Chlorophyll makes sugar, not hydrogen & oxygen. And its effectiveness depends on what color the incident light is, for instance it can absorb 90% of blue light, but only 30% of red light and 8% of green light, roughly. The real-life efficiency of plants (containing all different photosynthetic dyes including chlorophyll) has a theoretically maximum of 4.5 percent. However the crop yields, which give a harvested quantity in bushels per acre, often yield something like 0.4 percent, which isn't so great. So you might think that the stated efficiency of 0.66% for splitting water into hydrogen and oxygen is comparable, but we are comparing apples and oranges - the purpose of the process is different. We need to look at the efficiency of electrolysis, which is the current method of splitting water.

I have an article that I downloaded from the Home Power website detailing a home hydrogen electrolysis plant. I quote......

"A cubic meter (35.3 cubic feet) of hydrogen gas takes about 5.9 hours to produce in this electrolyzer, when operated at its rated input power of 1000 Watts. This means the energy required to produce a cubic meter of hydrogen and 0.5 cubic meters of oxygen is about 5.9 kW-hr. This translates to an efficiency of 51%, where 3 kW-hr/m^3 equals 100% efficiency at 20 deg.C. Typical industrial scale plants operate at about 4.5 kW-hr/m^3 or 67% efficiency at high current density. The efficiency is better at lower current density."

....The point here is that you need to first produce electricity to electrolyze water the conventional way, and that energy has to come from fossil fuels or nuclear power plants, etc. If the photocatalyst can be improved, say 50-100x greater than 0.66% stated in Nature, then we can eliminate the cost of electricity from the process. The big potential here is that it is based on semiconductor technology, which as you may know, has obeyed Moore's Law for the past 40 years. If it doubles in efficiency every 12-18 months, we may have automotive hydrogen supplies, based on clean, renewable solar energy, in just 6-10 years.

We gotta get Intel, IBM, and AMD in on this. Then we can really make sure Moores Law holds!