.’ You can see whether we have milliamps flowing or not. It’s very easy. You can measure that with one of those cheap meters. So, they asked the post doc to make up the solutions. Believe me or not, the guy was never able to make up the solutions, I don’t know why, maybe he didn’t want to. He had like two hours of time to make up a 0.1 molar iodide solution; you would think that’s not much of a challenge. So, but at the end, I asked him, ‘why don’t you show me your electrodes,’ because time was running out. And they looked very odd. Finally, what I said to the professor, ‘Look, we can’t do the experiment, I have to go on a plane, but since you are coming to Berkeley, we’ll do it together in Berkeley,’ because I had set up my own little system there. ‘You come to Berkeley, we do the experiment, I give you my electrode, you go back and repeat it in Texas.’ Fair deal, wouldn’t you think so? If I tricked them in my demo, she gets the electrode, she can check it out. So that’s what happened. Her name was Mary Anne Fox. I showed her the experiment, she said it looked great. No problem. The guy had made up the wrong electrode by mistake. So I had to do all this stuff to convince my colleagues. I was a traveling salesman for a while, traveling around with these films, showing that our paper was correct.


We are very grateful for everybody who comes on board and helps. We are very open. We feel this is a blessing. Competition is good, there is nothing wrong with it. The industrial companies have taken a huge step forward. They have invested the money, like g24 innovation is now doing the first mass production. These cells are now coming out in mass production, the flexible version, being put on the rucksacks and tennis bags. All sorts of electronic applications already now being pursued. So we should be very grateful for all, so what we are saying is, ‘welcome to the club, can we help you?’ That’s my attitude. So I have a lot of visitors. We have not had any quarrels, never any patent quarrels.

We have had patent quarrels… the others try to un-validate our patents. I’m talking about competition from other solar P-V technologies. That’s a different story. So there is the dye-sensitized field, and there are many other photovoltaic cells. Especially, now almost every day you hear something about a new photovoltaic converter. So there is this part to it. The folks, some are very close to us like OPV, organic PV. The inorganic also, there are lots of mesoscopic cells. Some are more or less closely related, others are not. Many are now using this mesoscopic or nanostructured theme, that’s for sure, there’s no doubt. So we are facing competition from within the community and were facing a lot of competition from other fields. So where’s our take? Well we can make some things the others cannot make. For example, transparent glass that produces electric power. Nobody can do it. So, here I have a unique selling proposition. That’s very important if you go out there and want to get investors involved. They will say, ‘so what’s so special about your solar cell, tell me Michael?’ You just say, ‘well it’s cheaper.’ That is not enough, everyone claims it’s cheap and so on. That’s not a sufficient argument. You have to have more, for example, flexible and light weight. That’s already something good. ..A lot of technologies cannot be flexible and lightweight. We have that advantage, we can easily go to thin films. We have some areas where we are very strong. That’s where actually the commercialization has started. And that’s correct. This is a technology that comes in through the niche application where we are very strong, where we have very little or no competition. I wouldn’t dare put DSSCs in the desert and competing with these very large systems that are being setup in Northern Africa. We are not there yet, maybe one day yes. We have to go through the building integrated applications, the portable electronics applications. That’s where we are strong, and in some fields we have no competition. If the investor looks at you and he says, ‘only this guy can do it,’ he is more comfortable putting his money there. That’s important.

Organic Dyes

The organic dyes have picked up since my lecture in Pittsburgh. They have gone up further, a lot. Now at eleven percent. Yeah, we have a porphryin dye that is at eleven percent, looks green, it’s beautiful. I would even be much stronger today and say yes we still have those Ruthenium dyes, but their days are counted, because some of those organic dyes are very stable too. Actually, I am hiring organic chemists, because I feel that in order to be competitive we have to make our own. We have our own ideas. We know how to improve those—we think we know. We have people doing calculations for us. So we have to get really involved in the preparation even of these compounds. Not only making solar cells. So I am hiring 2 or 3 organic chemists for that purpose. Just to cover our ideas. We have some new ideas how to put these donor-acceptor dyes into use. They work beautifully, really they do. They will certainly one day be the material of choice, because you know cost matters. They can be synthesized more cheaply, certainly. There is no metal like ruthenium. Ruthenium is not very expensive, there is plenty, you can go to the gigawatt scale. But, definitely if you have only one type of dye molecule, you expose yourself to fluctuations. Somebody discovers a use for ruthenium, so far there is very little, price goes up, then you’re in trouble. You don’t want to depend on one single element.

We are committed through our research projects to meet certain efficiency targets, and we better deliver on those targets. It’s not easy, that’s why we have to build up our strength in preparing the dyes, especially with enhanced near infrared response. That’s where the weakness is. Just give me anything that is efficient above 800 (nanometers), I’ll be very happy. Porphryn was a breakthrough, because it does have an enhanced near infrared response. It is a very nicely engineered dye. We have to do much more of that.


I think what is really is very much on my heart is the educational part…We need to educate the population. I think everybody hears about the climate problem, but nobody feels the temperature going up, it’s too slow. And now we have a cold winter!

We have to educate, especially the kids. I think this is a mission, so I am taking a lot of time especially with students like yourself. It’s very important to get the gospel spread. So that this whole thing is becoming a huge worldwide effort. People realize that its one problem that really is there. And that there is a planetary emergency developing, if we don’t deal with it, it’s something that seems (unnoticeable) that the temperature is going up, well we feel a little bit warmer.

We see all these traumatic events are already happening now. It’s these terrible storms and stuff. But people don’t take this to heart. It’s not as much of an emergency as the gasoline, if we would have no gasoline. But that moment will come anyway. And in addition, we haven’t any gasoline to fill the gap, there’s a gap opening up of 14 terawatts. We need to educate people. In the United States, still no CO2 tax. Here in Switzerland we have ordered a CO2 tax. I don’t think it will come even though Obama is very friendly towards renewable energies.

The message to the young people is the most important thing. That they take this on and get involved and that will be so much easier for the rest of us if we have young dedicated peoples helping out.

Environmental technology and the future

I think we really need to also be honest. Look, we need the wind power. Let’s not fight for heaven’s sake. Let’s not say I am the best, you will be dead. We better make sure that everybody is there to pick up the job to cover those 14 terawatts. We will have to collect on every little bit: geothermal, wave tides, wind, photovoltaics—photovoltaics cannot cover everything, water splitting by light. There has to be input from all sides, hydroelectric, biomass. I am the last one to say I can solve the problem…I am making an effort. We hope that a lot of people will join us and be helpful. But photovoltaics at best, people think maybe 25% of those 14 terawatts that will be missing in 2030 can be covered by photovoltaics. There’s still 75% left. So let’s be humble and try our best and especially educate the people so they realize that we have to do something. Those that are motivated come in. Like I said in my talk at Pittsburgh, we need that tsunami wave to move people forward. It’s not enough to have a little wriggle from time to time, when somebody says, ‘Oh yes we have that problem but its twenty years away.’


How to Make a Dye-sensitized Solar Cell

We decided that to best understand dye-sensitized solar cells, we would need to make some ourselves. For this reason, we bought a “Nanocrystalline Solar Cell Kit” from the Institute for Chemical Education at the University of Wisconsin. The procedures and materials provided by the kit are modeled after an article published in the Journal for Chemical Education by Michael Graetzel and Greg Smestad.

Preparation of Titanium Dioxide Film

In 1 mL increments, 9 mL of pH 4 aqueous acetic acid was added to 6g of Degussa P25 TiO2 powder. Using a mortar and pestle, a lump free paste was prepared (Figure at right). One drop of dish soap was added to act as a surfactant (a molecule with both hydrophobic and hydrophilic parts).

Two conductive glass plates were cleaned. One of these plates was oriented conductive side up (verified with a digital multi-meter), and bordered by scotch tape about 1mm on each side. The titanium dioxide solution was then added to the surface and smoothed with a glass stirring rod so that it was at the same height as the scotch tape (40-50 micrometers). The covered glass plate was then placed in a furnace at 450 degrees Celsius for 30 minutes to anneal the titanium dioxide film to it (see figure below).


Coating Film with Dye

Blackberries were crushed and dissolved in a methanol, acetic acid, and water solution (above). The titanium dioxide film on glass was then immersed in the solution for about 15 minutes. Upon removing, the film had turned a rich purple that remained after washing with cold water. This is because the carbonyl and hydroxyl groups of the anthocyanin dye of the blackberries had coordinated with the titanium metal centers of the film. This double coordination is called chelation, and is quite strong (See figure below).

Made with Chem draw

Assembling the Dye-Sensitized Solar Cell

The uncoated glass plate was covered with a layer of carbon (using a graphite pencil) on the conducting side. This layer of carbon serves as a catalyst for the electrolyte reaction of triiodide to iodide. The two glass layers were finally placed together, conductive sides in, and clamped with binder clips. The glass plates were offset to provide locations to attach alligator clips to measure power production. A few drops of a 0.5M potassium iodide with 0.05M iodine solution was applied to the side the apparatus and allowed to seep between the plates via capillary action. The solar cell is now complete.

Measuring Power of Cell

A digital multi-meter was used to test the power production of the solar cell. When we first tested it, it was dark outside and we only had fluorescent lights in the room where we were working. This resulted in readings of zero which was quite disheartening after many hours of preparing the solar cell. However, the next morning we got readings that significantly increased the closer we brought the cell to windows. We plotted a series of current vs. voltage readings for the cell induced by varying the resistance load. We found a maximum power output of 172 nanowatts, which is quite small. A graph of the I-V curve is shown below.

About UsAlex and Jacob

Alex Vargo (in purple) is a math major and a rising junior at Haverford College. His interests extend to the fields of computer science and chemistry, and he spent the summer of 2009 working with Joshua Schrier using computer simulations to examine heat transfer through nanowires. This summer, he will be examining symmetric function inequalities at Haverford under the guidance of Curtis Greene, and he plans to attend the Budapest Seminars in Mathematics in the fall.

Jacob Olshansky (in orange) is a chemistry and physics double major at Haverford College. He is currently a rising Junior with particular interest in the fields of electrochemistry and materials chemistry with an eye for creating greener energy options. He will be exploring these interests at Stanford University’s Center for Polymer Interfaces and Macromolecular Assemblies this summer by working on photo-catalyzed hydrogen production.


Thanks to Josh Schrier and Alex Norquist for generally helping us out and making sure that we were following the "rules" of the science community.

Thanks to Ruth Guyer for helping us with our interview, and to David Moore for his technical help.

Thanks to Robert Fairman and Kate Heston for coming up with this project and giving us the opportunity to make one ourselves.

Finally, we'd like to give immense thanks to Michael Graetzel for his interview and his support that made this project possible.